{
  "best_global_step": 36252,
  "best_metric": 0.4916197657585144,
  "best_model_checkpoint": "saves_multiple/ia3/llama-3-8b-instruct/train_codealpacapy_123_1762572064/checkpoint-36252",
  "epoch": 20.0,
  "eval_steps": 1908,
  "global_step": 38160,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.002620545073375262,
      "grad_norm": 0.7216964364051819,
      "learning_rate": 5.241090146750525e-08,
      "loss": 1.2578,
      "num_input_tokens_seen": 2944,
      "step": 5
    },
    {
      "epoch": 0.005241090146750524,
      "grad_norm": 0.34492501616477966,
      "learning_rate": 1.1792452830188679e-07,
      "loss": 1.1168,
      "num_input_tokens_seen": 6816,
      "step": 10
    },
    {
      "epoch": 0.007861635220125786,
      "grad_norm": 0.5016224980354309,
      "learning_rate": 1.8343815513626837e-07,
      "loss": 1.2249,
      "num_input_tokens_seen": 9760,
      "step": 15
    },
    {
      "epoch": 0.010482180293501049,
      "grad_norm": 0.34834542870521545,
      "learning_rate": 2.489517819706499e-07,
      "loss": 1.1638,
      "num_input_tokens_seen": 13472,
      "step": 20
    },
    {
      "epoch": 0.01310272536687631,
      "grad_norm": 0.6954377293586731,
      "learning_rate": 3.144654088050315e-07,
      "loss": 1.2351,
      "num_input_tokens_seen": 16864,
      "step": 25
    },
    {
      "epoch": 0.015723270440251572,
      "grad_norm": 0.5666912794113159,
      "learning_rate": 3.79979035639413e-07,
      "loss": 1.3578,
      "num_input_tokens_seen": 20352,
      "step": 30
    },
    {
      "epoch": 0.018343815513626835,
      "grad_norm": 0.46658873558044434,
      "learning_rate": 4.454926624737945e-07,
      "loss": 1.3356,
      "num_input_tokens_seen": 23008,
      "step": 35
    },
    {
      "epoch": 0.020964360587002098,
      "grad_norm": 0.5186488628387451,
      "learning_rate": 5.110062893081761e-07,
      "loss": 1.077,
      "num_input_tokens_seen": 25920,
      "step": 40
    },
    {
      "epoch": 0.02358490566037736,
      "grad_norm": 0.6623024940490723,
      "learning_rate": 5.765199161425577e-07,
      "loss": 1.2357,
      "num_input_tokens_seen": 28672,
      "step": 45
    },
    {
      "epoch": 0.02620545073375262,
      "grad_norm": 0.3720197379589081,
      "learning_rate": 6.420335429769392e-07,
      "loss": 1.0804,
      "num_input_tokens_seen": 31488,
      "step": 50
    },
    {
      "epoch": 0.028825995807127882,
      "grad_norm": 0.44508644938468933,
      "learning_rate": 7.075471698113208e-07,
      "loss": 1.2172,
      "num_input_tokens_seen": 34496,
      "step": 55
    },
    {
      "epoch": 0.031446540880503145,
      "grad_norm": 0.28141215443611145,
      "learning_rate": 7.730607966457024e-07,
      "loss": 1.2964,
      "num_input_tokens_seen": 38112,
      "step": 60
    },
    {
      "epoch": 0.034067085953878404,
      "grad_norm": 0.5175747871398926,
      "learning_rate": 8.38574423480084e-07,
      "loss": 1.3723,
      "num_input_tokens_seen": 40864,
      "step": 65
    },
    {
      "epoch": 0.03668763102725367,
      "grad_norm": 0.34982386231422424,
      "learning_rate": 9.040880503144654e-07,
      "loss": 1.2841,
      "num_input_tokens_seen": 44192,
      "step": 70
    },
    {
      "epoch": 0.03930817610062893,
      "grad_norm": 0.6717489957809448,
      "learning_rate": 9.69601677148847e-07,
      "loss": 1.4111,
      "num_input_tokens_seen": 46784,
      "step": 75
    },
    {
      "epoch": 0.041928721174004195,
      "grad_norm": 1.2376956939697266,
      "learning_rate": 1.0351153039832286e-06,
      "loss": 1.398,
      "num_input_tokens_seen": 50176,
      "step": 80
    },
    {
      "epoch": 0.044549266247379454,
      "grad_norm": 0.31769198179244995,
      "learning_rate": 1.10062893081761e-06,
      "loss": 1.0922,
      "num_input_tokens_seen": 54400,
      "step": 85
    },
    {
      "epoch": 0.04716981132075472,
      "grad_norm": 0.5534708499908447,
      "learning_rate": 1.1661425576519918e-06,
      "loss": 1.4445,
      "num_input_tokens_seen": 56992,
      "step": 90
    },
    {
      "epoch": 0.04979035639412998,
      "grad_norm": 0.6244670748710632,
      "learning_rate": 1.2316561844863732e-06,
      "loss": 1.0207,
      "num_input_tokens_seen": 60896,
      "step": 95
    },
    {
      "epoch": 0.05241090146750524,
      "grad_norm": 0.35429972410202026,
      "learning_rate": 1.2971698113207548e-06,
      "loss": 1.1544,
      "num_input_tokens_seen": 64576,
      "step": 100
    },
    {
      "epoch": 0.055031446540880505,
      "grad_norm": 0.8778157830238342,
      "learning_rate": 1.3626834381551364e-06,
      "loss": 1.3875,
      "num_input_tokens_seen": 67616,
      "step": 105
    },
    {
      "epoch": 0.057651991614255764,
      "grad_norm": 0.4613979458808899,
      "learning_rate": 1.4281970649895177e-06,
      "loss": 1.031,
      "num_input_tokens_seen": 72544,
      "step": 110
    },
    {
      "epoch": 0.06027253668763103,
      "grad_norm": 0.48877450823783875,
      "learning_rate": 1.4937106918238995e-06,
      "loss": 1.3083,
      "num_input_tokens_seen": 75648,
      "step": 115
    },
    {
      "epoch": 0.06289308176100629,
      "grad_norm": 0.5516355633735657,
      "learning_rate": 1.559224318658281e-06,
      "loss": 1.3933,
      "num_input_tokens_seen": 78368,
      "step": 120
    },
    {
      "epoch": 0.06551362683438156,
      "grad_norm": 0.5274444818496704,
      "learning_rate": 1.6247379454926625e-06,
      "loss": 1.2468,
      "num_input_tokens_seen": 81952,
      "step": 125
    },
    {
      "epoch": 0.06813417190775681,
      "grad_norm": 0.5234399437904358,
      "learning_rate": 1.690251572327044e-06,
      "loss": 1.1986,
      "num_input_tokens_seen": 84960,
      "step": 130
    },
    {
      "epoch": 0.07075471698113207,
      "grad_norm": 0.7185492515563965,
      "learning_rate": 1.7557651991614257e-06,
      "loss": 1.3141,
      "num_input_tokens_seen": 88576,
      "step": 135
    },
    {
      "epoch": 0.07337526205450734,
      "grad_norm": 0.5319799780845642,
      "learning_rate": 1.821278825995807e-06,
      "loss": 1.0275,
      "num_input_tokens_seen": 92160,
      "step": 140
    },
    {
      "epoch": 0.0759958071278826,
      "grad_norm": 0.38666588068008423,
      "learning_rate": 1.8867924528301887e-06,
      "loss": 1.1114,
      "num_input_tokens_seen": 95840,
      "step": 145
    },
    {
      "epoch": 0.07861635220125786,
      "grad_norm": 0.5166113972663879,
      "learning_rate": 1.9523060796645703e-06,
      "loss": 1.1988,
      "num_input_tokens_seen": 98816,
      "step": 150
    },
    {
      "epoch": 0.08123689727463312,
      "grad_norm": 0.37040674686431885,
      "learning_rate": 2.017819706498952e-06,
      "loss": 0.9972,
      "num_input_tokens_seen": 102880,
      "step": 155
    },
    {
      "epoch": 0.08385744234800839,
      "grad_norm": 0.6831836104393005,
      "learning_rate": 2.0833333333333334e-06,
      "loss": 1.1761,
      "num_input_tokens_seen": 105920,
      "step": 160
    },
    {
      "epoch": 0.08647798742138364,
      "grad_norm": 0.3428932726383209,
      "learning_rate": 2.148846960167715e-06,
      "loss": 1.3947,
      "num_input_tokens_seen": 108800,
      "step": 165
    },
    {
      "epoch": 0.08909853249475891,
      "grad_norm": 0.53214430809021,
      "learning_rate": 2.2143605870020966e-06,
      "loss": 1.1005,
      "num_input_tokens_seen": 111968,
      "step": 170
    },
    {
      "epoch": 0.09171907756813417,
      "grad_norm": 0.6421660780906677,
      "learning_rate": 2.279874213836478e-06,
      "loss": 1.1831,
      "num_input_tokens_seen": 114976,
      "step": 175
    },
    {
      "epoch": 0.09433962264150944,
      "grad_norm": 1.1537531614303589,
      "learning_rate": 2.34538784067086e-06,
      "loss": 1.4491,
      "num_input_tokens_seen": 117568,
      "step": 180
    },
    {
      "epoch": 0.09696016771488469,
      "grad_norm": 0.54970782995224,
      "learning_rate": 2.410901467505241e-06,
      "loss": 1.0558,
      "num_input_tokens_seen": 120384,
      "step": 185
    },
    {
      "epoch": 0.09958071278825996,
      "grad_norm": 0.30963578820228577,
      "learning_rate": 2.476415094339623e-06,
      "loss": 1.0173,
      "num_input_tokens_seen": 123680,
      "step": 190
    },
    {
      "epoch": 0.10220125786163523,
      "grad_norm": 0.529529333114624,
      "learning_rate": 2.5419287211740044e-06,
      "loss": 1.2604,
      "num_input_tokens_seen": 127168,
      "step": 195
    },
    {
      "epoch": 0.10482180293501048,
      "grad_norm": 0.487486332654953,
      "learning_rate": 2.6074423480083857e-06,
      "loss": 0.9988,
      "num_input_tokens_seen": 130176,
      "step": 200
    },
    {
      "epoch": 0.10744234800838574,
      "grad_norm": 0.46844395995140076,
      "learning_rate": 2.6729559748427675e-06,
      "loss": 1.1632,
      "num_input_tokens_seen": 133056,
      "step": 205
    },
    {
      "epoch": 0.11006289308176101,
      "grad_norm": 0.4961576759815216,
      "learning_rate": 2.738469601677149e-06,
      "loss": 1.0827,
      "num_input_tokens_seen": 136864,
      "step": 210
    },
    {
      "epoch": 0.11268343815513626,
      "grad_norm": 0.550585150718689,
      "learning_rate": 2.8039832285115303e-06,
      "loss": 1.5171,
      "num_input_tokens_seen": 140064,
      "step": 215
    },
    {
      "epoch": 0.11530398322851153,
      "grad_norm": 0.60359126329422,
      "learning_rate": 2.869496855345912e-06,
      "loss": 1.0757,
      "num_input_tokens_seen": 143712,
      "step": 220
    },
    {
      "epoch": 0.1179245283018868,
      "grad_norm": 0.44368457794189453,
      "learning_rate": 2.935010482180294e-06,
      "loss": 1.19,
      "num_input_tokens_seen": 146880,
      "step": 225
    },
    {
      "epoch": 0.12054507337526206,
      "grad_norm": 0.5072504878044128,
      "learning_rate": 3.0005241090146753e-06,
      "loss": 1.0497,
      "num_input_tokens_seen": 150560,
      "step": 230
    },
    {
      "epoch": 0.12316561844863731,
      "grad_norm": 0.4481230080127716,
      "learning_rate": 3.0660377358490567e-06,
      "loss": 1.296,
      "num_input_tokens_seen": 153664,
      "step": 235
    },
    {
      "epoch": 0.12578616352201258,
      "grad_norm": 0.5778910517692566,
      "learning_rate": 3.131551362683438e-06,
      "loss": 1.0011,
      "num_input_tokens_seen": 157568,
      "step": 240
    },
    {
      "epoch": 0.12840670859538783,
      "grad_norm": 0.5576554536819458,
      "learning_rate": 3.1970649895178203e-06,
      "loss": 1.1081,
      "num_input_tokens_seen": 160512,
      "step": 245
    },
    {
      "epoch": 0.1310272536687631,
      "grad_norm": 0.21731668710708618,
      "learning_rate": 3.2625786163522012e-06,
      "loss": 1.0199,
      "num_input_tokens_seen": 164512,
      "step": 250
    },
    {
      "epoch": 0.13364779874213836,
      "grad_norm": 0.7453848123550415,
      "learning_rate": 3.3280922431865826e-06,
      "loss": 1.253,
      "num_input_tokens_seen": 167200,
      "step": 255
    },
    {
      "epoch": 0.13626834381551362,
      "grad_norm": 0.7321975827217102,
      "learning_rate": 3.393605870020965e-06,
      "loss": 1.2668,
      "num_input_tokens_seen": 169984,
      "step": 260
    },
    {
      "epoch": 0.1388888888888889,
      "grad_norm": 0.4264134466648102,
      "learning_rate": 3.459119496855346e-06,
      "loss": 1.1256,
      "num_input_tokens_seen": 173856,
      "step": 265
    },
    {
      "epoch": 0.14150943396226415,
      "grad_norm": 0.576667308807373,
      "learning_rate": 3.524633123689727e-06,
      "loss": 1.4792,
      "num_input_tokens_seen": 176192,
      "step": 270
    },
    {
      "epoch": 0.1441299790356394,
      "grad_norm": 0.30293548107147217,
      "learning_rate": 3.5901467505241094e-06,
      "loss": 0.9046,
      "num_input_tokens_seen": 179840,
      "step": 275
    },
    {
      "epoch": 0.14675052410901468,
      "grad_norm": 0.5430877208709717,
      "learning_rate": 3.6556603773584908e-06,
      "loss": 1.0426,
      "num_input_tokens_seen": 182976,
      "step": 280
    },
    {
      "epoch": 0.14937106918238993,
      "grad_norm": 0.41233935952186584,
      "learning_rate": 3.7211740041928726e-06,
      "loss": 1.0563,
      "num_input_tokens_seen": 186400,
      "step": 285
    },
    {
      "epoch": 0.1519916142557652,
      "grad_norm": 0.47023579478263855,
      "learning_rate": 3.786687631027254e-06,
      "loss": 1.1188,
      "num_input_tokens_seen": 190144,
      "step": 290
    },
    {
      "epoch": 0.15461215932914046,
      "grad_norm": 0.6516395211219788,
      "learning_rate": 3.852201257861635e-06,
      "loss": 1.3831,
      "num_input_tokens_seen": 193344,
      "step": 295
    },
    {
      "epoch": 0.15723270440251572,
      "grad_norm": 0.7337394952774048,
      "learning_rate": 3.917714884696017e-06,
      "loss": 1.0468,
      "num_input_tokens_seen": 195904,
      "step": 300
    },
    {
      "epoch": 0.159853249475891,
      "grad_norm": 0.5355212092399597,
      "learning_rate": 3.9832285115303985e-06,
      "loss": 1.1889,
      "num_input_tokens_seen": 198944,
      "step": 305
    },
    {
      "epoch": 0.16247379454926625,
      "grad_norm": 0.8317180871963501,
      "learning_rate": 4.04874213836478e-06,
      "loss": 1.4262,
      "num_input_tokens_seen": 201792,
      "step": 310
    },
    {
      "epoch": 0.1650943396226415,
      "grad_norm": 0.3859707713127136,
      "learning_rate": 4.114255765199162e-06,
      "loss": 1.0166,
      "num_input_tokens_seen": 206080,
      "step": 315
    },
    {
      "epoch": 0.16771488469601678,
      "grad_norm": 0.5858093500137329,
      "learning_rate": 4.1797693920335435e-06,
      "loss": 1.1986,
      "num_input_tokens_seen": 209376,
      "step": 320
    },
    {
      "epoch": 0.17033542976939203,
      "grad_norm": 0.4470485746860504,
      "learning_rate": 4.245283018867925e-06,
      "loss": 1.3603,
      "num_input_tokens_seen": 212160,
      "step": 325
    },
    {
      "epoch": 0.17295597484276728,
      "grad_norm": 0.35548603534698486,
      "learning_rate": 4.310796645702306e-06,
      "loss": 0.9667,
      "num_input_tokens_seen": 216128,
      "step": 330
    },
    {
      "epoch": 0.17557651991614256,
      "grad_norm": 0.481240838766098,
      "learning_rate": 4.376310272536688e-06,
      "loss": 2.186,
      "num_input_tokens_seen": 219232,
      "step": 335
    },
    {
      "epoch": 0.17819706498951782,
      "grad_norm": 0.3752843737602234,
      "learning_rate": 4.441823899371069e-06,
      "loss": 1.0847,
      "num_input_tokens_seen": 222464,
      "step": 340
    },
    {
      "epoch": 0.18081761006289307,
      "grad_norm": 1.6001651287078857,
      "learning_rate": 4.507337526205451e-06,
      "loss": 1.4772,
      "num_input_tokens_seen": 225184,
      "step": 345
    },
    {
      "epoch": 0.18343815513626835,
      "grad_norm": 0.5260669589042664,
      "learning_rate": 4.572851153039833e-06,
      "loss": 1.2933,
      "num_input_tokens_seen": 228000,
      "step": 350
    },
    {
      "epoch": 0.1860587002096436,
      "grad_norm": 0.37100857496261597,
      "learning_rate": 4.638364779874214e-06,
      "loss": 1.1101,
      "num_input_tokens_seen": 230944,
      "step": 355
    },
    {
      "epoch": 0.18867924528301888,
      "grad_norm": 0.6833113431930542,
      "learning_rate": 4.703878406708595e-06,
      "loss": 1.1584,
      "num_input_tokens_seen": 233728,
      "step": 360
    },
    {
      "epoch": 0.19129979035639413,
      "grad_norm": 0.5918346643447876,
      "learning_rate": 4.769392033542977e-06,
      "loss": 1.1046,
      "num_input_tokens_seen": 236480,
      "step": 365
    },
    {
      "epoch": 0.19392033542976939,
      "grad_norm": 0.7305867075920105,
      "learning_rate": 4.834905660377359e-06,
      "loss": 1.3087,
      "num_input_tokens_seen": 239488,
      "step": 370
    },
    {
      "epoch": 0.19654088050314467,
      "grad_norm": 0.5267186164855957,
      "learning_rate": 4.90041928721174e-06,
      "loss": 1.3126,
      "num_input_tokens_seen": 242144,
      "step": 375
    },
    {
      "epoch": 0.19916142557651992,
      "grad_norm": 0.4238475561141968,
      "learning_rate": 4.965932914046122e-06,
      "loss": 1.1642,
      "num_input_tokens_seen": 245216,
      "step": 380
    },
    {
      "epoch": 0.20178197064989517,
      "grad_norm": 0.6099701523780823,
      "learning_rate": 5.031446540880504e-06,
      "loss": 1.0648,
      "num_input_tokens_seen": 249632,
      "step": 385
    },
    {
      "epoch": 0.20440251572327045,
      "grad_norm": 0.3424396514892578,
      "learning_rate": 5.096960167714885e-06,
      "loss": 1.2193,
      "num_input_tokens_seen": 253312,
      "step": 390
    },
    {
      "epoch": 0.2070230607966457,
      "grad_norm": 0.5516754388809204,
      "learning_rate": 5.162473794549266e-06,
      "loss": 1.1968,
      "num_input_tokens_seen": 256384,
      "step": 395
    },
    {
      "epoch": 0.20964360587002095,
      "grad_norm": 1.0295958518981934,
      "learning_rate": 5.227987421383648e-06,
      "loss": 1.4332,
      "num_input_tokens_seen": 259744,
      "step": 400
    },
    {
      "epoch": 0.21226415094339623,
      "grad_norm": 0.7402304410934448,
      "learning_rate": 5.2935010482180295e-06,
      "loss": 1.4679,
      "num_input_tokens_seen": 263456,
      "step": 405
    },
    {
      "epoch": 0.2148846960167715,
      "grad_norm": 0.5755616426467896,
      "learning_rate": 5.359014675052411e-06,
      "loss": 1.1311,
      "num_input_tokens_seen": 266496,
      "step": 410
    },
    {
      "epoch": 0.21750524109014674,
      "grad_norm": 0.5959190726280212,
      "learning_rate": 5.424528301886793e-06,
      "loss": 1.2536,
      "num_input_tokens_seen": 269600,
      "step": 415
    },
    {
      "epoch": 0.22012578616352202,
      "grad_norm": 0.511217474937439,
      "learning_rate": 5.4900419287211744e-06,
      "loss": 1.416,
      "num_input_tokens_seen": 273152,
      "step": 420
    },
    {
      "epoch": 0.22274633123689727,
      "grad_norm": 0.6326025128364563,
      "learning_rate": 5.555555555555556e-06,
      "loss": 1.6296,
      "num_input_tokens_seen": 276992,
      "step": 425
    },
    {
      "epoch": 0.22536687631027252,
      "grad_norm": 0.7170044183731079,
      "learning_rate": 5.621069182389937e-06,
      "loss": 1.5661,
      "num_input_tokens_seen": 279296,
      "step": 430
    },
    {
      "epoch": 0.2279874213836478,
      "grad_norm": 0.3545818328857422,
      "learning_rate": 5.686582809224319e-06,
      "loss": 1.006,
      "num_input_tokens_seen": 283296,
      "step": 435
    },
    {
      "epoch": 0.23060796645702306,
      "grad_norm": 0.7831827402114868,
      "learning_rate": 5.752096436058701e-06,
      "loss": 1.6407,
      "num_input_tokens_seen": 286016,
      "step": 440
    },
    {
      "epoch": 0.23322851153039834,
      "grad_norm": 0.6653702855110168,
      "learning_rate": 5.817610062893082e-06,
      "loss": 1.1514,
      "num_input_tokens_seen": 289248,
      "step": 445
    },
    {
      "epoch": 0.2358490566037736,
      "grad_norm": 0.3303171992301941,
      "learning_rate": 5.8831236897274636e-06,
      "loss": 1.6228,
      "num_input_tokens_seen": 292160,
      "step": 450
    },
    {
      "epoch": 0.23846960167714884,
      "grad_norm": 0.5361154079437256,
      "learning_rate": 5.948637316561846e-06,
      "loss": 1.2259,
      "num_input_tokens_seen": 295296,
      "step": 455
    },
    {
      "epoch": 0.24109014675052412,
      "grad_norm": 0.38453778624534607,
      "learning_rate": 6.014150943396227e-06,
      "loss": 1.1442,
      "num_input_tokens_seen": 299360,
      "step": 460
    },
    {
      "epoch": 0.24371069182389937,
      "grad_norm": 0.4447961747646332,
      "learning_rate": 6.079664570230608e-06,
      "loss": 1.0107,
      "num_input_tokens_seen": 303072,
      "step": 465
    },
    {
      "epoch": 0.24633123689727462,
      "grad_norm": 0.4021739363670349,
      "learning_rate": 6.14517819706499e-06,
      "loss": 0.984,
      "num_input_tokens_seen": 307616,
      "step": 470
    },
    {
      "epoch": 0.2489517819706499,
      "grad_norm": 0.4146757125854492,
      "learning_rate": 6.210691823899371e-06,
      "loss": 1.4762,
      "num_input_tokens_seen": 310112,
      "step": 475
    },
    {
      "epoch": 0.25157232704402516,
      "grad_norm": 0.5901830792427063,
      "learning_rate": 6.2762054507337535e-06,
      "loss": 1.1344,
      "num_input_tokens_seen": 313472,
      "step": 480
    },
    {
      "epoch": 0.25419287211740044,
      "grad_norm": 0.4956987798213959,
      "learning_rate": 6.341719077568134e-06,
      "loss": 1.3435,
      "num_input_tokens_seen": 316128,
      "step": 485
    },
    {
      "epoch": 0.25681341719077566,
      "grad_norm": 0.4853038191795349,
      "learning_rate": 6.407232704402516e-06,
      "loss": 1.0421,
      "num_input_tokens_seen": 319584,
      "step": 490
    },
    {
      "epoch": 0.25943396226415094,
      "grad_norm": 0.3987901210784912,
      "learning_rate": 6.472746331236898e-06,
      "loss": 1.0756,
      "num_input_tokens_seen": 322496,
      "step": 495
    },
    {
      "epoch": 0.2620545073375262,
      "grad_norm": 0.4531082808971405,
      "learning_rate": 6.538259958071278e-06,
      "loss": 1.6694,
      "num_input_tokens_seen": 325120,
      "step": 500
    },
    {
      "epoch": 0.26467505241090145,
      "grad_norm": 0.5327111482620239,
      "learning_rate": 6.60377358490566e-06,
      "loss": 1.4162,
      "num_input_tokens_seen": 327840,
      "step": 505
    },
    {
      "epoch": 0.2672955974842767,
      "grad_norm": 0.4680022597312927,
      "learning_rate": 6.669287211740043e-06,
      "loss": 0.8881,
      "num_input_tokens_seen": 331776,
      "step": 510
    },
    {
      "epoch": 0.269916142557652,
      "grad_norm": 0.7558777332305908,
      "learning_rate": 6.734800838574423e-06,
      "loss": 1.1643,
      "num_input_tokens_seen": 336544,
      "step": 515
    },
    {
      "epoch": 0.27253668763102723,
      "grad_norm": 0.33330103754997253,
      "learning_rate": 6.800314465408805e-06,
      "loss": 1.1721,
      "num_input_tokens_seen": 340768,
      "step": 520
    },
    {
      "epoch": 0.2751572327044025,
      "grad_norm": 0.5421535968780518,
      "learning_rate": 6.865828092243188e-06,
      "loss": 1.5452,
      "num_input_tokens_seen": 343296,
      "step": 525
    },
    {
      "epoch": 0.2777777777777778,
      "grad_norm": 0.3898059129714966,
      "learning_rate": 6.931341719077568e-06,
      "loss": 1.0561,
      "num_input_tokens_seen": 346848,
      "step": 530
    },
    {
      "epoch": 0.280398322851153,
      "grad_norm": 0.3756280243396759,
      "learning_rate": 6.9968553459119495e-06,
      "loss": 1.4111,
      "num_input_tokens_seen": 349632,
      "step": 535
    },
    {
      "epoch": 0.2830188679245283,
      "grad_norm": 0.4444721043109894,
      "learning_rate": 7.062368972746332e-06,
      "loss": 1.0692,
      "num_input_tokens_seen": 353024,
      "step": 540
    },
    {
      "epoch": 0.2856394129979036,
      "grad_norm": 0.517319917678833,
      "learning_rate": 7.127882599580712e-06,
      "loss": 1.4912,
      "num_input_tokens_seen": 355552,
      "step": 545
    },
    {
      "epoch": 0.2882599580712788,
      "grad_norm": 0.7537397146224976,
      "learning_rate": 7.1933962264150945e-06,
      "loss": 0.9065,
      "num_input_tokens_seen": 360128,
      "step": 550
    },
    {
      "epoch": 0.2908805031446541,
      "grad_norm": 0.34754103422164917,
      "learning_rate": 7.258909853249477e-06,
      "loss": 1.0456,
      "num_input_tokens_seen": 363360,
      "step": 555
    },
    {
      "epoch": 0.29350104821802936,
      "grad_norm": 0.5561425089836121,
      "learning_rate": 7.324423480083858e-06,
      "loss": 1.3493,
      "num_input_tokens_seen": 366144,
      "step": 560
    },
    {
      "epoch": 0.29612159329140464,
      "grad_norm": 0.5821226239204407,
      "learning_rate": 7.3899371069182395e-06,
      "loss": 1.1461,
      "num_input_tokens_seen": 369216,
      "step": 565
    },
    {
      "epoch": 0.29874213836477986,
      "grad_norm": 0.7163313031196594,
      "learning_rate": 7.455450733752621e-06,
      "loss": 1.1566,
      "num_input_tokens_seen": 372512,
      "step": 570
    },
    {
      "epoch": 0.30136268343815514,
      "grad_norm": 0.5769231915473938,
      "learning_rate": 7.520964360587003e-06,
      "loss": 1.4862,
      "num_input_tokens_seen": 374976,
      "step": 575
    },
    {
      "epoch": 0.3039832285115304,
      "grad_norm": 0.8740735650062561,
      "learning_rate": 7.586477987421384e-06,
      "loss": 1.3307,
      "num_input_tokens_seen": 377632,
      "step": 580
    },
    {
      "epoch": 0.30660377358490565,
      "grad_norm": 0.36442843079566956,
      "learning_rate": 7.651991614255765e-06,
      "loss": 1.3406,
      "num_input_tokens_seen": 380768,
      "step": 585
    },
    {
      "epoch": 0.30922431865828093,
      "grad_norm": 0.5510429739952087,
      "learning_rate": 7.717505241090148e-06,
      "loss": 1.2095,
      "num_input_tokens_seen": 383520,
      "step": 590
    },
    {
      "epoch": 0.3118448637316562,
      "grad_norm": 1.1646664142608643,
      "learning_rate": 7.783018867924528e-06,
      "loss": 1.5209,
      "num_input_tokens_seen": 386496,
      "step": 595
    },
    {
      "epoch": 0.31446540880503143,
      "grad_norm": 0.5275092720985413,
      "learning_rate": 7.848532494758911e-06,
      "loss": 1.113,
      "num_input_tokens_seen": 390464,
      "step": 600
    },
    {
      "epoch": 0.3170859538784067,
      "grad_norm": 0.5060479044914246,
      "learning_rate": 7.914046121593292e-06,
      "loss": 1.0524,
      "num_input_tokens_seen": 393440,
      "step": 605
    },
    {
      "epoch": 0.319706498951782,
      "grad_norm": 0.34906846284866333,
      "learning_rate": 7.979559748427674e-06,
      "loss": 0.9733,
      "num_input_tokens_seen": 398336,
      "step": 610
    },
    {
      "epoch": 0.3223270440251572,
      "grad_norm": 0.5911673903465271,
      "learning_rate": 8.045073375262055e-06,
      "loss": 1.1991,
      "num_input_tokens_seen": 401504,
      "step": 615
    },
    {
      "epoch": 0.3249475890985325,
      "grad_norm": 0.6935735940933228,
      "learning_rate": 8.110587002096436e-06,
      "loss": 1.1115,
      "num_input_tokens_seen": 404864,
      "step": 620
    },
    {
      "epoch": 0.3275681341719078,
      "grad_norm": 0.3731983006000519,
      "learning_rate": 8.176100628930818e-06,
      "loss": 1.1438,
      "num_input_tokens_seen": 409056,
      "step": 625
    },
    {
      "epoch": 0.330188679245283,
      "grad_norm": 0.6295798420906067,
      "learning_rate": 8.241614255765199e-06,
      "loss": 1.1596,
      "num_input_tokens_seen": 412544,
      "step": 630
    },
    {
      "epoch": 0.3328092243186583,
      "grad_norm": 0.5062088966369629,
      "learning_rate": 8.307127882599582e-06,
      "loss": 1.1177,
      "num_input_tokens_seen": 415520,
      "step": 635
    },
    {
      "epoch": 0.33542976939203356,
      "grad_norm": 0.606227695941925,
      "learning_rate": 8.372641509433962e-06,
      "loss": 2.0134,
      "num_input_tokens_seen": 417760,
      "step": 640
    },
    {
      "epoch": 0.3380503144654088,
      "grad_norm": 0.292998343706131,
      "learning_rate": 8.438155136268345e-06,
      "loss": 1.0691,
      "num_input_tokens_seen": 421728,
      "step": 645
    },
    {
      "epoch": 0.34067085953878407,
      "grad_norm": 1.4188380241394043,
      "learning_rate": 8.503668763102726e-06,
      "loss": 1.2841,
      "num_input_tokens_seen": 425664,
      "step": 650
    },
    {
      "epoch": 0.34329140461215935,
      "grad_norm": 0.5622422099113464,
      "learning_rate": 8.569182389937106e-06,
      "loss": 1.0346,
      "num_input_tokens_seen": 428896,
      "step": 655
    },
    {
      "epoch": 0.34591194968553457,
      "grad_norm": 0.6255545616149902,
      "learning_rate": 8.634696016771489e-06,
      "loss": 1.2648,
      "num_input_tokens_seen": 431744,
      "step": 660
    },
    {
      "epoch": 0.34853249475890985,
      "grad_norm": 0.43299660086631775,
      "learning_rate": 8.70020964360587e-06,
      "loss": 1.0353,
      "num_input_tokens_seen": 435520,
      "step": 665
    },
    {
      "epoch": 0.35115303983228513,
      "grad_norm": 0.4174851179122925,
      "learning_rate": 8.765723270440252e-06,
      "loss": 1.0902,
      "num_input_tokens_seen": 438592,
      "step": 670
    },
    {
      "epoch": 0.35377358490566035,
      "grad_norm": 0.4729178547859192,
      "learning_rate": 8.831236897274633e-06,
      "loss": 1.0547,
      "num_input_tokens_seen": 441664,
      "step": 675
    },
    {
      "epoch": 0.35639412997903563,
      "grad_norm": 0.6811114549636841,
      "learning_rate": 8.896750524109015e-06,
      "loss": 1.2885,
      "num_input_tokens_seen": 444832,
      "step": 680
    },
    {
      "epoch": 0.3590146750524109,
      "grad_norm": 0.5392375588417053,
      "learning_rate": 8.962264150943396e-06,
      "loss": 1.5815,
      "num_input_tokens_seen": 447392,
      "step": 685
    },
    {
      "epoch": 0.36163522012578614,
      "grad_norm": 0.7169814705848694,
      "learning_rate": 9.027777777777777e-06,
      "loss": 1.2013,
      "num_input_tokens_seen": 450528,
      "step": 690
    },
    {
      "epoch": 0.3642557651991614,
      "grad_norm": 0.5835467576980591,
      "learning_rate": 9.09329140461216e-06,
      "loss": 1.0927,
      "num_input_tokens_seen": 453504,
      "step": 695
    },
    {
      "epoch": 0.3668763102725367,
      "grad_norm": 0.7944832444190979,
      "learning_rate": 9.15880503144654e-06,
      "loss": 2.4245,
      "num_input_tokens_seen": 456288,
      "step": 700
    },
    {
      "epoch": 0.3694968553459119,
      "grad_norm": 0.628071129322052,
      "learning_rate": 9.224318658280923e-06,
      "loss": 0.9132,
      "num_input_tokens_seen": 459840,
      "step": 705
    },
    {
      "epoch": 0.3721174004192872,
      "grad_norm": 0.36372700333595276,
      "learning_rate": 9.289832285115305e-06,
      "loss": 1.0988,
      "num_input_tokens_seen": 463168,
      "step": 710
    },
    {
      "epoch": 0.3747379454926625,
      "grad_norm": 0.5679545402526855,
      "learning_rate": 9.355345911949686e-06,
      "loss": 1.1601,
      "num_input_tokens_seen": 466112,
      "step": 715
    },
    {
      "epoch": 0.37735849056603776,
      "grad_norm": 0.5550117492675781,
      "learning_rate": 9.420859538784067e-06,
      "loss": 1.1485,
      "num_input_tokens_seen": 469568,
      "step": 720
    },
    {
      "epoch": 0.379979035639413,
      "grad_norm": 0.34233763813972473,
      "learning_rate": 9.486373165618449e-06,
      "loss": 0.9739,
      "num_input_tokens_seen": 472896,
      "step": 725
    },
    {
      "epoch": 0.38259958071278827,
      "grad_norm": 0.612312376499176,
      "learning_rate": 9.551886792452832e-06,
      "loss": 1.0912,
      "num_input_tokens_seen": 475744,
      "step": 730
    },
    {
      "epoch": 0.38522012578616355,
      "grad_norm": 0.16308392584323883,
      "learning_rate": 9.617400419287211e-06,
      "loss": 1.1273,
      "num_input_tokens_seen": 481216,
      "step": 735
    },
    {
      "epoch": 0.38784067085953877,
      "grad_norm": 0.45836517214775085,
      "learning_rate": 9.682914046121595e-06,
      "loss": 1.2812,
      "num_input_tokens_seen": 483488,
      "step": 740
    },
    {
      "epoch": 0.39046121593291405,
      "grad_norm": 0.5059182643890381,
      "learning_rate": 9.748427672955976e-06,
      "loss": 1.2395,
      "num_input_tokens_seen": 487072,
      "step": 745
    },
    {
      "epoch": 0.39308176100628933,
      "grad_norm": 0.48566392064094543,
      "learning_rate": 9.813941299790357e-06,
      "loss": 1.0617,
      "num_input_tokens_seen": 490496,
      "step": 750
    },
    {
      "epoch": 0.39570230607966456,
      "grad_norm": 0.4548680782318115,
      "learning_rate": 9.879454926624739e-06,
      "loss": 1.0319,
      "num_input_tokens_seen": 493824,
      "step": 755
    },
    {
      "epoch": 0.39832285115303984,
      "grad_norm": 0.623921811580658,
      "learning_rate": 9.94496855345912e-06,
      "loss": 1.4834,
      "num_input_tokens_seen": 496192,
      "step": 760
    },
    {
      "epoch": 0.4009433962264151,
      "grad_norm": 0.42646437883377075,
      "learning_rate": 1.0010482180293501e-05,
      "loss": 1.3289,
      "num_input_tokens_seen": 498880,
      "step": 765
    },
    {
      "epoch": 0.40356394129979034,
      "grad_norm": 0.4827862083911896,
      "learning_rate": 1.0075995807127883e-05,
      "loss": 1.1122,
      "num_input_tokens_seen": 502016,
      "step": 770
    },
    {
      "epoch": 0.4061844863731656,
      "grad_norm": 0.27281495928764343,
      "learning_rate": 1.0141509433962266e-05,
      "loss": 1.0969,
      "num_input_tokens_seen": 505120,
      "step": 775
    },
    {
      "epoch": 0.4088050314465409,
      "grad_norm": 0.4635528326034546,
      "learning_rate": 1.0207023060796646e-05,
      "loss": 1.0532,
      "num_input_tokens_seen": 508256,
      "step": 780
    },
    {
      "epoch": 0.4114255765199161,
      "grad_norm": 0.48127612471580505,
      "learning_rate": 1.0272536687631027e-05,
      "loss": 1.0911,
      "num_input_tokens_seen": 511776,
      "step": 785
    },
    {
      "epoch": 0.4140461215932914,
      "grad_norm": 0.4962087571620941,
      "learning_rate": 1.033805031446541e-05,
      "loss": 1.3765,
      "num_input_tokens_seen": 514464,
      "step": 790
    },
    {
      "epoch": 0.4166666666666667,
      "grad_norm": 0.46620601415634155,
      "learning_rate": 1.040356394129979e-05,
      "loss": 1.0522,
      "num_input_tokens_seen": 518144,
      "step": 795
    },
    {
      "epoch": 0.4192872117400419,
      "grad_norm": 0.6702263951301575,
      "learning_rate": 1.0469077568134173e-05,
      "loss": 1.4112,
      "num_input_tokens_seen": 520768,
      "step": 800
    },
    {
      "epoch": 0.4219077568134172,
      "grad_norm": 0.49861466884613037,
      "learning_rate": 1.0534591194968554e-05,
      "loss": 1.3407,
      "num_input_tokens_seen": 523808,
      "step": 805
    },
    {
      "epoch": 0.42452830188679247,
      "grad_norm": 0.5024625062942505,
      "learning_rate": 1.0600104821802936e-05,
      "loss": 0.8425,
      "num_input_tokens_seen": 527136,
      "step": 810
    },
    {
      "epoch": 0.4271488469601677,
      "grad_norm": 0.706407368183136,
      "learning_rate": 1.0665618448637317e-05,
      "loss": 1.1627,
      "num_input_tokens_seen": 530240,
      "step": 815
    },
    {
      "epoch": 0.429769392033543,
      "grad_norm": 0.7140575051307678,
      "learning_rate": 1.0731132075471698e-05,
      "loss": 1.1076,
      "num_input_tokens_seen": 534496,
      "step": 820
    },
    {
      "epoch": 0.43238993710691825,
      "grad_norm": 0.6450242400169373,
      "learning_rate": 1.079664570230608e-05,
      "loss": 1.0701,
      "num_input_tokens_seen": 538624,
      "step": 825
    },
    {
      "epoch": 0.4350104821802935,
      "grad_norm": 0.6114311814308167,
      "learning_rate": 1.0862159329140461e-05,
      "loss": 1.1237,
      "num_input_tokens_seen": 542112,
      "step": 830
    },
    {
      "epoch": 0.43763102725366876,
      "grad_norm": 0.5072643756866455,
      "learning_rate": 1.0927672955974844e-05,
      "loss": 1.1319,
      "num_input_tokens_seen": 545408,
      "step": 835
    },
    {
      "epoch": 0.44025157232704404,
      "grad_norm": 0.3320441246032715,
      "learning_rate": 1.0993186582809224e-05,
      "loss": 1.0544,
      "num_input_tokens_seen": 548416,
      "step": 840
    },
    {
      "epoch": 0.44287211740041926,
      "grad_norm": 0.47253912687301636,
      "learning_rate": 1.1058700209643607e-05,
      "loss": 1.1082,
      "num_input_tokens_seen": 551264,
      "step": 845
    },
    {
      "epoch": 0.44549266247379454,
      "grad_norm": 0.35436657071113586,
      "learning_rate": 1.1124213836477988e-05,
      "loss": 0.8858,
      "num_input_tokens_seen": 554592,
      "step": 850
    },
    {
      "epoch": 0.4481132075471698,
      "grad_norm": 0.836014449596405,
      "learning_rate": 1.118972746331237e-05,
      "loss": 0.8762,
      "num_input_tokens_seen": 558304,
      "step": 855
    },
    {
      "epoch": 0.45073375262054505,
      "grad_norm": 0.4901266098022461,
      "learning_rate": 1.1255241090146751e-05,
      "loss": 1.1776,
      "num_input_tokens_seen": 561248,
      "step": 860
    },
    {
      "epoch": 0.4533542976939203,
      "grad_norm": 0.4057990610599518,
      "learning_rate": 1.1320754716981132e-05,
      "loss": 0.931,
      "num_input_tokens_seen": 564352,
      "step": 865
    },
    {
      "epoch": 0.4559748427672956,
      "grad_norm": 0.8029787540435791,
      "learning_rate": 1.1386268343815514e-05,
      "loss": 0.9946,
      "num_input_tokens_seen": 567712,
      "step": 870
    },
    {
      "epoch": 0.4585953878406709,
      "grad_norm": 0.6317062377929688,
      "learning_rate": 1.1451781970649895e-05,
      "loss": 1.1528,
      "num_input_tokens_seen": 570240,
      "step": 875
    },
    {
      "epoch": 0.4612159329140461,
      "grad_norm": 0.84505295753479,
      "learning_rate": 1.1517295597484278e-05,
      "loss": 1.2648,
      "num_input_tokens_seen": 572768,
      "step": 880
    },
    {
      "epoch": 0.4638364779874214,
      "grad_norm": 0.43877866864204407,
      "learning_rate": 1.158280922431866e-05,
      "loss": 1.025,
      "num_input_tokens_seen": 576512,
      "step": 885
    },
    {
      "epoch": 0.46645702306079667,
      "grad_norm": 0.406337171792984,
      "learning_rate": 1.164832285115304e-05,
      "loss": 1.1295,
      "num_input_tokens_seen": 579264,
      "step": 890
    },
    {
      "epoch": 0.4690775681341719,
      "grad_norm": 0.40448352694511414,
      "learning_rate": 1.1713836477987422e-05,
      "loss": 1.4583,
      "num_input_tokens_seen": 581440,
      "step": 895
    },
    {
      "epoch": 0.4716981132075472,
      "grad_norm": 0.3858742117881775,
      "learning_rate": 1.1779350104821804e-05,
      "loss": 0.8009,
      "num_input_tokens_seen": 584544,
      "step": 900
    },
    {
      "epoch": 0.47431865828092246,
      "grad_norm": 0.5132381319999695,
      "learning_rate": 1.1844863731656185e-05,
      "loss": 0.9976,
      "num_input_tokens_seen": 588160,
      "step": 905
    },
    {
      "epoch": 0.4769392033542977,
      "grad_norm": 1.0414471626281738,
      "learning_rate": 1.1910377358490566e-05,
      "loss": 1.2778,
      "num_input_tokens_seen": 590400,
      "step": 910
    },
    {
      "epoch": 0.47955974842767296,
      "grad_norm": 0.4193120300769806,
      "learning_rate": 1.197589098532495e-05,
      "loss": 0.9093,
      "num_input_tokens_seen": 593440,
      "step": 915
    },
    {
      "epoch": 0.48218029350104824,
      "grad_norm": 0.5144922137260437,
      "learning_rate": 1.204140461215933e-05,
      "loss": 1.098,
      "num_input_tokens_seen": 596480,
      "step": 920
    },
    {
      "epoch": 0.48480083857442346,
      "grad_norm": 0.5624714493751526,
      "learning_rate": 1.210691823899371e-05,
      "loss": 0.9874,
      "num_input_tokens_seen": 599392,
      "step": 925
    },
    {
      "epoch": 0.48742138364779874,
      "grad_norm": 0.4579547643661499,
      "learning_rate": 1.2172431865828094e-05,
      "loss": 0.9442,
      "num_input_tokens_seen": 602432,
      "step": 930
    },
    {
      "epoch": 0.490041928721174,
      "grad_norm": 0.6309740543365479,
      "learning_rate": 1.2237945492662473e-05,
      "loss": 1.2741,
      "num_input_tokens_seen": 604512,
      "step": 935
    },
    {
      "epoch": 0.49266247379454925,
      "grad_norm": 0.2913280129432678,
      "learning_rate": 1.2303459119496856e-05,
      "loss": 0.8424,
      "num_input_tokens_seen": 608864,
      "step": 940
    },
    {
      "epoch": 0.49528301886792453,
      "grad_norm": 0.5513325929641724,
      "learning_rate": 1.2368972746331238e-05,
      "loss": 0.9356,
      "num_input_tokens_seen": 612448,
      "step": 945
    },
    {
      "epoch": 0.4979035639412998,
      "grad_norm": 1.6853828430175781,
      "learning_rate": 1.243448637316562e-05,
      "loss": 1.4435,
      "num_input_tokens_seen": 614944,
      "step": 950
    },
    {
      "epoch": 0.500524109014675,
      "grad_norm": 0.44786009192466736,
      "learning_rate": 1.25e-05,
      "loss": 0.9754,
      "num_input_tokens_seen": 617472,
      "step": 955
    },
    {
      "epoch": 0.5031446540880503,
      "grad_norm": 0.695514976978302,
      "learning_rate": 1.2565513626834382e-05,
      "loss": 0.9569,
      "num_input_tokens_seen": 620192,
      "step": 960
    },
    {
      "epoch": 0.5057651991614256,
      "grad_norm": 0.39288297295570374,
      "learning_rate": 1.2631027253668765e-05,
      "loss": 0.9952,
      "num_input_tokens_seen": 623232,
      "step": 965
    },
    {
      "epoch": 0.5083857442348009,
      "grad_norm": 0.7787977457046509,
      "learning_rate": 1.2696540880503146e-05,
      "loss": 1.233,
      "num_input_tokens_seen": 626016,
      "step": 970
    },
    {
      "epoch": 0.5110062893081762,
      "grad_norm": 0.3991059362888336,
      "learning_rate": 1.2762054507337526e-05,
      "loss": 0.9577,
      "num_input_tokens_seen": 629472,
      "step": 975
    },
    {
      "epoch": 0.5136268343815513,
      "grad_norm": 0.5484543442726135,
      "learning_rate": 1.2827568134171907e-05,
      "loss": 1.1177,
      "num_input_tokens_seen": 632608,
      "step": 980
    },
    {
      "epoch": 0.5162473794549266,
      "grad_norm": 0.5143454670906067,
      "learning_rate": 1.289308176100629e-05,
      "loss": 1.0447,
      "num_input_tokens_seen": 636192,
      "step": 985
    },
    {
      "epoch": 0.5188679245283019,
      "grad_norm": 0.35655200481414795,
      "learning_rate": 1.2958595387840672e-05,
      "loss": 0.8575,
      "num_input_tokens_seen": 640544,
      "step": 990
    },
    {
      "epoch": 0.5214884696016772,
      "grad_norm": 0.49763596057891846,
      "learning_rate": 1.3024109014675053e-05,
      "loss": 0.8823,
      "num_input_tokens_seen": 644448,
      "step": 995
    },
    {
      "epoch": 0.5241090146750524,
      "grad_norm": 0.35441404581069946,
      "learning_rate": 1.3089622641509436e-05,
      "loss": 0.8459,
      "num_input_tokens_seen": 648288,
      "step": 1000
    },
    {
      "epoch": 0.5267295597484277,
      "grad_norm": 0.8856920003890991,
      "learning_rate": 1.3155136268343814e-05,
      "loss": 1.1436,
      "num_input_tokens_seen": 651296,
      "step": 1005
    },
    {
      "epoch": 0.5293501048218029,
      "grad_norm": 0.3003148138523102,
      "learning_rate": 1.3220649895178197e-05,
      "loss": 0.9248,
      "num_input_tokens_seen": 654176,
      "step": 1010
    },
    {
      "epoch": 0.5319706498951782,
      "grad_norm": 0.41880083084106445,
      "learning_rate": 1.3286163522012579e-05,
      "loss": 1.0441,
      "num_input_tokens_seen": 657664,
      "step": 1015
    },
    {
      "epoch": 0.5345911949685535,
      "grad_norm": 0.47481369972229004,
      "learning_rate": 1.3351677148846962e-05,
      "loss": 1.1763,
      "num_input_tokens_seen": 660064,
      "step": 1020
    },
    {
      "epoch": 0.5372117400419287,
      "grad_norm": 0.6679515242576599,
      "learning_rate": 1.3417190775681343e-05,
      "loss": 1.0763,
      "num_input_tokens_seen": 662880,
      "step": 1025
    },
    {
      "epoch": 0.539832285115304,
      "grad_norm": 0.5416033267974854,
      "learning_rate": 1.3482704402515725e-05,
      "loss": 0.8589,
      "num_input_tokens_seen": 665920,
      "step": 1030
    },
    {
      "epoch": 0.5424528301886793,
      "grad_norm": 0.5498341917991638,
      "learning_rate": 1.3548218029350104e-05,
      "loss": 0.8887,
      "num_input_tokens_seen": 668864,
      "step": 1035
    },
    {
      "epoch": 0.5450733752620545,
      "grad_norm": 0.4155057966709137,
      "learning_rate": 1.3613731656184486e-05,
      "loss": 0.9214,
      "num_input_tokens_seen": 671776,
      "step": 1040
    },
    {
      "epoch": 0.5476939203354297,
      "grad_norm": 0.5185301899909973,
      "learning_rate": 1.3679245283018869e-05,
      "loss": 0.9572,
      "num_input_tokens_seen": 674912,
      "step": 1045
    },
    {
      "epoch": 0.550314465408805,
      "grad_norm": 0.37846025824546814,
      "learning_rate": 1.374475890985325e-05,
      "loss": 0.7878,
      "num_input_tokens_seen": 678368,
      "step": 1050
    },
    {
      "epoch": 0.5529350104821803,
      "grad_norm": 0.2989298105239868,
      "learning_rate": 1.3810272536687632e-05,
      "loss": 1.0044,
      "num_input_tokens_seen": 682336,
      "step": 1055
    },
    {
      "epoch": 0.5555555555555556,
      "grad_norm": 0.812520444393158,
      "learning_rate": 1.3875786163522015e-05,
      "loss": 1.024,
      "num_input_tokens_seen": 685760,
      "step": 1060
    },
    {
      "epoch": 0.5581761006289309,
      "grad_norm": 0.7283947467803955,
      "learning_rate": 1.3941299790356394e-05,
      "loss": 0.9467,
      "num_input_tokens_seen": 688480,
      "step": 1065
    },
    {
      "epoch": 0.560796645702306,
      "grad_norm": 0.4394959509372711,
      "learning_rate": 1.4006813417190776e-05,
      "loss": 1.1005,
      "num_input_tokens_seen": 692064,
      "step": 1070
    },
    {
      "epoch": 0.5634171907756813,
      "grad_norm": 0.9607936143875122,
      "learning_rate": 1.4072327044025157e-05,
      "loss": 1.3,
      "num_input_tokens_seen": 694464,
      "step": 1075
    },
    {
      "epoch": 0.5660377358490566,
      "grad_norm": 0.5124691724777222,
      "learning_rate": 1.413784067085954e-05,
      "loss": 0.9896,
      "num_input_tokens_seen": 697344,
      "step": 1080
    },
    {
      "epoch": 0.5686582809224319,
      "grad_norm": 0.681376039981842,
      "learning_rate": 1.4203354297693922e-05,
      "loss": 1.0487,
      "num_input_tokens_seen": 700256,
      "step": 1085
    },
    {
      "epoch": 0.5712788259958071,
      "grad_norm": 0.4484342634677887,
      "learning_rate": 1.4268867924528303e-05,
      "loss": 0.8867,
      "num_input_tokens_seen": 703968,
      "step": 1090
    },
    {
      "epoch": 0.5738993710691824,
      "grad_norm": 0.36188432574272156,
      "learning_rate": 1.4334381551362683e-05,
      "loss": 0.9019,
      "num_input_tokens_seen": 708160,
      "step": 1095
    },
    {
      "epoch": 0.5765199161425576,
      "grad_norm": 0.4732229709625244,
      "learning_rate": 1.4399895178197064e-05,
      "loss": 1.2425,
      "num_input_tokens_seen": 710976,
      "step": 1100
    },
    {
      "epoch": 0.5791404612159329,
      "grad_norm": 1.0519061088562012,
      "learning_rate": 1.4465408805031447e-05,
      "loss": 1.2203,
      "num_input_tokens_seen": 713696,
      "step": 1105
    },
    {
      "epoch": 0.5817610062893082,
      "grad_norm": 0.7025429606437683,
      "learning_rate": 1.4530922431865828e-05,
      "loss": 1.0146,
      "num_input_tokens_seen": 716608,
      "step": 1110
    },
    {
      "epoch": 0.5843815513626834,
      "grad_norm": 0.4694790542125702,
      "learning_rate": 1.4596436058700211e-05,
      "loss": 0.9652,
      "num_input_tokens_seen": 719488,
      "step": 1115
    },
    {
      "epoch": 0.5870020964360587,
      "grad_norm": 0.37730488181114197,
      "learning_rate": 1.4661949685534593e-05,
      "loss": 0.924,
      "num_input_tokens_seen": 722304,
      "step": 1120
    },
    {
      "epoch": 0.589622641509434,
      "grad_norm": 0.5408967137336731,
      "learning_rate": 1.4727463312368974e-05,
      "loss": 0.9323,
      "num_input_tokens_seen": 725472,
      "step": 1125
    },
    {
      "epoch": 0.5922431865828093,
      "grad_norm": 0.6793172955513,
      "learning_rate": 1.4792976939203354e-05,
      "loss": 0.7344,
      "num_input_tokens_seen": 728896,
      "step": 1130
    },
    {
      "epoch": 0.5948637316561844,
      "grad_norm": 0.3291223347187042,
      "learning_rate": 1.4858490566037735e-05,
      "loss": 0.9581,
      "num_input_tokens_seen": 732000,
      "step": 1135
    },
    {
      "epoch": 0.5974842767295597,
      "grad_norm": 0.7166542410850525,
      "learning_rate": 1.4924004192872118e-05,
      "loss": 0.9665,
      "num_input_tokens_seen": 735328,
      "step": 1140
    },
    {
      "epoch": 0.600104821802935,
      "grad_norm": 0.3379591107368469,
      "learning_rate": 1.49895178197065e-05,
      "loss": 0.7967,
      "num_input_tokens_seen": 738528,
      "step": 1145
    },
    {
      "epoch": 0.6027253668763103,
      "grad_norm": 0.4831031858921051,
      "learning_rate": 1.5055031446540883e-05,
      "loss": 1.1536,
      "num_input_tokens_seen": 741376,
      "step": 1150
    },
    {
      "epoch": 0.6053459119496856,
      "grad_norm": 0.629763662815094,
      "learning_rate": 1.5120545073375264e-05,
      "loss": 1.0355,
      "num_input_tokens_seen": 743872,
      "step": 1155
    },
    {
      "epoch": 0.6079664570230608,
      "grad_norm": 0.5228995680809021,
      "learning_rate": 1.5186058700209644e-05,
      "loss": 1.1568,
      "num_input_tokens_seen": 746720,
      "step": 1160
    },
    {
      "epoch": 0.610587002096436,
      "grad_norm": 0.44927698373794556,
      "learning_rate": 1.5251572327044025e-05,
      "loss": 0.9854,
      "num_input_tokens_seen": 750240,
      "step": 1165
    },
    {
      "epoch": 0.6132075471698113,
      "grad_norm": 0.6730298399925232,
      "learning_rate": 1.5317085953878407e-05,
      "loss": 0.9373,
      "num_input_tokens_seen": 753344,
      "step": 1170
    },
    {
      "epoch": 0.6158280922431866,
      "grad_norm": 0.43992170691490173,
      "learning_rate": 1.538259958071279e-05,
      "loss": 1.1082,
      "num_input_tokens_seen": 755968,
      "step": 1175
    },
    {
      "epoch": 0.6184486373165619,
      "grad_norm": 0.32227322459220886,
      "learning_rate": 1.5448113207547173e-05,
      "loss": 1.0945,
      "num_input_tokens_seen": 759520,
      "step": 1180
    },
    {
      "epoch": 0.6210691823899371,
      "grad_norm": 0.5578993558883667,
      "learning_rate": 1.5513626834381552e-05,
      "loss": 0.9786,
      "num_input_tokens_seen": 762048,
      "step": 1185
    },
    {
      "epoch": 0.6236897274633124,
      "grad_norm": 0.6992300152778625,
      "learning_rate": 1.5579140461215932e-05,
      "loss": 1.0697,
      "num_input_tokens_seen": 765504,
      "step": 1190
    },
    {
      "epoch": 0.6263102725366876,
      "grad_norm": 0.3891379237174988,
      "learning_rate": 1.5644654088050315e-05,
      "loss": 0.9465,
      "num_input_tokens_seen": 768192,
      "step": 1195
    },
    {
      "epoch": 0.6289308176100629,
      "grad_norm": 0.461365669965744,
      "learning_rate": 1.5710167714884695e-05,
      "loss": 1.0084,
      "num_input_tokens_seen": 771168,
      "step": 1200
    },
    {
      "epoch": 0.6315513626834381,
      "grad_norm": 0.4189107120037079,
      "learning_rate": 1.5775681341719078e-05,
      "loss": 0.6093,
      "num_input_tokens_seen": 777504,
      "step": 1205
    },
    {
      "epoch": 0.6341719077568134,
      "grad_norm": 0.7258732914924622,
      "learning_rate": 1.584119496855346e-05,
      "loss": 0.8854,
      "num_input_tokens_seen": 780512,
      "step": 1210
    },
    {
      "epoch": 0.6367924528301887,
      "grad_norm": 0.49903181195259094,
      "learning_rate": 1.5906708595387844e-05,
      "loss": 0.8977,
      "num_input_tokens_seen": 784192,
      "step": 1215
    },
    {
      "epoch": 0.639412997903564,
      "grad_norm": 0.7804319858551025,
      "learning_rate": 1.597222222222222e-05,
      "loss": 1.0116,
      "num_input_tokens_seen": 787552,
      "step": 1220
    },
    {
      "epoch": 0.6420335429769392,
      "grad_norm": 0.3384166359901428,
      "learning_rate": 1.6037735849056604e-05,
      "loss": 1.0425,
      "num_input_tokens_seen": 790720,
      "step": 1225
    },
    {
      "epoch": 0.6446540880503144,
      "grad_norm": 0.6310428380966187,
      "learning_rate": 1.6103249475890987e-05,
      "loss": 0.975,
      "num_input_tokens_seen": 793696,
      "step": 1230
    },
    {
      "epoch": 0.6472746331236897,
      "grad_norm": 0.5797524452209473,
      "learning_rate": 1.6168763102725366e-05,
      "loss": 0.9664,
      "num_input_tokens_seen": 796640,
      "step": 1235
    },
    {
      "epoch": 0.649895178197065,
      "grad_norm": 0.7608315348625183,
      "learning_rate": 1.623427672955975e-05,
      "loss": 0.9622,
      "num_input_tokens_seen": 799648,
      "step": 1240
    },
    {
      "epoch": 0.6525157232704403,
      "grad_norm": 0.6117333769798279,
      "learning_rate": 1.6299790356394132e-05,
      "loss": 0.6879,
      "num_input_tokens_seen": 802496,
      "step": 1245
    },
    {
      "epoch": 0.6551362683438156,
      "grad_norm": 0.6499700546264648,
      "learning_rate": 1.6365303983228512e-05,
      "loss": 0.9457,
      "num_input_tokens_seen": 805760,
      "step": 1250
    },
    {
      "epoch": 0.6577568134171907,
      "grad_norm": 0.7300409078598022,
      "learning_rate": 1.6430817610062892e-05,
      "loss": 1.0276,
      "num_input_tokens_seen": 808832,
      "step": 1255
    },
    {
      "epoch": 0.660377358490566,
      "grad_norm": 0.45413732528686523,
      "learning_rate": 1.6496331236897275e-05,
      "loss": 0.768,
      "num_input_tokens_seen": 812256,
      "step": 1260
    },
    {
      "epoch": 0.6629979035639413,
      "grad_norm": 0.5108651518821716,
      "learning_rate": 1.6561844863731658e-05,
      "loss": 0.8512,
      "num_input_tokens_seen": 815680,
      "step": 1265
    },
    {
      "epoch": 0.6656184486373166,
      "grad_norm": 0.3523840010166168,
      "learning_rate": 1.6627358490566038e-05,
      "loss": 0.8768,
      "num_input_tokens_seen": 818400,
      "step": 1270
    },
    {
      "epoch": 0.6682389937106918,
      "grad_norm": 0.35606443881988525,
      "learning_rate": 1.669287211740042e-05,
      "loss": 0.855,
      "num_input_tokens_seen": 821536,
      "step": 1275
    },
    {
      "epoch": 0.6708595387840671,
      "grad_norm": 0.5332791805267334,
      "learning_rate": 1.6758385744234804e-05,
      "loss": 0.7896,
      "num_input_tokens_seen": 825024,
      "step": 1280
    },
    {
      "epoch": 0.6734800838574424,
      "grad_norm": 0.34594854712486267,
      "learning_rate": 1.6823899371069183e-05,
      "loss": 0.7595,
      "num_input_tokens_seen": 829248,
      "step": 1285
    },
    {
      "epoch": 0.6761006289308176,
      "grad_norm": 0.4295627772808075,
      "learning_rate": 1.6889412997903563e-05,
      "loss": 0.9179,
      "num_input_tokens_seen": 831872,
      "step": 1290
    },
    {
      "epoch": 0.6787211740041929,
      "grad_norm": 0.6636160612106323,
      "learning_rate": 1.6954926624737946e-05,
      "loss": 1.1362,
      "num_input_tokens_seen": 834880,
      "step": 1295
    },
    {
      "epoch": 0.6813417190775681,
      "grad_norm": 0.36296263337135315,
      "learning_rate": 1.702044025157233e-05,
      "loss": 0.9671,
      "num_input_tokens_seen": 837824,
      "step": 1300
    },
    {
      "epoch": 0.6839622641509434,
      "grad_norm": 0.4524725377559662,
      "learning_rate": 1.708595387840671e-05,
      "loss": 1.0089,
      "num_input_tokens_seen": 840896,
      "step": 1305
    },
    {
      "epoch": 0.6865828092243187,
      "grad_norm": 0.4894102215766907,
      "learning_rate": 1.7151467505241092e-05,
      "loss": 0.9962,
      "num_input_tokens_seen": 844000,
      "step": 1310
    },
    {
      "epoch": 0.689203354297694,
      "grad_norm": 0.2682334780693054,
      "learning_rate": 1.7216981132075472e-05,
      "loss": 0.7965,
      "num_input_tokens_seen": 847840,
      "step": 1315
    },
    {
      "epoch": 0.6918238993710691,
      "grad_norm": 0.5316897034645081,
      "learning_rate": 1.7282494758909855e-05,
      "loss": 1.0282,
      "num_input_tokens_seen": 850976,
      "step": 1320
    },
    {
      "epoch": 0.6944444444444444,
      "grad_norm": 0.3451741933822632,
      "learning_rate": 1.7348008385744234e-05,
      "loss": 0.8614,
      "num_input_tokens_seen": 855616,
      "step": 1325
    },
    {
      "epoch": 0.6970649895178197,
      "grad_norm": 0.3874368965625763,
      "learning_rate": 1.7413522012578618e-05,
      "loss": 0.6803,
      "num_input_tokens_seen": 858528,
      "step": 1330
    },
    {
      "epoch": 0.699685534591195,
      "grad_norm": 0.4741744101047516,
      "learning_rate": 1.7479035639413e-05,
      "loss": 0.8395,
      "num_input_tokens_seen": 862144,
      "step": 1335
    },
    {
      "epoch": 0.7023060796645703,
      "grad_norm": 0.3425808250904083,
      "learning_rate": 1.754454926624738e-05,
      "loss": 0.913,
      "num_input_tokens_seen": 865440,
      "step": 1340
    },
    {
      "epoch": 0.7049266247379455,
      "grad_norm": 0.3463025689125061,
      "learning_rate": 1.761006289308176e-05,
      "loss": 0.8653,
      "num_input_tokens_seen": 868992,
      "step": 1345
    },
    {
      "epoch": 0.7075471698113207,
      "grad_norm": 0.7573238015174866,
      "learning_rate": 1.7675576519916143e-05,
      "loss": 0.8274,
      "num_input_tokens_seen": 871968,
      "step": 1350
    },
    {
      "epoch": 0.710167714884696,
      "grad_norm": 0.41685155034065247,
      "learning_rate": 1.7741090146750526e-05,
      "loss": 0.8537,
      "num_input_tokens_seen": 874912,
      "step": 1355
    },
    {
      "epoch": 0.7127882599580713,
      "grad_norm": 0.47678256034851074,
      "learning_rate": 1.7806603773584906e-05,
      "loss": 0.8558,
      "num_input_tokens_seen": 878112,
      "step": 1360
    },
    {
      "epoch": 0.7154088050314465,
      "grad_norm": 0.7794644832611084,
      "learning_rate": 1.787211740041929e-05,
      "loss": 0.8832,
      "num_input_tokens_seen": 880928,
      "step": 1365
    },
    {
      "epoch": 0.7180293501048218,
      "grad_norm": 0.27170926332473755,
      "learning_rate": 1.7937631027253672e-05,
      "loss": 0.7393,
      "num_input_tokens_seen": 884832,
      "step": 1370
    },
    {
      "epoch": 0.7206498951781971,
      "grad_norm": 0.27845290303230286,
      "learning_rate": 1.8003144654088048e-05,
      "loss": 0.8645,
      "num_input_tokens_seen": 888800,
      "step": 1375
    },
    {
      "epoch": 0.7232704402515723,
      "grad_norm": 0.38210463523864746,
      "learning_rate": 1.806865828092243e-05,
      "loss": 0.7882,
      "num_input_tokens_seen": 892416,
      "step": 1380
    },
    {
      "epoch": 0.7258909853249476,
      "grad_norm": 0.7268489003181458,
      "learning_rate": 1.8134171907756814e-05,
      "loss": 0.8944,
      "num_input_tokens_seen": 895840,
      "step": 1385
    },
    {
      "epoch": 0.7285115303983228,
      "grad_norm": 0.43863847851753235,
      "learning_rate": 1.8199685534591197e-05,
      "loss": 0.8177,
      "num_input_tokens_seen": 899392,
      "step": 1390
    },
    {
      "epoch": 0.7311320754716981,
      "grad_norm": 0.4724295735359192,
      "learning_rate": 1.8265199161425577e-05,
      "loss": 0.7657,
      "num_input_tokens_seen": 903264,
      "step": 1395
    },
    {
      "epoch": 0.7337526205450734,
      "grad_norm": 0.5505765080451965,
      "learning_rate": 1.833071278825996e-05,
      "loss": 0.9543,
      "num_input_tokens_seen": 906016,
      "step": 1400
    },
    {
      "epoch": 0.7363731656184487,
      "grad_norm": 0.5474266409873962,
      "learning_rate": 1.839622641509434e-05,
      "loss": 0.9625,
      "num_input_tokens_seen": 909152,
      "step": 1405
    },
    {
      "epoch": 0.7389937106918238,
      "grad_norm": 0.7477304935455322,
      "learning_rate": 1.846174004192872e-05,
      "loss": 0.9126,
      "num_input_tokens_seen": 912416,
      "step": 1410
    },
    {
      "epoch": 0.7416142557651991,
      "grad_norm": 0.6361988186836243,
      "learning_rate": 1.8527253668763103e-05,
      "loss": 1.0856,
      "num_input_tokens_seen": 915616,
      "step": 1415
    },
    {
      "epoch": 0.7442348008385744,
      "grad_norm": 0.9503481984138489,
      "learning_rate": 1.8592767295597486e-05,
      "loss": 0.9058,
      "num_input_tokens_seen": 921568,
      "step": 1420
    },
    {
      "epoch": 0.7468553459119497,
      "grad_norm": 0.37432435154914856,
      "learning_rate": 1.865828092243187e-05,
      "loss": 0.7242,
      "num_input_tokens_seen": 924864,
      "step": 1425
    },
    {
      "epoch": 0.749475890985325,
      "grad_norm": 0.985359251499176,
      "learning_rate": 1.872379454926625e-05,
      "loss": 0.9384,
      "num_input_tokens_seen": 927968,
      "step": 1430
    },
    {
      "epoch": 0.7520964360587002,
      "grad_norm": 0.482819527387619,
      "learning_rate": 1.878930817610063e-05,
      "loss": 0.7895,
      "num_input_tokens_seen": 931584,
      "step": 1435
    },
    {
      "epoch": 0.7547169811320755,
      "grad_norm": 0.8244794607162476,
      "learning_rate": 1.885482180293501e-05,
      "loss": 0.8772,
      "num_input_tokens_seen": 934816,
      "step": 1440
    },
    {
      "epoch": 0.7573375262054507,
      "grad_norm": 0.5569032430648804,
      "learning_rate": 1.892033542976939e-05,
      "loss": 0.8303,
      "num_input_tokens_seen": 937856,
      "step": 1445
    },
    {
      "epoch": 0.759958071278826,
      "grad_norm": 0.5226762890815735,
      "learning_rate": 1.8985849056603774e-05,
      "loss": 0.767,
      "num_input_tokens_seen": 945632,
      "step": 1450
    },
    {
      "epoch": 0.7625786163522013,
      "grad_norm": 0.46799060702323914,
      "learning_rate": 1.9051362683438157e-05,
      "loss": 0.9562,
      "num_input_tokens_seen": 948608,
      "step": 1455
    },
    {
      "epoch": 0.7651991614255765,
      "grad_norm": 0.6254369020462036,
      "learning_rate": 1.9116876310272537e-05,
      "loss": 0.9293,
      "num_input_tokens_seen": 951488,
      "step": 1460
    },
    {
      "epoch": 0.7678197064989518,
      "grad_norm": 0.4250691533088684,
      "learning_rate": 1.918238993710692e-05,
      "loss": 0.8123,
      "num_input_tokens_seen": 955072,
      "step": 1465
    },
    {
      "epoch": 0.7704402515723271,
      "grad_norm": 0.54888916015625,
      "learning_rate": 1.92479035639413e-05,
      "loss": 0.8436,
      "num_input_tokens_seen": 958272,
      "step": 1470
    },
    {
      "epoch": 0.7730607966457023,
      "grad_norm": 0.2580413222312927,
      "learning_rate": 1.9313417190775683e-05,
      "loss": 0.9277,
      "num_input_tokens_seen": 961888,
      "step": 1475
    },
    {
      "epoch": 0.7756813417190775,
      "grad_norm": 0.8835746049880981,
      "learning_rate": 1.9378930817610062e-05,
      "loss": 1.1487,
      "num_input_tokens_seen": 964256,
      "step": 1480
    },
    {
      "epoch": 0.7783018867924528,
      "grad_norm": 0.5668450593948364,
      "learning_rate": 1.9444444444444445e-05,
      "loss": 0.7616,
      "num_input_tokens_seen": 968224,
      "step": 1485
    },
    {
      "epoch": 0.7809224318658281,
      "grad_norm": 0.3405549228191376,
      "learning_rate": 1.950995807127883e-05,
      "loss": 0.8011,
      "num_input_tokens_seen": 971136,
      "step": 1490
    },
    {
      "epoch": 0.7835429769392034,
      "grad_norm": 0.466417133808136,
      "learning_rate": 1.9575471698113208e-05,
      "loss": 0.758,
      "num_input_tokens_seen": 975936,
      "step": 1495
    },
    {
      "epoch": 0.7861635220125787,
      "grad_norm": 0.7094401717185974,
      "learning_rate": 1.9640985324947588e-05,
      "loss": 0.8225,
      "num_input_tokens_seen": 979360,
      "step": 1500
    },
    {
      "epoch": 0.7887840670859538,
      "grad_norm": 0.27430853247642517,
      "learning_rate": 1.970649895178197e-05,
      "loss": 0.8011,
      "num_input_tokens_seen": 983616,
      "step": 1505
    },
    {
      "epoch": 0.7914046121593291,
      "grad_norm": 0.4838595986366272,
      "learning_rate": 1.9772012578616354e-05,
      "loss": 0.8034,
      "num_input_tokens_seen": 986944,
      "step": 1510
    },
    {
      "epoch": 0.7940251572327044,
      "grad_norm": 0.5617373585700989,
      "learning_rate": 1.9837526205450734e-05,
      "loss": 0.9113,
      "num_input_tokens_seen": 989888,
      "step": 1515
    },
    {
      "epoch": 0.7966457023060797,
      "grad_norm": 0.6383323669433594,
      "learning_rate": 1.9903039832285117e-05,
      "loss": 0.8373,
      "num_input_tokens_seen": 992768,
      "step": 1520
    },
    {
      "epoch": 0.799266247379455,
      "grad_norm": 0.4001499116420746,
      "learning_rate": 1.99685534591195e-05,
      "loss": 0.7507,
      "num_input_tokens_seen": 996896,
      "step": 1525
    },
    {
      "epoch": 0.8018867924528302,
      "grad_norm": 0.38751986622810364,
      "learning_rate": 2.003406708595388e-05,
      "loss": 0.7942,
      "num_input_tokens_seen": 999840,
      "step": 1530
    },
    {
      "epoch": 0.8045073375262054,
      "grad_norm": 0.35130801796913147,
      "learning_rate": 2.009958071278826e-05,
      "loss": 0.6846,
      "num_input_tokens_seen": 1003680,
      "step": 1535
    },
    {
      "epoch": 0.8071278825995807,
      "grad_norm": 0.4237540364265442,
      "learning_rate": 2.0165094339622642e-05,
      "loss": 0.7988,
      "num_input_tokens_seen": 1006656,
      "step": 1540
    },
    {
      "epoch": 0.809748427672956,
      "grad_norm": 1.3065630197525024,
      "learning_rate": 2.0230607966457025e-05,
      "loss": 0.8898,
      "num_input_tokens_seen": 1008768,
      "step": 1545
    },
    {
      "epoch": 0.8123689727463312,
      "grad_norm": 0.26505768299102783,
      "learning_rate": 2.0296121593291405e-05,
      "loss": 0.8197,
      "num_input_tokens_seen": 1011968,
      "step": 1550
    },
    {
      "epoch": 0.8149895178197065,
      "grad_norm": 0.6027921438217163,
      "learning_rate": 2.0361635220125788e-05,
      "loss": 0.8492,
      "num_input_tokens_seen": 1018784,
      "step": 1555
    },
    {
      "epoch": 0.8176100628930818,
      "grad_norm": 0.34913134574890137,
      "learning_rate": 2.0427148846960168e-05,
      "loss": 0.5714,
      "num_input_tokens_seen": 1023520,
      "step": 1560
    },
    {
      "epoch": 0.820230607966457,
      "grad_norm": 0.41463860869407654,
      "learning_rate": 2.049266247379455e-05,
      "loss": 0.6883,
      "num_input_tokens_seen": 1027200,
      "step": 1565
    },
    {
      "epoch": 0.8228511530398323,
      "grad_norm": 0.27313849329948425,
      "learning_rate": 2.055817610062893e-05,
      "loss": 1.0693,
      "num_input_tokens_seen": 1030848,
      "step": 1570
    },
    {
      "epoch": 0.8254716981132075,
      "grad_norm": 0.48806053400039673,
      "learning_rate": 2.0623689727463314e-05,
      "loss": 0.7896,
      "num_input_tokens_seen": 1035232,
      "step": 1575
    },
    {
      "epoch": 0.8280922431865828,
      "grad_norm": 0.610421359539032,
      "learning_rate": 2.0689203354297697e-05,
      "loss": 0.7786,
      "num_input_tokens_seen": 1038176,
      "step": 1580
    },
    {
      "epoch": 0.8307127882599581,
      "grad_norm": 0.9365735650062561,
      "learning_rate": 2.0754716981132076e-05,
      "loss": 0.801,
      "num_input_tokens_seen": 1041216,
      "step": 1585
    },
    {
      "epoch": 0.8333333333333334,
      "grad_norm": 0.5737321376800537,
      "learning_rate": 2.0820230607966456e-05,
      "loss": 0.7997,
      "num_input_tokens_seen": 1043840,
      "step": 1590
    },
    {
      "epoch": 0.8359538784067087,
      "grad_norm": 0.29974251985549927,
      "learning_rate": 2.088574423480084e-05,
      "loss": 0.6996,
      "num_input_tokens_seen": 1046656,
      "step": 1595
    },
    {
      "epoch": 0.8385744234800838,
      "grad_norm": 0.380540132522583,
      "learning_rate": 2.0951257861635222e-05,
      "loss": 0.6634,
      "num_input_tokens_seen": 1051616,
      "step": 1600
    },
    {
      "epoch": 0.8411949685534591,
      "grad_norm": 0.7956778407096863,
      "learning_rate": 2.1016771488469602e-05,
      "loss": 0.9996,
      "num_input_tokens_seen": 1053824,
      "step": 1605
    },
    {
      "epoch": 0.8438155136268344,
      "grad_norm": 0.47959059476852417,
      "learning_rate": 2.1082285115303985e-05,
      "loss": 0.7911,
      "num_input_tokens_seen": 1056512,
      "step": 1610
    },
    {
      "epoch": 0.8464360587002097,
      "grad_norm": 0.3409917652606964,
      "learning_rate": 2.1147798742138368e-05,
      "loss": 0.6959,
      "num_input_tokens_seen": 1060160,
      "step": 1615
    },
    {
      "epoch": 0.8490566037735849,
      "grad_norm": 0.7309471368789673,
      "learning_rate": 2.1213312368972748e-05,
      "loss": 0.9179,
      "num_input_tokens_seen": 1062880,
      "step": 1620
    },
    {
      "epoch": 0.8516771488469602,
      "grad_norm": 0.6224607825279236,
      "learning_rate": 2.1278825995807127e-05,
      "loss": 0.8707,
      "num_input_tokens_seen": 1066528,
      "step": 1625
    },
    {
      "epoch": 0.8542976939203354,
      "grad_norm": 0.4693185091018677,
      "learning_rate": 2.134433962264151e-05,
      "loss": 0.6845,
      "num_input_tokens_seen": 1069472,
      "step": 1630
    },
    {
      "epoch": 0.8569182389937107,
      "grad_norm": 0.5225701332092285,
      "learning_rate": 2.1409853249475894e-05,
      "loss": 0.704,
      "num_input_tokens_seen": 1074304,
      "step": 1635
    },
    {
      "epoch": 0.859538784067086,
      "grad_norm": 0.7280882596969604,
      "learning_rate": 2.1475366876310273e-05,
      "loss": 0.7045,
      "num_input_tokens_seen": 1077376,
      "step": 1640
    },
    {
      "epoch": 0.8621593291404612,
      "grad_norm": 0.4066319465637207,
      "learning_rate": 2.1540880503144656e-05,
      "loss": 0.7348,
      "num_input_tokens_seen": 1080672,
      "step": 1645
    },
    {
      "epoch": 0.8647798742138365,
      "grad_norm": 0.4844609200954437,
      "learning_rate": 2.160639412997904e-05,
      "loss": 0.79,
      "num_input_tokens_seen": 1084224,
      "step": 1650
    },
    {
      "epoch": 0.8674004192872118,
      "grad_norm": 0.5878341197967529,
      "learning_rate": 2.1671907756813416e-05,
      "loss": 0.9954,
      "num_input_tokens_seen": 1086912,
      "step": 1655
    },
    {
      "epoch": 0.870020964360587,
      "grad_norm": 0.3797484338283539,
      "learning_rate": 2.17374213836478e-05,
      "loss": 0.8993,
      "num_input_tokens_seen": 1089696,
      "step": 1660
    },
    {
      "epoch": 0.8726415094339622,
      "grad_norm": 0.494465172290802,
      "learning_rate": 2.1802935010482182e-05,
      "loss": 0.7156,
      "num_input_tokens_seen": 1092608,
      "step": 1665
    },
    {
      "epoch": 0.8752620545073375,
      "grad_norm": 0.2773623764514923,
      "learning_rate": 2.186844863731656e-05,
      "loss": 0.9599,
      "num_input_tokens_seen": 1095808,
      "step": 1670
    },
    {
      "epoch": 0.8778825995807128,
      "grad_norm": 0.2600119411945343,
      "learning_rate": 2.1933962264150945e-05,
      "loss": 0.8605,
      "num_input_tokens_seen": 1098816,
      "step": 1675
    },
    {
      "epoch": 0.8805031446540881,
      "grad_norm": 0.20950450003147125,
      "learning_rate": 2.1999475890985328e-05,
      "loss": 0.7591,
      "num_input_tokens_seen": 1103136,
      "step": 1680
    },
    {
      "epoch": 0.8831236897274634,
      "grad_norm": 0.39760515093803406,
      "learning_rate": 2.2064989517819707e-05,
      "loss": 0.6561,
      "num_input_tokens_seen": 1107872,
      "step": 1685
    },
    {
      "epoch": 0.8857442348008385,
      "grad_norm": 0.4011513590812683,
      "learning_rate": 2.2130503144654087e-05,
      "loss": 0.8817,
      "num_input_tokens_seen": 1111072,
      "step": 1690
    },
    {
      "epoch": 0.8883647798742138,
      "grad_norm": 0.3891417384147644,
      "learning_rate": 2.219601677148847e-05,
      "loss": 0.8301,
      "num_input_tokens_seen": 1113792,
      "step": 1695
    },
    {
      "epoch": 0.8909853249475891,
      "grad_norm": 0.572274923324585,
      "learning_rate": 2.2261530398322853e-05,
      "loss": 0.8069,
      "num_input_tokens_seen": 1116384,
      "step": 1700
    },
    {
      "epoch": 0.8936058700209644,
      "grad_norm": 0.29355600476264954,
      "learning_rate": 2.2327044025157233e-05,
      "loss": 0.7151,
      "num_input_tokens_seen": 1119296,
      "step": 1705
    },
    {
      "epoch": 0.8962264150943396,
      "grad_norm": 0.23985031247138977,
      "learning_rate": 2.2392557651991616e-05,
      "loss": 0.8124,
      "num_input_tokens_seen": 1122720,
      "step": 1710
    },
    {
      "epoch": 0.8988469601677149,
      "grad_norm": 1.022184133529663,
      "learning_rate": 2.2458071278825996e-05,
      "loss": 1.1253,
      "num_input_tokens_seen": 1125088,
      "step": 1715
    },
    {
      "epoch": 0.9014675052410901,
      "grad_norm": 0.3364733159542084,
      "learning_rate": 2.252358490566038e-05,
      "loss": 0.8268,
      "num_input_tokens_seen": 1128288,
      "step": 1720
    },
    {
      "epoch": 0.9040880503144654,
      "grad_norm": 0.9141489267349243,
      "learning_rate": 2.258909853249476e-05,
      "loss": 0.9272,
      "num_input_tokens_seen": 1131040,
      "step": 1725
    },
    {
      "epoch": 0.9067085953878407,
      "grad_norm": 0.3849252164363861,
      "learning_rate": 2.265461215932914e-05,
      "loss": 0.7285,
      "num_input_tokens_seen": 1135136,
      "step": 1730
    },
    {
      "epoch": 0.9093291404612159,
      "grad_norm": 0.27575042843818665,
      "learning_rate": 2.2720125786163524e-05,
      "loss": 0.885,
      "num_input_tokens_seen": 1139008,
      "step": 1735
    },
    {
      "epoch": 0.9119496855345912,
      "grad_norm": 0.2625947594642639,
      "learning_rate": 2.2785639412997904e-05,
      "loss": 0.7207,
      "num_input_tokens_seen": 1141920,
      "step": 1740
    },
    {
      "epoch": 0.9145702306079665,
      "grad_norm": 0.3945213258266449,
      "learning_rate": 2.2851153039832284e-05,
      "loss": 0.8439,
      "num_input_tokens_seen": 1144960,
      "step": 1745
    },
    {
      "epoch": 0.9171907756813418,
      "grad_norm": 0.5239211320877075,
      "learning_rate": 2.2916666666666667e-05,
      "loss": 0.823,
      "num_input_tokens_seen": 1147712,
      "step": 1750
    },
    {
      "epoch": 0.9198113207547169,
      "grad_norm": 0.26826953887939453,
      "learning_rate": 2.298218029350105e-05,
      "loss": 0.7461,
      "num_input_tokens_seen": 1150528,
      "step": 1755
    },
    {
      "epoch": 0.9224318658280922,
      "grad_norm": 0.5652328729629517,
      "learning_rate": 2.304769392033543e-05,
      "loss": 0.8533,
      "num_input_tokens_seen": 1154272,
      "step": 1760
    },
    {
      "epoch": 0.9250524109014675,
      "grad_norm": 0.923331081867218,
      "learning_rate": 2.3113207547169813e-05,
      "loss": 0.7983,
      "num_input_tokens_seen": 1157088,
      "step": 1765
    },
    {
      "epoch": 0.9276729559748428,
      "grad_norm": 0.34153836965560913,
      "learning_rate": 2.3178721174004196e-05,
      "loss": 0.8705,
      "num_input_tokens_seen": 1160192,
      "step": 1770
    },
    {
      "epoch": 0.9302935010482181,
      "grad_norm": 0.32567697763442993,
      "learning_rate": 2.3244234800838576e-05,
      "loss": 0.7191,
      "num_input_tokens_seen": 1163424,
      "step": 1775
    },
    {
      "epoch": 0.9329140461215933,
      "grad_norm": 0.321990042924881,
      "learning_rate": 2.3309748427672955e-05,
      "loss": 0.7622,
      "num_input_tokens_seen": 1167104,
      "step": 1780
    },
    {
      "epoch": 0.9355345911949685,
      "grad_norm": 0.5200710892677307,
      "learning_rate": 2.3375262054507338e-05,
      "loss": 0.7283,
      "num_input_tokens_seen": 1170368,
      "step": 1785
    },
    {
      "epoch": 0.9381551362683438,
      "grad_norm": 0.2619456350803375,
      "learning_rate": 2.344077568134172e-05,
      "loss": 0.8257,
      "num_input_tokens_seen": 1173888,
      "step": 1790
    },
    {
      "epoch": 0.9407756813417191,
      "grad_norm": 0.2625054121017456,
      "learning_rate": 2.35062893081761e-05,
      "loss": 0.6322,
      "num_input_tokens_seen": 1177408,
      "step": 1795
    },
    {
      "epoch": 0.9433962264150944,
      "grad_norm": 0.45680564641952515,
      "learning_rate": 2.3571802935010484e-05,
      "loss": 0.638,
      "num_input_tokens_seen": 1181056,
      "step": 1800
    },
    {
      "epoch": 0.9460167714884696,
      "grad_norm": 0.38979312777519226,
      "learning_rate": 2.3637316561844867e-05,
      "loss": 0.9595,
      "num_input_tokens_seen": 1184256,
      "step": 1805
    },
    {
      "epoch": 0.9486373165618449,
      "grad_norm": 0.3534207046031952,
      "learning_rate": 2.3702830188679247e-05,
      "loss": 0.6589,
      "num_input_tokens_seen": 1187168,
      "step": 1810
    },
    {
      "epoch": 0.9512578616352201,
      "grad_norm": 0.379721075296402,
      "learning_rate": 2.3768343815513627e-05,
      "loss": 0.6308,
      "num_input_tokens_seen": 1190784,
      "step": 1815
    },
    {
      "epoch": 0.9538784067085954,
      "grad_norm": 0.3232034742832184,
      "learning_rate": 2.383385744234801e-05,
      "loss": 0.8493,
      "num_input_tokens_seen": 1193312,
      "step": 1820
    },
    {
      "epoch": 0.9564989517819706,
      "grad_norm": 0.3618949055671692,
      "learning_rate": 2.3899371069182393e-05,
      "loss": 0.7466,
      "num_input_tokens_seen": 1196608,
      "step": 1825
    },
    {
      "epoch": 0.9591194968553459,
      "grad_norm": 0.432960569858551,
      "learning_rate": 2.3964884696016772e-05,
      "loss": 0.8237,
      "num_input_tokens_seen": 1200160,
      "step": 1830
    },
    {
      "epoch": 0.9617400419287212,
      "grad_norm": 0.7547792196273804,
      "learning_rate": 2.4030398322851155e-05,
      "loss": 0.6989,
      "num_input_tokens_seen": 1203040,
      "step": 1835
    },
    {
      "epoch": 0.9643605870020965,
      "grad_norm": 0.3299584686756134,
      "learning_rate": 2.4095911949685535e-05,
      "loss": 0.5969,
      "num_input_tokens_seen": 1205408,
      "step": 1840
    },
    {
      "epoch": 0.9669811320754716,
      "grad_norm": 0.8720060586929321,
      "learning_rate": 2.4161425576519918e-05,
      "loss": 0.6665,
      "num_input_tokens_seen": 1208896,
      "step": 1845
    },
    {
      "epoch": 0.9696016771488469,
      "grad_norm": 0.20096178352832794,
      "learning_rate": 2.4226939203354298e-05,
      "loss": 0.6455,
      "num_input_tokens_seen": 1212576,
      "step": 1850
    },
    {
      "epoch": 0.9722222222222222,
      "grad_norm": 0.3095814287662506,
      "learning_rate": 2.429245283018868e-05,
      "loss": 0.6666,
      "num_input_tokens_seen": 1215872,
      "step": 1855
    },
    {
      "epoch": 0.9748427672955975,
      "grad_norm": 0.43872708082199097,
      "learning_rate": 2.4357966457023064e-05,
      "loss": 0.6605,
      "num_input_tokens_seen": 1219968,
      "step": 1860
    },
    {
      "epoch": 0.9774633123689728,
      "grad_norm": 0.2877185344696045,
      "learning_rate": 2.4423480083857444e-05,
      "loss": 0.7984,
      "num_input_tokens_seen": 1222048,
      "step": 1865
    },
    {
      "epoch": 0.980083857442348,
      "grad_norm": 0.42481884360313416,
      "learning_rate": 2.4488993710691823e-05,
      "loss": 0.7677,
      "num_input_tokens_seen": 1225120,
      "step": 1870
    },
    {
      "epoch": 0.9827044025157232,
      "grad_norm": 0.2259686440229416,
      "learning_rate": 2.4554507337526206e-05,
      "loss": 0.6823,
      "num_input_tokens_seen": 1229056,
      "step": 1875
    },
    {
      "epoch": 0.9853249475890985,
      "grad_norm": 0.5509381294250488,
      "learning_rate": 2.4620020964360586e-05,
      "loss": 0.6416,
      "num_input_tokens_seen": 1231776,
      "step": 1880
    },
    {
      "epoch": 0.9879454926624738,
      "grad_norm": 0.32879507541656494,
      "learning_rate": 2.468553459119497e-05,
      "loss": 0.7308,
      "num_input_tokens_seen": 1235040,
      "step": 1885
    },
    {
      "epoch": 0.9905660377358491,
      "grad_norm": 0.5346120595932007,
      "learning_rate": 2.4751048218029352e-05,
      "loss": 0.6759,
      "num_input_tokens_seen": 1238048,
      "step": 1890
    },
    {
      "epoch": 0.9931865828092243,
      "grad_norm": 0.28786781430244446,
      "learning_rate": 2.4816561844863735e-05,
      "loss": 0.7161,
      "num_input_tokens_seen": 1241344,
      "step": 1895
    },
    {
      "epoch": 0.9958071278825996,
      "grad_norm": 0.4217551052570343,
      "learning_rate": 2.488207547169811e-05,
      "loss": 0.5741,
      "num_input_tokens_seen": 1243840,
      "step": 1900
    },
    {
      "epoch": 0.9984276729559748,
      "grad_norm": 0.1762140840291977,
      "learning_rate": 2.4947589098532495e-05,
      "loss": 0.7537,
      "num_input_tokens_seen": 1246816,
      "step": 1905
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.7072458267211914,
      "eval_runtime": 9.8949,
      "eval_samples_per_second": 85.701,
      "eval_steps_per_second": 21.425,
      "num_input_tokens_seen": 1248304,
      "step": 1908
    },
    {
      "epoch": 1.00104821802935,
      "grad_norm": 0.38779303431510925,
      "learning_rate": 2.5013102725366878e-05,
      "loss": 0.7354,
      "num_input_tokens_seen": 1249200,
      "step": 1910
    },
    {
      "epoch": 1.0036687631027255,
      "grad_norm": 0.4270458519458771,
      "learning_rate": 2.507861635220126e-05,
      "loss": 0.8665,
      "num_input_tokens_seen": 1251664,
      "step": 1915
    },
    {
      "epoch": 1.0062893081761006,
      "grad_norm": 0.19739660620689392,
      "learning_rate": 2.514412997903564e-05,
      "loss": 0.6064,
      "num_input_tokens_seen": 1254928,
      "step": 1920
    },
    {
      "epoch": 1.0089098532494758,
      "grad_norm": 0.7388924956321716,
      "learning_rate": 2.520964360587002e-05,
      "loss": 0.6706,
      "num_input_tokens_seen": 1257744,
      "step": 1925
    },
    {
      "epoch": 1.0115303983228512,
      "grad_norm": 0.5749493837356567,
      "learning_rate": 2.5275157232704407e-05,
      "loss": 0.6268,
      "num_input_tokens_seen": 1261904,
      "step": 1930
    },
    {
      "epoch": 1.0141509433962264,
      "grad_norm": 0.3705325126647949,
      "learning_rate": 2.5340670859538783e-05,
      "loss": 0.5757,
      "num_input_tokens_seen": 1264432,
      "step": 1935
    },
    {
      "epoch": 1.0167714884696017,
      "grad_norm": 0.4873795807361603,
      "learning_rate": 2.540618448637317e-05,
      "loss": 0.745,
      "num_input_tokens_seen": 1268400,
      "step": 1940
    },
    {
      "epoch": 1.019392033542977,
      "grad_norm": 0.31739819049835205,
      "learning_rate": 2.547169811320755e-05,
      "loss": 0.5995,
      "num_input_tokens_seen": 1272752,
      "step": 1945
    },
    {
      "epoch": 1.0220125786163523,
      "grad_norm": 0.30870485305786133,
      "learning_rate": 2.553721174004193e-05,
      "loss": 0.7554,
      "num_input_tokens_seen": 1275568,
      "step": 1950
    },
    {
      "epoch": 1.0246331236897275,
      "grad_norm": 0.15778301656246185,
      "learning_rate": 2.5602725366876312e-05,
      "loss": 0.7591,
      "num_input_tokens_seen": 1278160,
      "step": 1955
    },
    {
      "epoch": 1.0272536687631026,
      "grad_norm": 0.45691439509391785,
      "learning_rate": 2.566823899371069e-05,
      "loss": 0.6453,
      "num_input_tokens_seen": 1281424,
      "step": 1960
    },
    {
      "epoch": 1.029874213836478,
      "grad_norm": 0.5578243732452393,
      "learning_rate": 2.5733752620545075e-05,
      "loss": 0.6894,
      "num_input_tokens_seen": 1283920,
      "step": 1965
    },
    {
      "epoch": 1.0324947589098532,
      "grad_norm": 0.5494023561477661,
      "learning_rate": 2.5799266247379454e-05,
      "loss": 0.6187,
      "num_input_tokens_seen": 1287920,
      "step": 1970
    },
    {
      "epoch": 1.0351153039832286,
      "grad_norm": 0.5668577551841736,
      "learning_rate": 2.586477987421384e-05,
      "loss": 0.7255,
      "num_input_tokens_seen": 1290960,
      "step": 1975
    },
    {
      "epoch": 1.0377358490566038,
      "grad_norm": 0.3337642252445221,
      "learning_rate": 2.593029350104822e-05,
      "loss": 0.5457,
      "num_input_tokens_seen": 1294544,
      "step": 1980
    },
    {
      "epoch": 1.040356394129979,
      "grad_norm": 0.20938237011432648,
      "learning_rate": 2.59958071278826e-05,
      "loss": 0.7951,
      "num_input_tokens_seen": 1298160,
      "step": 1985
    },
    {
      "epoch": 1.0429769392033543,
      "grad_norm": 0.3567543029785156,
      "learning_rate": 2.6061320754716983e-05,
      "loss": 0.7384,
      "num_input_tokens_seen": 1300368,
      "step": 1990
    },
    {
      "epoch": 1.0455974842767295,
      "grad_norm": 0.13182514905929565,
      "learning_rate": 2.6126834381551363e-05,
      "loss": 0.539,
      "num_input_tokens_seen": 1307184,
      "step": 1995
    },
    {
      "epoch": 1.0482180293501049,
      "grad_norm": 0.30401811003685,
      "learning_rate": 2.6192348008385746e-05,
      "loss": 1.0234,
      "num_input_tokens_seen": 1309840,
      "step": 2000
    },
    {
      "epoch": 1.05083857442348,
      "grad_norm": 0.41542139649391174,
      "learning_rate": 2.6257861635220126e-05,
      "loss": 0.6238,
      "num_input_tokens_seen": 1313648,
      "step": 2005
    },
    {
      "epoch": 1.0534591194968554,
      "grad_norm": 0.2633415758609772,
      "learning_rate": 2.6323375262054505e-05,
      "loss": 0.6211,
      "num_input_tokens_seen": 1316496,
      "step": 2010
    },
    {
      "epoch": 1.0560796645702306,
      "grad_norm": 0.5686927437782288,
      "learning_rate": 2.6388888888888892e-05,
      "loss": 0.6638,
      "num_input_tokens_seen": 1319952,
      "step": 2015
    },
    {
      "epoch": 1.0587002096436058,
      "grad_norm": 0.29825761914253235,
      "learning_rate": 2.645440251572327e-05,
      "loss": 0.6317,
      "num_input_tokens_seen": 1323344,
      "step": 2020
    },
    {
      "epoch": 1.0613207547169812,
      "grad_norm": 0.3566904664039612,
      "learning_rate": 2.6519916142557655e-05,
      "loss": 0.7619,
      "num_input_tokens_seen": 1326704,
      "step": 2025
    },
    {
      "epoch": 1.0639412997903563,
      "grad_norm": 0.2449316829442978,
      "learning_rate": 2.6585429769392034e-05,
      "loss": 0.6269,
      "num_input_tokens_seen": 1330128,
      "step": 2030
    },
    {
      "epoch": 1.0665618448637317,
      "grad_norm": 0.4169636368751526,
      "learning_rate": 2.6650943396226417e-05,
      "loss": 0.5927,
      "num_input_tokens_seen": 1333040,
      "step": 2035
    },
    {
      "epoch": 1.069182389937107,
      "grad_norm": 0.3292876183986664,
      "learning_rate": 2.6716457023060797e-05,
      "loss": 0.6994,
      "num_input_tokens_seen": 1335504,
      "step": 2040
    },
    {
      "epoch": 1.0718029350104823,
      "grad_norm": 0.45653411746025085,
      "learning_rate": 2.6781970649895177e-05,
      "loss": 0.6769,
      "num_input_tokens_seen": 1339280,
      "step": 2045
    },
    {
      "epoch": 1.0744234800838575,
      "grad_norm": 0.4294579029083252,
      "learning_rate": 2.6847484276729563e-05,
      "loss": 0.6955,
      "num_input_tokens_seen": 1342576,
      "step": 2050
    },
    {
      "epoch": 1.0770440251572326,
      "grad_norm": 0.38918110728263855,
      "learning_rate": 2.6912997903563943e-05,
      "loss": 0.6391,
      "num_input_tokens_seen": 1346256,
      "step": 2055
    },
    {
      "epoch": 1.079664570230608,
      "grad_norm": 0.4659879505634308,
      "learning_rate": 2.6978511530398326e-05,
      "loss": 0.7475,
      "num_input_tokens_seen": 1349008,
      "step": 2060
    },
    {
      "epoch": 1.0822851153039832,
      "grad_norm": 0.35708847641944885,
      "learning_rate": 2.7044025157232706e-05,
      "loss": 0.6515,
      "num_input_tokens_seen": 1353552,
      "step": 2065
    },
    {
      "epoch": 1.0849056603773586,
      "grad_norm": 0.4358104467391968,
      "learning_rate": 2.710953878406709e-05,
      "loss": 0.6737,
      "num_input_tokens_seen": 1356400,
      "step": 2070
    },
    {
      "epoch": 1.0875262054507338,
      "grad_norm": 0.36165547370910645,
      "learning_rate": 2.717505241090147e-05,
      "loss": 0.6542,
      "num_input_tokens_seen": 1361008,
      "step": 2075
    },
    {
      "epoch": 1.090146750524109,
      "grad_norm": 0.7130039930343628,
      "learning_rate": 2.7240566037735848e-05,
      "loss": 0.82,
      "num_input_tokens_seen": 1363184,
      "step": 2080
    },
    {
      "epoch": 1.0927672955974843,
      "grad_norm": 0.5353054404258728,
      "learning_rate": 2.7306079664570235e-05,
      "loss": 0.6667,
      "num_input_tokens_seen": 1365872,
      "step": 2085
    },
    {
      "epoch": 1.0953878406708595,
      "grad_norm": 0.33861270546913147,
      "learning_rate": 2.737159329140461e-05,
      "loss": 0.7211,
      "num_input_tokens_seen": 1369168,
      "step": 2090
    },
    {
      "epoch": 1.0980083857442349,
      "grad_norm": 0.27300313115119934,
      "learning_rate": 2.7437106918238997e-05,
      "loss": 0.6323,
      "num_input_tokens_seen": 1372208,
      "step": 2095
    },
    {
      "epoch": 1.10062893081761,
      "grad_norm": 0.3452147841453552,
      "learning_rate": 2.7502620545073377e-05,
      "loss": 0.5125,
      "num_input_tokens_seen": 1377360,
      "step": 2100
    },
    {
      "epoch": 1.1032494758909852,
      "grad_norm": 0.486378937959671,
      "learning_rate": 2.7568134171907757e-05,
      "loss": 0.6406,
      "num_input_tokens_seen": 1380400,
      "step": 2105
    },
    {
      "epoch": 1.1058700209643606,
      "grad_norm": 0.40273019671440125,
      "learning_rate": 2.763364779874214e-05,
      "loss": 0.6828,
      "num_input_tokens_seen": 1383056,
      "step": 2110
    },
    {
      "epoch": 1.1084905660377358,
      "grad_norm": 0.329985111951828,
      "learning_rate": 2.769916142557652e-05,
      "loss": 0.742,
      "num_input_tokens_seen": 1386160,
      "step": 2115
    },
    {
      "epoch": 1.1111111111111112,
      "grad_norm": 0.2694031894207001,
      "learning_rate": 2.7764675052410906e-05,
      "loss": 0.5819,
      "num_input_tokens_seen": 1389232,
      "step": 2120
    },
    {
      "epoch": 1.1137316561844863,
      "grad_norm": 0.26810723543167114,
      "learning_rate": 2.7830188679245282e-05,
      "loss": 0.7549,
      "num_input_tokens_seen": 1393328,
      "step": 2125
    },
    {
      "epoch": 1.1163522012578617,
      "grad_norm": 0.296525776386261,
      "learning_rate": 2.789570230607967e-05,
      "loss": 0.5597,
      "num_input_tokens_seen": 1396816,
      "step": 2130
    },
    {
      "epoch": 1.118972746331237,
      "grad_norm": 0.25546449422836304,
      "learning_rate": 2.796121593291405e-05,
      "loss": 0.612,
      "num_input_tokens_seen": 1400176,
      "step": 2135
    },
    {
      "epoch": 1.121593291404612,
      "grad_norm": 0.24599328637123108,
      "learning_rate": 2.8026729559748428e-05,
      "loss": 0.7797,
      "num_input_tokens_seen": 1403408,
      "step": 2140
    },
    {
      "epoch": 1.1242138364779874,
      "grad_norm": 0.22193920612335205,
      "learning_rate": 2.809224318658281e-05,
      "loss": 0.6809,
      "num_input_tokens_seen": 1406448,
      "step": 2145
    },
    {
      "epoch": 1.1268343815513626,
      "grad_norm": 0.7556794881820679,
      "learning_rate": 2.815775681341719e-05,
      "loss": 0.7585,
      "num_input_tokens_seen": 1409584,
      "step": 2150
    },
    {
      "epoch": 1.129454926624738,
      "grad_norm": 0.30433884263038635,
      "learning_rate": 2.8223270440251577e-05,
      "loss": 0.5789,
      "num_input_tokens_seen": 1413872,
      "step": 2155
    },
    {
      "epoch": 1.1320754716981132,
      "grad_norm": 0.30414465069770813,
      "learning_rate": 2.8288784067085954e-05,
      "loss": 0.867,
      "num_input_tokens_seen": 1416528,
      "step": 2160
    },
    {
      "epoch": 1.1346960167714886,
      "grad_norm": 0.4505849778652191,
      "learning_rate": 2.8354297693920333e-05,
      "loss": 0.6485,
      "num_input_tokens_seen": 1418896,
      "step": 2165
    },
    {
      "epoch": 1.1373165618448637,
      "grad_norm": 0.36208876967430115,
      "learning_rate": 2.841981132075472e-05,
      "loss": 0.6954,
      "num_input_tokens_seen": 1421712,
      "step": 2170
    },
    {
      "epoch": 1.139937106918239,
      "grad_norm": 0.2805896997451782,
      "learning_rate": 2.84853249475891e-05,
      "loss": 0.6452,
      "num_input_tokens_seen": 1425072,
      "step": 2175
    },
    {
      "epoch": 1.1425576519916143,
      "grad_norm": 0.168817400932312,
      "learning_rate": 2.8550838574423482e-05,
      "loss": 0.7639,
      "num_input_tokens_seen": 1428848,
      "step": 2180
    },
    {
      "epoch": 1.1451781970649895,
      "grad_norm": 0.4264901876449585,
      "learning_rate": 2.8616352201257862e-05,
      "loss": 0.6769,
      "num_input_tokens_seen": 1432432,
      "step": 2185
    },
    {
      "epoch": 1.1477987421383649,
      "grad_norm": 0.1703544557094574,
      "learning_rate": 2.868186582809225e-05,
      "loss": 0.7748,
      "num_input_tokens_seen": 1436048,
      "step": 2190
    },
    {
      "epoch": 1.15041928721174,
      "grad_norm": 0.3849943280220032,
      "learning_rate": 2.8747379454926625e-05,
      "loss": 0.8101,
      "num_input_tokens_seen": 1440144,
      "step": 2195
    },
    {
      "epoch": 1.1530398322851152,
      "grad_norm": 0.2645401954650879,
      "learning_rate": 2.8812893081761005e-05,
      "loss": 0.7181,
      "num_input_tokens_seen": 1443760,
      "step": 2200
    },
    {
      "epoch": 1.1556603773584906,
      "grad_norm": 0.2240818589925766,
      "learning_rate": 2.887840670859539e-05,
      "loss": 0.9936,
      "num_input_tokens_seen": 1446864,
      "step": 2205
    },
    {
      "epoch": 1.1582809224318658,
      "grad_norm": 0.3040007948875427,
      "learning_rate": 2.894392033542977e-05,
      "loss": 0.4621,
      "num_input_tokens_seen": 1449456,
      "step": 2210
    },
    {
      "epoch": 1.1609014675052411,
      "grad_norm": 0.38323673605918884,
      "learning_rate": 2.9009433962264154e-05,
      "loss": 0.7465,
      "num_input_tokens_seen": 1452944,
      "step": 2215
    },
    {
      "epoch": 1.1635220125786163,
      "grad_norm": 0.15721428394317627,
      "learning_rate": 2.9074947589098533e-05,
      "loss": 0.6092,
      "num_input_tokens_seen": 1455952,
      "step": 2220
    },
    {
      "epoch": 1.1661425576519917,
      "grad_norm": 0.2581445574760437,
      "learning_rate": 2.9140461215932913e-05,
      "loss": 0.7697,
      "num_input_tokens_seen": 1459376,
      "step": 2225
    },
    {
      "epoch": 1.1687631027253669,
      "grad_norm": 0.24116995930671692,
      "learning_rate": 2.9205974842767296e-05,
      "loss": 0.6047,
      "num_input_tokens_seen": 1462352,
      "step": 2230
    },
    {
      "epoch": 1.171383647798742,
      "grad_norm": 0.3326542377471924,
      "learning_rate": 2.9271488469601676e-05,
      "loss": 0.7815,
      "num_input_tokens_seen": 1465104,
      "step": 2235
    },
    {
      "epoch": 1.1740041928721174,
      "grad_norm": 0.17764152586460114,
      "learning_rate": 2.9337002096436062e-05,
      "loss": 0.7033,
      "num_input_tokens_seen": 1468624,
      "step": 2240
    },
    {
      "epoch": 1.1766247379454926,
      "grad_norm": 0.3504980802536011,
      "learning_rate": 2.9402515723270442e-05,
      "loss": 0.581,
      "num_input_tokens_seen": 1475184,
      "step": 2245
    },
    {
      "epoch": 1.179245283018868,
      "grad_norm": 0.11385639011859894,
      "learning_rate": 2.9468029350104825e-05,
      "loss": 0.549,
      "num_input_tokens_seen": 1478640,
      "step": 2250
    },
    {
      "epoch": 1.1818658280922432,
      "grad_norm": 0.2812000513076782,
      "learning_rate": 2.9533542976939205e-05,
      "loss": 0.7222,
      "num_input_tokens_seen": 1481552,
      "step": 2255
    },
    {
      "epoch": 1.1844863731656186,
      "grad_norm": 0.30726855993270874,
      "learning_rate": 2.9599056603773584e-05,
      "loss": 0.6347,
      "num_input_tokens_seen": 1485104,
      "step": 2260
    },
    {
      "epoch": 1.1871069182389937,
      "grad_norm": 0.34347212314605713,
      "learning_rate": 2.9664570230607968e-05,
      "loss": 0.773,
      "num_input_tokens_seen": 1487728,
      "step": 2265
    },
    {
      "epoch": 1.189727463312369,
      "grad_norm": 0.16114485263824463,
      "learning_rate": 2.9730083857442347e-05,
      "loss": 0.5472,
      "num_input_tokens_seen": 1491312,
      "step": 2270
    },
    {
      "epoch": 1.1923480083857443,
      "grad_norm": 0.30426403880119324,
      "learning_rate": 2.9795597484276734e-05,
      "loss": 0.735,
      "num_input_tokens_seen": 1494192,
      "step": 2275
    },
    {
      "epoch": 1.1949685534591195,
      "grad_norm": 0.26742681860923767,
      "learning_rate": 2.9861111111111113e-05,
      "loss": 0.8152,
      "num_input_tokens_seen": 1497168,
      "step": 2280
    },
    {
      "epoch": 1.1975890985324948,
      "grad_norm": 0.2045978158712387,
      "learning_rate": 2.9926624737945496e-05,
      "loss": 0.606,
      "num_input_tokens_seen": 1501936,
      "step": 2285
    },
    {
      "epoch": 1.20020964360587,
      "grad_norm": 0.17807307839393616,
      "learning_rate": 2.9992138364779876e-05,
      "loss": 0.5586,
      "num_input_tokens_seen": 1505040,
      "step": 2290
    },
    {
      "epoch": 1.2028301886792452,
      "grad_norm": 0.46331796050071716,
      "learning_rate": 3.0057651991614256e-05,
      "loss": 0.7962,
      "num_input_tokens_seen": 1507856,
      "step": 2295
    },
    {
      "epoch": 1.2054507337526206,
      "grad_norm": 0.40816980600357056,
      "learning_rate": 3.012316561844864e-05,
      "loss": 0.6437,
      "num_input_tokens_seen": 1510768,
      "step": 2300
    },
    {
      "epoch": 1.2080712788259957,
      "grad_norm": 0.24285192787647247,
      "learning_rate": 3.018867924528302e-05,
      "loss": 0.7669,
      "num_input_tokens_seen": 1513808,
      "step": 2305
    },
    {
      "epoch": 1.2106918238993711,
      "grad_norm": 0.18570417165756226,
      "learning_rate": 3.0254192872117405e-05,
      "loss": 0.6059,
      "num_input_tokens_seen": 1517392,
      "step": 2310
    },
    {
      "epoch": 1.2133123689727463,
      "grad_norm": 0.13912251591682434,
      "learning_rate": 3.0319706498951785e-05,
      "loss": 0.6354,
      "num_input_tokens_seen": 1521648,
      "step": 2315
    },
    {
      "epoch": 1.2159329140461215,
      "grad_norm": 0.1836262345314026,
      "learning_rate": 3.038522012578616e-05,
      "loss": 0.5917,
      "num_input_tokens_seen": 1524496,
      "step": 2320
    },
    {
      "epoch": 1.2185534591194969,
      "grad_norm": 0.2401469349861145,
      "learning_rate": 3.0450733752620547e-05,
      "loss": 0.5914,
      "num_input_tokens_seen": 1528144,
      "step": 2325
    },
    {
      "epoch": 1.221174004192872,
      "grad_norm": 0.2382930964231491,
      "learning_rate": 3.0516247379454927e-05,
      "loss": 0.5985,
      "num_input_tokens_seen": 1531664,
      "step": 2330
    },
    {
      "epoch": 1.2237945492662474,
      "grad_norm": 0.32092753052711487,
      "learning_rate": 3.0581761006289314e-05,
      "loss": 0.6913,
      "num_input_tokens_seen": 1535408,
      "step": 2335
    },
    {
      "epoch": 1.2264150943396226,
      "grad_norm": 0.29872164130210876,
      "learning_rate": 3.064727463312369e-05,
      "loss": 0.6398,
      "num_input_tokens_seen": 1538448,
      "step": 2340
    },
    {
      "epoch": 1.229035639412998,
      "grad_norm": 0.3116190731525421,
      "learning_rate": 3.071278825995807e-05,
      "loss": 0.6767,
      "num_input_tokens_seen": 1541968,
      "step": 2345
    },
    {
      "epoch": 1.2316561844863732,
      "grad_norm": 0.1845983862876892,
      "learning_rate": 3.0778301886792456e-05,
      "loss": 0.6732,
      "num_input_tokens_seen": 1544560,
      "step": 2350
    },
    {
      "epoch": 1.2342767295597485,
      "grad_norm": 0.44856685400009155,
      "learning_rate": 3.084381551362683e-05,
      "loss": 0.6691,
      "num_input_tokens_seen": 1548080,
      "step": 2355
    },
    {
      "epoch": 1.2368972746331237,
      "grad_norm": 0.30172231793403625,
      "learning_rate": 3.0909329140461215e-05,
      "loss": 0.5769,
      "num_input_tokens_seen": 1551472,
      "step": 2360
    },
    {
      "epoch": 1.2395178197064989,
      "grad_norm": 0.18307532370090485,
      "learning_rate": 3.09748427672956e-05,
      "loss": 0.7658,
      "num_input_tokens_seen": 1553904,
      "step": 2365
    },
    {
      "epoch": 1.2421383647798743,
      "grad_norm": 0.233571395277977,
      "learning_rate": 3.104035639412998e-05,
      "loss": 0.5758,
      "num_input_tokens_seen": 1558352,
      "step": 2370
    },
    {
      "epoch": 1.2447589098532494,
      "grad_norm": 0.29167330265045166,
      "learning_rate": 3.1105870020964365e-05,
      "loss": 0.5536,
      "num_input_tokens_seen": 1561616,
      "step": 2375
    },
    {
      "epoch": 1.2473794549266248,
      "grad_norm": 0.25581085681915283,
      "learning_rate": 3.117138364779874e-05,
      "loss": 0.4907,
      "num_input_tokens_seen": 1566000,
      "step": 2380
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.20756588876247406,
      "learning_rate": 3.1236897274633124e-05,
      "loss": 0.6424,
      "num_input_tokens_seen": 1569776,
      "step": 2385
    },
    {
      "epoch": 1.2526205450733752,
      "grad_norm": 0.3925889730453491,
      "learning_rate": 3.130241090146751e-05,
      "loss": 0.8045,
      "num_input_tokens_seen": 1572112,
      "step": 2390
    },
    {
      "epoch": 1.2552410901467506,
      "grad_norm": 0.2388410121202469,
      "learning_rate": 3.136792452830189e-05,
      "loss": 0.6365,
      "num_input_tokens_seen": 1575312,
      "step": 2395
    },
    {
      "epoch": 1.2578616352201257,
      "grad_norm": 0.17310717701911926,
      "learning_rate": 3.1433438155136266e-05,
      "loss": 0.7034,
      "num_input_tokens_seen": 1578256,
      "step": 2400
    },
    {
      "epoch": 1.2604821802935011,
      "grad_norm": 0.1832849085330963,
      "learning_rate": 3.1498951781970656e-05,
      "loss": 0.812,
      "num_input_tokens_seen": 1581328,
      "step": 2405
    },
    {
      "epoch": 1.2631027253668763,
      "grad_norm": 0.22802752256393433,
      "learning_rate": 3.156446540880503e-05,
      "loss": 0.7029,
      "num_input_tokens_seen": 1583952,
      "step": 2410
    },
    {
      "epoch": 1.2657232704402515,
      "grad_norm": 0.5026324391365051,
      "learning_rate": 3.162997903563941e-05,
      "loss": 0.6571,
      "num_input_tokens_seen": 1587472,
      "step": 2415
    },
    {
      "epoch": 1.2683438155136268,
      "grad_norm": 0.3409280776977539,
      "learning_rate": 3.16954926624738e-05,
      "loss": 0.5638,
      "num_input_tokens_seen": 1590384,
      "step": 2420
    },
    {
      "epoch": 1.270964360587002,
      "grad_norm": 0.24790845811367035,
      "learning_rate": 3.1761006289308175e-05,
      "loss": 0.6187,
      "num_input_tokens_seen": 1593264,
      "step": 2425
    },
    {
      "epoch": 1.2735849056603774,
      "grad_norm": 0.2389829009771347,
      "learning_rate": 3.182651991614256e-05,
      "loss": 0.5686,
      "num_input_tokens_seen": 1597680,
      "step": 2430
    },
    {
      "epoch": 1.2762054507337526,
      "grad_norm": 0.4011847972869873,
      "learning_rate": 3.189203354297694e-05,
      "loss": 0.655,
      "num_input_tokens_seen": 1600912,
      "step": 2435
    },
    {
      "epoch": 1.2788259958071277,
      "grad_norm": 0.2898082733154297,
      "learning_rate": 3.1957547169811324e-05,
      "loss": 0.5954,
      "num_input_tokens_seen": 1603472,
      "step": 2440
    },
    {
      "epoch": 1.2814465408805031,
      "grad_norm": 0.6602475047111511,
      "learning_rate": 3.202306079664571e-05,
      "loss": 0.6213,
      "num_input_tokens_seen": 1606384,
      "step": 2445
    },
    {
      "epoch": 1.2840670859538785,
      "grad_norm": 0.22532162070274353,
      "learning_rate": 3.2088574423480084e-05,
      "loss": 0.6189,
      "num_input_tokens_seen": 1609360,
      "step": 2450
    },
    {
      "epoch": 1.2866876310272537,
      "grad_norm": 0.34905996918678284,
      "learning_rate": 3.215408805031447e-05,
      "loss": 0.6569,
      "num_input_tokens_seen": 1612336,
      "step": 2455
    },
    {
      "epoch": 1.2893081761006289,
      "grad_norm": 0.18759477138519287,
      "learning_rate": 3.221960167714885e-05,
      "loss": 0.5578,
      "num_input_tokens_seen": 1616048,
      "step": 2460
    },
    {
      "epoch": 1.2919287211740043,
      "grad_norm": 0.4631894528865814,
      "learning_rate": 3.228511530398323e-05,
      "loss": 0.5377,
      "num_input_tokens_seen": 1618960,
      "step": 2465
    },
    {
      "epoch": 1.2945492662473794,
      "grad_norm": 0.15094557404518127,
      "learning_rate": 3.235062893081761e-05,
      "loss": 0.5166,
      "num_input_tokens_seen": 1622544,
      "step": 2470
    },
    {
      "epoch": 1.2971698113207548,
      "grad_norm": 0.17350712418556213,
      "learning_rate": 3.241614255765199e-05,
      "loss": 0.5536,
      "num_input_tokens_seen": 1625872,
      "step": 2475
    },
    {
      "epoch": 1.29979035639413,
      "grad_norm": 0.2105402797460556,
      "learning_rate": 3.2481656184486375e-05,
      "loss": 0.6338,
      "num_input_tokens_seen": 1629456,
      "step": 2480
    },
    {
      "epoch": 1.3024109014675052,
      "grad_norm": 0.1608578860759735,
      "learning_rate": 3.254716981132075e-05,
      "loss": 0.7643,
      "num_input_tokens_seen": 1632432,
      "step": 2485
    },
    {
      "epoch": 1.3050314465408805,
      "grad_norm": 0.20104636251926422,
      "learning_rate": 3.261268343815514e-05,
      "loss": 0.5689,
      "num_input_tokens_seen": 1636080,
      "step": 2490
    },
    {
      "epoch": 1.3076519916142557,
      "grad_norm": 0.16667599976062775,
      "learning_rate": 3.267819706498952e-05,
      "loss": 0.6333,
      "num_input_tokens_seen": 1638960,
      "step": 2495
    },
    {
      "epoch": 1.310272536687631,
      "grad_norm": 0.109183169901371,
      "learning_rate": 3.27437106918239e-05,
      "loss": 0.5812,
      "num_input_tokens_seen": 1643120,
      "step": 2500
    },
    {
      "epoch": 1.3128930817610063,
      "grad_norm": 0.3758978545665741,
      "learning_rate": 3.2809224318658284e-05,
      "loss": 0.5189,
      "num_input_tokens_seen": 1646768,
      "step": 2505
    },
    {
      "epoch": 1.3155136268343814,
      "grad_norm": 0.16196171939373016,
      "learning_rate": 3.287473794549266e-05,
      "loss": 0.5928,
      "num_input_tokens_seen": 1650256,
      "step": 2510
    },
    {
      "epoch": 1.3181341719077568,
      "grad_norm": 0.3992875814437866,
      "learning_rate": 3.294025157232704e-05,
      "loss": 0.6851,
      "num_input_tokens_seen": 1652912,
      "step": 2515
    },
    {
      "epoch": 1.320754716981132,
      "grad_norm": 0.13877877593040466,
      "learning_rate": 3.3005765199161426e-05,
      "loss": 0.6031,
      "num_input_tokens_seen": 1656432,
      "step": 2520
    },
    {
      "epoch": 1.3233752620545074,
      "grad_norm": 0.299739807844162,
      "learning_rate": 3.307127882599581e-05,
      "loss": 0.5343,
      "num_input_tokens_seen": 1659472,
      "step": 2525
    },
    {
      "epoch": 1.3259958071278826,
      "grad_norm": 0.39294520020484924,
      "learning_rate": 3.313679245283019e-05,
      "loss": 0.6383,
      "num_input_tokens_seen": 1662480,
      "step": 2530
    },
    {
      "epoch": 1.3286163522012577,
      "grad_norm": 0.22237874567508698,
      "learning_rate": 3.320230607966457e-05,
      "loss": 0.6387,
      "num_input_tokens_seen": 1665424,
      "step": 2535
    },
    {
      "epoch": 1.3312368972746331,
      "grad_norm": 0.2166832536458969,
      "learning_rate": 3.326781970649895e-05,
      "loss": 0.6726,
      "num_input_tokens_seen": 1667760,
      "step": 2540
    },
    {
      "epoch": 1.3338574423480085,
      "grad_norm": 0.3190220594406128,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 0.5429,
      "num_input_tokens_seen": 1670800,
      "step": 2545
    },
    {
      "epoch": 1.3364779874213837,
      "grad_norm": 0.24078711867332458,
      "learning_rate": 3.339884696016772e-05,
      "loss": 0.6481,
      "num_input_tokens_seen": 1675760,
      "step": 2550
    },
    {
      "epoch": 1.3390985324947589,
      "grad_norm": 0.5286165475845337,
      "learning_rate": 3.3464360587002094e-05,
      "loss": 0.7605,
      "num_input_tokens_seen": 1677936,
      "step": 2555
    },
    {
      "epoch": 1.3417190775681342,
      "grad_norm": 0.23920495808124542,
      "learning_rate": 3.3529874213836484e-05,
      "loss": 0.7087,
      "num_input_tokens_seen": 1685008,
      "step": 2560
    },
    {
      "epoch": 1.3443396226415094,
      "grad_norm": 0.3169342279434204,
      "learning_rate": 3.359538784067086e-05,
      "loss": 0.8138,
      "num_input_tokens_seen": 1688240,
      "step": 2565
    },
    {
      "epoch": 1.3469601677148848,
      "grad_norm": 0.2707183361053467,
      "learning_rate": 3.3660901467505244e-05,
      "loss": 0.8018,
      "num_input_tokens_seen": 1690832,
      "step": 2570
    },
    {
      "epoch": 1.34958071278826,
      "grad_norm": 0.19579987227916718,
      "learning_rate": 3.3726415094339627e-05,
      "loss": 0.7298,
      "num_input_tokens_seen": 1695248,
      "step": 2575
    },
    {
      "epoch": 1.3522012578616351,
      "grad_norm": 0.2280801236629486,
      "learning_rate": 3.3791928721174e-05,
      "loss": 0.6365,
      "num_input_tokens_seen": 1698896,
      "step": 2580
    },
    {
      "epoch": 1.3548218029350105,
      "grad_norm": 0.4949784576892853,
      "learning_rate": 3.3857442348008386e-05,
      "loss": 0.7654,
      "num_input_tokens_seen": 1701296,
      "step": 2585
    },
    {
      "epoch": 1.3574423480083857,
      "grad_norm": 0.2900424897670746,
      "learning_rate": 3.392295597484277e-05,
      "loss": 0.4787,
      "num_input_tokens_seen": 1705136,
      "step": 2590
    },
    {
      "epoch": 1.360062893081761,
      "grad_norm": 0.2080165594816208,
      "learning_rate": 3.398846960167715e-05,
      "loss": 0.5724,
      "num_input_tokens_seen": 1708016,
      "step": 2595
    },
    {
      "epoch": 1.3626834381551363,
      "grad_norm": 1.0398157835006714,
      "learning_rate": 3.4053983228511535e-05,
      "loss": 0.8012,
      "num_input_tokens_seen": 1710864,
      "step": 2600
    },
    {
      "epoch": 1.3653039832285114,
      "grad_norm": 0.22577577829360962,
      "learning_rate": 3.411949685534591e-05,
      "loss": 0.5631,
      "num_input_tokens_seen": 1713552,
      "step": 2605
    },
    {
      "epoch": 1.3679245283018868,
      "grad_norm": 0.22516585886478424,
      "learning_rate": 3.4185010482180295e-05,
      "loss": 0.5448,
      "num_input_tokens_seen": 1716528,
      "step": 2610
    },
    {
      "epoch": 1.370545073375262,
      "grad_norm": 0.7753145694732666,
      "learning_rate": 3.425052410901468e-05,
      "loss": 0.6428,
      "num_input_tokens_seen": 1719216,
      "step": 2615
    },
    {
      "epoch": 1.3731656184486374,
      "grad_norm": 0.20000678300857544,
      "learning_rate": 3.431603773584906e-05,
      "loss": 0.6377,
      "num_input_tokens_seen": 1722384,
      "step": 2620
    },
    {
      "epoch": 1.3757861635220126,
      "grad_norm": 0.1831120401620865,
      "learning_rate": 3.438155136268344e-05,
      "loss": 0.5406,
      "num_input_tokens_seen": 1725104,
      "step": 2625
    },
    {
      "epoch": 1.3784067085953877,
      "grad_norm": 0.3071554899215698,
      "learning_rate": 3.444706498951782e-05,
      "loss": 0.5221,
      "num_input_tokens_seen": 1728112,
      "step": 2630
    },
    {
      "epoch": 1.381027253668763,
      "grad_norm": 0.2960919737815857,
      "learning_rate": 3.45125786163522e-05,
      "loss": 0.7264,
      "num_input_tokens_seen": 1731536,
      "step": 2635
    },
    {
      "epoch": 1.3836477987421385,
      "grad_norm": 0.1737969070672989,
      "learning_rate": 3.457809224318658e-05,
      "loss": 0.681,
      "num_input_tokens_seen": 1734896,
      "step": 2640
    },
    {
      "epoch": 1.3862683438155137,
      "grad_norm": 0.3599955439567566,
      "learning_rate": 3.464360587002097e-05,
      "loss": 0.6716,
      "num_input_tokens_seen": 1737584,
      "step": 2645
    },
    {
      "epoch": 1.3888888888888888,
      "grad_norm": 0.20090270042419434,
      "learning_rate": 3.4709119496855346e-05,
      "loss": 0.701,
      "num_input_tokens_seen": 1740720,
      "step": 2650
    },
    {
      "epoch": 1.3915094339622642,
      "grad_norm": 0.1673928201198578,
      "learning_rate": 3.477463312368973e-05,
      "loss": 0.497,
      "num_input_tokens_seen": 1743216,
      "step": 2655
    },
    {
      "epoch": 1.3941299790356394,
      "grad_norm": 0.2771523594856262,
      "learning_rate": 3.484014675052411e-05,
      "loss": 0.6806,
      "num_input_tokens_seen": 1746000,
      "step": 2660
    },
    {
      "epoch": 1.3967505241090148,
      "grad_norm": 0.28555169701576233,
      "learning_rate": 3.490566037735849e-05,
      "loss": 0.5422,
      "num_input_tokens_seen": 1748560,
      "step": 2665
    },
    {
      "epoch": 1.39937106918239,
      "grad_norm": 0.19936688244342804,
      "learning_rate": 3.497117400419288e-05,
      "loss": 0.7949,
      "num_input_tokens_seen": 1751504,
      "step": 2670
    },
    {
      "epoch": 1.4019916142557651,
      "grad_norm": 0.24548573791980743,
      "learning_rate": 3.5036687631027254e-05,
      "loss": 0.5651,
      "num_input_tokens_seen": 1754864,
      "step": 2675
    },
    {
      "epoch": 1.4046121593291405,
      "grad_norm": 0.7693271040916443,
      "learning_rate": 3.510220125786164e-05,
      "loss": 0.6728,
      "num_input_tokens_seen": 1757648,
      "step": 2680
    },
    {
      "epoch": 1.4072327044025157,
      "grad_norm": 0.18622463941574097,
      "learning_rate": 3.516771488469602e-05,
      "loss": 0.6543,
      "num_input_tokens_seen": 1760432,
      "step": 2685
    },
    {
      "epoch": 1.409853249475891,
      "grad_norm": 0.11892703175544739,
      "learning_rate": 3.5233228511530397e-05,
      "loss": 0.6244,
      "num_input_tokens_seen": 1763664,
      "step": 2690
    },
    {
      "epoch": 1.4124737945492662,
      "grad_norm": 0.20145182311534882,
      "learning_rate": 3.529874213836478e-05,
      "loss": 0.5141,
      "num_input_tokens_seen": 1766640,
      "step": 2695
    },
    {
      "epoch": 1.4150943396226414,
      "grad_norm": 0.22144775092601776,
      "learning_rate": 3.536425576519916e-05,
      "loss": 0.6268,
      "num_input_tokens_seen": 1769776,
      "step": 2700
    },
    {
      "epoch": 1.4177148846960168,
      "grad_norm": 0.2950643301010132,
      "learning_rate": 3.5429769392033546e-05,
      "loss": 0.672,
      "num_input_tokens_seen": 1772112,
      "step": 2705
    },
    {
      "epoch": 1.420335429769392,
      "grad_norm": 0.19326114654541016,
      "learning_rate": 3.549528301886792e-05,
      "loss": 0.5671,
      "num_input_tokens_seen": 1775312,
      "step": 2710
    },
    {
      "epoch": 1.4229559748427674,
      "grad_norm": 0.17795901000499725,
      "learning_rate": 3.556079664570231e-05,
      "loss": 0.7017,
      "num_input_tokens_seen": 1779504,
      "step": 2715
    },
    {
      "epoch": 1.4255765199161425,
      "grad_norm": 0.19894693791866302,
      "learning_rate": 3.562631027253669e-05,
      "loss": 0.5091,
      "num_input_tokens_seen": 1783248,
      "step": 2720
    },
    {
      "epoch": 1.4281970649895177,
      "grad_norm": 0.3566473722457886,
      "learning_rate": 3.569182389937107e-05,
      "loss": 0.4504,
      "num_input_tokens_seen": 1789168,
      "step": 2725
    },
    {
      "epoch": 1.430817610062893,
      "grad_norm": 0.23489902913570404,
      "learning_rate": 3.5757337526205454e-05,
      "loss": 0.5584,
      "num_input_tokens_seen": 1792592,
      "step": 2730
    },
    {
      "epoch": 1.4334381551362683,
      "grad_norm": 0.38872289657592773,
      "learning_rate": 3.582285115303983e-05,
      "loss": 0.6167,
      "num_input_tokens_seen": 1795536,
      "step": 2735
    },
    {
      "epoch": 1.4360587002096437,
      "grad_norm": 0.32345855236053467,
      "learning_rate": 3.588836477987422e-05,
      "loss": 0.524,
      "num_input_tokens_seen": 1798064,
      "step": 2740
    },
    {
      "epoch": 1.4386792452830188,
      "grad_norm": 0.21224910020828247,
      "learning_rate": 3.59538784067086e-05,
      "loss": 0.6126,
      "num_input_tokens_seen": 1801392,
      "step": 2745
    },
    {
      "epoch": 1.441299790356394,
      "grad_norm": 0.2603282630443573,
      "learning_rate": 3.601939203354298e-05,
      "loss": 0.6355,
      "num_input_tokens_seen": 1804368,
      "step": 2750
    },
    {
      "epoch": 1.4439203354297694,
      "grad_norm": 0.5813952088356018,
      "learning_rate": 3.608490566037736e-05,
      "loss": 0.6855,
      "num_input_tokens_seen": 1807504,
      "step": 2755
    },
    {
      "epoch": 1.4465408805031448,
      "grad_norm": 0.3359467685222626,
      "learning_rate": 3.615041928721174e-05,
      "loss": 0.4559,
      "num_input_tokens_seen": 1810160,
      "step": 2760
    },
    {
      "epoch": 1.44916142557652,
      "grad_norm": 0.263943612575531,
      "learning_rate": 3.621593291404612e-05,
      "loss": 0.6076,
      "num_input_tokens_seen": 1813616,
      "step": 2765
    },
    {
      "epoch": 1.4517819706498951,
      "grad_norm": 0.23925159871578217,
      "learning_rate": 3.6281446540880505e-05,
      "loss": 0.5494,
      "num_input_tokens_seen": 1816752,
      "step": 2770
    },
    {
      "epoch": 1.4544025157232705,
      "grad_norm": 0.24475190043449402,
      "learning_rate": 3.634696016771489e-05,
      "loss": 0.5685,
      "num_input_tokens_seen": 1820784,
      "step": 2775
    },
    {
      "epoch": 1.4570230607966457,
      "grad_norm": 0.1947847455739975,
      "learning_rate": 3.6412473794549265e-05,
      "loss": 0.5421,
      "num_input_tokens_seen": 1823024,
      "step": 2780
    },
    {
      "epoch": 1.459643605870021,
      "grad_norm": 0.2729569673538208,
      "learning_rate": 3.647798742138365e-05,
      "loss": 0.6238,
      "num_input_tokens_seen": 1828016,
      "step": 2785
    },
    {
      "epoch": 1.4622641509433962,
      "grad_norm": 0.1825701892375946,
      "learning_rate": 3.654350104821803e-05,
      "loss": 0.6217,
      "num_input_tokens_seen": 1831280,
      "step": 2790
    },
    {
      "epoch": 1.4648846960167714,
      "grad_norm": 0.16244155168533325,
      "learning_rate": 3.6609014675052414e-05,
      "loss": 0.4974,
      "num_input_tokens_seen": 1834896,
      "step": 2795
    },
    {
      "epoch": 1.4675052410901468,
      "grad_norm": 0.1547158807516098,
      "learning_rate": 3.66745283018868e-05,
      "loss": 0.5889,
      "num_input_tokens_seen": 1837488,
      "step": 2800
    },
    {
      "epoch": 1.470125786163522,
      "grad_norm": 0.18901865184307098,
      "learning_rate": 3.6740041928721173e-05,
      "loss": 0.644,
      "num_input_tokens_seen": 1841584,
      "step": 2805
    },
    {
      "epoch": 1.4727463312368974,
      "grad_norm": 0.21802574396133423,
      "learning_rate": 3.6805555555555556e-05,
      "loss": 0.6655,
      "num_input_tokens_seen": 1844688,
      "step": 2810
    },
    {
      "epoch": 1.4753668763102725,
      "grad_norm": 0.2378256767988205,
      "learning_rate": 3.687106918238994e-05,
      "loss": 0.6015,
      "num_input_tokens_seen": 1847472,
      "step": 2815
    },
    {
      "epoch": 1.4779874213836477,
      "grad_norm": 0.26022443175315857,
      "learning_rate": 3.6936582809224316e-05,
      "loss": 0.8549,
      "num_input_tokens_seen": 1851088,
      "step": 2820
    },
    {
      "epoch": 1.480607966457023,
      "grad_norm": 0.18769079446792603,
      "learning_rate": 3.7002096436058706e-05,
      "loss": 0.6034,
      "num_input_tokens_seen": 1855216,
      "step": 2825
    },
    {
      "epoch": 1.4832285115303983,
      "grad_norm": 0.12982119619846344,
      "learning_rate": 3.706761006289308e-05,
      "loss": 0.5715,
      "num_input_tokens_seen": 1857744,
      "step": 2830
    },
    {
      "epoch": 1.4858490566037736,
      "grad_norm": 0.21862652897834778,
      "learning_rate": 3.7133123689727465e-05,
      "loss": 0.5365,
      "num_input_tokens_seen": 1860400,
      "step": 2835
    },
    {
      "epoch": 1.4884696016771488,
      "grad_norm": 0.2665807902812958,
      "learning_rate": 3.719863731656185e-05,
      "loss": 0.6249,
      "num_input_tokens_seen": 1863248,
      "step": 2840
    },
    {
      "epoch": 1.491090146750524,
      "grad_norm": 0.23453150689601898,
      "learning_rate": 3.7264150943396224e-05,
      "loss": 0.5777,
      "num_input_tokens_seen": 1866576,
      "step": 2845
    },
    {
      "epoch": 1.4937106918238994,
      "grad_norm": 0.2286229431629181,
      "learning_rate": 3.732966457023061e-05,
      "loss": 0.5529,
      "num_input_tokens_seen": 1869840,
      "step": 2850
    },
    {
      "epoch": 1.4963312368972748,
      "grad_norm": 0.16957947611808777,
      "learning_rate": 3.739517819706499e-05,
      "loss": 0.7135,
      "num_input_tokens_seen": 1872624,
      "step": 2855
    },
    {
      "epoch": 1.49895178197065,
      "grad_norm": 0.13918916881084442,
      "learning_rate": 3.7460691823899374e-05,
      "loss": 0.6377,
      "num_input_tokens_seen": 1876016,
      "step": 2860
    },
    {
      "epoch": 1.501572327044025,
      "grad_norm": 0.19929097592830658,
      "learning_rate": 3.752620545073376e-05,
      "loss": 0.4501,
      "num_input_tokens_seen": 1880976,
      "step": 2865
    },
    {
      "epoch": 1.5041928721174003,
      "grad_norm": 0.36921226978302,
      "learning_rate": 3.759171907756814e-05,
      "loss": 0.688,
      "num_input_tokens_seen": 1883568,
      "step": 2870
    },
    {
      "epoch": 1.5068134171907757,
      "grad_norm": 0.261046439409256,
      "learning_rate": 3.7657232704402516e-05,
      "loss": 0.566,
      "num_input_tokens_seen": 1886800,
      "step": 2875
    },
    {
      "epoch": 1.509433962264151,
      "grad_norm": 0.2747538387775421,
      "learning_rate": 3.77227463312369e-05,
      "loss": 0.5761,
      "num_input_tokens_seen": 1890672,
      "step": 2880
    },
    {
      "epoch": 1.5120545073375262,
      "grad_norm": 0.6136439442634583,
      "learning_rate": 3.778825995807128e-05,
      "loss": 0.441,
      "num_input_tokens_seen": 1893456,
      "step": 2885
    },
    {
      "epoch": 1.5146750524109014,
      "grad_norm": 0.2735607326030731,
      "learning_rate": 3.785377358490566e-05,
      "loss": 0.604,
      "num_input_tokens_seen": 1896752,
      "step": 2890
    },
    {
      "epoch": 1.5172955974842768,
      "grad_norm": 0.16903607547283173,
      "learning_rate": 3.791928721174005e-05,
      "loss": 0.5088,
      "num_input_tokens_seen": 1899664,
      "step": 2895
    },
    {
      "epoch": 1.519916142557652,
      "grad_norm": 0.1658587008714676,
      "learning_rate": 3.7984800838574425e-05,
      "loss": 0.4797,
      "num_input_tokens_seen": 1902320,
      "step": 2900
    },
    {
      "epoch": 1.5225366876310273,
      "grad_norm": 0.1681608408689499,
      "learning_rate": 3.805031446540881e-05,
      "loss": 0.6172,
      "num_input_tokens_seen": 1905488,
      "step": 2905
    },
    {
      "epoch": 1.5251572327044025,
      "grad_norm": 0.184820294380188,
      "learning_rate": 3.811582809224319e-05,
      "loss": 0.5797,
      "num_input_tokens_seen": 1908144,
      "step": 2910
    },
    {
      "epoch": 1.5277777777777777,
      "grad_norm": 0.17820237576961517,
      "learning_rate": 3.818134171907757e-05,
      "loss": 0.603,
      "num_input_tokens_seen": 1911536,
      "step": 2915
    },
    {
      "epoch": 1.530398322851153,
      "grad_norm": 0.13188765943050385,
      "learning_rate": 3.824685534591195e-05,
      "loss": 0.5608,
      "num_input_tokens_seen": 1916176,
      "step": 2920
    },
    {
      "epoch": 1.5330188679245285,
      "grad_norm": 0.20786520838737488,
      "learning_rate": 3.831236897274633e-05,
      "loss": 0.5675,
      "num_input_tokens_seen": 1920592,
      "step": 2925
    },
    {
      "epoch": 1.5356394129979036,
      "grad_norm": 0.22530382871627808,
      "learning_rate": 3.8377882599580716e-05,
      "loss": 0.5406,
      "num_input_tokens_seen": 1923152,
      "step": 2930
    },
    {
      "epoch": 1.5382599580712788,
      "grad_norm": 0.27108314633369446,
      "learning_rate": 3.844339622641509e-05,
      "loss": 0.6268,
      "num_input_tokens_seen": 1926000,
      "step": 2935
    },
    {
      "epoch": 1.540880503144654,
      "grad_norm": 0.23771947622299194,
      "learning_rate": 3.8508909853249476e-05,
      "loss": 0.7277,
      "num_input_tokens_seen": 1929072,
      "step": 2940
    },
    {
      "epoch": 1.5435010482180294,
      "grad_norm": 0.16058854758739471,
      "learning_rate": 3.857442348008386e-05,
      "loss": 0.4417,
      "num_input_tokens_seen": 1932208,
      "step": 2945
    },
    {
      "epoch": 1.5461215932914047,
      "grad_norm": 0.20007535815238953,
      "learning_rate": 3.863993710691824e-05,
      "loss": 0.739,
      "num_input_tokens_seen": 1935152,
      "step": 2950
    },
    {
      "epoch": 1.54874213836478,
      "grad_norm": 0.2343435436487198,
      "learning_rate": 3.8705450733752625e-05,
      "loss": 0.6178,
      "num_input_tokens_seen": 1938704,
      "step": 2955
    },
    {
      "epoch": 1.551362683438155,
      "grad_norm": 0.1441684365272522,
      "learning_rate": 3.8770964360587e-05,
      "loss": 0.6807,
      "num_input_tokens_seen": 1942192,
      "step": 2960
    },
    {
      "epoch": 1.5539832285115303,
      "grad_norm": 0.492066890001297,
      "learning_rate": 3.883647798742139e-05,
      "loss": 0.5514,
      "num_input_tokens_seen": 1945744,
      "step": 2965
    },
    {
      "epoch": 1.5566037735849056,
      "grad_norm": 0.21294142305850983,
      "learning_rate": 3.890199161425577e-05,
      "loss": 0.5736,
      "num_input_tokens_seen": 1948944,
      "step": 2970
    },
    {
      "epoch": 1.559224318658281,
      "grad_norm": 0.2836361825466156,
      "learning_rate": 3.8967505241090144e-05,
      "loss": 0.5329,
      "num_input_tokens_seen": 1951792,
      "step": 2975
    },
    {
      "epoch": 1.5618448637316562,
      "grad_norm": 0.2331097424030304,
      "learning_rate": 3.9033018867924534e-05,
      "loss": 0.6358,
      "num_input_tokens_seen": 1955376,
      "step": 2980
    },
    {
      "epoch": 1.5644654088050314,
      "grad_norm": 0.16012650728225708,
      "learning_rate": 3.909853249475891e-05,
      "loss": 0.5935,
      "num_input_tokens_seen": 1959120,
      "step": 2985
    },
    {
      "epoch": 1.5670859538784065,
      "grad_norm": 0.11899362504482269,
      "learning_rate": 3.916404612159329e-05,
      "loss": 0.4523,
      "num_input_tokens_seen": 1964304,
      "step": 2990
    },
    {
      "epoch": 1.569706498951782,
      "grad_norm": 0.3201289474964142,
      "learning_rate": 3.9229559748427676e-05,
      "loss": 0.7536,
      "num_input_tokens_seen": 1966832,
      "step": 2995
    },
    {
      "epoch": 1.5723270440251573,
      "grad_norm": 0.188313826918602,
      "learning_rate": 3.929507337526205e-05,
      "loss": 0.7106,
      "num_input_tokens_seen": 1970224,
      "step": 3000
    },
    {
      "epoch": 1.5749475890985325,
      "grad_norm": 0.14889413118362427,
      "learning_rate": 3.9360587002096435e-05,
      "loss": 0.5985,
      "num_input_tokens_seen": 1974288,
      "step": 3005
    },
    {
      "epoch": 1.5775681341719077,
      "grad_norm": 0.16683706641197205,
      "learning_rate": 3.942610062893082e-05,
      "loss": 0.5098,
      "num_input_tokens_seen": 1978064,
      "step": 3010
    },
    {
      "epoch": 1.580188679245283,
      "grad_norm": 0.2748569846153259,
      "learning_rate": 3.94916142557652e-05,
      "loss": 0.6285,
      "num_input_tokens_seen": 1981744,
      "step": 3015
    },
    {
      "epoch": 1.5828092243186582,
      "grad_norm": 0.408591628074646,
      "learning_rate": 3.9557127882599585e-05,
      "loss": 0.6365,
      "num_input_tokens_seen": 1984016,
      "step": 3020
    },
    {
      "epoch": 1.5854297693920336,
      "grad_norm": 0.21540632843971252,
      "learning_rate": 3.962264150943397e-05,
      "loss": 0.4541,
      "num_input_tokens_seen": 1987056,
      "step": 3025
    },
    {
      "epoch": 1.5880503144654088,
      "grad_norm": 0.17844150960445404,
      "learning_rate": 3.9688155136268344e-05,
      "loss": 0.6027,
      "num_input_tokens_seen": 1992048,
      "step": 3030
    },
    {
      "epoch": 1.590670859538784,
      "grad_norm": 0.1881033480167389,
      "learning_rate": 3.975366876310273e-05,
      "loss": 0.7535,
      "num_input_tokens_seen": 1995472,
      "step": 3035
    },
    {
      "epoch": 1.5932914046121593,
      "grad_norm": 0.12984390556812286,
      "learning_rate": 3.981918238993711e-05,
      "loss": 0.4437,
      "num_input_tokens_seen": 1999152,
      "step": 3040
    },
    {
      "epoch": 1.5959119496855347,
      "grad_norm": 0.20338360965251923,
      "learning_rate": 3.9884696016771486e-05,
      "loss": 0.6668,
      "num_input_tokens_seen": 2001744,
      "step": 3045
    },
    {
      "epoch": 1.59853249475891,
      "grad_norm": 0.1912604719400406,
      "learning_rate": 3.9950209643605876e-05,
      "loss": 0.689,
      "num_input_tokens_seen": 2004848,
      "step": 3050
    },
    {
      "epoch": 1.601153039832285,
      "grad_norm": 0.4567301273345947,
      "learning_rate": 4.001572327044025e-05,
      "loss": 0.6354,
      "num_input_tokens_seen": 2007920,
      "step": 3055
    },
    {
      "epoch": 1.6037735849056602,
      "grad_norm": 0.23516181111335754,
      "learning_rate": 4.0081236897274636e-05,
      "loss": 0.5704,
      "num_input_tokens_seen": 2011152,
      "step": 3060
    },
    {
      "epoch": 1.6063941299790356,
      "grad_norm": 0.26267683506011963,
      "learning_rate": 4.014675052410902e-05,
      "loss": 0.7285,
      "num_input_tokens_seen": 2013968,
      "step": 3065
    },
    {
      "epoch": 1.609014675052411,
      "grad_norm": 0.25067979097366333,
      "learning_rate": 4.0212264150943395e-05,
      "loss": 0.596,
      "num_input_tokens_seen": 2017936,
      "step": 3070
    },
    {
      "epoch": 1.6116352201257862,
      "grad_norm": 0.12045088410377502,
      "learning_rate": 4.027777777777778e-05,
      "loss": 0.5981,
      "num_input_tokens_seen": 2020528,
      "step": 3075
    },
    {
      "epoch": 1.6142557651991614,
      "grad_norm": 0.36347243189811707,
      "learning_rate": 4.034329140461216e-05,
      "loss": 0.7432,
      "num_input_tokens_seen": 2023792,
      "step": 3080
    },
    {
      "epoch": 1.6168763102725365,
      "grad_norm": 0.18610987067222595,
      "learning_rate": 4.0408805031446544e-05,
      "loss": 0.6893,
      "num_input_tokens_seen": 2027152,
      "step": 3085
    },
    {
      "epoch": 1.619496855345912,
      "grad_norm": 0.2010204792022705,
      "learning_rate": 4.047431865828093e-05,
      "loss": 0.5339,
      "num_input_tokens_seen": 2029968,
      "step": 3090
    },
    {
      "epoch": 1.6221174004192873,
      "grad_norm": 0.24311195313930511,
      "learning_rate": 4.0539832285115304e-05,
      "loss": 0.5757,
      "num_input_tokens_seen": 2033488,
      "step": 3095
    },
    {
      "epoch": 1.6247379454926625,
      "grad_norm": 0.4318805932998657,
      "learning_rate": 4.0605345911949687e-05,
      "loss": 0.7001,
      "num_input_tokens_seen": 2037136,
      "step": 3100
    },
    {
      "epoch": 1.6273584905660377,
      "grad_norm": 0.20165060460567474,
      "learning_rate": 4.067085953878407e-05,
      "loss": 0.5923,
      "num_input_tokens_seen": 2039888,
      "step": 3105
    },
    {
      "epoch": 1.629979035639413,
      "grad_norm": 0.17864173650741577,
      "learning_rate": 4.073637316561845e-05,
      "loss": 0.7569,
      "num_input_tokens_seen": 2043024,
      "step": 3110
    },
    {
      "epoch": 1.6325995807127882,
      "grad_norm": 0.22970709204673767,
      "learning_rate": 4.080188679245283e-05,
      "loss": 0.5038,
      "num_input_tokens_seen": 2047056,
      "step": 3115
    },
    {
      "epoch": 1.6352201257861636,
      "grad_norm": 0.4349408447742462,
      "learning_rate": 4.086740041928722e-05,
      "loss": 0.6366,
      "num_input_tokens_seen": 2050000,
      "step": 3120
    },
    {
      "epoch": 1.6378406708595388,
      "grad_norm": 0.17803725600242615,
      "learning_rate": 4.0932914046121595e-05,
      "loss": 0.5567,
      "num_input_tokens_seen": 2054032,
      "step": 3125
    },
    {
      "epoch": 1.640461215932914,
      "grad_norm": 0.18202456831932068,
      "learning_rate": 4.099842767295597e-05,
      "loss": 0.5559,
      "num_input_tokens_seen": 2057072,
      "step": 3130
    },
    {
      "epoch": 1.6430817610062893,
      "grad_norm": 0.14211434125900269,
      "learning_rate": 4.106394129979036e-05,
      "loss": 0.5225,
      "num_input_tokens_seen": 2060656,
      "step": 3135
    },
    {
      "epoch": 1.6457023060796647,
      "grad_norm": 0.41483303904533386,
      "learning_rate": 4.112945492662474e-05,
      "loss": 0.6672,
      "num_input_tokens_seen": 2063568,
      "step": 3140
    },
    {
      "epoch": 1.64832285115304,
      "grad_norm": 0.3575991988182068,
      "learning_rate": 4.119496855345912e-05,
      "loss": 0.4878,
      "num_input_tokens_seen": 2066960,
      "step": 3145
    },
    {
      "epoch": 1.650943396226415,
      "grad_norm": 0.1627785712480545,
      "learning_rate": 4.1260482180293504e-05,
      "loss": 0.5818,
      "num_input_tokens_seen": 2070096,
      "step": 3150
    },
    {
      "epoch": 1.6535639412997902,
      "grad_norm": 0.16916991770267487,
      "learning_rate": 4.132599580712788e-05,
      "loss": 0.5733,
      "num_input_tokens_seen": 2073072,
      "step": 3155
    },
    {
      "epoch": 1.6561844863731656,
      "grad_norm": 0.16684071719646454,
      "learning_rate": 4.139150943396227e-05,
      "loss": 0.5184,
      "num_input_tokens_seen": 2075632,
      "step": 3160
    },
    {
      "epoch": 1.658805031446541,
      "grad_norm": 0.5103742480278015,
      "learning_rate": 4.1457023060796646e-05,
      "loss": 0.58,
      "num_input_tokens_seen": 2079024,
      "step": 3165
    },
    {
      "epoch": 1.6614255765199162,
      "grad_norm": 0.3390767276287079,
      "learning_rate": 4.152253668763103e-05,
      "loss": 0.7402,
      "num_input_tokens_seen": 2081744,
      "step": 3170
    },
    {
      "epoch": 1.6640461215932913,
      "grad_norm": 0.17809677124023438,
      "learning_rate": 4.158805031446541e-05,
      "loss": 0.6538,
      "num_input_tokens_seen": 2084784,
      "step": 3175
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 0.16959461569786072,
      "learning_rate": 4.1653563941299795e-05,
      "loss": 0.497,
      "num_input_tokens_seen": 2087984,
      "step": 3180
    },
    {
      "epoch": 1.669287211740042,
      "grad_norm": 0.14567600190639496,
      "learning_rate": 4.171907756813417e-05,
      "loss": 0.5694,
      "num_input_tokens_seen": 2091344,
      "step": 3185
    },
    {
      "epoch": 1.6719077568134173,
      "grad_norm": 0.1193731278181076,
      "learning_rate": 4.1784591194968555e-05,
      "loss": 0.4918,
      "num_input_tokens_seen": 2095312,
      "step": 3190
    },
    {
      "epoch": 1.6745283018867925,
      "grad_norm": 0.1310415416955948,
      "learning_rate": 4.185010482180294e-05,
      "loss": 0.5193,
      "num_input_tokens_seen": 2098832,
      "step": 3195
    },
    {
      "epoch": 1.6771488469601676,
      "grad_norm": 0.24627353250980377,
      "learning_rate": 4.1915618448637314e-05,
      "loss": 0.7719,
      "num_input_tokens_seen": 2101168,
      "step": 3200
    },
    {
      "epoch": 1.679769392033543,
      "grad_norm": 0.27190983295440674,
      "learning_rate": 4.1981132075471704e-05,
      "loss": 0.8175,
      "num_input_tokens_seen": 2104816,
      "step": 3205
    },
    {
      "epoch": 1.6823899371069182,
      "grad_norm": 0.40247493982315063,
      "learning_rate": 4.204664570230608e-05,
      "loss": 0.5702,
      "num_input_tokens_seen": 2107728,
      "step": 3210
    },
    {
      "epoch": 1.6850104821802936,
      "grad_norm": 0.4733574390411377,
      "learning_rate": 4.2112159329140463e-05,
      "loss": 0.8336,
      "num_input_tokens_seen": 2110992,
      "step": 3215
    },
    {
      "epoch": 1.6876310272536688,
      "grad_norm": 0.31548887491226196,
      "learning_rate": 4.2177672955974846e-05,
      "loss": 0.5681,
      "num_input_tokens_seen": 2113840,
      "step": 3220
    },
    {
      "epoch": 1.690251572327044,
      "grad_norm": 0.25386300683021545,
      "learning_rate": 4.224318658280922e-05,
      "loss": 0.5757,
      "num_input_tokens_seen": 2116784,
      "step": 3225
    },
    {
      "epoch": 1.6928721174004193,
      "grad_norm": 0.15457089245319366,
      "learning_rate": 4.2308700209643606e-05,
      "loss": 0.7,
      "num_input_tokens_seen": 2119696,
      "step": 3230
    },
    {
      "epoch": 1.6954926624737947,
      "grad_norm": 0.22581031918525696,
      "learning_rate": 4.237421383647799e-05,
      "loss": 0.6299,
      "num_input_tokens_seen": 2122608,
      "step": 3235
    },
    {
      "epoch": 1.6981132075471699,
      "grad_norm": 0.19909602403640747,
      "learning_rate": 4.243972746331237e-05,
      "loss": 0.5911,
      "num_input_tokens_seen": 2125808,
      "step": 3240
    },
    {
      "epoch": 1.700733752620545,
      "grad_norm": 0.2086244821548462,
      "learning_rate": 4.2505241090146755e-05,
      "loss": 0.5149,
      "num_input_tokens_seen": 2129584,
      "step": 3245
    },
    {
      "epoch": 1.7033542976939202,
      "grad_norm": 0.32144591212272644,
      "learning_rate": 4.257075471698113e-05,
      "loss": 0.575,
      "num_input_tokens_seen": 2132208,
      "step": 3250
    },
    {
      "epoch": 1.7059748427672956,
      "grad_norm": 0.1669386923313141,
      "learning_rate": 4.2636268343815514e-05,
      "loss": 0.5621,
      "num_input_tokens_seen": 2135632,
      "step": 3255
    },
    {
      "epoch": 1.708595387840671,
      "grad_norm": 0.27797991037368774,
      "learning_rate": 4.27017819706499e-05,
      "loss": 0.7834,
      "num_input_tokens_seen": 2139760,
      "step": 3260
    },
    {
      "epoch": 1.7112159329140462,
      "grad_norm": 0.22698073089122772,
      "learning_rate": 4.276729559748428e-05,
      "loss": 0.5776,
      "num_input_tokens_seen": 2143280,
      "step": 3265
    },
    {
      "epoch": 1.7138364779874213,
      "grad_norm": 0.20458531379699707,
      "learning_rate": 4.283280922431866e-05,
      "loss": 0.6005,
      "num_input_tokens_seen": 2145648,
      "step": 3270
    },
    {
      "epoch": 1.7164570230607965,
      "grad_norm": 0.2139437347650528,
      "learning_rate": 4.289832285115305e-05,
      "loss": 0.5678,
      "num_input_tokens_seen": 2149040,
      "step": 3275
    },
    {
      "epoch": 1.719077568134172,
      "grad_norm": 0.27923357486724854,
      "learning_rate": 4.296383647798742e-05,
      "loss": 0.6714,
      "num_input_tokens_seen": 2151792,
      "step": 3280
    },
    {
      "epoch": 1.7216981132075473,
      "grad_norm": 0.2655605673789978,
      "learning_rate": 4.3029350104821806e-05,
      "loss": 0.5398,
      "num_input_tokens_seen": 2155600,
      "step": 3285
    },
    {
      "epoch": 1.7243186582809225,
      "grad_norm": 0.1590365320444107,
      "learning_rate": 4.309486373165619e-05,
      "loss": 0.4817,
      "num_input_tokens_seen": 2159088,
      "step": 3290
    },
    {
      "epoch": 1.7269392033542976,
      "grad_norm": 0.2732340097427368,
      "learning_rate": 4.3160377358490565e-05,
      "loss": 0.5179,
      "num_input_tokens_seen": 2162800,
      "step": 3295
    },
    {
      "epoch": 1.7295597484276728,
      "grad_norm": 0.21749530732631683,
      "learning_rate": 4.322589098532495e-05,
      "loss": 0.6215,
      "num_input_tokens_seen": 2165872,
      "step": 3300
    },
    {
      "epoch": 1.7321802935010482,
      "grad_norm": 0.18265195190906525,
      "learning_rate": 4.329140461215933e-05,
      "loss": 0.6771,
      "num_input_tokens_seen": 2169456,
      "step": 3305
    },
    {
      "epoch": 1.7348008385744236,
      "grad_norm": 0.201838880777359,
      "learning_rate": 4.335691823899371e-05,
      "loss": 0.5647,
      "num_input_tokens_seen": 2172368,
      "step": 3310
    },
    {
      "epoch": 1.7374213836477987,
      "grad_norm": 0.16238678991794586,
      "learning_rate": 4.34224318658281e-05,
      "loss": 0.6851,
      "num_input_tokens_seen": 2175504,
      "step": 3315
    },
    {
      "epoch": 1.740041928721174,
      "grad_norm": 0.2830530107021332,
      "learning_rate": 4.3487945492662474e-05,
      "loss": 0.5725,
      "num_input_tokens_seen": 2179440,
      "step": 3320
    },
    {
      "epoch": 1.7426624737945493,
      "grad_norm": 0.3266308903694153,
      "learning_rate": 4.355345911949686e-05,
      "loss": 0.7278,
      "num_input_tokens_seen": 2181904,
      "step": 3325
    },
    {
      "epoch": 1.7452830188679245,
      "grad_norm": 0.1789105087518692,
      "learning_rate": 4.361897274633124e-05,
      "loss": 0.4841,
      "num_input_tokens_seen": 2184816,
      "step": 3330
    },
    {
      "epoch": 1.7479035639412999,
      "grad_norm": 0.32555249333381653,
      "learning_rate": 4.368448637316562e-05,
      "loss": 0.7077,
      "num_input_tokens_seen": 2187920,
      "step": 3335
    },
    {
      "epoch": 1.750524109014675,
      "grad_norm": 0.10908674448728561,
      "learning_rate": 4.375e-05,
      "loss": 0.6246,
      "num_input_tokens_seen": 2191216,
      "step": 3340
    },
    {
      "epoch": 1.7531446540880502,
      "grad_norm": 0.23010259866714478,
      "learning_rate": 4.381551362683438e-05,
      "loss": 0.6025,
      "num_input_tokens_seen": 2193744,
      "step": 3345
    },
    {
      "epoch": 1.7557651991614256,
      "grad_norm": 0.23882876336574554,
      "learning_rate": 4.3881027253668766e-05,
      "loss": 0.5405,
      "num_input_tokens_seen": 2196368,
      "step": 3350
    },
    {
      "epoch": 1.758385744234801,
      "grad_norm": 0.1655389666557312,
      "learning_rate": 4.394654088050314e-05,
      "loss": 0.4281,
      "num_input_tokens_seen": 2199632,
      "step": 3355
    },
    {
      "epoch": 1.7610062893081762,
      "grad_norm": 0.1537657380104065,
      "learning_rate": 4.401205450733753e-05,
      "loss": 0.57,
      "num_input_tokens_seen": 2202704,
      "step": 3360
    },
    {
      "epoch": 1.7636268343815513,
      "grad_norm": 0.2357429414987564,
      "learning_rate": 4.407756813417191e-05,
      "loss": 0.795,
      "num_input_tokens_seen": 2205616,
      "step": 3365
    },
    {
      "epoch": 1.7662473794549265,
      "grad_norm": 0.1496199667453766,
      "learning_rate": 4.414308176100629e-05,
      "loss": 0.531,
      "num_input_tokens_seen": 2209008,
      "step": 3370
    },
    {
      "epoch": 1.7688679245283019,
      "grad_norm": 0.16405031085014343,
      "learning_rate": 4.4208595387840674e-05,
      "loss": 0.5152,
      "num_input_tokens_seen": 2211824,
      "step": 3375
    },
    {
      "epoch": 1.7714884696016773,
      "grad_norm": 0.10641821473836899,
      "learning_rate": 4.427410901467505e-05,
      "loss": 0.5459,
      "num_input_tokens_seen": 2215632,
      "step": 3380
    },
    {
      "epoch": 1.7741090146750524,
      "grad_norm": 0.28993645310401917,
      "learning_rate": 4.433962264150944e-05,
      "loss": 0.6285,
      "num_input_tokens_seen": 2218512,
      "step": 3385
    },
    {
      "epoch": 1.7767295597484276,
      "grad_norm": 0.19157549738883972,
      "learning_rate": 4.440513626834382e-05,
      "loss": 0.4789,
      "num_input_tokens_seen": 2221776,
      "step": 3390
    },
    {
      "epoch": 1.7793501048218028,
      "grad_norm": 0.11153203248977661,
      "learning_rate": 4.44706498951782e-05,
      "loss": 0.5031,
      "num_input_tokens_seen": 2224784,
      "step": 3395
    },
    {
      "epoch": 1.7819706498951782,
      "grad_norm": 0.3232632875442505,
      "learning_rate": 4.453616352201258e-05,
      "loss": 0.5952,
      "num_input_tokens_seen": 2227408,
      "step": 3400
    },
    {
      "epoch": 1.7845911949685536,
      "grad_norm": 0.25448814034461975,
      "learning_rate": 4.460167714884696e-05,
      "loss": 0.5571,
      "num_input_tokens_seen": 2230064,
      "step": 3405
    },
    {
      "epoch": 1.7872117400419287,
      "grad_norm": 0.19861839711666107,
      "learning_rate": 4.466719077568134e-05,
      "loss": 0.4908,
      "num_input_tokens_seen": 2234288,
      "step": 3410
    },
    {
      "epoch": 1.789832285115304,
      "grad_norm": 0.1453595757484436,
      "learning_rate": 4.4732704402515725e-05,
      "loss": 0.483,
      "num_input_tokens_seen": 2237072,
      "step": 3415
    },
    {
      "epoch": 1.7924528301886793,
      "grad_norm": 0.12983712553977966,
      "learning_rate": 4.479821802935011e-05,
      "loss": 0.4379,
      "num_input_tokens_seen": 2240016,
      "step": 3420
    },
    {
      "epoch": 1.7950733752620545,
      "grad_norm": 0.20206144452095032,
      "learning_rate": 4.4863731656184485e-05,
      "loss": 0.5149,
      "num_input_tokens_seen": 2242800,
      "step": 3425
    },
    {
      "epoch": 1.7976939203354299,
      "grad_norm": 0.33537790179252625,
      "learning_rate": 4.4929245283018875e-05,
      "loss": 0.6671,
      "num_input_tokens_seen": 2245680,
      "step": 3430
    },
    {
      "epoch": 1.800314465408805,
      "grad_norm": 0.2591547667980194,
      "learning_rate": 4.499475890985325e-05,
      "loss": 0.613,
      "num_input_tokens_seen": 2248272,
      "step": 3435
    },
    {
      "epoch": 1.8029350104821802,
      "grad_norm": 0.2673072814941406,
      "learning_rate": 4.5060272536687634e-05,
      "loss": 0.4382,
      "num_input_tokens_seen": 2251376,
      "step": 3440
    },
    {
      "epoch": 1.8055555555555556,
      "grad_norm": 0.34239137172698975,
      "learning_rate": 4.512578616352202e-05,
      "loss": 0.6601,
      "num_input_tokens_seen": 2254256,
      "step": 3445
    },
    {
      "epoch": 1.808176100628931,
      "grad_norm": 0.2575056850910187,
      "learning_rate": 4.519129979035639e-05,
      "loss": 0.6448,
      "num_input_tokens_seen": 2257104,
      "step": 3450
    },
    {
      "epoch": 1.8107966457023061,
      "grad_norm": 0.23772484064102173,
      "learning_rate": 4.525681341719078e-05,
      "loss": 0.6811,
      "num_input_tokens_seen": 2259920,
      "step": 3455
    },
    {
      "epoch": 1.8134171907756813,
      "grad_norm": 0.14128726720809937,
      "learning_rate": 4.532232704402516e-05,
      "loss": 0.4901,
      "num_input_tokens_seen": 2262736,
      "step": 3460
    },
    {
      "epoch": 1.8160377358490565,
      "grad_norm": 0.08639707416296005,
      "learning_rate": 4.5387840670859536e-05,
      "loss": 0.5228,
      "num_input_tokens_seen": 2266864,
      "step": 3465
    },
    {
      "epoch": 1.8186582809224319,
      "grad_norm": 0.34170591831207275,
      "learning_rate": 4.5453354297693926e-05,
      "loss": 0.7867,
      "num_input_tokens_seen": 2269584,
      "step": 3470
    },
    {
      "epoch": 1.8212788259958073,
      "grad_norm": 0.12828446924686432,
      "learning_rate": 4.55188679245283e-05,
      "loss": 0.5376,
      "num_input_tokens_seen": 2272720,
      "step": 3475
    },
    {
      "epoch": 1.8238993710691824,
      "grad_norm": 0.15150244534015656,
      "learning_rate": 4.5584381551362685e-05,
      "loss": 0.6405,
      "num_input_tokens_seen": 2276880,
      "step": 3480
    },
    {
      "epoch": 1.8265199161425576,
      "grad_norm": 0.6726819276809692,
      "learning_rate": 4.564989517819707e-05,
      "loss": 0.4924,
      "num_input_tokens_seen": 2280016,
      "step": 3485
    },
    {
      "epoch": 1.8291404612159328,
      "grad_norm": 0.1879269927740097,
      "learning_rate": 4.571540880503145e-05,
      "loss": 0.5291,
      "num_input_tokens_seen": 2282640,
      "step": 3490
    },
    {
      "epoch": 1.8317610062893082,
      "grad_norm": 0.1697889268398285,
      "learning_rate": 4.578092243186583e-05,
      "loss": 0.5255,
      "num_input_tokens_seen": 2286032,
      "step": 3495
    },
    {
      "epoch": 1.8343815513626835,
      "grad_norm": 0.2082970142364502,
      "learning_rate": 4.584643605870021e-05,
      "loss": 0.6537,
      "num_input_tokens_seen": 2289136,
      "step": 3500
    },
    {
      "epoch": 1.8370020964360587,
      "grad_norm": 0.20397113263607025,
      "learning_rate": 4.5911949685534594e-05,
      "loss": 0.4507,
      "num_input_tokens_seen": 2292944,
      "step": 3505
    },
    {
      "epoch": 1.8396226415094339,
      "grad_norm": 0.3750011622905731,
      "learning_rate": 4.5977463312368977e-05,
      "loss": 0.5472,
      "num_input_tokens_seen": 2296240,
      "step": 3510
    },
    {
      "epoch": 1.8422431865828093,
      "grad_norm": 0.5603168606758118,
      "learning_rate": 4.604297693920336e-05,
      "loss": 0.5756,
      "num_input_tokens_seen": 2299088,
      "step": 3515
    },
    {
      "epoch": 1.8448637316561844,
      "grad_norm": 0.29232779145240784,
      "learning_rate": 4.6108490566037736e-05,
      "loss": 0.5671,
      "num_input_tokens_seen": 2302736,
      "step": 3520
    },
    {
      "epoch": 1.8474842767295598,
      "grad_norm": 0.2196652740240097,
      "learning_rate": 4.617400419287212e-05,
      "loss": 0.6328,
      "num_input_tokens_seen": 2306640,
      "step": 3525
    },
    {
      "epoch": 1.850104821802935,
      "grad_norm": 0.21781904995441437,
      "learning_rate": 4.62395178197065e-05,
      "loss": 0.6411,
      "num_input_tokens_seen": 2309840,
      "step": 3530
    },
    {
      "epoch": 1.8527253668763102,
      "grad_norm": 0.2872043550014496,
      "learning_rate": 4.630503144654088e-05,
      "loss": 0.6172,
      "num_input_tokens_seen": 2312208,
      "step": 3535
    },
    {
      "epoch": 1.8553459119496856,
      "grad_norm": 0.6338785290718079,
      "learning_rate": 4.637054507337527e-05,
      "loss": 0.791,
      "num_input_tokens_seen": 2315088,
      "step": 3540
    },
    {
      "epoch": 1.857966457023061,
      "grad_norm": 0.3036263585090637,
      "learning_rate": 4.6436058700209645e-05,
      "loss": 0.6079,
      "num_input_tokens_seen": 2318224,
      "step": 3545
    },
    {
      "epoch": 1.8605870020964361,
      "grad_norm": 0.10705530643463135,
      "learning_rate": 4.650157232704403e-05,
      "loss": 0.4668,
      "num_input_tokens_seen": 2322800,
      "step": 3550
    },
    {
      "epoch": 1.8632075471698113,
      "grad_norm": 0.28191784024238586,
      "learning_rate": 4.656708595387841e-05,
      "loss": 0.3868,
      "num_input_tokens_seen": 2329008,
      "step": 3555
    },
    {
      "epoch": 1.8658280922431865,
      "grad_norm": 0.35992610454559326,
      "learning_rate": 4.663259958071279e-05,
      "loss": 0.7667,
      "num_input_tokens_seen": 2331344,
      "step": 3560
    },
    {
      "epoch": 1.8684486373165619,
      "grad_norm": 0.16643838584423065,
      "learning_rate": 4.669811320754717e-05,
      "loss": 0.5199,
      "num_input_tokens_seen": 2334352,
      "step": 3565
    },
    {
      "epoch": 1.8710691823899372,
      "grad_norm": 0.14197629690170288,
      "learning_rate": 4.676362683438155e-05,
      "loss": 0.6394,
      "num_input_tokens_seen": 2338512,
      "step": 3570
    },
    {
      "epoch": 1.8736897274633124,
      "grad_norm": 0.20250023901462555,
      "learning_rate": 4.6829140461215936e-05,
      "loss": 0.6553,
      "num_input_tokens_seen": 2342224,
      "step": 3575
    },
    {
      "epoch": 1.8763102725366876,
      "grad_norm": 0.47178545594215393,
      "learning_rate": 4.689465408805032e-05,
      "loss": 0.55,
      "num_input_tokens_seen": 2345808,
      "step": 3580
    },
    {
      "epoch": 1.8789308176100628,
      "grad_norm": 0.12432249635457993,
      "learning_rate": 4.69601677148847e-05,
      "loss": 0.6312,
      "num_input_tokens_seen": 2348560,
      "step": 3585
    },
    {
      "epoch": 1.8815513626834381,
      "grad_norm": 0.22249062359333038,
      "learning_rate": 4.702568134171908e-05,
      "loss": 0.4507,
      "num_input_tokens_seen": 2351824,
      "step": 3590
    },
    {
      "epoch": 1.8841719077568135,
      "grad_norm": 0.574161171913147,
      "learning_rate": 4.709119496855346e-05,
      "loss": 0.441,
      "num_input_tokens_seen": 2355696,
      "step": 3595
    },
    {
      "epoch": 1.8867924528301887,
      "grad_norm": 0.16986577212810516,
      "learning_rate": 4.7156708595387845e-05,
      "loss": 0.4991,
      "num_input_tokens_seen": 2358352,
      "step": 3600
    },
    {
      "epoch": 1.8894129979035639,
      "grad_norm": 0.13100649416446686,
      "learning_rate": 4.722222222222222e-05,
      "loss": 0.4832,
      "num_input_tokens_seen": 2361776,
      "step": 3605
    },
    {
      "epoch": 1.892033542976939,
      "grad_norm": 0.2662994861602783,
      "learning_rate": 4.728773584905661e-05,
      "loss": 0.6217,
      "num_input_tokens_seen": 2365264,
      "step": 3610
    },
    {
      "epoch": 1.8946540880503144,
      "grad_norm": 0.2521533966064453,
      "learning_rate": 4.735324947589099e-05,
      "loss": 0.8296,
      "num_input_tokens_seen": 2368080,
      "step": 3615
    },
    {
      "epoch": 1.8972746331236898,
      "grad_norm": 0.164041206240654,
      "learning_rate": 4.7418763102725364e-05,
      "loss": 0.5875,
      "num_input_tokens_seen": 2371696,
      "step": 3620
    },
    {
      "epoch": 1.899895178197065,
      "grad_norm": 0.21007420122623444,
      "learning_rate": 4.7484276729559753e-05,
      "loss": 0.593,
      "num_input_tokens_seen": 2374160,
      "step": 3625
    },
    {
      "epoch": 1.9025157232704402,
      "grad_norm": 0.19312812387943268,
      "learning_rate": 4.754979035639413e-05,
      "loss": 0.4951,
      "num_input_tokens_seen": 2376656,
      "step": 3630
    },
    {
      "epoch": 1.9051362683438156,
      "grad_norm": 0.20282921195030212,
      "learning_rate": 4.761530398322851e-05,
      "loss": 0.5141,
      "num_input_tokens_seen": 2379856,
      "step": 3635
    },
    {
      "epoch": 1.9077568134171907,
      "grad_norm": 0.21298614144325256,
      "learning_rate": 4.7680817610062896e-05,
      "loss": 0.6663,
      "num_input_tokens_seen": 2382736,
      "step": 3640
    },
    {
      "epoch": 1.9103773584905661,
      "grad_norm": 0.2359444499015808,
      "learning_rate": 4.774633123689728e-05,
      "loss": 0.5397,
      "num_input_tokens_seen": 2386672,
      "step": 3645
    },
    {
      "epoch": 1.9129979035639413,
      "grad_norm": 0.1553792953491211,
      "learning_rate": 4.7811844863731655e-05,
      "loss": 0.4899,
      "num_input_tokens_seen": 2389296,
      "step": 3650
    },
    {
      "epoch": 1.9156184486373165,
      "grad_norm": 0.22143056988716125,
      "learning_rate": 4.787735849056604e-05,
      "loss": 0.6521,
      "num_input_tokens_seen": 2392080,
      "step": 3655
    },
    {
      "epoch": 1.9182389937106918,
      "grad_norm": 0.1196957603096962,
      "learning_rate": 4.794287211740042e-05,
      "loss": 0.446,
      "num_input_tokens_seen": 2395952,
      "step": 3660
    },
    {
      "epoch": 1.9208595387840672,
      "grad_norm": 0.14969347417354584,
      "learning_rate": 4.8008385744234804e-05,
      "loss": 0.4872,
      "num_input_tokens_seen": 2398512,
      "step": 3665
    },
    {
      "epoch": 1.9234800838574424,
      "grad_norm": 0.17281480133533478,
      "learning_rate": 4.807389937106919e-05,
      "loss": 0.6277,
      "num_input_tokens_seen": 2401904,
      "step": 3670
    },
    {
      "epoch": 1.9261006289308176,
      "grad_norm": 0.1187439039349556,
      "learning_rate": 4.8139412997903564e-05,
      "loss": 0.5703,
      "num_input_tokens_seen": 2405584,
      "step": 3675
    },
    {
      "epoch": 1.9287211740041927,
      "grad_norm": 0.11101225018501282,
      "learning_rate": 4.8204926624737954e-05,
      "loss": 0.6649,
      "num_input_tokens_seen": 2409712,
      "step": 3680
    },
    {
      "epoch": 1.9313417190775681,
      "grad_norm": 0.17884312570095062,
      "learning_rate": 4.827044025157233e-05,
      "loss": 0.7103,
      "num_input_tokens_seen": 2412240,
      "step": 3685
    },
    {
      "epoch": 1.9339622641509435,
      "grad_norm": 0.17582325637340546,
      "learning_rate": 4.8335953878406706e-05,
      "loss": 0.5309,
      "num_input_tokens_seen": 2415056,
      "step": 3690
    },
    {
      "epoch": 1.9365828092243187,
      "grad_norm": 0.23162120580673218,
      "learning_rate": 4.8401467505241096e-05,
      "loss": 0.5186,
      "num_input_tokens_seen": 2419184,
      "step": 3695
    },
    {
      "epoch": 1.9392033542976939,
      "grad_norm": 0.30133530497550964,
      "learning_rate": 4.846698113207547e-05,
      "loss": 0.5796,
      "num_input_tokens_seen": 2422512,
      "step": 3700
    },
    {
      "epoch": 1.941823899371069,
      "grad_norm": 0.2644193470478058,
      "learning_rate": 4.8532494758909855e-05,
      "loss": 0.5491,
      "num_input_tokens_seen": 2425936,
      "step": 3705
    },
    {
      "epoch": 1.9444444444444444,
      "grad_norm": 0.28271809220314026,
      "learning_rate": 4.859800838574424e-05,
      "loss": 0.6895,
      "num_input_tokens_seen": 2428944,
      "step": 3710
    },
    {
      "epoch": 1.9470649895178198,
      "grad_norm": 0.19315490126609802,
      "learning_rate": 4.8663522012578615e-05,
      "loss": 0.4413,
      "num_input_tokens_seen": 2432080,
      "step": 3715
    },
    {
      "epoch": 1.949685534591195,
      "grad_norm": 0.25529518723487854,
      "learning_rate": 4.8729035639413e-05,
      "loss": 0.5173,
      "num_input_tokens_seen": 2434864,
      "step": 3720
    },
    {
      "epoch": 1.9523060796645701,
      "grad_norm": 0.2032879739999771,
      "learning_rate": 4.879454926624738e-05,
      "loss": 0.5529,
      "num_input_tokens_seen": 2437104,
      "step": 3725
    },
    {
      "epoch": 1.9549266247379455,
      "grad_norm": 0.24451684951782227,
      "learning_rate": 4.8860062893081764e-05,
      "loss": 0.725,
      "num_input_tokens_seen": 2439792,
      "step": 3730
    },
    {
      "epoch": 1.9575471698113207,
      "grad_norm": 0.1548883467912674,
      "learning_rate": 4.892557651991615e-05,
      "loss": 0.7167,
      "num_input_tokens_seen": 2442320,
      "step": 3735
    },
    {
      "epoch": 1.960167714884696,
      "grad_norm": 0.18486560881137848,
      "learning_rate": 4.899109014675053e-05,
      "loss": 0.5006,
      "num_input_tokens_seen": 2445136,
      "step": 3740
    },
    {
      "epoch": 1.9627882599580713,
      "grad_norm": 0.1321464627981186,
      "learning_rate": 4.9056603773584906e-05,
      "loss": 0.5662,
      "num_input_tokens_seen": 2448080,
      "step": 3745
    },
    {
      "epoch": 1.9654088050314464,
      "grad_norm": 0.14490632712841034,
      "learning_rate": 4.912211740041929e-05,
      "loss": 0.6131,
      "num_input_tokens_seen": 2451376,
      "step": 3750
    },
    {
      "epoch": 1.9680293501048218,
      "grad_norm": 0.17073142528533936,
      "learning_rate": 4.918763102725367e-05,
      "loss": 0.6684,
      "num_input_tokens_seen": 2454672,
      "step": 3755
    },
    {
      "epoch": 1.9706498951781972,
      "grad_norm": 0.24143894016742706,
      "learning_rate": 4.925314465408805e-05,
      "loss": 0.5118,
      "num_input_tokens_seen": 2457712,
      "step": 3760
    },
    {
      "epoch": 1.9732704402515724,
      "grad_norm": 0.2427196353673935,
      "learning_rate": 4.931865828092244e-05,
      "loss": 0.5371,
      "num_input_tokens_seen": 2461936,
      "step": 3765
    },
    {
      "epoch": 1.9758909853249476,
      "grad_norm": 0.17046324908733368,
      "learning_rate": 4.9384171907756815e-05,
      "loss": 0.5164,
      "num_input_tokens_seen": 2465296,
      "step": 3770
    },
    {
      "epoch": 1.9785115303983227,
      "grad_norm": 0.23041148483753204,
      "learning_rate": 4.944968553459119e-05,
      "loss": 0.5626,
      "num_input_tokens_seen": 2469872,
      "step": 3775
    },
    {
      "epoch": 1.9811320754716981,
      "grad_norm": 0.22888687252998352,
      "learning_rate": 4.951519916142558e-05,
      "loss": 0.4462,
      "num_input_tokens_seen": 2473840,
      "step": 3780
    },
    {
      "epoch": 1.9837526205450735,
      "grad_norm": 0.255147784948349,
      "learning_rate": 4.958071278825996e-05,
      "loss": 0.5822,
      "num_input_tokens_seen": 2476944,
      "step": 3785
    },
    {
      "epoch": 1.9863731656184487,
      "grad_norm": 0.19360268115997314,
      "learning_rate": 4.964622641509434e-05,
      "loss": 0.5587,
      "num_input_tokens_seen": 2479792,
      "step": 3790
    },
    {
      "epoch": 1.9889937106918238,
      "grad_norm": 0.21665294468402863,
      "learning_rate": 4.9711740041928724e-05,
      "loss": 0.662,
      "num_input_tokens_seen": 2484464,
      "step": 3795
    },
    {
      "epoch": 1.991614255765199,
      "grad_norm": 0.19283010065555573,
      "learning_rate": 4.977725366876311e-05,
      "loss": 0.5186,
      "num_input_tokens_seen": 2487312,
      "step": 3800
    },
    {
      "epoch": 1.9942348008385744,
      "grad_norm": 0.22487054765224457,
      "learning_rate": 4.984276729559749e-05,
      "loss": 0.5368,
      "num_input_tokens_seen": 2490288,
      "step": 3805
    },
    {
      "epoch": 1.9968553459119498,
      "grad_norm": 0.1827460676431656,
      "learning_rate": 4.9908280922431866e-05,
      "loss": 0.595,
      "num_input_tokens_seen": 2494128,
      "step": 3810
    },
    {
      "epoch": 1.999475890985325,
      "grad_norm": 0.1675104796886444,
      "learning_rate": 4.997379454926625e-05,
      "loss": 0.5438,
      "num_input_tokens_seen": 2496848,
      "step": 3815
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.5793213248252869,
      "eval_runtime": 9.9374,
      "eval_samples_per_second": 85.335,
      "eval_steps_per_second": 21.334,
      "num_input_tokens_seen": 2497016,
      "step": 3816
    },
    {
      "epoch": 2.0020964360587,
      "grad_norm": 0.19449985027313232,
      "learning_rate": 4.999999905865133e-05,
      "loss": 0.4605,
      "num_input_tokens_seen": 2499256,
      "step": 3820
    },
    {
      "epoch": 2.0047169811320753,
      "grad_norm": 0.24465259909629822,
      "learning_rate": 4.999999330596526e-05,
      "loss": 0.5869,
      "num_input_tokens_seen": 2502200,
      "step": 3825
    },
    {
      "epoch": 2.007337526205451,
      "grad_norm": 0.351024329662323,
      "learning_rate": 4.9999982323565806e-05,
      "loss": 0.7439,
      "num_input_tokens_seen": 2505080,
      "step": 3830
    },
    {
      "epoch": 2.009958071278826,
      "grad_norm": 0.25242879986763,
      "learning_rate": 4.999996611145527e-05,
      "loss": 0.5798,
      "num_input_tokens_seen": 2508216,
      "step": 3835
    },
    {
      "epoch": 2.0125786163522013,
      "grad_norm": 0.16839496791362762,
      "learning_rate": 4.9999944669637024e-05,
      "loss": 0.4139,
      "num_input_tokens_seen": 2511128,
      "step": 3840
    },
    {
      "epoch": 2.0151991614255764,
      "grad_norm": 0.20628784596920013,
      "learning_rate": 4.999991799811557e-05,
      "loss": 0.5659,
      "num_input_tokens_seen": 2514776,
      "step": 3845
    },
    {
      "epoch": 2.0178197064989516,
      "grad_norm": 0.3333597779273987,
      "learning_rate": 4.99998860968965e-05,
      "loss": 0.7673,
      "num_input_tokens_seen": 2517528,
      "step": 3850
    },
    {
      "epoch": 2.020440251572327,
      "grad_norm": 0.12071157991886139,
      "learning_rate": 4.999984896598646e-05,
      "loss": 0.6056,
      "num_input_tokens_seen": 2523032,
      "step": 3855
    },
    {
      "epoch": 2.0230607966457024,
      "grad_norm": 0.15287144482135773,
      "learning_rate": 4.999980660539323e-05,
      "loss": 0.5464,
      "num_input_tokens_seen": 2526200,
      "step": 3860
    },
    {
      "epoch": 2.0256813417190775,
      "grad_norm": 0.18033352494239807,
      "learning_rate": 4.999975901512567e-05,
      "loss": 0.6507,
      "num_input_tokens_seen": 2532760,
      "step": 3865
    },
    {
      "epoch": 2.0283018867924527,
      "grad_norm": 0.16539660096168518,
      "learning_rate": 4.999970619519374e-05,
      "loss": 0.5394,
      "num_input_tokens_seen": 2535640,
      "step": 3870
    },
    {
      "epoch": 2.030922431865828,
      "grad_norm": 0.1383124440908432,
      "learning_rate": 4.999964814560849e-05,
      "loss": 0.5381,
      "num_input_tokens_seen": 2539288,
      "step": 3875
    },
    {
      "epoch": 2.0335429769392035,
      "grad_norm": 0.3425218462944031,
      "learning_rate": 4.999958486638205e-05,
      "loss": 0.6776,
      "num_input_tokens_seen": 2541880,
      "step": 3880
    },
    {
      "epoch": 2.0361635220125787,
      "grad_norm": 0.17353114485740662,
      "learning_rate": 4.999951635752767e-05,
      "loss": 0.5595,
      "num_input_tokens_seen": 2545080,
      "step": 3885
    },
    {
      "epoch": 2.038784067085954,
      "grad_norm": 0.15688447654247284,
      "learning_rate": 4.999944261905968e-05,
      "loss": 0.4167,
      "num_input_tokens_seen": 2548376,
      "step": 3890
    },
    {
      "epoch": 2.041404612159329,
      "grad_norm": 0.177053302526474,
      "learning_rate": 4.99993636509935e-05,
      "loss": 0.4884,
      "num_input_tokens_seen": 2550616,
      "step": 3895
    },
    {
      "epoch": 2.0440251572327046,
      "grad_norm": 0.2112322598695755,
      "learning_rate": 4.999927945334565e-05,
      "loss": 0.6455,
      "num_input_tokens_seen": 2555000,
      "step": 3900
    },
    {
      "epoch": 2.04664570230608,
      "grad_norm": 0.36097726225852966,
      "learning_rate": 4.9999190026133745e-05,
      "loss": 0.5679,
      "num_input_tokens_seen": 2558328,
      "step": 3905
    },
    {
      "epoch": 2.049266247379455,
      "grad_norm": 0.17867060005664825,
      "learning_rate": 4.9999095369376506e-05,
      "loss": 0.6548,
      "num_input_tokens_seen": 2561592,
      "step": 3910
    },
    {
      "epoch": 2.05188679245283,
      "grad_norm": 0.18889671564102173,
      "learning_rate": 4.99989954830937e-05,
      "loss": 0.4207,
      "num_input_tokens_seen": 2564472,
      "step": 3915
    },
    {
      "epoch": 2.0545073375262053,
      "grad_norm": 0.3160642385482788,
      "learning_rate": 4.9998890367306255e-05,
      "loss": 0.5515,
      "num_input_tokens_seen": 2567352,
      "step": 3920
    },
    {
      "epoch": 2.057127882599581,
      "grad_norm": 0.266102135181427,
      "learning_rate": 4.999878002203614e-05,
      "loss": 0.4162,
      "num_input_tokens_seen": 2571448,
      "step": 3925
    },
    {
      "epoch": 2.059748427672956,
      "grad_norm": 0.26654863357543945,
      "learning_rate": 4.999866444730645e-05,
      "loss": 0.4946,
      "num_input_tokens_seen": 2574264,
      "step": 3930
    },
    {
      "epoch": 2.0623689727463312,
      "grad_norm": 0.22647742927074432,
      "learning_rate": 4.999854364314136e-05,
      "loss": 0.6843,
      "num_input_tokens_seen": 2577592,
      "step": 3935
    },
    {
      "epoch": 2.0649895178197064,
      "grad_norm": 0.24468041956424713,
      "learning_rate": 4.999841760956614e-05,
      "loss": 0.5631,
      "num_input_tokens_seen": 2581624,
      "step": 3940
    },
    {
      "epoch": 2.0676100628930816,
      "grad_norm": 0.21723292768001556,
      "learning_rate": 4.999828634660715e-05,
      "loss": 0.5803,
      "num_input_tokens_seen": 2584888,
      "step": 3945
    },
    {
      "epoch": 2.070230607966457,
      "grad_norm": 0.3108595907688141,
      "learning_rate": 4.9998149854291856e-05,
      "loss": 0.5524,
      "num_input_tokens_seen": 2587672,
      "step": 3950
    },
    {
      "epoch": 2.0728511530398324,
      "grad_norm": 0.2372923046350479,
      "learning_rate": 4.99980081326488e-05,
      "loss": 0.55,
      "num_input_tokens_seen": 2590136,
      "step": 3955
    },
    {
      "epoch": 2.0754716981132075,
      "grad_norm": 0.16356781125068665,
      "learning_rate": 4.999786118170763e-05,
      "loss": 0.5368,
      "num_input_tokens_seen": 2593976,
      "step": 3960
    },
    {
      "epoch": 2.0780922431865827,
      "grad_norm": 0.19205883145332336,
      "learning_rate": 4.99977090014991e-05,
      "loss": 1.1229,
      "num_input_tokens_seen": 2597208,
      "step": 3965
    },
    {
      "epoch": 2.080712788259958,
      "grad_norm": 0.21624022722244263,
      "learning_rate": 4.9997551592055034e-05,
      "loss": 0.567,
      "num_input_tokens_seen": 2600536,
      "step": 3970
    },
    {
      "epoch": 2.0833333333333335,
      "grad_norm": 0.20295873284339905,
      "learning_rate": 4.999738895340836e-05,
      "loss": 0.5457,
      "num_input_tokens_seen": 2603608,
      "step": 3975
    },
    {
      "epoch": 2.0859538784067087,
      "grad_norm": 0.24153585731983185,
      "learning_rate": 4.999722108559311e-05,
      "loss": 0.5689,
      "num_input_tokens_seen": 2607288,
      "step": 3980
    },
    {
      "epoch": 2.088574423480084,
      "grad_norm": 0.165012925863266,
      "learning_rate": 4.9997047988644385e-05,
      "loss": 0.6764,
      "num_input_tokens_seen": 2610392,
      "step": 3985
    },
    {
      "epoch": 2.091194968553459,
      "grad_norm": 0.16908755898475647,
      "learning_rate": 4.99968696625984e-05,
      "loss": 0.7145,
      "num_input_tokens_seen": 2613208,
      "step": 3990
    },
    {
      "epoch": 2.0938155136268346,
      "grad_norm": 0.1540287584066391,
      "learning_rate": 4.999668610749246e-05,
      "loss": 0.7092,
      "num_input_tokens_seen": 2616472,
      "step": 3995
    },
    {
      "epoch": 2.0964360587002098,
      "grad_norm": 0.16380023956298828,
      "learning_rate": 4.999649732336498e-05,
      "loss": 0.5546,
      "num_input_tokens_seen": 2620504,
      "step": 4000
    },
    {
      "epoch": 2.099056603773585,
      "grad_norm": 0.17821268737316132,
      "learning_rate": 4.999630331025542e-05,
      "loss": 0.5865,
      "num_input_tokens_seen": 2624632,
      "step": 4005
    },
    {
      "epoch": 2.10167714884696,
      "grad_norm": 0.2202167958021164,
      "learning_rate": 4.9996104068204385e-05,
      "loss": 0.5555,
      "num_input_tokens_seen": 2627608,
      "step": 4010
    },
    {
      "epoch": 2.1042976939203353,
      "grad_norm": 0.11672813445329666,
      "learning_rate": 4.999589959725355e-05,
      "loss": 0.5129,
      "num_input_tokens_seen": 2630232,
      "step": 4015
    },
    {
      "epoch": 2.106918238993711,
      "grad_norm": 0.13238902390003204,
      "learning_rate": 4.999568989744568e-05,
      "loss": 0.5003,
      "num_input_tokens_seen": 2633624,
      "step": 4020
    },
    {
      "epoch": 2.109538784067086,
      "grad_norm": 0.5528309941291809,
      "learning_rate": 4.999547496882466e-05,
      "loss": 0.6371,
      "num_input_tokens_seen": 2636280,
      "step": 4025
    },
    {
      "epoch": 2.1121593291404612,
      "grad_norm": 0.1958792805671692,
      "learning_rate": 4.999525481143543e-05,
      "loss": 0.6028,
      "num_input_tokens_seen": 2640152,
      "step": 4030
    },
    {
      "epoch": 2.1147798742138364,
      "grad_norm": 0.1332084685564041,
      "learning_rate": 4.999502942532406e-05,
      "loss": 0.5486,
      "num_input_tokens_seen": 2643800,
      "step": 4035
    },
    {
      "epoch": 2.1174004192872116,
      "grad_norm": 0.15046599507331848,
      "learning_rate": 4.999479881053769e-05,
      "loss": 0.5102,
      "num_input_tokens_seen": 2647608,
      "step": 4040
    },
    {
      "epoch": 2.120020964360587,
      "grad_norm": 0.18746806681156158,
      "learning_rate": 4.999456296712457e-05,
      "loss": 0.6487,
      "num_input_tokens_seen": 2650200,
      "step": 4045
    },
    {
      "epoch": 2.1226415094339623,
      "grad_norm": 0.17142881453037262,
      "learning_rate": 4.9994321895134025e-05,
      "loss": 0.6816,
      "num_input_tokens_seen": 2653784,
      "step": 4050
    },
    {
      "epoch": 2.1252620545073375,
      "grad_norm": 0.11574537307024002,
      "learning_rate": 4.999407559461649e-05,
      "loss": 0.5805,
      "num_input_tokens_seen": 2658104,
      "step": 4055
    },
    {
      "epoch": 2.1278825995807127,
      "grad_norm": 0.23913410305976868,
      "learning_rate": 4.999382406562349e-05,
      "loss": 0.5202,
      "num_input_tokens_seen": 2661080,
      "step": 4060
    },
    {
      "epoch": 2.130503144654088,
      "grad_norm": 0.17755503952503204,
      "learning_rate": 4.999356730820765e-05,
      "loss": 0.4745,
      "num_input_tokens_seen": 2664312,
      "step": 4065
    },
    {
      "epoch": 2.1331236897274635,
      "grad_norm": 0.5249155163764954,
      "learning_rate": 4.999330532242266e-05,
      "loss": 0.6811,
      "num_input_tokens_seen": 2667704,
      "step": 4070
    },
    {
      "epoch": 2.1357442348008386,
      "grad_norm": 0.1892002373933792,
      "learning_rate": 4.999303810832333e-05,
      "loss": 0.4525,
      "num_input_tokens_seen": 2671160,
      "step": 4075
    },
    {
      "epoch": 2.138364779874214,
      "grad_norm": 0.12049844861030579,
      "learning_rate": 4.999276566596557e-05,
      "loss": 0.6101,
      "num_input_tokens_seen": 2675544,
      "step": 4080
    },
    {
      "epoch": 2.140985324947589,
      "grad_norm": 0.196145698428154,
      "learning_rate": 4.9992487995406364e-05,
      "loss": 0.525,
      "num_input_tokens_seen": 2678264,
      "step": 4085
    },
    {
      "epoch": 2.1436058700209646,
      "grad_norm": 0.23833103477954865,
      "learning_rate": 4.9992205096703806e-05,
      "loss": 0.479,
      "num_input_tokens_seen": 2684120,
      "step": 4090
    },
    {
      "epoch": 2.1462264150943398,
      "grad_norm": 0.39245274662971497,
      "learning_rate": 4.9991916969917054e-05,
      "loss": 0.755,
      "num_input_tokens_seen": 2686872,
      "step": 4095
    },
    {
      "epoch": 2.148846960167715,
      "grad_norm": 0.15931633114814758,
      "learning_rate": 4.99916236151064e-05,
      "loss": 0.5226,
      "num_input_tokens_seen": 2690456,
      "step": 4100
    },
    {
      "epoch": 2.15146750524109,
      "grad_norm": 0.16941091418266296,
      "learning_rate": 4.999132503233321e-05,
      "loss": 0.5684,
      "num_input_tokens_seen": 2692824,
      "step": 4105
    },
    {
      "epoch": 2.1540880503144653,
      "grad_norm": 0.22545014321804047,
      "learning_rate": 4.999102122165994e-05,
      "loss": 0.6706,
      "num_input_tokens_seen": 2695704,
      "step": 4110
    },
    {
      "epoch": 2.156708595387841,
      "grad_norm": 0.15288467705249786,
      "learning_rate": 4.9990712183150135e-05,
      "loss": 0.5393,
      "num_input_tokens_seen": 2699480,
      "step": 4115
    },
    {
      "epoch": 2.159329140461216,
      "grad_norm": 0.18652179837226868,
      "learning_rate": 4.9990397916868456e-05,
      "loss": 0.6169,
      "num_input_tokens_seen": 2703960,
      "step": 4120
    },
    {
      "epoch": 2.161949685534591,
      "grad_norm": 0.1496136337518692,
      "learning_rate": 4.999007842288064e-05,
      "loss": 0.5241,
      "num_input_tokens_seen": 2706552,
      "step": 4125
    },
    {
      "epoch": 2.1645702306079664,
      "grad_norm": 0.17529448866844177,
      "learning_rate": 4.9989753701253514e-05,
      "loss": 0.617,
      "num_input_tokens_seen": 2709368,
      "step": 4130
    },
    {
      "epoch": 2.1671907756813416,
      "grad_norm": 0.2389903962612152,
      "learning_rate": 4.998942375205502e-05,
      "loss": 0.618,
      "num_input_tokens_seen": 2712248,
      "step": 4135
    },
    {
      "epoch": 2.169811320754717,
      "grad_norm": 0.17285238206386566,
      "learning_rate": 4.998908857535416e-05,
      "loss": 0.56,
      "num_input_tokens_seen": 2715448,
      "step": 4140
    },
    {
      "epoch": 2.1724318658280923,
      "grad_norm": 0.12814339995384216,
      "learning_rate": 4.9988748171221065e-05,
      "loss": 0.4569,
      "num_input_tokens_seen": 2719064,
      "step": 4145
    },
    {
      "epoch": 2.1750524109014675,
      "grad_norm": 0.1425209790468216,
      "learning_rate": 4.998840253972695e-05,
      "loss": 0.4978,
      "num_input_tokens_seen": 2722488,
      "step": 4150
    },
    {
      "epoch": 2.1776729559748427,
      "grad_norm": 0.2557077407836914,
      "learning_rate": 4.9988051680944094e-05,
      "loss": 0.4943,
      "num_input_tokens_seen": 2725528,
      "step": 4155
    },
    {
      "epoch": 2.180293501048218,
      "grad_norm": 0.12327254563570023,
      "learning_rate": 4.998769559494591e-05,
      "loss": 0.4781,
      "num_input_tokens_seen": 2728824,
      "step": 4160
    },
    {
      "epoch": 2.1829140461215935,
      "grad_norm": 0.182267427444458,
      "learning_rate": 4.998733428180689e-05,
      "loss": 0.6944,
      "num_input_tokens_seen": 2732472,
      "step": 4165
    },
    {
      "epoch": 2.1855345911949686,
      "grad_norm": 0.11855469644069672,
      "learning_rate": 4.998696774160259e-05,
      "loss": 0.5533,
      "num_input_tokens_seen": 2735544,
      "step": 4170
    },
    {
      "epoch": 2.188155136268344,
      "grad_norm": 0.13490603864192963,
      "learning_rate": 4.9986595974409714e-05,
      "loss": 0.5632,
      "num_input_tokens_seen": 2739032,
      "step": 4175
    },
    {
      "epoch": 2.190775681341719,
      "grad_norm": 0.21515314280986786,
      "learning_rate": 4.998621898030603e-05,
      "loss": 0.7208,
      "num_input_tokens_seen": 2742456,
      "step": 4180
    },
    {
      "epoch": 2.1933962264150946,
      "grad_norm": 0.27070075273513794,
      "learning_rate": 4.998583675937039e-05,
      "loss": 0.488,
      "num_input_tokens_seen": 2744888,
      "step": 4185
    },
    {
      "epoch": 2.1960167714884697,
      "grad_norm": 0.19607198238372803,
      "learning_rate": 4.998544931168274e-05,
      "loss": 0.6695,
      "num_input_tokens_seen": 2747512,
      "step": 4190
    },
    {
      "epoch": 2.198637316561845,
      "grad_norm": 0.3549392521381378,
      "learning_rate": 4.9985056637324154e-05,
      "loss": 0.6052,
      "num_input_tokens_seen": 2750648,
      "step": 4195
    },
    {
      "epoch": 2.20125786163522,
      "grad_norm": 0.17501980066299438,
      "learning_rate": 4.998465873637677e-05,
      "loss": 0.4439,
      "num_input_tokens_seen": 2753368,
      "step": 4200
    },
    {
      "epoch": 2.2038784067085953,
      "grad_norm": 0.24566097557544708,
      "learning_rate": 4.9984255608923806e-05,
      "loss": 0.5171,
      "num_input_tokens_seen": 2756696,
      "step": 4205
    },
    {
      "epoch": 2.2064989517819704,
      "grad_norm": 0.26491478085517883,
      "learning_rate": 4.998384725504961e-05,
      "loss": 0.584,
      "num_input_tokens_seen": 2759256,
      "step": 4210
    },
    {
      "epoch": 2.209119496855346,
      "grad_norm": 0.20138002932071686,
      "learning_rate": 4.9983433674839596e-05,
      "loss": 0.5923,
      "num_input_tokens_seen": 2762296,
      "step": 4215
    },
    {
      "epoch": 2.211740041928721,
      "grad_norm": 0.2762216031551361,
      "learning_rate": 4.998301486838028e-05,
      "loss": 0.5393,
      "num_input_tokens_seen": 2765400,
      "step": 4220
    },
    {
      "epoch": 2.2143605870020964,
      "grad_norm": 0.328377902507782,
      "learning_rate": 4.998259083575929e-05,
      "loss": 0.5893,
      "num_input_tokens_seen": 2768120,
      "step": 4225
    },
    {
      "epoch": 2.2169811320754715,
      "grad_norm": 0.17632973194122314,
      "learning_rate": 4.99821615770653e-05,
      "loss": 0.604,
      "num_input_tokens_seen": 2771096,
      "step": 4230
    },
    {
      "epoch": 2.219601677148847,
      "grad_norm": 0.22862277925014496,
      "learning_rate": 4.998172709238813e-05,
      "loss": 0.6282,
      "num_input_tokens_seen": 2774328,
      "step": 4235
    },
    {
      "epoch": 2.2222222222222223,
      "grad_norm": 0.12649179995059967,
      "learning_rate": 4.9981287381818654e-05,
      "loss": 0.5963,
      "num_input_tokens_seen": 2778008,
      "step": 4240
    },
    {
      "epoch": 2.2248427672955975,
      "grad_norm": 0.5964694023132324,
      "learning_rate": 4.9980842445448856e-05,
      "loss": 0.653,
      "num_input_tokens_seen": 2780216,
      "step": 4245
    },
    {
      "epoch": 2.2274633123689727,
      "grad_norm": 0.18201784789562225,
      "learning_rate": 4.9980392283371815e-05,
      "loss": 0.5455,
      "num_input_tokens_seen": 2784088,
      "step": 4250
    },
    {
      "epoch": 2.230083857442348,
      "grad_norm": 0.14869056642055511,
      "learning_rate": 4.99799368956817e-05,
      "loss": 0.4539,
      "num_input_tokens_seen": 2786488,
      "step": 4255
    },
    {
      "epoch": 2.2327044025157234,
      "grad_norm": 0.28734248876571655,
      "learning_rate": 4.997947628247378e-05,
      "loss": 0.5034,
      "num_input_tokens_seen": 2788920,
      "step": 4260
    },
    {
      "epoch": 2.2353249475890986,
      "grad_norm": 0.19088907539844513,
      "learning_rate": 4.9979010443844395e-05,
      "loss": 0.7374,
      "num_input_tokens_seen": 2791640,
      "step": 4265
    },
    {
      "epoch": 2.237945492662474,
      "grad_norm": 0.20580680668354034,
      "learning_rate": 4.9978539379891e-05,
      "loss": 0.4696,
      "num_input_tokens_seen": 2794904,
      "step": 4270
    },
    {
      "epoch": 2.240566037735849,
      "grad_norm": 0.17539188265800476,
      "learning_rate": 4.997806309071214e-05,
      "loss": 0.4673,
      "num_input_tokens_seen": 2798840,
      "step": 4275
    },
    {
      "epoch": 2.243186582809224,
      "grad_norm": 0.1892538070678711,
      "learning_rate": 4.997758157640745e-05,
      "loss": 0.5933,
      "num_input_tokens_seen": 2802616,
      "step": 4280
    },
    {
      "epoch": 2.2458071278825997,
      "grad_norm": 0.2909959554672241,
      "learning_rate": 4.997709483707765e-05,
      "loss": 0.5307,
      "num_input_tokens_seen": 2805432,
      "step": 4285
    },
    {
      "epoch": 2.248427672955975,
      "grad_norm": 0.25527894496917725,
      "learning_rate": 4.997660287282456e-05,
      "loss": 0.5907,
      "num_input_tokens_seen": 2808696,
      "step": 4290
    },
    {
      "epoch": 2.25104821802935,
      "grad_norm": 0.1497332900762558,
      "learning_rate": 4.99761056837511e-05,
      "loss": 0.4852,
      "num_input_tokens_seen": 2811640,
      "step": 4295
    },
    {
      "epoch": 2.2536687631027252,
      "grad_norm": 0.17872123420238495,
      "learning_rate": 4.997560326996128e-05,
      "loss": 0.5206,
      "num_input_tokens_seen": 2815224,
      "step": 4300
    },
    {
      "epoch": 2.2562893081761004,
      "grad_norm": 0.16927208006381989,
      "learning_rate": 4.997509563156018e-05,
      "loss": 0.5516,
      "num_input_tokens_seen": 2817688,
      "step": 4305
    },
    {
      "epoch": 2.258909853249476,
      "grad_norm": 0.13638313114643097,
      "learning_rate": 4.997458276865401e-05,
      "loss": 0.4632,
      "num_input_tokens_seen": 2821016,
      "step": 4310
    },
    {
      "epoch": 2.261530398322851,
      "grad_norm": 0.18468528985977173,
      "learning_rate": 4.997406468135005e-05,
      "loss": 0.6521,
      "num_input_tokens_seen": 2823608,
      "step": 4315
    },
    {
      "epoch": 2.2641509433962264,
      "grad_norm": 0.21714916825294495,
      "learning_rate": 4.9973541369756686e-05,
      "loss": 0.4703,
      "num_input_tokens_seen": 2828248,
      "step": 4320
    },
    {
      "epoch": 2.2667714884696015,
      "grad_norm": 0.18649111688137054,
      "learning_rate": 4.9973012833983365e-05,
      "loss": 0.5661,
      "num_input_tokens_seen": 2832344,
      "step": 4325
    },
    {
      "epoch": 2.269392033542977,
      "grad_norm": 0.2118178755044937,
      "learning_rate": 4.9972479074140684e-05,
      "loss": 0.4876,
      "num_input_tokens_seen": 2836408,
      "step": 4330
    },
    {
      "epoch": 2.2720125786163523,
      "grad_norm": 0.18490396440029144,
      "learning_rate": 4.9971940090340276e-05,
      "loss": 0.6718,
      "num_input_tokens_seen": 2839320,
      "step": 4335
    },
    {
      "epoch": 2.2746331236897275,
      "grad_norm": 0.2679949700832367,
      "learning_rate": 4.997139588269489e-05,
      "loss": 0.6322,
      "num_input_tokens_seen": 2842456,
      "step": 4340
    },
    {
      "epoch": 2.2772536687631026,
      "grad_norm": 0.301643967628479,
      "learning_rate": 4.997084645131838e-05,
      "loss": 0.7111,
      "num_input_tokens_seen": 2845432,
      "step": 4345
    },
    {
      "epoch": 2.279874213836478,
      "grad_norm": 0.36210396885871887,
      "learning_rate": 4.9970291796325666e-05,
      "loss": 0.593,
      "num_input_tokens_seen": 2848536,
      "step": 4350
    },
    {
      "epoch": 2.2824947589098534,
      "grad_norm": 0.1752099096775055,
      "learning_rate": 4.996973191783279e-05,
      "loss": 0.6287,
      "num_input_tokens_seen": 2851544,
      "step": 4355
    },
    {
      "epoch": 2.2851153039832286,
      "grad_norm": 0.20414572954177856,
      "learning_rate": 4.996916681595687e-05,
      "loss": 0.5344,
      "num_input_tokens_seen": 2854392,
      "step": 4360
    },
    {
      "epoch": 2.2877358490566038,
      "grad_norm": 0.11355189979076385,
      "learning_rate": 4.996859649081611e-05,
      "loss": 0.4882,
      "num_input_tokens_seen": 2858424,
      "step": 4365
    },
    {
      "epoch": 2.290356394129979,
      "grad_norm": 0.13140782713890076,
      "learning_rate": 4.996802094252983e-05,
      "loss": 0.5658,
      "num_input_tokens_seen": 2861368,
      "step": 4370
    },
    {
      "epoch": 2.2929769392033545,
      "grad_norm": 0.2589550018310547,
      "learning_rate": 4.996744017121841e-05,
      "loss": 0.5998,
      "num_input_tokens_seen": 2863960,
      "step": 4375
    },
    {
      "epoch": 2.2955974842767297,
      "grad_norm": 0.3700113296508789,
      "learning_rate": 4.9966854177003355e-05,
      "loss": 0.6334,
      "num_input_tokens_seen": 2867352,
      "step": 4380
    },
    {
      "epoch": 2.298218029350105,
      "grad_norm": 0.2457696497440338,
      "learning_rate": 4.996626296000724e-05,
      "loss": 0.5893,
      "num_input_tokens_seen": 2871992,
      "step": 4385
    },
    {
      "epoch": 2.30083857442348,
      "grad_norm": 0.1724703460931778,
      "learning_rate": 4.996566652035374e-05,
      "loss": 0.5045,
      "num_input_tokens_seen": 2874680,
      "step": 4390
    },
    {
      "epoch": 2.3034591194968552,
      "grad_norm": 0.33341923356056213,
      "learning_rate": 4.996506485816763e-05,
      "loss": 0.417,
      "num_input_tokens_seen": 2877688,
      "step": 4395
    },
    {
      "epoch": 2.3060796645702304,
      "grad_norm": 0.1539616733789444,
      "learning_rate": 4.996445797357477e-05,
      "loss": 0.6071,
      "num_input_tokens_seen": 2885528,
      "step": 4400
    },
    {
      "epoch": 2.308700209643606,
      "grad_norm": 0.16675472259521484,
      "learning_rate": 4.996384586670211e-05,
      "loss": 0.6691,
      "num_input_tokens_seen": 2889464,
      "step": 4405
    },
    {
      "epoch": 2.311320754716981,
      "grad_norm": 0.2359144389629364,
      "learning_rate": 4.99632285376777e-05,
      "loss": 0.4338,
      "num_input_tokens_seen": 2893464,
      "step": 4410
    },
    {
      "epoch": 2.3139412997903563,
      "grad_norm": 0.21750353276729584,
      "learning_rate": 4.9962605986630674e-05,
      "loss": 0.5986,
      "num_input_tokens_seen": 2896792,
      "step": 4415
    },
    {
      "epoch": 2.3165618448637315,
      "grad_norm": 0.26544129848480225,
      "learning_rate": 4.996197821369126e-05,
      "loss": 0.5437,
      "num_input_tokens_seen": 2900664,
      "step": 4420
    },
    {
      "epoch": 2.319182389937107,
      "grad_norm": 0.23091909289360046,
      "learning_rate": 4.9961345218990796e-05,
      "loss": 0.6303,
      "num_input_tokens_seen": 2904152,
      "step": 4425
    },
    {
      "epoch": 2.3218029350104823,
      "grad_norm": 0.15353824198246002,
      "learning_rate": 4.996070700266168e-05,
      "loss": 0.5694,
      "num_input_tokens_seen": 2906776,
      "step": 4430
    },
    {
      "epoch": 2.3244234800838575,
      "grad_norm": 0.12943534553050995,
      "learning_rate": 4.9960063564837424e-05,
      "loss": 0.5722,
      "num_input_tokens_seen": 2909720,
      "step": 4435
    },
    {
      "epoch": 2.3270440251572326,
      "grad_norm": 0.23035654425621033,
      "learning_rate": 4.995941490565262e-05,
      "loss": 0.3889,
      "num_input_tokens_seen": 2916536,
      "step": 4440
    },
    {
      "epoch": 2.329664570230608,
      "grad_norm": 0.19919265806674957,
      "learning_rate": 4.995876102524299e-05,
      "loss": 0.5921,
      "num_input_tokens_seen": 2919608,
      "step": 4445
    },
    {
      "epoch": 2.3322851153039834,
      "grad_norm": 0.15625131130218506,
      "learning_rate": 4.995810192374529e-05,
      "loss": 0.6763,
      "num_input_tokens_seen": 2923352,
      "step": 4450
    },
    {
      "epoch": 2.3349056603773586,
      "grad_norm": 0.20577314496040344,
      "learning_rate": 4.99574376012974e-05,
      "loss": 0.5507,
      "num_input_tokens_seen": 2926232,
      "step": 4455
    },
    {
      "epoch": 2.3375262054507338,
      "grad_norm": 0.26860862970352173,
      "learning_rate": 4.995676805803829e-05,
      "loss": 0.4874,
      "num_input_tokens_seen": 2929656,
      "step": 4460
    },
    {
      "epoch": 2.340146750524109,
      "grad_norm": 0.20483672618865967,
      "learning_rate": 4.995609329410804e-05,
      "loss": 0.6094,
      "num_input_tokens_seen": 2932344,
      "step": 4465
    },
    {
      "epoch": 2.342767295597484,
      "grad_norm": 0.1824646145105362,
      "learning_rate": 4.9955413309647776e-05,
      "loss": 0.464,
      "num_input_tokens_seen": 2936152,
      "step": 4470
    },
    {
      "epoch": 2.3453878406708597,
      "grad_norm": 0.17765261232852936,
      "learning_rate": 4.9954728104799756e-05,
      "loss": 0.512,
      "num_input_tokens_seen": 2939224,
      "step": 4475
    },
    {
      "epoch": 2.348008385744235,
      "grad_norm": 0.1342320591211319,
      "learning_rate": 4.995403767970731e-05,
      "loss": 0.4572,
      "num_input_tokens_seen": 2941912,
      "step": 4480
    },
    {
      "epoch": 2.35062893081761,
      "grad_norm": 0.16416345536708832,
      "learning_rate": 4.995334203451488e-05,
      "loss": 0.4803,
      "num_input_tokens_seen": 2944568,
      "step": 4485
    },
    {
      "epoch": 2.353249475890985,
      "grad_norm": 0.175105020403862,
      "learning_rate": 4.995264116936798e-05,
      "loss": 0.4675,
      "num_input_tokens_seen": 2947480,
      "step": 4490
    },
    {
      "epoch": 2.3558700209643604,
      "grad_norm": 0.23171298205852509,
      "learning_rate": 4.995193508441322e-05,
      "loss": 0.5848,
      "num_input_tokens_seen": 2951192,
      "step": 4495
    },
    {
      "epoch": 2.358490566037736,
      "grad_norm": 0.28566113114356995,
      "learning_rate": 4.99512237797983e-05,
      "loss": 0.5633,
      "num_input_tokens_seen": 2954648,
      "step": 4500
    },
    {
      "epoch": 2.361111111111111,
      "grad_norm": 0.12965252995491028,
      "learning_rate": 4.995050725567203e-05,
      "loss": 0.4995,
      "num_input_tokens_seen": 2958552,
      "step": 4505
    },
    {
      "epoch": 2.3637316561844863,
      "grad_norm": 0.1272067278623581,
      "learning_rate": 4.994978551218429e-05,
      "loss": 0.4978,
      "num_input_tokens_seen": 2962040,
      "step": 4510
    },
    {
      "epoch": 2.3663522012578615,
      "grad_norm": 0.15256230533123016,
      "learning_rate": 4.9949058549486066e-05,
      "loss": 0.5144,
      "num_input_tokens_seen": 2964824,
      "step": 4515
    },
    {
      "epoch": 2.368972746331237,
      "grad_norm": 0.12783552706241608,
      "learning_rate": 4.994832636772943e-05,
      "loss": 0.5006,
      "num_input_tokens_seen": 2967448,
      "step": 4520
    },
    {
      "epoch": 2.3715932914046123,
      "grad_norm": 0.18466055393218994,
      "learning_rate": 4.994758896706754e-05,
      "loss": 0.6264,
      "num_input_tokens_seen": 2971000,
      "step": 4525
    },
    {
      "epoch": 2.3742138364779874,
      "grad_norm": 0.27617475390434265,
      "learning_rate": 4.994684634765465e-05,
      "loss": 0.556,
      "num_input_tokens_seen": 2973592,
      "step": 4530
    },
    {
      "epoch": 2.3768343815513626,
      "grad_norm": 0.14491568505764008,
      "learning_rate": 4.9946098509646125e-05,
      "loss": 0.6214,
      "num_input_tokens_seen": 2977560,
      "step": 4535
    },
    {
      "epoch": 2.379454926624738,
      "grad_norm": 0.17889107763767242,
      "learning_rate": 4.994534545319838e-05,
      "loss": 0.5562,
      "num_input_tokens_seen": 2980280,
      "step": 4540
    },
    {
      "epoch": 2.3820754716981134,
      "grad_norm": 0.18095320463180542,
      "learning_rate": 4.994458717846896e-05,
      "loss": 0.5549,
      "num_input_tokens_seen": 2983352,
      "step": 4545
    },
    {
      "epoch": 2.3846960167714886,
      "grad_norm": 0.22530461847782135,
      "learning_rate": 4.9943823685616485e-05,
      "loss": 0.5759,
      "num_input_tokens_seen": 2986648,
      "step": 4550
    },
    {
      "epoch": 2.3873165618448637,
      "grad_norm": 0.2535705864429474,
      "learning_rate": 4.994305497480067e-05,
      "loss": 0.5304,
      "num_input_tokens_seen": 2989880,
      "step": 4555
    },
    {
      "epoch": 2.389937106918239,
      "grad_norm": 0.19790537655353546,
      "learning_rate": 4.994228104618233e-05,
      "loss": 0.5087,
      "num_input_tokens_seen": 2992536,
      "step": 4560
    },
    {
      "epoch": 2.392557651991614,
      "grad_norm": 0.2780994772911072,
      "learning_rate": 4.994150189992334e-05,
      "loss": 0.5672,
      "num_input_tokens_seen": 2996152,
      "step": 4565
    },
    {
      "epoch": 2.3951781970649897,
      "grad_norm": 0.4232412874698639,
      "learning_rate": 4.99407175361867e-05,
      "loss": 0.5198,
      "num_input_tokens_seen": 2999288,
      "step": 4570
    },
    {
      "epoch": 2.397798742138365,
      "grad_norm": 0.21552059054374695,
      "learning_rate": 4.99399279551365e-05,
      "loss": 0.4234,
      "num_input_tokens_seen": 3002520,
      "step": 4575
    },
    {
      "epoch": 2.40041928721174,
      "grad_norm": 0.20093107223510742,
      "learning_rate": 4.9939133156937886e-05,
      "loss": 0.6217,
      "num_input_tokens_seen": 3005752,
      "step": 4580
    },
    {
      "epoch": 2.403039832285115,
      "grad_norm": 0.1837342232465744,
      "learning_rate": 4.993833314175716e-05,
      "loss": 0.5666,
      "num_input_tokens_seen": 3009432,
      "step": 4585
    },
    {
      "epoch": 2.4056603773584904,
      "grad_norm": 0.15775416791439056,
      "learning_rate": 4.993752790976163e-05,
      "loss": 0.5481,
      "num_input_tokens_seen": 3014136,
      "step": 4590
    },
    {
      "epoch": 2.408280922431866,
      "grad_norm": 0.16124515235424042,
      "learning_rate": 4.993671746111977e-05,
      "loss": 0.589,
      "num_input_tokens_seen": 3017144,
      "step": 4595
    },
    {
      "epoch": 2.410901467505241,
      "grad_norm": 0.9814425110816956,
      "learning_rate": 4.993590179600111e-05,
      "loss": 0.7481,
      "num_input_tokens_seen": 3019448,
      "step": 4600
    },
    {
      "epoch": 2.4135220125786163,
      "grad_norm": 0.1764144003391266,
      "learning_rate": 4.9935080914576284e-05,
      "loss": 0.4571,
      "num_input_tokens_seen": 3022104,
      "step": 4605
    },
    {
      "epoch": 2.4161425576519915,
      "grad_norm": 0.19275425374507904,
      "learning_rate": 4.9934254817016995e-05,
      "loss": 0.5962,
      "num_input_tokens_seen": 3025688,
      "step": 4610
    },
    {
      "epoch": 2.418763102725367,
      "grad_norm": 0.15990829467773438,
      "learning_rate": 4.993342350349607e-05,
      "loss": 0.4121,
      "num_input_tokens_seen": 3029816,
      "step": 4615
    },
    {
      "epoch": 2.4213836477987423,
      "grad_norm": 0.14037202298641205,
      "learning_rate": 4.9932586974187404e-05,
      "loss": 0.3982,
      "num_input_tokens_seen": 3032824,
      "step": 4620
    },
    {
      "epoch": 2.4240041928721174,
      "grad_norm": 0.17263446748256683,
      "learning_rate": 4.9931745229265986e-05,
      "loss": 0.6733,
      "num_input_tokens_seen": 3036632,
      "step": 4625
    },
    {
      "epoch": 2.4266247379454926,
      "grad_norm": 0.2113834172487259,
      "learning_rate": 4.99308982689079e-05,
      "loss": 0.5969,
      "num_input_tokens_seen": 3039384,
      "step": 4630
    },
    {
      "epoch": 2.4292452830188678,
      "grad_norm": 0.13098129630088806,
      "learning_rate": 4.993004609329033e-05,
      "loss": 0.5508,
      "num_input_tokens_seen": 3042296,
      "step": 4635
    },
    {
      "epoch": 2.431865828092243,
      "grad_norm": 0.18418657779693604,
      "learning_rate": 4.992918870259153e-05,
      "loss": 0.4925,
      "num_input_tokens_seen": 3045304,
      "step": 4640
    },
    {
      "epoch": 2.4344863731656186,
      "grad_norm": 0.22864530980587006,
      "learning_rate": 4.992832609699086e-05,
      "loss": 0.4443,
      "num_input_tokens_seen": 3048824,
      "step": 4645
    },
    {
      "epoch": 2.4371069182389937,
      "grad_norm": 0.13290788233280182,
      "learning_rate": 4.992745827666877e-05,
      "loss": 0.6046,
      "num_input_tokens_seen": 3052120,
      "step": 4650
    },
    {
      "epoch": 2.439727463312369,
      "grad_norm": 0.14405398070812225,
      "learning_rate": 4.99265852418068e-05,
      "loss": 0.5584,
      "num_input_tokens_seen": 3055384,
      "step": 4655
    },
    {
      "epoch": 2.442348008385744,
      "grad_norm": 0.12851689755916595,
      "learning_rate": 4.992570699258757e-05,
      "loss": 0.445,
      "num_input_tokens_seen": 3058328,
      "step": 4660
    },
    {
      "epoch": 2.4449685534591197,
      "grad_norm": 0.20466171205043793,
      "learning_rate": 4.99248235291948e-05,
      "loss": 0.5977,
      "num_input_tokens_seen": 3063512,
      "step": 4665
    },
    {
      "epoch": 2.447589098532495,
      "grad_norm": 0.18217210471630096,
      "learning_rate": 4.9923934851813314e-05,
      "loss": 0.5666,
      "num_input_tokens_seen": 3068056,
      "step": 4670
    },
    {
      "epoch": 2.45020964360587,
      "grad_norm": 0.17912468314170837,
      "learning_rate": 4.9923040960628996e-05,
      "loss": 0.4892,
      "num_input_tokens_seen": 3071608,
      "step": 4675
    },
    {
      "epoch": 2.452830188679245,
      "grad_norm": 0.32185354828834534,
      "learning_rate": 4.9922141855828856e-05,
      "loss": 0.5423,
      "num_input_tokens_seen": 3074872,
      "step": 4680
    },
    {
      "epoch": 2.4554507337526204,
      "grad_norm": 0.17795902490615845,
      "learning_rate": 4.992123753760096e-05,
      "loss": 0.5005,
      "num_input_tokens_seen": 3077720,
      "step": 4685
    },
    {
      "epoch": 2.458071278825996,
      "grad_norm": 0.1882069855928421,
      "learning_rate": 4.9920328006134484e-05,
      "loss": 0.6549,
      "num_input_tokens_seen": 3080664,
      "step": 4690
    },
    {
      "epoch": 2.460691823899371,
      "grad_norm": 0.29612475633621216,
      "learning_rate": 4.991941326161971e-05,
      "loss": 0.6577,
      "num_input_tokens_seen": 3084408,
      "step": 4695
    },
    {
      "epoch": 2.4633123689727463,
      "grad_norm": 0.14895449578762054,
      "learning_rate": 4.991849330424796e-05,
      "loss": 0.597,
      "num_input_tokens_seen": 3087640,
      "step": 4700
    },
    {
      "epoch": 2.4659329140461215,
      "grad_norm": 0.32350048422813416,
      "learning_rate": 4.9917568134211703e-05,
      "loss": 0.625,
      "num_input_tokens_seen": 3090520,
      "step": 4705
    },
    {
      "epoch": 2.468553459119497,
      "grad_norm": 0.17291124165058136,
      "learning_rate": 4.991663775170447e-05,
      "loss": 0.4571,
      "num_input_tokens_seen": 3093624,
      "step": 4710
    },
    {
      "epoch": 2.4711740041928723,
      "grad_norm": 0.1373242288827896,
      "learning_rate": 4.991570215692088e-05,
      "loss": 0.4962,
      "num_input_tokens_seen": 3096184,
      "step": 4715
    },
    {
      "epoch": 2.4737945492662474,
      "grad_norm": 0.18572978675365448,
      "learning_rate": 4.991476135005665e-05,
      "loss": 0.6272,
      "num_input_tokens_seen": 3099928,
      "step": 4720
    },
    {
      "epoch": 2.4764150943396226,
      "grad_norm": 0.3035193383693695,
      "learning_rate": 4.991381533130859e-05,
      "loss": 0.5715,
      "num_input_tokens_seen": 3102456,
      "step": 4725
    },
    {
      "epoch": 2.4790356394129978,
      "grad_norm": 0.5561485886573792,
      "learning_rate": 4.99128641008746e-05,
      "loss": 0.6564,
      "num_input_tokens_seen": 3104728,
      "step": 4730
    },
    {
      "epoch": 2.481656184486373,
      "grad_norm": 0.32556378841400146,
      "learning_rate": 4.9911907658953664e-05,
      "loss": 0.4903,
      "num_input_tokens_seen": 3107704,
      "step": 4735
    },
    {
      "epoch": 2.4842767295597485,
      "grad_norm": 0.15502867102622986,
      "learning_rate": 4.991094600574584e-05,
      "loss": 0.4691,
      "num_input_tokens_seen": 3110328,
      "step": 4740
    },
    {
      "epoch": 2.4868972746331237,
      "grad_norm": 0.14795690774917603,
      "learning_rate": 4.9909979141452324e-05,
      "loss": 0.562,
      "num_input_tokens_seen": 3113464,
      "step": 4745
    },
    {
      "epoch": 2.489517819706499,
      "grad_norm": 0.13571849465370178,
      "learning_rate": 4.9909007066275346e-05,
      "loss": 0.4606,
      "num_input_tokens_seen": 3117112,
      "step": 4750
    },
    {
      "epoch": 2.492138364779874,
      "grad_norm": 0.1438831388950348,
      "learning_rate": 4.990802978041828e-05,
      "loss": 0.5095,
      "num_input_tokens_seen": 3120280,
      "step": 4755
    },
    {
      "epoch": 2.4947589098532497,
      "grad_norm": 0.14734888076782227,
      "learning_rate": 4.990704728408555e-05,
      "loss": 0.5332,
      "num_input_tokens_seen": 3123928,
      "step": 4760
    },
    {
      "epoch": 2.497379454926625,
      "grad_norm": 0.1680797040462494,
      "learning_rate": 4.990605957748268e-05,
      "loss": 0.4637,
      "num_input_tokens_seen": 3126584,
      "step": 4765
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.27684473991394043,
      "learning_rate": 4.990506666081628e-05,
      "loss": 0.6905,
      "num_input_tokens_seen": 3129368,
      "step": 4770
    },
    {
      "epoch": 2.502620545073375,
      "grad_norm": 0.4948260486125946,
      "learning_rate": 4.990406853429408e-05,
      "loss": 0.6141,
      "num_input_tokens_seen": 3132248,
      "step": 4775
    },
    {
      "epoch": 2.5052410901467503,
      "grad_norm": 0.15455913543701172,
      "learning_rate": 4.990306519812485e-05,
      "loss": 0.4139,
      "num_input_tokens_seen": 3135320,
      "step": 4780
    },
    {
      "epoch": 2.507861635220126,
      "grad_norm": 0.21602225303649902,
      "learning_rate": 4.99020566525185e-05,
      "loss": 0.7393,
      "num_input_tokens_seen": 3139032,
      "step": 4785
    },
    {
      "epoch": 2.510482180293501,
      "grad_norm": 0.20285511016845703,
      "learning_rate": 4.9901042897685996e-05,
      "loss": 0.7142,
      "num_input_tokens_seen": 3141944,
      "step": 4790
    },
    {
      "epoch": 2.5131027253668763,
      "grad_norm": 0.29504361748695374,
      "learning_rate": 4.99000239338394e-05,
      "loss": 0.587,
      "num_input_tokens_seen": 3145400,
      "step": 4795
    },
    {
      "epoch": 2.5157232704402515,
      "grad_norm": 0.12135568261146545,
      "learning_rate": 4.989899976119187e-05,
      "loss": 0.6409,
      "num_input_tokens_seen": 3149496,
      "step": 4800
    },
    {
      "epoch": 2.518343815513627,
      "grad_norm": 0.15776917338371277,
      "learning_rate": 4.989797037995766e-05,
      "loss": 0.6513,
      "num_input_tokens_seen": 3152344,
      "step": 4805
    },
    {
      "epoch": 2.5209643605870022,
      "grad_norm": 0.16105860471725464,
      "learning_rate": 4.989693579035209e-05,
      "loss": 0.4587,
      "num_input_tokens_seen": 3155192,
      "step": 4810
    },
    {
      "epoch": 2.5235849056603774,
      "grad_norm": 0.20254339277744293,
      "learning_rate": 4.989589599259159e-05,
      "loss": 0.4249,
      "num_input_tokens_seen": 3157688,
      "step": 4815
    },
    {
      "epoch": 2.5262054507337526,
      "grad_norm": 0.08982880413532257,
      "learning_rate": 4.9894850986893685e-05,
      "loss": 0.5381,
      "num_input_tokens_seen": 3161592,
      "step": 4820
    },
    {
      "epoch": 2.5288259958071277,
      "grad_norm": 0.2423134744167328,
      "learning_rate": 4.9893800773476965e-05,
      "loss": 0.5268,
      "num_input_tokens_seen": 3164024,
      "step": 4825
    },
    {
      "epoch": 2.531446540880503,
      "grad_norm": 0.23665232956409454,
      "learning_rate": 4.9892745352561134e-05,
      "loss": 0.4907,
      "num_input_tokens_seen": 3166424,
      "step": 4830
    },
    {
      "epoch": 2.5340670859538785,
      "grad_norm": 0.2353820949792862,
      "learning_rate": 4.9891684724366955e-05,
      "loss": 0.7441,
      "num_input_tokens_seen": 3169240,
      "step": 4835
    },
    {
      "epoch": 2.5366876310272537,
      "grad_norm": 0.23696815967559814,
      "learning_rate": 4.989061888911631e-05,
      "loss": 0.5608,
      "num_input_tokens_seen": 3172088,
      "step": 4840
    },
    {
      "epoch": 2.539308176100629,
      "grad_norm": 0.23276285827159882,
      "learning_rate": 4.988954784703217e-05,
      "loss": 0.5819,
      "num_input_tokens_seen": 3175480,
      "step": 4845
    },
    {
      "epoch": 2.541928721174004,
      "grad_norm": 0.17062059044837952,
      "learning_rate": 4.9888471598338574e-05,
      "loss": 0.5917,
      "num_input_tokens_seen": 3179480,
      "step": 4850
    },
    {
      "epoch": 2.5445492662473796,
      "grad_norm": 0.13451500236988068,
      "learning_rate": 4.988739014326066e-05,
      "loss": 0.5212,
      "num_input_tokens_seen": 3182520,
      "step": 4855
    },
    {
      "epoch": 2.547169811320755,
      "grad_norm": 0.21446633338928223,
      "learning_rate": 4.9886303482024656e-05,
      "loss": 0.5825,
      "num_input_tokens_seen": 3185560,
      "step": 4860
    },
    {
      "epoch": 2.54979035639413,
      "grad_norm": 0.16312311589717865,
      "learning_rate": 4.988521161485789e-05,
      "loss": 0.5363,
      "num_input_tokens_seen": 3189592,
      "step": 4865
    },
    {
      "epoch": 2.552410901467505,
      "grad_norm": 0.27326318621635437,
      "learning_rate": 4.9884114541988747e-05,
      "loss": 0.4686,
      "num_input_tokens_seen": 3192760,
      "step": 4870
    },
    {
      "epoch": 2.5550314465408803,
      "grad_norm": 0.5858509540557861,
      "learning_rate": 4.988301226364674e-05,
      "loss": 0.4713,
      "num_input_tokens_seen": 3195992,
      "step": 4875
    },
    {
      "epoch": 2.5576519916142555,
      "grad_norm": 0.2623160183429718,
      "learning_rate": 4.988190478006245e-05,
      "loss": 0.5293,
      "num_input_tokens_seen": 3198200,
      "step": 4880
    },
    {
      "epoch": 2.560272536687631,
      "grad_norm": 0.13763396441936493,
      "learning_rate": 4.988079209146754e-05,
      "loss": 0.5276,
      "num_input_tokens_seen": 3201848,
      "step": 4885
    },
    {
      "epoch": 2.5628930817610063,
      "grad_norm": 0.5128800272941589,
      "learning_rate": 4.9879674198094794e-05,
      "loss": 0.7245,
      "num_input_tokens_seen": 3204504,
      "step": 4890
    },
    {
      "epoch": 2.5655136268343814,
      "grad_norm": 0.210112527012825,
      "learning_rate": 4.987855110017804e-05,
      "loss": 0.6924,
      "num_input_tokens_seen": 3207320,
      "step": 4895
    },
    {
      "epoch": 2.568134171907757,
      "grad_norm": 0.1798069179058075,
      "learning_rate": 4.9877422797952224e-05,
      "loss": 0.502,
      "num_input_tokens_seen": 3210584,
      "step": 4900
    },
    {
      "epoch": 2.5707547169811322,
      "grad_norm": 0.15442593395709991,
      "learning_rate": 4.987628929165338e-05,
      "loss": 0.4314,
      "num_input_tokens_seen": 3213496,
      "step": 4905
    },
    {
      "epoch": 2.5733752620545074,
      "grad_norm": 0.1414920538663864,
      "learning_rate": 4.9875150581518615e-05,
      "loss": 0.5165,
      "num_input_tokens_seen": 3217016,
      "step": 4910
    },
    {
      "epoch": 2.5759958071278826,
      "grad_norm": 0.1816764622926712,
      "learning_rate": 4.987400666778614e-05,
      "loss": 0.6721,
      "num_input_tokens_seen": 3219800,
      "step": 4915
    },
    {
      "epoch": 2.5786163522012577,
      "grad_norm": 0.24908120930194855,
      "learning_rate": 4.987285755069524e-05,
      "loss": 0.5574,
      "num_input_tokens_seen": 3222424,
      "step": 4920
    },
    {
      "epoch": 2.581236897274633,
      "grad_norm": 0.22848233580589294,
      "learning_rate": 4.987170323048632e-05,
      "loss": 0.5463,
      "num_input_tokens_seen": 3225272,
      "step": 4925
    },
    {
      "epoch": 2.5838574423480085,
      "grad_norm": 0.12899158895015717,
      "learning_rate": 4.987054370740083e-05,
      "loss": 0.5119,
      "num_input_tokens_seen": 3228088,
      "step": 4930
    },
    {
      "epoch": 2.5864779874213837,
      "grad_norm": 0.1288922280073166,
      "learning_rate": 4.986937898168134e-05,
      "loss": 0.5211,
      "num_input_tokens_seen": 3231096,
      "step": 4935
    },
    {
      "epoch": 2.589098532494759,
      "grad_norm": 0.20535406470298767,
      "learning_rate": 4.986820905357148e-05,
      "loss": 0.6316,
      "num_input_tokens_seen": 3234552,
      "step": 4940
    },
    {
      "epoch": 2.591719077568134,
      "grad_norm": 0.3658965229988098,
      "learning_rate": 4.9867033923315995e-05,
      "loss": 0.433,
      "num_input_tokens_seen": 3238072,
      "step": 4945
    },
    {
      "epoch": 2.5943396226415096,
      "grad_norm": 0.23066486418247223,
      "learning_rate": 4.9865853591160724e-05,
      "loss": 0.5939,
      "num_input_tokens_seen": 3241624,
      "step": 4950
    },
    {
      "epoch": 2.596960167714885,
      "grad_norm": 0.19529950618743896,
      "learning_rate": 4.9864668057352556e-05,
      "loss": 0.8317,
      "num_input_tokens_seen": 3245080,
      "step": 4955
    },
    {
      "epoch": 2.59958071278826,
      "grad_norm": 0.16667501628398895,
      "learning_rate": 4.9863477322139517e-05,
      "loss": 0.601,
      "num_input_tokens_seen": 3247800,
      "step": 4960
    },
    {
      "epoch": 2.602201257861635,
      "grad_norm": 0.12480061501264572,
      "learning_rate": 4.986228138577067e-05,
      "loss": 0.4089,
      "num_input_tokens_seen": 3251000,
      "step": 4965
    },
    {
      "epoch": 2.6048218029350103,
      "grad_norm": 0.17270779609680176,
      "learning_rate": 4.986108024849619e-05,
      "loss": 0.4372,
      "num_input_tokens_seen": 3254872,
      "step": 4970
    },
    {
      "epoch": 2.6074423480083855,
      "grad_norm": 0.2192661315202713,
      "learning_rate": 4.985987391056737e-05,
      "loss": 0.5547,
      "num_input_tokens_seen": 3258520,
      "step": 4975
    },
    {
      "epoch": 2.610062893081761,
      "grad_norm": 0.2144969403743744,
      "learning_rate": 4.9858662372236544e-05,
      "loss": 0.5629,
      "num_input_tokens_seen": 3261944,
      "step": 4980
    },
    {
      "epoch": 2.6126834381551363,
      "grad_norm": 0.15817749500274658,
      "learning_rate": 4.985744563375714e-05,
      "loss": 0.6679,
      "num_input_tokens_seen": 3265432,
      "step": 4985
    },
    {
      "epoch": 2.6153039832285114,
      "grad_norm": 0.28907257318496704,
      "learning_rate": 4.985622369538371e-05,
      "loss": 0.5735,
      "num_input_tokens_seen": 3268632,
      "step": 4990
    },
    {
      "epoch": 2.617924528301887,
      "grad_norm": 0.3044603168964386,
      "learning_rate": 4.985499655737185e-05,
      "loss": 0.5274,
      "num_input_tokens_seen": 3271416,
      "step": 4995
    },
    {
      "epoch": 2.620545073375262,
      "grad_norm": 0.2371654361486435,
      "learning_rate": 4.985376421997827e-05,
      "loss": 0.4207,
      "num_input_tokens_seen": 3274424,
      "step": 5000
    },
    {
      "epoch": 2.6231656184486374,
      "grad_norm": 0.2006535530090332,
      "learning_rate": 4.985252668346076e-05,
      "loss": 0.5164,
      "num_input_tokens_seen": 3277688,
      "step": 5005
    },
    {
      "epoch": 2.6257861635220126,
      "grad_norm": 0.2281840443611145,
      "learning_rate": 4.9851283948078206e-05,
      "loss": 0.4427,
      "num_input_tokens_seen": 3280440,
      "step": 5010
    },
    {
      "epoch": 2.6284067085953877,
      "grad_norm": 0.3030897080898285,
      "learning_rate": 4.985003601409057e-05,
      "loss": 0.7065,
      "num_input_tokens_seen": 3283704,
      "step": 5015
    },
    {
      "epoch": 2.631027253668763,
      "grad_norm": 0.1618473082780838,
      "learning_rate": 4.9848782881758906e-05,
      "loss": 0.5304,
      "num_input_tokens_seen": 3286808,
      "step": 5020
    },
    {
      "epoch": 2.6336477987421385,
      "grad_norm": 0.1349976509809494,
      "learning_rate": 4.9847524551345345e-05,
      "loss": 0.6564,
      "num_input_tokens_seen": 3289592,
      "step": 5025
    },
    {
      "epoch": 2.6362683438155137,
      "grad_norm": 0.1879965215921402,
      "learning_rate": 4.984626102311313e-05,
      "loss": 0.5748,
      "num_input_tokens_seen": 3292696,
      "step": 5030
    },
    {
      "epoch": 2.638888888888889,
      "grad_norm": 0.16415823996067047,
      "learning_rate": 4.984499229732656e-05,
      "loss": 0.3747,
      "num_input_tokens_seen": 3296632,
      "step": 5035
    },
    {
      "epoch": 2.641509433962264,
      "grad_norm": 0.228614941239357,
      "learning_rate": 4.984371837425106e-05,
      "loss": 0.4557,
      "num_input_tokens_seen": 3300536,
      "step": 5040
    },
    {
      "epoch": 2.6441299790356396,
      "grad_norm": 0.18359452486038208,
      "learning_rate": 4.9842439254153106e-05,
      "loss": 0.5125,
      "num_input_tokens_seen": 3304728,
      "step": 5045
    },
    {
      "epoch": 2.646750524109015,
      "grad_norm": 0.20619283616542816,
      "learning_rate": 4.984115493730027e-05,
      "loss": 0.5996,
      "num_input_tokens_seen": 3307928,
      "step": 5050
    },
    {
      "epoch": 2.64937106918239,
      "grad_norm": 0.2513817250728607,
      "learning_rate": 4.983986542396123e-05,
      "loss": 0.6172,
      "num_input_tokens_seen": 3310552,
      "step": 5055
    },
    {
      "epoch": 2.651991614255765,
      "grad_norm": 0.18185825645923615,
      "learning_rate": 4.983857071440573e-05,
      "loss": 0.6559,
      "num_input_tokens_seen": 3314904,
      "step": 5060
    },
    {
      "epoch": 2.6546121593291403,
      "grad_norm": 0.1893380582332611,
      "learning_rate": 4.9837270808904616e-05,
      "loss": 0.5565,
      "num_input_tokens_seen": 3320184,
      "step": 5065
    },
    {
      "epoch": 2.6572327044025155,
      "grad_norm": 0.2390594631433487,
      "learning_rate": 4.9835965707729807e-05,
      "loss": 0.4857,
      "num_input_tokens_seen": 3322904,
      "step": 5070
    },
    {
      "epoch": 2.659853249475891,
      "grad_norm": 0.13592272996902466,
      "learning_rate": 4.983465541115431e-05,
      "loss": 0.4809,
      "num_input_tokens_seen": 3325784,
      "step": 5075
    },
    {
      "epoch": 2.6624737945492662,
      "grad_norm": 0.1577116996049881,
      "learning_rate": 4.983333991945223e-05,
      "loss": 0.601,
      "num_input_tokens_seen": 3328760,
      "step": 5080
    },
    {
      "epoch": 2.6650943396226414,
      "grad_norm": 0.25634023547172546,
      "learning_rate": 4.983201923289875e-05,
      "loss": 0.5666,
      "num_input_tokens_seen": 3331576,
      "step": 5085
    },
    {
      "epoch": 2.667714884696017,
      "grad_norm": 0.28389838337898254,
      "learning_rate": 4.983069335177014e-05,
      "loss": 0.6254,
      "num_input_tokens_seen": 3333880,
      "step": 5090
    },
    {
      "epoch": 2.670335429769392,
      "grad_norm": 0.27650898694992065,
      "learning_rate": 4.982936227634378e-05,
      "loss": 0.551,
      "num_input_tokens_seen": 3337880,
      "step": 5095
    },
    {
      "epoch": 2.6729559748427674,
      "grad_norm": 0.1752854287624359,
      "learning_rate": 4.982802600689809e-05,
      "loss": 0.5566,
      "num_input_tokens_seen": 3340920,
      "step": 5100
    },
    {
      "epoch": 2.6755765199161425,
      "grad_norm": 0.2359677106142044,
      "learning_rate": 4.982668454371261e-05,
      "loss": 0.5309,
      "num_input_tokens_seen": 3344312,
      "step": 5105
    },
    {
      "epoch": 2.6781970649895177,
      "grad_norm": 0.16136693954467773,
      "learning_rate": 4.9825337887067966e-05,
      "loss": 0.4626,
      "num_input_tokens_seen": 3347672,
      "step": 5110
    },
    {
      "epoch": 2.680817610062893,
      "grad_norm": 0.14028611779212952,
      "learning_rate": 4.982398603724586e-05,
      "loss": 0.552,
      "num_input_tokens_seen": 3351032,
      "step": 5115
    },
    {
      "epoch": 2.6834381551362685,
      "grad_norm": 0.21488098800182343,
      "learning_rate": 4.982262899452909e-05,
      "loss": 0.7192,
      "num_input_tokens_seen": 3353752,
      "step": 5120
    },
    {
      "epoch": 2.6860587002096437,
      "grad_norm": 0.2104182243347168,
      "learning_rate": 4.982126675920151e-05,
      "loss": 0.4807,
      "num_input_tokens_seen": 3356600,
      "step": 5125
    },
    {
      "epoch": 2.688679245283019,
      "grad_norm": 0.2783065438270569,
      "learning_rate": 4.98198993315481e-05,
      "loss": 0.5858,
      "num_input_tokens_seen": 3360280,
      "step": 5130
    },
    {
      "epoch": 2.691299790356394,
      "grad_norm": 0.16745951771736145,
      "learning_rate": 4.981852671185492e-05,
      "loss": 0.644,
      "num_input_tokens_seen": 3363768,
      "step": 5135
    },
    {
      "epoch": 2.6939203354297696,
      "grad_norm": 0.274097204208374,
      "learning_rate": 4.9817148900409093e-05,
      "loss": 0.4646,
      "num_input_tokens_seen": 3367480,
      "step": 5140
    },
    {
      "epoch": 2.6965408805031448,
      "grad_norm": 0.38218408823013306,
      "learning_rate": 4.981576589749885e-05,
      "loss": 0.5696,
      "num_input_tokens_seen": 3370520,
      "step": 5145
    },
    {
      "epoch": 2.69916142557652,
      "grad_norm": 0.343828409910202,
      "learning_rate": 4.981437770341348e-05,
      "loss": 0.493,
      "num_input_tokens_seen": 3372984,
      "step": 5150
    },
    {
      "epoch": 2.701781970649895,
      "grad_norm": 0.485529363155365,
      "learning_rate": 4.9812984318443414e-05,
      "loss": 0.5241,
      "num_input_tokens_seen": 3375288,
      "step": 5155
    },
    {
      "epoch": 2.7044025157232703,
      "grad_norm": 0.19717003405094147,
      "learning_rate": 4.98115857428801e-05,
      "loss": 0.4465,
      "num_input_tokens_seen": 3378488,
      "step": 5160
    },
    {
      "epoch": 2.7070230607966455,
      "grad_norm": 0.157302588224411,
      "learning_rate": 4.9810181977016104e-05,
      "loss": 0.4954,
      "num_input_tokens_seen": 3381368,
      "step": 5165
    },
    {
      "epoch": 2.709643605870021,
      "grad_norm": 0.41158822178840637,
      "learning_rate": 4.9808773021145096e-05,
      "loss": 0.5405,
      "num_input_tokens_seen": 3385016,
      "step": 5170
    },
    {
      "epoch": 2.7122641509433962,
      "grad_norm": 0.1590348333120346,
      "learning_rate": 4.98073588755618e-05,
      "loss": 0.6054,
      "num_input_tokens_seen": 3387992,
      "step": 5175
    },
    {
      "epoch": 2.7148846960167714,
      "grad_norm": 0.17711110413074493,
      "learning_rate": 4.980593954056205e-05,
      "loss": 0.599,
      "num_input_tokens_seen": 3390872,
      "step": 5180
    },
    {
      "epoch": 2.717505241090147,
      "grad_norm": 0.18998660147190094,
      "learning_rate": 4.9804515016442754e-05,
      "loss": 0.6581,
      "num_input_tokens_seen": 3393496,
      "step": 5185
    },
    {
      "epoch": 2.720125786163522,
      "grad_norm": 0.3109906315803528,
      "learning_rate": 4.9803085303501896e-05,
      "loss": 0.7762,
      "num_input_tokens_seen": 3395992,
      "step": 5190
    },
    {
      "epoch": 2.7227463312368974,
      "grad_norm": 0.19552798569202423,
      "learning_rate": 4.9801650402038555e-05,
      "loss": 0.6548,
      "num_input_tokens_seen": 3399160,
      "step": 5195
    },
    {
      "epoch": 2.7253668763102725,
      "grad_norm": 0.19212988018989563,
      "learning_rate": 4.980021031235291e-05,
      "loss": 0.5362,
      "num_input_tokens_seen": 3402072,
      "step": 5200
    },
    {
      "epoch": 2.7279874213836477,
      "grad_norm": 0.29620927572250366,
      "learning_rate": 4.9798765034746206e-05,
      "loss": 0.5531,
      "num_input_tokens_seen": 3404344,
      "step": 5205
    },
    {
      "epoch": 2.730607966457023,
      "grad_norm": 0.18025889992713928,
      "learning_rate": 4.979731456952077e-05,
      "loss": 0.5196,
      "num_input_tokens_seen": 3407640,
      "step": 5210
    },
    {
      "epoch": 2.7332285115303985,
      "grad_norm": 0.2408881038427353,
      "learning_rate": 4.9795858916980034e-05,
      "loss": 0.4213,
      "num_input_tokens_seen": 3410200,
      "step": 5215
    },
    {
      "epoch": 2.7358490566037736,
      "grad_norm": 0.5496697425842285,
      "learning_rate": 4.9794398077428495e-05,
      "loss": 0.4951,
      "num_input_tokens_seen": 3412856,
      "step": 5220
    },
    {
      "epoch": 2.738469601677149,
      "grad_norm": 0.1075938269495964,
      "learning_rate": 4.979293205117175e-05,
      "loss": 0.5119,
      "num_input_tokens_seen": 3415864,
      "step": 5225
    },
    {
      "epoch": 2.741090146750524,
      "grad_norm": 0.1333715319633484,
      "learning_rate": 4.979146083851647e-05,
      "loss": 0.5512,
      "num_input_tokens_seen": 3419224,
      "step": 5230
    },
    {
      "epoch": 2.7437106918238996,
      "grad_norm": 0.2648145854473114,
      "learning_rate": 4.978998443977042e-05,
      "loss": 0.4868,
      "num_input_tokens_seen": 3422520,
      "step": 5235
    },
    {
      "epoch": 2.7463312368972748,
      "grad_norm": 0.21819467842578888,
      "learning_rate": 4.9788502855242436e-05,
      "loss": 0.507,
      "num_input_tokens_seen": 3425976,
      "step": 5240
    },
    {
      "epoch": 2.74895178197065,
      "grad_norm": 0.1475064605474472,
      "learning_rate": 4.978701608524247e-05,
      "loss": 0.5366,
      "num_input_tokens_seen": 3429368,
      "step": 5245
    },
    {
      "epoch": 2.751572327044025,
      "grad_norm": 0.21297402679920197,
      "learning_rate": 4.978552413008152e-05,
      "loss": 0.5773,
      "num_input_tokens_seen": 3434680,
      "step": 5250
    },
    {
      "epoch": 2.7541928721174003,
      "grad_norm": 0.4070872664451599,
      "learning_rate": 4.9784026990071687e-05,
      "loss": 0.6404,
      "num_input_tokens_seen": 3437752,
      "step": 5255
    },
    {
      "epoch": 2.7568134171907754,
      "grad_norm": 0.12593621015548706,
      "learning_rate": 4.978252466552615e-05,
      "loss": 0.6571,
      "num_input_tokens_seen": 3440760,
      "step": 5260
    },
    {
      "epoch": 2.759433962264151,
      "grad_norm": 0.22161827981472015,
      "learning_rate": 4.9781017156759205e-05,
      "loss": 0.5481,
      "num_input_tokens_seen": 3443736,
      "step": 5265
    },
    {
      "epoch": 2.762054507337526,
      "grad_norm": 0.3033123016357422,
      "learning_rate": 4.977950446408618e-05,
      "loss": 0.6599,
      "num_input_tokens_seen": 3446904,
      "step": 5270
    },
    {
      "epoch": 2.7646750524109014,
      "grad_norm": 0.18063236773014069,
      "learning_rate": 4.9777986587823523e-05,
      "loss": 0.4795,
      "num_input_tokens_seen": 3450104,
      "step": 5275
    },
    {
      "epoch": 2.767295597484277,
      "grad_norm": 0.14093194901943207,
      "learning_rate": 4.977646352828875e-05,
      "loss": 0.7122,
      "num_input_tokens_seen": 3453592,
      "step": 5280
    },
    {
      "epoch": 2.769916142557652,
      "grad_norm": 0.3062164783477783,
      "learning_rate": 4.9774935285800476e-05,
      "loss": 0.545,
      "num_input_tokens_seen": 3456824,
      "step": 5285
    },
    {
      "epoch": 2.7725366876310273,
      "grad_norm": 0.3482477068901062,
      "learning_rate": 4.977340186067838e-05,
      "loss": 0.5564,
      "num_input_tokens_seen": 3460792,
      "step": 5290
    },
    {
      "epoch": 2.7751572327044025,
      "grad_norm": 0.17622068524360657,
      "learning_rate": 4.977186325324325e-05,
      "loss": 0.6146,
      "num_input_tokens_seen": 3464280,
      "step": 5295
    },
    {
      "epoch": 2.7777777777777777,
      "grad_norm": 0.4419034421443939,
      "learning_rate": 4.977031946381694e-05,
      "loss": 0.6955,
      "num_input_tokens_seen": 3467320,
      "step": 5300
    },
    {
      "epoch": 2.780398322851153,
      "grad_norm": 0.19524917006492615,
      "learning_rate": 4.9768770492722394e-05,
      "loss": 0.5404,
      "num_input_tokens_seen": 3470520,
      "step": 5305
    },
    {
      "epoch": 2.7830188679245285,
      "grad_norm": 0.18999026715755463,
      "learning_rate": 4.976721634028363e-05,
      "loss": 0.5429,
      "num_input_tokens_seen": 3474008,
      "step": 5310
    },
    {
      "epoch": 2.7856394129979036,
      "grad_norm": 0.48916956782341003,
      "learning_rate": 4.976565700682578e-05,
      "loss": 0.508,
      "num_input_tokens_seen": 3477208,
      "step": 5315
    },
    {
      "epoch": 2.788259958071279,
      "grad_norm": 0.23268717527389526,
      "learning_rate": 4.976409249267502e-05,
      "loss": 0.5792,
      "num_input_tokens_seen": 3480088,
      "step": 5320
    },
    {
      "epoch": 2.790880503144654,
      "grad_norm": 0.10665013641119003,
      "learning_rate": 4.976252279815863e-05,
      "loss": 0.5448,
      "num_input_tokens_seen": 3484504,
      "step": 5325
    },
    {
      "epoch": 2.7935010482180296,
      "grad_norm": 0.23526600003242493,
      "learning_rate": 4.976094792360498e-05,
      "loss": 0.4926,
      "num_input_tokens_seen": 3487000,
      "step": 5330
    },
    {
      "epoch": 2.7961215932914047,
      "grad_norm": 0.2464539110660553,
      "learning_rate": 4.975936786934351e-05,
      "loss": 0.5079,
      "num_input_tokens_seen": 3490712,
      "step": 5335
    },
    {
      "epoch": 2.79874213836478,
      "grad_norm": 0.1809893548488617,
      "learning_rate": 4.975778263570475e-05,
      "loss": 0.5927,
      "num_input_tokens_seen": 3494296,
      "step": 5340
    },
    {
      "epoch": 2.801362683438155,
      "grad_norm": 0.14518924057483673,
      "learning_rate": 4.975619222302032e-05,
      "loss": 0.4758,
      "num_input_tokens_seen": 3498232,
      "step": 5345
    },
    {
      "epoch": 2.8039832285115303,
      "grad_norm": 0.2536754906177521,
      "learning_rate": 4.975459663162291e-05,
      "loss": 0.5217,
      "num_input_tokens_seen": 3501080,
      "step": 5350
    },
    {
      "epoch": 2.8066037735849054,
      "grad_norm": 0.21917900443077087,
      "learning_rate": 4.9752995861846296e-05,
      "loss": 0.4772,
      "num_input_tokens_seen": 3504024,
      "step": 5355
    },
    {
      "epoch": 2.809224318658281,
      "grad_norm": 0.222678080201149,
      "learning_rate": 4.975138991402535e-05,
      "loss": 0.4282,
      "num_input_tokens_seen": 3507256,
      "step": 5360
    },
    {
      "epoch": 2.811844863731656,
      "grad_norm": 0.1276940256357193,
      "learning_rate": 4.974977878849601e-05,
      "loss": 0.5791,
      "num_input_tokens_seen": 3510392,
      "step": 5365
    },
    {
      "epoch": 2.8144654088050314,
      "grad_norm": 0.19827067852020264,
      "learning_rate": 4.9748162485595303e-05,
      "loss": 0.5938,
      "num_input_tokens_seen": 3513272,
      "step": 5370
    },
    {
      "epoch": 2.8170859538784065,
      "grad_norm": 0.12525035440921783,
      "learning_rate": 4.974654100566135e-05,
      "loss": 0.6065,
      "num_input_tokens_seen": 3516312,
      "step": 5375
    },
    {
      "epoch": 2.819706498951782,
      "grad_norm": 0.1964031159877777,
      "learning_rate": 4.974491434903335e-05,
      "loss": 0.5555,
      "num_input_tokens_seen": 3519032,
      "step": 5380
    },
    {
      "epoch": 2.8223270440251573,
      "grad_norm": 0.3037385642528534,
      "learning_rate": 4.9743282516051556e-05,
      "loss": 0.4308,
      "num_input_tokens_seen": 3521720,
      "step": 5385
    },
    {
      "epoch": 2.8249475890985325,
      "grad_norm": 0.2761027216911316,
      "learning_rate": 4.974164550705736e-05,
      "loss": 0.5168,
      "num_input_tokens_seen": 3524888,
      "step": 5390
    },
    {
      "epoch": 2.8275681341719077,
      "grad_norm": 0.10071144998073578,
      "learning_rate": 4.974000332239319e-05,
      "loss": 0.4456,
      "num_input_tokens_seen": 3529688,
      "step": 5395
    },
    {
      "epoch": 2.830188679245283,
      "grad_norm": 0.150055393576622,
      "learning_rate": 4.973835596240257e-05,
      "loss": 0.5578,
      "num_input_tokens_seen": 3532440,
      "step": 5400
    },
    {
      "epoch": 2.832809224318658,
      "grad_norm": 0.20058925449848175,
      "learning_rate": 4.9736703427430114e-05,
      "loss": 0.5612,
      "num_input_tokens_seen": 3535288,
      "step": 5405
    },
    {
      "epoch": 2.8354297693920336,
      "grad_norm": 0.35558193922042847,
      "learning_rate": 4.9735045717821514e-05,
      "loss": 0.5616,
      "num_input_tokens_seen": 3538840,
      "step": 5410
    },
    {
      "epoch": 2.838050314465409,
      "grad_norm": 0.17069107294082642,
      "learning_rate": 4.973338283392354e-05,
      "loss": 0.4154,
      "num_input_tokens_seen": 3541880,
      "step": 5415
    },
    {
      "epoch": 2.840670859538784,
      "grad_norm": 0.19458119571208954,
      "learning_rate": 4.9731714776084046e-05,
      "loss": 0.565,
      "num_input_tokens_seen": 3545208,
      "step": 5420
    },
    {
      "epoch": 2.8432914046121596,
      "grad_norm": 0.2180483639240265,
      "learning_rate": 4.973004154465198e-05,
      "loss": 0.4865,
      "num_input_tokens_seen": 3548120,
      "step": 5425
    },
    {
      "epoch": 2.8459119496855347,
      "grad_norm": 0.20868085324764252,
      "learning_rate": 4.972836313997736e-05,
      "loss": 0.5849,
      "num_input_tokens_seen": 3551512,
      "step": 5430
    },
    {
      "epoch": 2.84853249475891,
      "grad_norm": 0.25992047786712646,
      "learning_rate": 4.9726679562411285e-05,
      "loss": 0.4397,
      "num_input_tokens_seen": 3554520,
      "step": 5435
    },
    {
      "epoch": 2.851153039832285,
      "grad_norm": 0.15713615715503693,
      "learning_rate": 4.972499081230595e-05,
      "loss": 0.636,
      "num_input_tokens_seen": 3557208,
      "step": 5440
    },
    {
      "epoch": 2.8537735849056602,
      "grad_norm": 0.31343376636505127,
      "learning_rate": 4.9723296890014605e-05,
      "loss": 0.7014,
      "num_input_tokens_seen": 3560600,
      "step": 5445
    },
    {
      "epoch": 2.8563941299790354,
      "grad_norm": 0.1319926232099533,
      "learning_rate": 4.9721597795891605e-05,
      "loss": 0.5117,
      "num_input_tokens_seen": 3565752,
      "step": 5450
    },
    {
      "epoch": 2.859014675052411,
      "grad_norm": 0.14631019532680511,
      "learning_rate": 4.9719893530292396e-05,
      "loss": 0.448,
      "num_input_tokens_seen": 3569400,
      "step": 5455
    },
    {
      "epoch": 2.861635220125786,
      "grad_norm": 0.7728410363197327,
      "learning_rate": 4.9718184093573475e-05,
      "loss": 0.5694,
      "num_input_tokens_seen": 3572088,
      "step": 5460
    },
    {
      "epoch": 2.8642557651991614,
      "grad_norm": 0.12339439988136292,
      "learning_rate": 4.9716469486092435e-05,
      "loss": 0.6383,
      "num_input_tokens_seen": 3575032,
      "step": 5465
    },
    {
      "epoch": 2.8668763102725365,
      "grad_norm": 0.20353026688098907,
      "learning_rate": 4.971474970820797e-05,
      "loss": 0.5268,
      "num_input_tokens_seen": 3578232,
      "step": 5470
    },
    {
      "epoch": 2.869496855345912,
      "grad_norm": 0.15554796159267426,
      "learning_rate": 4.971302476027982e-05,
      "loss": 0.5728,
      "num_input_tokens_seen": 3581304,
      "step": 5475
    },
    {
      "epoch": 2.8721174004192873,
      "grad_norm": 0.1422307938337326,
      "learning_rate": 4.971129464266884e-05,
      "loss": 0.4647,
      "num_input_tokens_seen": 3584664,
      "step": 5480
    },
    {
      "epoch": 2.8747379454926625,
      "grad_norm": 0.22500449419021606,
      "learning_rate": 4.970955935573693e-05,
      "loss": 0.6245,
      "num_input_tokens_seen": 3587544,
      "step": 5485
    },
    {
      "epoch": 2.8773584905660377,
      "grad_norm": 0.3439403772354126,
      "learning_rate": 4.9707818899847106e-05,
      "loss": 0.6095,
      "num_input_tokens_seen": 3590360,
      "step": 5490
    },
    {
      "epoch": 2.879979035639413,
      "grad_norm": 0.17850379645824432,
      "learning_rate": 4.970607327536346e-05,
      "loss": 0.4501,
      "num_input_tokens_seen": 3597208,
      "step": 5495
    },
    {
      "epoch": 2.882599580712788,
      "grad_norm": 0.2345161736011505,
      "learning_rate": 4.970432248265114e-05,
      "loss": 0.4286,
      "num_input_tokens_seen": 3600568,
      "step": 5500
    },
    {
      "epoch": 2.8852201257861636,
      "grad_norm": 0.530044674873352,
      "learning_rate": 4.97025665220764e-05,
      "loss": 0.6249,
      "num_input_tokens_seen": 3603480,
      "step": 5505
    },
    {
      "epoch": 2.8878406708595388,
      "grad_norm": 0.1800822913646698,
      "learning_rate": 4.970080539400656e-05,
      "loss": 0.6034,
      "num_input_tokens_seen": 3606360,
      "step": 5510
    },
    {
      "epoch": 2.890461215932914,
      "grad_norm": 0.37153249979019165,
      "learning_rate": 4.969903909881004e-05,
      "loss": 0.4848,
      "num_input_tokens_seen": 3608856,
      "step": 5515
    },
    {
      "epoch": 2.8930817610062896,
      "grad_norm": 0.1575583517551422,
      "learning_rate": 4.9697267636856317e-05,
      "loss": 0.4365,
      "num_input_tokens_seen": 3611384,
      "step": 5520
    },
    {
      "epoch": 2.8957023060796647,
      "grad_norm": 0.1930665224790573,
      "learning_rate": 4.969549100851597e-05,
      "loss": 0.7645,
      "num_input_tokens_seen": 3614488,
      "step": 5525
    },
    {
      "epoch": 2.89832285115304,
      "grad_norm": 0.2354651391506195,
      "learning_rate": 4.9693709214160645e-05,
      "loss": 0.5304,
      "num_input_tokens_seen": 3617080,
      "step": 5530
    },
    {
      "epoch": 2.900943396226415,
      "grad_norm": 0.29613035917282104,
      "learning_rate": 4.969192225416307e-05,
      "loss": 0.6917,
      "num_input_tokens_seen": 3620728,
      "step": 5535
    },
    {
      "epoch": 2.9035639412997902,
      "grad_norm": 0.19069264829158783,
      "learning_rate": 4.969013012889706e-05,
      "loss": 0.5887,
      "num_input_tokens_seen": 3624184,
      "step": 5540
    },
    {
      "epoch": 2.9061844863731654,
      "grad_norm": 0.15302002429962158,
      "learning_rate": 4.9688332838737504e-05,
      "loss": 0.4348,
      "num_input_tokens_seen": 3628856,
      "step": 5545
    },
    {
      "epoch": 2.908805031446541,
      "grad_norm": 0.1689635068178177,
      "learning_rate": 4.9686530384060386e-05,
      "loss": 0.631,
      "num_input_tokens_seen": 3631800,
      "step": 5550
    },
    {
      "epoch": 2.911425576519916,
      "grad_norm": 0.4627755880355835,
      "learning_rate": 4.968472276524275e-05,
      "loss": 0.5259,
      "num_input_tokens_seen": 3634680,
      "step": 5555
    },
    {
      "epoch": 2.9140461215932913,
      "grad_norm": 0.14026138186454773,
      "learning_rate": 4.9682909982662725e-05,
      "loss": 0.4559,
      "num_input_tokens_seen": 3638264,
      "step": 5560
    },
    {
      "epoch": 2.9166666666666665,
      "grad_norm": 0.21168199181556702,
      "learning_rate": 4.968109203669953e-05,
      "loss": 0.4887,
      "num_input_tokens_seen": 3641240,
      "step": 5565
    },
    {
      "epoch": 2.919287211740042,
      "grad_norm": 0.15447571873664856,
      "learning_rate": 4.967926892773346e-05,
      "loss": 0.6589,
      "num_input_tokens_seen": 3643640,
      "step": 5570
    },
    {
      "epoch": 2.9219077568134173,
      "grad_norm": 0.13826793432235718,
      "learning_rate": 4.967744065614588e-05,
      "loss": 0.5401,
      "num_input_tokens_seen": 3648088,
      "step": 5575
    },
    {
      "epoch": 2.9245283018867925,
      "grad_norm": 0.13572581112384796,
      "learning_rate": 4.9675607222319255e-05,
      "loss": 0.4906,
      "num_input_tokens_seen": 3651352,
      "step": 5580
    },
    {
      "epoch": 2.9271488469601676,
      "grad_norm": 0.1567004919052124,
      "learning_rate": 4.967376862663711e-05,
      "loss": 0.5544,
      "num_input_tokens_seen": 3655288,
      "step": 5585
    },
    {
      "epoch": 2.929769392033543,
      "grad_norm": 0.17770683765411377,
      "learning_rate": 4.9671924869484064e-05,
      "loss": 0.4981,
      "num_input_tokens_seen": 3657624,
      "step": 5590
    },
    {
      "epoch": 2.932389937106918,
      "grad_norm": 0.1701856106519699,
      "learning_rate": 4.967007595124581e-05,
      "loss": 0.4285,
      "num_input_tokens_seen": 3660472,
      "step": 5595
    },
    {
      "epoch": 2.9350104821802936,
      "grad_norm": 0.13465195894241333,
      "learning_rate": 4.966822187230911e-05,
      "loss": 0.6329,
      "num_input_tokens_seen": 3663992,
      "step": 5600
    },
    {
      "epoch": 2.9376310272536688,
      "grad_norm": 0.21256372332572937,
      "learning_rate": 4.966636263306183e-05,
      "loss": 0.5753,
      "num_input_tokens_seen": 3666424,
      "step": 5605
    },
    {
      "epoch": 2.940251572327044,
      "grad_norm": 0.20333606004714966,
      "learning_rate": 4.9664498233892895e-05,
      "loss": 0.5919,
      "num_input_tokens_seen": 3669400,
      "step": 5610
    },
    {
      "epoch": 2.9428721174004195,
      "grad_norm": 0.2003977745771408,
      "learning_rate": 4.966262867519232e-05,
      "loss": 0.6694,
      "num_input_tokens_seen": 3672440,
      "step": 5615
    },
    {
      "epoch": 2.9454926624737947,
      "grad_norm": 0.38744738698005676,
      "learning_rate": 4.966075395735119e-05,
      "loss": 0.447,
      "num_input_tokens_seen": 3675704,
      "step": 5620
    },
    {
      "epoch": 2.94811320754717,
      "grad_norm": 0.22707951068878174,
      "learning_rate": 4.965887408076167e-05,
      "loss": 0.7138,
      "num_input_tokens_seen": 3679160,
      "step": 5625
    },
    {
      "epoch": 2.950733752620545,
      "grad_norm": 0.14909060299396515,
      "learning_rate": 4.965698904581703e-05,
      "loss": 0.5045,
      "num_input_tokens_seen": 3682328,
      "step": 5630
    },
    {
      "epoch": 2.95335429769392,
      "grad_norm": 0.22221484780311584,
      "learning_rate": 4.965509885291157e-05,
      "loss": 0.4889,
      "num_input_tokens_seen": 3685240,
      "step": 5635
    },
    {
      "epoch": 2.9559748427672954,
      "grad_norm": 0.20435240864753723,
      "learning_rate": 4.9653203502440716e-05,
      "loss": 0.7006,
      "num_input_tokens_seen": 3687960,
      "step": 5640
    },
    {
      "epoch": 2.958595387840671,
      "grad_norm": 0.17057465016841888,
      "learning_rate": 4.965130299480095e-05,
      "loss": 0.4491,
      "num_input_tokens_seen": 3691704,
      "step": 5645
    },
    {
      "epoch": 2.961215932914046,
      "grad_norm": 0.22819432616233826,
      "learning_rate": 4.9649397330389834e-05,
      "loss": 0.5769,
      "num_input_tokens_seen": 3695416,
      "step": 5650
    },
    {
      "epoch": 2.9638364779874213,
      "grad_norm": 0.31380435824394226,
      "learning_rate": 4.964748650960601e-05,
      "loss": 0.507,
      "num_input_tokens_seen": 3698168,
      "step": 5655
    },
    {
      "epoch": 2.9664570230607965,
      "grad_norm": 0.2667488157749176,
      "learning_rate": 4.96455705328492e-05,
      "loss": 0.6797,
      "num_input_tokens_seen": 3700984,
      "step": 5660
    },
    {
      "epoch": 2.969077568134172,
      "grad_norm": 0.22493256628513336,
      "learning_rate": 4.964364940052021e-05,
      "loss": 0.7233,
      "num_input_tokens_seen": 3704472,
      "step": 5665
    },
    {
      "epoch": 2.9716981132075473,
      "grad_norm": 0.2808436453342438,
      "learning_rate": 4.964172311302091e-05,
      "loss": 0.4293,
      "num_input_tokens_seen": 3707960,
      "step": 5670
    },
    {
      "epoch": 2.9743186582809225,
      "grad_norm": 0.28673985600471497,
      "learning_rate": 4.9639791670754265e-05,
      "loss": 0.5168,
      "num_input_tokens_seen": 3710520,
      "step": 5675
    },
    {
      "epoch": 2.9769392033542976,
      "grad_norm": 0.1857220083475113,
      "learning_rate": 4.9637855074124305e-05,
      "loss": 0.5568,
      "num_input_tokens_seen": 3713688,
      "step": 5680
    },
    {
      "epoch": 2.979559748427673,
      "grad_norm": 0.16747218370437622,
      "learning_rate": 4.963591332353615e-05,
      "loss": 0.4758,
      "num_input_tokens_seen": 3717304,
      "step": 5685
    },
    {
      "epoch": 2.982180293501048,
      "grad_norm": 0.13526812195777893,
      "learning_rate": 4.9633966419396e-05,
      "loss": 0.6062,
      "num_input_tokens_seen": 3720888,
      "step": 5690
    },
    {
      "epoch": 2.9848008385744236,
      "grad_norm": 0.1924818754196167,
      "learning_rate": 4.96320143621111e-05,
      "loss": 0.6024,
      "num_input_tokens_seen": 3724888,
      "step": 5695
    },
    {
      "epoch": 2.9874213836477987,
      "grad_norm": 0.22534741461277008,
      "learning_rate": 4.963005715208981e-05,
      "loss": 0.4459,
      "num_input_tokens_seen": 3727288,
      "step": 5700
    },
    {
      "epoch": 2.990041928721174,
      "grad_norm": 0.26666897535324097,
      "learning_rate": 4.962809478974157e-05,
      "loss": 0.6277,
      "num_input_tokens_seen": 3730328,
      "step": 5705
    },
    {
      "epoch": 2.9926624737945495,
      "grad_norm": 0.2594793140888214,
      "learning_rate": 4.962612727547686e-05,
      "loss": 0.4846,
      "num_input_tokens_seen": 3733912,
      "step": 5710
    },
    {
      "epoch": 2.9952830188679247,
      "grad_norm": 0.172962948679924,
      "learning_rate": 4.962415460970728e-05,
      "loss": 0.4984,
      "num_input_tokens_seen": 3737464,
      "step": 5715
    },
    {
      "epoch": 2.9979035639413,
      "grad_norm": 0.16008783876895905,
      "learning_rate": 4.962217679284548e-05,
      "loss": 0.4952,
      "num_input_tokens_seen": 3740152,
      "step": 5720
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.5438814759254456,
      "eval_runtime": 10.0209,
      "eval_samples_per_second": 84.623,
      "eval_steps_per_second": 21.156,
      "num_input_tokens_seen": 3742552,
      "step": 5724
    },
    {
      "epoch": 3.000524109014675,
      "grad_norm": 0.2688823640346527,
      "learning_rate": 4.962019382530521e-05,
      "loss": 0.4829,
      "num_input_tokens_seen": 3742968,
      "step": 5725
    },
    {
      "epoch": 3.00314465408805,
      "grad_norm": 0.15417318046092987,
      "learning_rate": 4.9618205707501256e-05,
      "loss": 0.5346,
      "num_input_tokens_seen": 3745880,
      "step": 5730
    },
    {
      "epoch": 3.0057651991614254,
      "grad_norm": 0.1282328963279724,
      "learning_rate": 4.961621243984954e-05,
      "loss": 0.626,
      "num_input_tokens_seen": 3749464,
      "step": 5735
    },
    {
      "epoch": 3.008385744234801,
      "grad_norm": 0.25924673676490784,
      "learning_rate": 4.9614214022767014e-05,
      "loss": 0.4999,
      "num_input_tokens_seen": 3752248,
      "step": 5740
    },
    {
      "epoch": 3.011006289308176,
      "grad_norm": 0.23259757459163666,
      "learning_rate": 4.9612210456671735e-05,
      "loss": 0.5369,
      "num_input_tokens_seen": 3755480,
      "step": 5745
    },
    {
      "epoch": 3.0136268343815513,
      "grad_norm": 0.13589568436145782,
      "learning_rate": 4.961020174198281e-05,
      "loss": 0.4737,
      "num_input_tokens_seen": 3758040,
      "step": 5750
    },
    {
      "epoch": 3.0162473794549265,
      "grad_norm": 0.5279970169067383,
      "learning_rate": 4.9608187879120454e-05,
      "loss": 0.5078,
      "num_input_tokens_seen": 3761368,
      "step": 5755
    },
    {
      "epoch": 3.018867924528302,
      "grad_norm": 0.22131891548633575,
      "learning_rate": 4.9606168868505936e-05,
      "loss": 0.5333,
      "num_input_tokens_seen": 3764824,
      "step": 5760
    },
    {
      "epoch": 3.0214884696016773,
      "grad_norm": 0.16204112768173218,
      "learning_rate": 4.9604144710561616e-05,
      "loss": 0.4502,
      "num_input_tokens_seen": 3768056,
      "step": 5765
    },
    {
      "epoch": 3.0241090146750524,
      "grad_norm": 0.23489825427532196,
      "learning_rate": 4.960211540571092e-05,
      "loss": 0.5917,
      "num_input_tokens_seen": 3770168,
      "step": 5770
    },
    {
      "epoch": 3.0267295597484276,
      "grad_norm": 0.21002423763275146,
      "learning_rate": 4.960008095437836e-05,
      "loss": 0.5469,
      "num_input_tokens_seen": 3773112,
      "step": 5775
    },
    {
      "epoch": 3.029350104821803,
      "grad_norm": 0.16026277840137482,
      "learning_rate": 4.959804135698951e-05,
      "loss": 0.5709,
      "num_input_tokens_seen": 3776568,
      "step": 5780
    },
    {
      "epoch": 3.0319706498951784,
      "grad_norm": 0.11610930413007736,
      "learning_rate": 4.959599661397104e-05,
      "loss": 0.7122,
      "num_input_tokens_seen": 3780920,
      "step": 5785
    },
    {
      "epoch": 3.0345911949685536,
      "grad_norm": 0.17344605922698975,
      "learning_rate": 4.959394672575068e-05,
      "loss": 0.6243,
      "num_input_tokens_seen": 3783960,
      "step": 5790
    },
    {
      "epoch": 3.0372117400419287,
      "grad_norm": 0.236676424741745,
      "learning_rate": 4.959189169275725e-05,
      "loss": 0.5154,
      "num_input_tokens_seen": 3786968,
      "step": 5795
    },
    {
      "epoch": 3.039832285115304,
      "grad_norm": 0.13599739968776703,
      "learning_rate": 4.958983151542065e-05,
      "loss": 0.604,
      "num_input_tokens_seen": 3790808,
      "step": 5800
    },
    {
      "epoch": 3.042452830188679,
      "grad_norm": 0.14703775942325592,
      "learning_rate": 4.958776619417182e-05,
      "loss": 0.5258,
      "num_input_tokens_seen": 3794264,
      "step": 5805
    },
    {
      "epoch": 3.0450733752620547,
      "grad_norm": 0.174672931432724,
      "learning_rate": 4.9585695729442814e-05,
      "loss": 0.591,
      "num_input_tokens_seen": 3797496,
      "step": 5810
    },
    {
      "epoch": 3.04769392033543,
      "grad_norm": 0.2183014154434204,
      "learning_rate": 4.9583620121666754e-05,
      "loss": 0.6668,
      "num_input_tokens_seen": 3800888,
      "step": 5815
    },
    {
      "epoch": 3.050314465408805,
      "grad_norm": 0.24586859345436096,
      "learning_rate": 4.958153937127783e-05,
      "loss": 0.4713,
      "num_input_tokens_seen": 3803512,
      "step": 5820
    },
    {
      "epoch": 3.05293501048218,
      "grad_norm": 0.35399535298347473,
      "learning_rate": 4.95794534787113e-05,
      "loss": 0.6271,
      "num_input_tokens_seen": 3806552,
      "step": 5825
    },
    {
      "epoch": 3.0555555555555554,
      "grad_norm": 0.14285854995250702,
      "learning_rate": 4.957736244440353e-05,
      "loss": 0.5913,
      "num_input_tokens_seen": 3810648,
      "step": 5830
    },
    {
      "epoch": 3.058176100628931,
      "grad_norm": 0.1532488316297531,
      "learning_rate": 4.957526626879193e-05,
      "loss": 0.4393,
      "num_input_tokens_seen": 3813976,
      "step": 5835
    },
    {
      "epoch": 3.060796645702306,
      "grad_norm": 0.4410422742366791,
      "learning_rate": 4.957316495231499e-05,
      "loss": 0.582,
      "num_input_tokens_seen": 3816824,
      "step": 5840
    },
    {
      "epoch": 3.0634171907756813,
      "grad_norm": 0.09564317762851715,
      "learning_rate": 4.9571058495412294e-05,
      "loss": 0.6496,
      "num_input_tokens_seen": 3820376,
      "step": 5845
    },
    {
      "epoch": 3.0660377358490565,
      "grad_norm": 0.2484472095966339,
      "learning_rate": 4.9568946898524485e-05,
      "loss": 0.4665,
      "num_input_tokens_seen": 3823736,
      "step": 5850
    },
    {
      "epoch": 3.068658280922432,
      "grad_norm": 0.17834265530109406,
      "learning_rate": 4.956683016209327e-05,
      "loss": 0.695,
      "num_input_tokens_seen": 3826104,
      "step": 5855
    },
    {
      "epoch": 3.0712788259958073,
      "grad_norm": 0.15999162197113037,
      "learning_rate": 4.956470828656147e-05,
      "loss": 0.5964,
      "num_input_tokens_seen": 3828632,
      "step": 5860
    },
    {
      "epoch": 3.0738993710691824,
      "grad_norm": 0.1338832825422287,
      "learning_rate": 4.956258127237294e-05,
      "loss": 0.3666,
      "num_input_tokens_seen": 3831576,
      "step": 5865
    },
    {
      "epoch": 3.0765199161425576,
      "grad_norm": 0.17260536551475525,
      "learning_rate": 4.956044911997263e-05,
      "loss": 0.3862,
      "num_input_tokens_seen": 3834392,
      "step": 5870
    },
    {
      "epoch": 3.0791404612159328,
      "grad_norm": 0.21816058456897736,
      "learning_rate": 4.955831182980657e-05,
      "loss": 0.6118,
      "num_input_tokens_seen": 3837144,
      "step": 5875
    },
    {
      "epoch": 3.0817610062893084,
      "grad_norm": 0.18427251279354095,
      "learning_rate": 4.955616940232185e-05,
      "loss": 0.6033,
      "num_input_tokens_seen": 3840504,
      "step": 5880
    },
    {
      "epoch": 3.0843815513626835,
      "grad_norm": 0.14134499430656433,
      "learning_rate": 4.9554021837966636e-05,
      "loss": 0.5677,
      "num_input_tokens_seen": 3844408,
      "step": 5885
    },
    {
      "epoch": 3.0870020964360587,
      "grad_norm": 0.18116672337055206,
      "learning_rate": 4.955186913719019e-05,
      "loss": 0.563,
      "num_input_tokens_seen": 3847736,
      "step": 5890
    },
    {
      "epoch": 3.089622641509434,
      "grad_norm": 0.18254190683364868,
      "learning_rate": 4.954971130044281e-05,
      "loss": 0.5862,
      "num_input_tokens_seen": 3850360,
      "step": 5895
    },
    {
      "epoch": 3.092243186582809,
      "grad_norm": 0.19990473985671997,
      "learning_rate": 4.954754832817592e-05,
      "loss": 0.5112,
      "num_input_tokens_seen": 3853624,
      "step": 5900
    },
    {
      "epoch": 3.0948637316561847,
      "grad_norm": 0.20208264887332916,
      "learning_rate": 4.954538022084196e-05,
      "loss": 0.4191,
      "num_input_tokens_seen": 3856664,
      "step": 5905
    },
    {
      "epoch": 3.09748427672956,
      "grad_norm": 0.23358522355556488,
      "learning_rate": 4.9543206978894496e-05,
      "loss": 0.5784,
      "num_input_tokens_seen": 3859640,
      "step": 5910
    },
    {
      "epoch": 3.100104821802935,
      "grad_norm": 0.26742663979530334,
      "learning_rate": 4.954102860278813e-05,
      "loss": 0.4607,
      "num_input_tokens_seen": 3862712,
      "step": 5915
    },
    {
      "epoch": 3.10272536687631,
      "grad_norm": 0.15471400320529938,
      "learning_rate": 4.9538845092978567e-05,
      "loss": 0.5318,
      "num_input_tokens_seen": 3866328,
      "step": 5920
    },
    {
      "epoch": 3.1053459119496853,
      "grad_norm": 0.16454139351844788,
      "learning_rate": 4.953665644992256e-05,
      "loss": 0.5126,
      "num_input_tokens_seen": 3869592,
      "step": 5925
    },
    {
      "epoch": 3.107966457023061,
      "grad_norm": 0.22247132658958435,
      "learning_rate": 4.953446267407795e-05,
      "loss": 0.4331,
      "num_input_tokens_seen": 3875032,
      "step": 5930
    },
    {
      "epoch": 3.110587002096436,
      "grad_norm": 0.09170828014612198,
      "learning_rate": 4.953226376590365e-05,
      "loss": 0.594,
      "num_input_tokens_seen": 3879288,
      "step": 5935
    },
    {
      "epoch": 3.1132075471698113,
      "grad_norm": 0.1428421139717102,
      "learning_rate": 4.953005972585965e-05,
      "loss": 0.4371,
      "num_input_tokens_seen": 3882936,
      "step": 5940
    },
    {
      "epoch": 3.1158280922431865,
      "grad_norm": 0.11682827025651932,
      "learning_rate": 4.952785055440701e-05,
      "loss": 0.5796,
      "num_input_tokens_seen": 3886296,
      "step": 5945
    },
    {
      "epoch": 3.1184486373165616,
      "grad_norm": 0.30461186170578003,
      "learning_rate": 4.9525636252007864e-05,
      "loss": 0.6025,
      "num_input_tokens_seen": 3889816,
      "step": 5950
    },
    {
      "epoch": 3.1210691823899372,
      "grad_norm": 0.24239134788513184,
      "learning_rate": 4.9523416819125425e-05,
      "loss": 0.4525,
      "num_input_tokens_seen": 3892888,
      "step": 5955
    },
    {
      "epoch": 3.1236897274633124,
      "grad_norm": 0.4165615737438202,
      "learning_rate": 4.952119225622396e-05,
      "loss": 0.6134,
      "num_input_tokens_seen": 3896152,
      "step": 5960
    },
    {
      "epoch": 3.1263102725366876,
      "grad_norm": 0.1416611522436142,
      "learning_rate": 4.951896256376882e-05,
      "loss": 0.4008,
      "num_input_tokens_seen": 3898776,
      "step": 5965
    },
    {
      "epoch": 3.1289308176100628,
      "grad_norm": 0.19084464013576508,
      "learning_rate": 4.9516727742226444e-05,
      "loss": 0.5722,
      "num_input_tokens_seen": 3901912,
      "step": 5970
    },
    {
      "epoch": 3.131551362683438,
      "grad_norm": 0.12927278876304626,
      "learning_rate": 4.9514487792064323e-05,
      "loss": 0.6203,
      "num_input_tokens_seen": 3905688,
      "step": 5975
    },
    {
      "epoch": 3.1341719077568135,
      "grad_norm": 0.16594019532203674,
      "learning_rate": 4.951224271375104e-05,
      "loss": 0.545,
      "num_input_tokens_seen": 3908888,
      "step": 5980
    },
    {
      "epoch": 3.1367924528301887,
      "grad_norm": 0.30173107981681824,
      "learning_rate": 4.950999250775623e-05,
      "loss": 0.4805,
      "num_input_tokens_seen": 3912280,
      "step": 5985
    },
    {
      "epoch": 3.139412997903564,
      "grad_norm": 0.2408144325017929,
      "learning_rate": 4.950773717455061e-05,
      "loss": 0.4849,
      "num_input_tokens_seen": 3915256,
      "step": 5990
    },
    {
      "epoch": 3.142033542976939,
      "grad_norm": 0.12474607676267624,
      "learning_rate": 4.950547671460596e-05,
      "loss": 0.4991,
      "num_input_tokens_seen": 3917912,
      "step": 5995
    },
    {
      "epoch": 3.1446540880503147,
      "grad_norm": 0.198663130402565,
      "learning_rate": 4.950321112839517e-05,
      "loss": 0.6122,
      "num_input_tokens_seen": 3920824,
      "step": 6000
    },
    {
      "epoch": 3.14727463312369,
      "grad_norm": 0.2601509392261505,
      "learning_rate": 4.9500940416392156e-05,
      "loss": 0.4327,
      "num_input_tokens_seen": 3923736,
      "step": 6005
    },
    {
      "epoch": 3.149895178197065,
      "grad_norm": 0.118253193795681,
      "learning_rate": 4.949866457907193e-05,
      "loss": 0.5825,
      "num_input_tokens_seen": 3926584,
      "step": 6010
    },
    {
      "epoch": 3.15251572327044,
      "grad_norm": 0.17814847826957703,
      "learning_rate": 4.9496383616910565e-05,
      "loss": 0.7089,
      "num_input_tokens_seen": 3929432,
      "step": 6015
    },
    {
      "epoch": 3.1551362683438153,
      "grad_norm": 0.1837228387594223,
      "learning_rate": 4.949409753038522e-05,
      "loss": 0.5898,
      "num_input_tokens_seen": 3932088,
      "step": 6020
    },
    {
      "epoch": 3.157756813417191,
      "grad_norm": 0.19067004323005676,
      "learning_rate": 4.949180631997411e-05,
      "loss": 0.4776,
      "num_input_tokens_seen": 3935384,
      "step": 6025
    },
    {
      "epoch": 3.160377358490566,
      "grad_norm": 0.159529447555542,
      "learning_rate": 4.948950998615654e-05,
      "loss": 0.4894,
      "num_input_tokens_seen": 3938424,
      "step": 6030
    },
    {
      "epoch": 3.1629979035639413,
      "grad_norm": 0.1765158772468567,
      "learning_rate": 4.9487208529412876e-05,
      "loss": 0.5505,
      "num_input_tokens_seen": 3940984,
      "step": 6035
    },
    {
      "epoch": 3.1656184486373165,
      "grad_norm": 0.15879835188388824,
      "learning_rate": 4.9484901950224546e-05,
      "loss": 0.5265,
      "num_input_tokens_seen": 3945144,
      "step": 6040
    },
    {
      "epoch": 3.1682389937106916,
      "grad_norm": 0.11376780271530151,
      "learning_rate": 4.948259024907407e-05,
      "loss": 0.5546,
      "num_input_tokens_seen": 3948344,
      "step": 6045
    },
    {
      "epoch": 3.1708595387840672,
      "grad_norm": 0.14890417456626892,
      "learning_rate": 4.948027342644502e-05,
      "loss": 0.5893,
      "num_input_tokens_seen": 3951000,
      "step": 6050
    },
    {
      "epoch": 3.1734800838574424,
      "grad_norm": 0.1917613446712494,
      "learning_rate": 4.947795148282206e-05,
      "loss": 0.5046,
      "num_input_tokens_seen": 3954584,
      "step": 6055
    },
    {
      "epoch": 3.1761006289308176,
      "grad_norm": 0.16687282919883728,
      "learning_rate": 4.947562441869091e-05,
      "loss": 0.3964,
      "num_input_tokens_seen": 3957944,
      "step": 6060
    },
    {
      "epoch": 3.1787211740041927,
      "grad_norm": 0.12885800004005432,
      "learning_rate": 4.947329223453836e-05,
      "loss": 0.4843,
      "num_input_tokens_seen": 3961816,
      "step": 6065
    },
    {
      "epoch": 3.181341719077568,
      "grad_norm": 0.16820840537548065,
      "learning_rate": 4.947095493085229e-05,
      "loss": 0.591,
      "num_input_tokens_seen": 3964824,
      "step": 6070
    },
    {
      "epoch": 3.1839622641509435,
      "grad_norm": 0.20637460052967072,
      "learning_rate": 4.9468612508121616e-05,
      "loss": 0.4065,
      "num_input_tokens_seen": 3968056,
      "step": 6075
    },
    {
      "epoch": 3.1865828092243187,
      "grad_norm": 0.1592218279838562,
      "learning_rate": 4.9466264966836365e-05,
      "loss": 0.541,
      "num_input_tokens_seen": 3970424,
      "step": 6080
    },
    {
      "epoch": 3.189203354297694,
      "grad_norm": 0.49624061584472656,
      "learning_rate": 4.94639123074876e-05,
      "loss": 0.5094,
      "num_input_tokens_seen": 3973208,
      "step": 6085
    },
    {
      "epoch": 3.191823899371069,
      "grad_norm": 0.18424080312252045,
      "learning_rate": 4.946155453056749e-05,
      "loss": 0.5749,
      "num_input_tokens_seen": 3976216,
      "step": 6090
    },
    {
      "epoch": 3.1944444444444446,
      "grad_norm": 0.2865707278251648,
      "learning_rate": 4.945919163656923e-05,
      "loss": 0.5811,
      "num_input_tokens_seen": 3980824,
      "step": 6095
    },
    {
      "epoch": 3.19706498951782,
      "grad_norm": 0.2676544189453125,
      "learning_rate": 4.945682362598715e-05,
      "loss": 0.5939,
      "num_input_tokens_seen": 3987480,
      "step": 6100
    },
    {
      "epoch": 3.199685534591195,
      "grad_norm": 0.14972631633281708,
      "learning_rate": 4.945445049931656e-05,
      "loss": 0.5027,
      "num_input_tokens_seen": 3991288,
      "step": 6105
    },
    {
      "epoch": 3.20230607966457,
      "grad_norm": 0.15760977566242218,
      "learning_rate": 4.945207225705393e-05,
      "loss": 0.6111,
      "num_input_tokens_seen": 3993848,
      "step": 6110
    },
    {
      "epoch": 3.2049266247379453,
      "grad_norm": 0.17664553225040436,
      "learning_rate": 4.944968889969674e-05,
      "loss": 0.5469,
      "num_input_tokens_seen": 3997848,
      "step": 6115
    },
    {
      "epoch": 3.207547169811321,
      "grad_norm": 0.16368629038333893,
      "learning_rate": 4.944730042774358e-05,
      "loss": 0.461,
      "num_input_tokens_seen": 4001784,
      "step": 6120
    },
    {
      "epoch": 3.210167714884696,
      "grad_norm": 0.11707063764333725,
      "learning_rate": 4.944490684169408e-05,
      "loss": 0.5414,
      "num_input_tokens_seen": 4005048,
      "step": 6125
    },
    {
      "epoch": 3.2127882599580713,
      "grad_norm": 0.3856685757637024,
      "learning_rate": 4.9442508142048945e-05,
      "loss": 0.573,
      "num_input_tokens_seen": 4007864,
      "step": 6130
    },
    {
      "epoch": 3.2154088050314464,
      "grad_norm": 0.29181748628616333,
      "learning_rate": 4.9440104329309956e-05,
      "loss": 0.5894,
      "num_input_tokens_seen": 4011832,
      "step": 6135
    },
    {
      "epoch": 3.2180293501048216,
      "grad_norm": 0.1464267522096634,
      "learning_rate": 4.943769540397998e-05,
      "loss": 0.5102,
      "num_input_tokens_seen": 4015032,
      "step": 6140
    },
    {
      "epoch": 3.220649895178197,
      "grad_norm": 0.18311236798763275,
      "learning_rate": 4.943528136656292e-05,
      "loss": 0.6464,
      "num_input_tokens_seen": 4017432,
      "step": 6145
    },
    {
      "epoch": 3.2232704402515724,
      "grad_norm": 0.25673606991767883,
      "learning_rate": 4.943286221756377e-05,
      "loss": 0.4556,
      "num_input_tokens_seen": 4020216,
      "step": 6150
    },
    {
      "epoch": 3.2258909853249476,
      "grad_norm": 0.22020255029201508,
      "learning_rate": 4.9430437957488586e-05,
      "loss": 0.5894,
      "num_input_tokens_seen": 4023096,
      "step": 6155
    },
    {
      "epoch": 3.2285115303983227,
      "grad_norm": 0.33070939779281616,
      "learning_rate": 4.94280085868445e-05,
      "loss": 0.4419,
      "num_input_tokens_seen": 4026712,
      "step": 6160
    },
    {
      "epoch": 3.231132075471698,
      "grad_norm": 0.28657475113868713,
      "learning_rate": 4.942557410613972e-05,
      "loss": 0.5793,
      "num_input_tokens_seen": 4030872,
      "step": 6165
    },
    {
      "epoch": 3.2337526205450735,
      "grad_norm": 0.15473473072052002,
      "learning_rate": 4.942313451588348e-05,
      "loss": 0.4749,
      "num_input_tokens_seen": 4034744,
      "step": 6170
    },
    {
      "epoch": 3.2363731656184487,
      "grad_norm": 0.1873379349708557,
      "learning_rate": 4.942068981658614e-05,
      "loss": 0.4927,
      "num_input_tokens_seen": 4037752,
      "step": 6175
    },
    {
      "epoch": 3.238993710691824,
      "grad_norm": 0.5294644832611084,
      "learning_rate": 4.9418240008759097e-05,
      "loss": 0.4662,
      "num_input_tokens_seen": 4041048,
      "step": 6180
    },
    {
      "epoch": 3.241614255765199,
      "grad_norm": 0.25690507888793945,
      "learning_rate": 4.9415785092914826e-05,
      "loss": 0.6463,
      "num_input_tokens_seen": 4043832,
      "step": 6185
    },
    {
      "epoch": 3.2442348008385746,
      "grad_norm": 0.21023911237716675,
      "learning_rate": 4.941332506956686e-05,
      "loss": 0.4601,
      "num_input_tokens_seen": 4047512,
      "step": 6190
    },
    {
      "epoch": 3.24685534591195,
      "grad_norm": 0.14354676008224487,
      "learning_rate": 4.941085993922981e-05,
      "loss": 0.5405,
      "num_input_tokens_seen": 4050584,
      "step": 6195
    },
    {
      "epoch": 3.249475890985325,
      "grad_norm": 0.2144630402326584,
      "learning_rate": 4.9408389702419354e-05,
      "loss": 0.4256,
      "num_input_tokens_seen": 4053752,
      "step": 6200
    },
    {
      "epoch": 3.2520964360587,
      "grad_norm": 0.3033960461616516,
      "learning_rate": 4.940591435965224e-05,
      "loss": 0.5547,
      "num_input_tokens_seen": 4056824,
      "step": 6205
    },
    {
      "epoch": 3.2547169811320753,
      "grad_norm": 0.4604698419570923,
      "learning_rate": 4.9403433911446285e-05,
      "loss": 0.7258,
      "num_input_tokens_seen": 4059864,
      "step": 6210
    },
    {
      "epoch": 3.257337526205451,
      "grad_norm": 0.2662343978881836,
      "learning_rate": 4.9400948358320356e-05,
      "loss": 0.4995,
      "num_input_tokens_seen": 4063416,
      "step": 6215
    },
    {
      "epoch": 3.259958071278826,
      "grad_norm": 0.09114550054073334,
      "learning_rate": 4.939845770079442e-05,
      "loss": 0.4554,
      "num_input_tokens_seen": 4066264,
      "step": 6220
    },
    {
      "epoch": 3.2625786163522013,
      "grad_norm": 0.14329573512077332,
      "learning_rate": 4.9395961939389475e-05,
      "loss": 0.578,
      "num_input_tokens_seen": 4068664,
      "step": 6225
    },
    {
      "epoch": 3.2651991614255764,
      "grad_norm": 0.19630615413188934,
      "learning_rate": 4.939346107462762e-05,
      "loss": 0.7363,
      "num_input_tokens_seen": 4072088,
      "step": 6230
    },
    {
      "epoch": 3.2678197064989516,
      "grad_norm": 0.3772164285182953,
      "learning_rate": 4.9390955107032e-05,
      "loss": 0.7198,
      "num_input_tokens_seen": 4075064,
      "step": 6235
    },
    {
      "epoch": 3.270440251572327,
      "grad_norm": 0.21394218504428864,
      "learning_rate": 4.938844403712685e-05,
      "loss": 0.5645,
      "num_input_tokens_seen": 4077752,
      "step": 6240
    },
    {
      "epoch": 3.2730607966457024,
      "grad_norm": 0.1629781723022461,
      "learning_rate": 4.938592786543743e-05,
      "loss": 0.5482,
      "num_input_tokens_seen": 4080472,
      "step": 6245
    },
    {
      "epoch": 3.2756813417190775,
      "grad_norm": 0.202707901597023,
      "learning_rate": 4.938340659249012e-05,
      "loss": 0.4342,
      "num_input_tokens_seen": 4083384,
      "step": 6250
    },
    {
      "epoch": 3.2783018867924527,
      "grad_norm": 0.13368090987205505,
      "learning_rate": 4.938088021881233e-05,
      "loss": 0.5664,
      "num_input_tokens_seen": 4086488,
      "step": 6255
    },
    {
      "epoch": 3.280922431865828,
      "grad_norm": 0.1725648194551468,
      "learning_rate": 4.937834874493255e-05,
      "loss": 0.3826,
      "num_input_tokens_seen": 4089464,
      "step": 6260
    },
    {
      "epoch": 3.2835429769392035,
      "grad_norm": 0.21166007220745087,
      "learning_rate": 4.9375812171380334e-05,
      "loss": 0.5846,
      "num_input_tokens_seen": 4092088,
      "step": 6265
    },
    {
      "epoch": 3.2861635220125787,
      "grad_norm": 0.23605014383792877,
      "learning_rate": 4.9373270498686306e-05,
      "loss": 0.5221,
      "num_input_tokens_seen": 4095544,
      "step": 6270
    },
    {
      "epoch": 3.288784067085954,
      "grad_norm": 0.11301878094673157,
      "learning_rate": 4.937072372738216e-05,
      "loss": 0.5329,
      "num_input_tokens_seen": 4098264,
      "step": 6275
    },
    {
      "epoch": 3.291404612159329,
      "grad_norm": 0.15254992246627808,
      "learning_rate": 4.9368171858000635e-05,
      "loss": 0.4906,
      "num_input_tokens_seen": 4100792,
      "step": 6280
    },
    {
      "epoch": 3.2940251572327046,
      "grad_norm": 0.14408913254737854,
      "learning_rate": 4.9365614891075576e-05,
      "loss": 0.5069,
      "num_input_tokens_seen": 4103896,
      "step": 6285
    },
    {
      "epoch": 3.29664570230608,
      "grad_norm": 0.139274001121521,
      "learning_rate": 4.936305282714185e-05,
      "loss": 0.4257,
      "num_input_tokens_seen": 4106968,
      "step": 6290
    },
    {
      "epoch": 3.299266247379455,
      "grad_norm": 0.1553727090358734,
      "learning_rate": 4.936048566673542e-05,
      "loss": 0.5952,
      "num_input_tokens_seen": 4110040,
      "step": 6295
    },
    {
      "epoch": 3.30188679245283,
      "grad_norm": 0.15786436200141907,
      "learning_rate": 4.935791341039332e-05,
      "loss": 0.4911,
      "num_input_tokens_seen": 4112664,
      "step": 6300
    },
    {
      "epoch": 3.3045073375262053,
      "grad_norm": 0.2781450152397156,
      "learning_rate": 4.935533605865361e-05,
      "loss": 0.5323,
      "num_input_tokens_seen": 4115832,
      "step": 6305
    },
    {
      "epoch": 3.307127882599581,
      "grad_norm": 0.1963469535112381,
      "learning_rate": 4.935275361205547e-05,
      "loss": 0.5096,
      "num_input_tokens_seen": 4119160,
      "step": 6310
    },
    {
      "epoch": 3.309748427672956,
      "grad_norm": 0.24517598748207092,
      "learning_rate": 4.93501660711391e-05,
      "loss": 0.8874,
      "num_input_tokens_seen": 4122520,
      "step": 6315
    },
    {
      "epoch": 3.3123689727463312,
      "grad_norm": 0.2677709758281708,
      "learning_rate": 4.934757343644578e-05,
      "loss": 0.4878,
      "num_input_tokens_seen": 4124888,
      "step": 6320
    },
    {
      "epoch": 3.3149895178197064,
      "grad_norm": 0.1803373247385025,
      "learning_rate": 4.934497570851788e-05,
      "loss": 0.57,
      "num_input_tokens_seen": 4127832,
      "step": 6325
    },
    {
      "epoch": 3.3176100628930816,
      "grad_norm": 0.16979239881038666,
      "learning_rate": 4.9342372887898804e-05,
      "loss": 0.6722,
      "num_input_tokens_seen": 4130232,
      "step": 6330
    },
    {
      "epoch": 3.320230607966457,
      "grad_norm": 0.13335970044136047,
      "learning_rate": 4.9339764975133026e-05,
      "loss": 0.4551,
      "num_input_tokens_seen": 4133080,
      "step": 6335
    },
    {
      "epoch": 3.3228511530398324,
      "grad_norm": 0.09070432931184769,
      "learning_rate": 4.93371519707661e-05,
      "loss": 0.4808,
      "num_input_tokens_seen": 4136536,
      "step": 6340
    },
    {
      "epoch": 3.3254716981132075,
      "grad_norm": 0.11413509398698807,
      "learning_rate": 4.933453387534464e-05,
      "loss": 0.4896,
      "num_input_tokens_seen": 4139608,
      "step": 6345
    },
    {
      "epoch": 3.3280922431865827,
      "grad_norm": 0.4076246917247772,
      "learning_rate": 4.933191068941631e-05,
      "loss": 0.6756,
      "num_input_tokens_seen": 4142200,
      "step": 6350
    },
    {
      "epoch": 3.330712788259958,
      "grad_norm": 0.3480856418609619,
      "learning_rate": 4.932928241352987e-05,
      "loss": 0.8725,
      "num_input_tokens_seen": 4146200,
      "step": 6355
    },
    {
      "epoch": 3.3333333333333335,
      "grad_norm": 0.18306899070739746,
      "learning_rate": 4.9326649048235095e-05,
      "loss": 0.4481,
      "num_input_tokens_seen": 4149464,
      "step": 6360
    },
    {
      "epoch": 3.3359538784067087,
      "grad_norm": 0.1311909407377243,
      "learning_rate": 4.932401059408287e-05,
      "loss": 0.609,
      "num_input_tokens_seen": 4151992,
      "step": 6365
    },
    {
      "epoch": 3.338574423480084,
      "grad_norm": 0.13838478922843933,
      "learning_rate": 4.932136705162514e-05,
      "loss": 0.5275,
      "num_input_tokens_seen": 4155576,
      "step": 6370
    },
    {
      "epoch": 3.341194968553459,
      "grad_norm": 0.19304578006267548,
      "learning_rate": 4.93187184214149e-05,
      "loss": 0.6504,
      "num_input_tokens_seen": 4159672,
      "step": 6375
    },
    {
      "epoch": 3.3438155136268346,
      "grad_norm": 0.20008952915668488,
      "learning_rate": 4.9316064704006205e-05,
      "loss": 0.4775,
      "num_input_tokens_seen": 4162360,
      "step": 6380
    },
    {
      "epoch": 3.3464360587002098,
      "grad_norm": 0.18059128522872925,
      "learning_rate": 4.931340589995419e-05,
      "loss": 0.4913,
      "num_input_tokens_seen": 4165304,
      "step": 6385
    },
    {
      "epoch": 3.349056603773585,
      "grad_norm": 0.17117470502853394,
      "learning_rate": 4.9310742009815035e-05,
      "loss": 0.4956,
      "num_input_tokens_seen": 4168728,
      "step": 6390
    },
    {
      "epoch": 3.35167714884696,
      "grad_norm": 0.2175646871328354,
      "learning_rate": 4.930807303414601e-05,
      "loss": 0.4233,
      "num_input_tokens_seen": 4171960,
      "step": 6395
    },
    {
      "epoch": 3.3542976939203353,
      "grad_norm": 0.10999969393014908,
      "learning_rate": 4.930539897350542e-05,
      "loss": 0.5135,
      "num_input_tokens_seen": 4175544,
      "step": 6400
    },
    {
      "epoch": 3.3569182389937104,
      "grad_norm": 0.10729072242975235,
      "learning_rate": 4.930271982845266e-05,
      "loss": 0.4579,
      "num_input_tokens_seen": 4178264,
      "step": 6405
    },
    {
      "epoch": 3.359538784067086,
      "grad_norm": 0.20853348076343536,
      "learning_rate": 4.930003559954817e-05,
      "loss": 0.5292,
      "num_input_tokens_seen": 4180984,
      "step": 6410
    },
    {
      "epoch": 3.3621593291404612,
      "grad_norm": 0.17756420373916626,
      "learning_rate": 4.929734628735346e-05,
      "loss": 0.6239,
      "num_input_tokens_seen": 4183768,
      "step": 6415
    },
    {
      "epoch": 3.3647798742138364,
      "grad_norm": 0.16921480000019073,
      "learning_rate": 4.9294651892431115e-05,
      "loss": 0.4868,
      "num_input_tokens_seen": 4187000,
      "step": 6420
    },
    {
      "epoch": 3.3674004192872116,
      "grad_norm": 0.2169916331768036,
      "learning_rate": 4.929195241534476e-05,
      "loss": 0.5473,
      "num_input_tokens_seen": 4189304,
      "step": 6425
    },
    {
      "epoch": 3.370020964360587,
      "grad_norm": 0.1268264800310135,
      "learning_rate": 4.9289247856659095e-05,
      "loss": 0.5327,
      "num_input_tokens_seen": 4193080,
      "step": 6430
    },
    {
      "epoch": 3.3726415094339623,
      "grad_norm": 0.1729055792093277,
      "learning_rate": 4.9286538216939883e-05,
      "loss": 0.5348,
      "num_input_tokens_seen": 4195608,
      "step": 6435
    },
    {
      "epoch": 3.3752620545073375,
      "grad_norm": 0.2002473771572113,
      "learning_rate": 4.9283823496753944e-05,
      "loss": 0.417,
      "num_input_tokens_seen": 4198744,
      "step": 6440
    },
    {
      "epoch": 3.3778825995807127,
      "grad_norm": 0.2557746469974518,
      "learning_rate": 4.928110369666918e-05,
      "loss": 0.7635,
      "num_input_tokens_seen": 4202680,
      "step": 6445
    },
    {
      "epoch": 3.380503144654088,
      "grad_norm": 0.24854841828346252,
      "learning_rate": 4.927837881725455e-05,
      "loss": 0.5163,
      "num_input_tokens_seen": 4205144,
      "step": 6450
    },
    {
      "epoch": 3.3831236897274635,
      "grad_norm": 0.1886710673570633,
      "learning_rate": 4.927564885908003e-05,
      "loss": 0.5046,
      "num_input_tokens_seen": 4209112,
      "step": 6455
    },
    {
      "epoch": 3.3857442348008386,
      "grad_norm": 0.15031355619430542,
      "learning_rate": 4.9272913822716735e-05,
      "loss": 0.5064,
      "num_input_tokens_seen": 4213048,
      "step": 6460
    },
    {
      "epoch": 3.388364779874214,
      "grad_norm": 0.09375160187482834,
      "learning_rate": 4.927017370873678e-05,
      "loss": 0.4722,
      "num_input_tokens_seen": 4216472,
      "step": 6465
    },
    {
      "epoch": 3.390985324947589,
      "grad_norm": 0.16620837152004242,
      "learning_rate": 4.926742851771338e-05,
      "loss": 0.4542,
      "num_input_tokens_seen": 4219672,
      "step": 6470
    },
    {
      "epoch": 3.3936058700209646,
      "grad_norm": 0.22092756628990173,
      "learning_rate": 4.926467825022079e-05,
      "loss": 0.558,
      "num_input_tokens_seen": 4222872,
      "step": 6475
    },
    {
      "epoch": 3.3962264150943398,
      "grad_norm": 0.15363958477973938,
      "learning_rate": 4.926192290683434e-05,
      "loss": 0.557,
      "num_input_tokens_seen": 4225528,
      "step": 6480
    },
    {
      "epoch": 3.398846960167715,
      "grad_norm": 0.2657897472381592,
      "learning_rate": 4.9259162488130404e-05,
      "loss": 0.4595,
      "num_input_tokens_seen": 4229144,
      "step": 6485
    },
    {
      "epoch": 3.40146750524109,
      "grad_norm": 0.14703333377838135,
      "learning_rate": 4.925639699468645e-05,
      "loss": 0.5233,
      "num_input_tokens_seen": 4232792,
      "step": 6490
    },
    {
      "epoch": 3.4040880503144653,
      "grad_norm": 0.14590153098106384,
      "learning_rate": 4.925362642708097e-05,
      "loss": 0.4761,
      "num_input_tokens_seen": 4236600,
      "step": 6495
    },
    {
      "epoch": 3.4067085953878404,
      "grad_norm": 0.2150430679321289,
      "learning_rate": 4.9250850785893536e-05,
      "loss": 0.6755,
      "num_input_tokens_seen": 4239896,
      "step": 6500
    },
    {
      "epoch": 3.409329140461216,
      "grad_norm": 0.13100558519363403,
      "learning_rate": 4.924807007170479e-05,
      "loss": 0.5442,
      "num_input_tokens_seen": 4243640,
      "step": 6505
    },
    {
      "epoch": 3.411949685534591,
      "grad_norm": 0.16540756821632385,
      "learning_rate": 4.924528428509643e-05,
      "loss": 0.6103,
      "num_input_tokens_seen": 4247160,
      "step": 6510
    },
    {
      "epoch": 3.4145702306079664,
      "grad_norm": 0.18645627796649933,
      "learning_rate": 4.9242493426651196e-05,
      "loss": 0.644,
      "num_input_tokens_seen": 4251032,
      "step": 6515
    },
    {
      "epoch": 3.4171907756813416,
      "grad_norm": 0.32188156247138977,
      "learning_rate": 4.92396974969529e-05,
      "loss": 0.5979,
      "num_input_tokens_seen": 4254680,
      "step": 6520
    },
    {
      "epoch": 3.419811320754717,
      "grad_norm": 0.14305198192596436,
      "learning_rate": 4.923689649658644e-05,
      "loss": 0.4763,
      "num_input_tokens_seen": 4257720,
      "step": 6525
    },
    {
      "epoch": 3.4224318658280923,
      "grad_norm": 0.1633310467004776,
      "learning_rate": 4.923409042613774e-05,
      "loss": 0.5617,
      "num_input_tokens_seen": 4260696,
      "step": 6530
    },
    {
      "epoch": 3.4250524109014675,
      "grad_norm": 0.17387863993644714,
      "learning_rate": 4.92312792861938e-05,
      "loss": 0.507,
      "num_input_tokens_seen": 4263416,
      "step": 6535
    },
    {
      "epoch": 3.4276729559748427,
      "grad_norm": 0.17658019065856934,
      "learning_rate": 4.922846307734268e-05,
      "loss": 0.6276,
      "num_input_tokens_seen": 4266232,
      "step": 6540
    },
    {
      "epoch": 3.430293501048218,
      "grad_norm": 0.36185386776924133,
      "learning_rate": 4.9225641800173496e-05,
      "loss": 0.4816,
      "num_input_tokens_seen": 4269400,
      "step": 6545
    },
    {
      "epoch": 3.4329140461215935,
      "grad_norm": 0.19992408156394958,
      "learning_rate": 4.922281545527643e-05,
      "loss": 0.5313,
      "num_input_tokens_seen": 4271896,
      "step": 6550
    },
    {
      "epoch": 3.4355345911949686,
      "grad_norm": 0.17742620408535004,
      "learning_rate": 4.9219984043242715e-05,
      "loss": 0.7281,
      "num_input_tokens_seen": 4275416,
      "step": 6555
    },
    {
      "epoch": 3.438155136268344,
      "grad_norm": 0.12203500419855118,
      "learning_rate": 4.921714756466466e-05,
      "loss": 0.4749,
      "num_input_tokens_seen": 4278936,
      "step": 6560
    },
    {
      "epoch": 3.440775681341719,
      "grad_norm": 0.17247624695301056,
      "learning_rate": 4.921430602013561e-05,
      "loss": 0.525,
      "num_input_tokens_seen": 4282808,
      "step": 6565
    },
    {
      "epoch": 3.4433962264150946,
      "grad_norm": 0.1599963903427124,
      "learning_rate": 4.9211459410249995e-05,
      "loss": 0.5508,
      "num_input_tokens_seen": 4285400,
      "step": 6570
    },
    {
      "epoch": 3.4460167714884697,
      "grad_norm": 0.3103216290473938,
      "learning_rate": 4.92086077356033e-05,
      "loss": 0.4284,
      "num_input_tokens_seen": 4288280,
      "step": 6575
    },
    {
      "epoch": 3.448637316561845,
      "grad_norm": 0.14778101444244385,
      "learning_rate": 4.920575099679204e-05,
      "loss": 0.5177,
      "num_input_tokens_seen": 4291576,
      "step": 6580
    },
    {
      "epoch": 3.45125786163522,
      "grad_norm": 0.1912824511528015,
      "learning_rate": 4.920288919441383e-05,
      "loss": 0.5191,
      "num_input_tokens_seen": 4295288,
      "step": 6585
    },
    {
      "epoch": 3.4538784067085953,
      "grad_norm": 0.18032397329807281,
      "learning_rate": 4.9200022329067316e-05,
      "loss": 0.5741,
      "num_input_tokens_seen": 4298808,
      "step": 6590
    },
    {
      "epoch": 3.4564989517819704,
      "grad_norm": 0.19620342552661896,
      "learning_rate": 4.9197150401352224e-05,
      "loss": 0.4189,
      "num_input_tokens_seen": 4304280,
      "step": 6595
    },
    {
      "epoch": 3.459119496855346,
      "grad_norm": 0.28092697262763977,
      "learning_rate": 4.919427341186933e-05,
      "loss": 0.7156,
      "num_input_tokens_seen": 4306616,
      "step": 6600
    },
    {
      "epoch": 3.461740041928721,
      "grad_norm": 0.24205051362514496,
      "learning_rate": 4.919139136122045e-05,
      "loss": 0.5124,
      "num_input_tokens_seen": 4309112,
      "step": 6605
    },
    {
      "epoch": 3.4643605870020964,
      "grad_norm": 0.29489266872406006,
      "learning_rate": 4.918850425000849e-05,
      "loss": 0.6187,
      "num_input_tokens_seen": 4311832,
      "step": 6610
    },
    {
      "epoch": 3.4669811320754715,
      "grad_norm": 0.0431131012737751,
      "learning_rate": 4.918561207883741e-05,
      "loss": 0.4971,
      "num_input_tokens_seen": 4318328,
      "step": 6615
    },
    {
      "epoch": 3.469601677148847,
      "grad_norm": 0.26293399930000305,
      "learning_rate": 4.9182714848312195e-05,
      "loss": 0.506,
      "num_input_tokens_seen": 4321848,
      "step": 6620
    },
    {
      "epoch": 3.4722222222222223,
      "grad_norm": 0.10728922486305237,
      "learning_rate": 4.917981255903893e-05,
      "loss": 0.3866,
      "num_input_tokens_seen": 4324984,
      "step": 6625
    },
    {
      "epoch": 3.4748427672955975,
      "grad_norm": 0.23154638707637787,
      "learning_rate": 4.917690521162474e-05,
      "loss": 0.5527,
      "num_input_tokens_seen": 4327704,
      "step": 6630
    },
    {
      "epoch": 3.4774633123689727,
      "grad_norm": 0.13181142508983612,
      "learning_rate": 4.9173992806677796e-05,
      "loss": 0.4632,
      "num_input_tokens_seen": 4330904,
      "step": 6635
    },
    {
      "epoch": 3.480083857442348,
      "grad_norm": 0.21483322978019714,
      "learning_rate": 4.917107534480735e-05,
      "loss": 0.5756,
      "num_input_tokens_seen": 4333592,
      "step": 6640
    },
    {
      "epoch": 3.4827044025157234,
      "grad_norm": 0.3076287508010864,
      "learning_rate": 4.916815282662369e-05,
      "loss": 0.5075,
      "num_input_tokens_seen": 4336696,
      "step": 6645
    },
    {
      "epoch": 3.4853249475890986,
      "grad_norm": 0.38978520035743713,
      "learning_rate": 4.91652252527382e-05,
      "loss": 0.8042,
      "num_input_tokens_seen": 4340088,
      "step": 6650
    },
    {
      "epoch": 3.487945492662474,
      "grad_norm": 0.3724929392337799,
      "learning_rate": 4.9162292623763275e-05,
      "loss": 0.6013,
      "num_input_tokens_seen": 4343640,
      "step": 6655
    },
    {
      "epoch": 3.490566037735849,
      "grad_norm": 0.12249955534934998,
      "learning_rate": 4.9159354940312386e-05,
      "loss": 0.4734,
      "num_input_tokens_seen": 4348568,
      "step": 6660
    },
    {
      "epoch": 3.4931865828092246,
      "grad_norm": 0.23631539940834045,
      "learning_rate": 4.915641220300007e-05,
      "loss": 0.4578,
      "num_input_tokens_seen": 4352376,
      "step": 6665
    },
    {
      "epoch": 3.4958071278825997,
      "grad_norm": 0.1655597686767578,
      "learning_rate": 4.9153464412441915e-05,
      "loss": 0.5621,
      "num_input_tokens_seen": 4356120,
      "step": 6670
    },
    {
      "epoch": 3.498427672955975,
      "grad_norm": 0.170710027217865,
      "learning_rate": 4.915051156925456e-05,
      "loss": 0.5197,
      "num_input_tokens_seen": 4359448,
      "step": 6675
    },
    {
      "epoch": 3.50104821802935,
      "grad_norm": 0.14882667362689972,
      "learning_rate": 4.91475536740557e-05,
      "loss": 0.4049,
      "num_input_tokens_seen": 4362936,
      "step": 6680
    },
    {
      "epoch": 3.5036687631027252,
      "grad_norm": 0.1512797474861145,
      "learning_rate": 4.914459072746411e-05,
      "loss": 0.4787,
      "num_input_tokens_seen": 4366968,
      "step": 6685
    },
    {
      "epoch": 3.5062893081761004,
      "grad_norm": 0.13575570285320282,
      "learning_rate": 4.91416227300996e-05,
      "loss": 0.4013,
      "num_input_tokens_seen": 4370264,
      "step": 6690
    },
    {
      "epoch": 3.508909853249476,
      "grad_norm": 0.15677398443222046,
      "learning_rate": 4.913864968258303e-05,
      "loss": 0.3874,
      "num_input_tokens_seen": 4373432,
      "step": 6695
    },
    {
      "epoch": 3.511530398322851,
      "grad_norm": 0.15345770120620728,
      "learning_rate": 4.913567158553634e-05,
      "loss": 0.6815,
      "num_input_tokens_seen": 4376824,
      "step": 6700
    },
    {
      "epoch": 3.5141509433962264,
      "grad_norm": 0.13085338473320007,
      "learning_rate": 4.9132688439582505e-05,
      "loss": 0.5803,
      "num_input_tokens_seen": 4380504,
      "step": 6705
    },
    {
      "epoch": 3.5167714884696015,
      "grad_norm": 0.16060011088848114,
      "learning_rate": 4.9129700245345565e-05,
      "loss": 0.4992,
      "num_input_tokens_seen": 4383704,
      "step": 6710
    },
    {
      "epoch": 3.519392033542977,
      "grad_norm": 0.27530568838119507,
      "learning_rate": 4.912670700345063e-05,
      "loss": 0.6489,
      "num_input_tokens_seen": 4387128,
      "step": 6715
    },
    {
      "epoch": 3.5220125786163523,
      "grad_norm": 0.21151357889175415,
      "learning_rate": 4.912370871452383e-05,
      "loss": 0.5493,
      "num_input_tokens_seen": 4390296,
      "step": 6720
    },
    {
      "epoch": 3.5246331236897275,
      "grad_norm": 0.1955179125070572,
      "learning_rate": 4.91207053791924e-05,
      "loss": 0.5823,
      "num_input_tokens_seen": 4393624,
      "step": 6725
    },
    {
      "epoch": 3.5272536687631026,
      "grad_norm": 0.1668822318315506,
      "learning_rate": 4.911769699808457e-05,
      "loss": 0.4605,
      "num_input_tokens_seen": 4397944,
      "step": 6730
    },
    {
      "epoch": 3.529874213836478,
      "grad_norm": 0.17098093032836914,
      "learning_rate": 4.911468357182969e-05,
      "loss": 0.5334,
      "num_input_tokens_seen": 4400696,
      "step": 6735
    },
    {
      "epoch": 3.532494758909853,
      "grad_norm": 0.5478280782699585,
      "learning_rate": 4.9111665101058116e-05,
      "loss": 0.5171,
      "num_input_tokens_seen": 4403416,
      "step": 6740
    },
    {
      "epoch": 3.5351153039832286,
      "grad_norm": 0.11886532604694366,
      "learning_rate": 4.910864158640128e-05,
      "loss": 0.4521,
      "num_input_tokens_seen": 4407320,
      "step": 6745
    },
    {
      "epoch": 3.5377358490566038,
      "grad_norm": 0.230941042304039,
      "learning_rate": 4.910561302849168e-05,
      "loss": 0.4818,
      "num_input_tokens_seen": 4409848,
      "step": 6750
    },
    {
      "epoch": 3.540356394129979,
      "grad_norm": 0.2001042664051056,
      "learning_rate": 4.910257942796284e-05,
      "loss": 0.5368,
      "num_input_tokens_seen": 4412920,
      "step": 6755
    },
    {
      "epoch": 3.5429769392033545,
      "grad_norm": 0.2539082169532776,
      "learning_rate": 4.909954078544936e-05,
      "loss": 0.4626,
      "num_input_tokens_seen": 4415384,
      "step": 6760
    },
    {
      "epoch": 3.5455974842767297,
      "grad_norm": 0.2900504171848297,
      "learning_rate": 4.909649710158689e-05,
      "loss": 0.443,
      "num_input_tokens_seen": 4418040,
      "step": 6765
    },
    {
      "epoch": 3.548218029350105,
      "grad_norm": 0.13722646236419678,
      "learning_rate": 4.909344837701214e-05,
      "loss": 0.5288,
      "num_input_tokens_seen": 4421464,
      "step": 6770
    },
    {
      "epoch": 3.55083857442348,
      "grad_norm": 0.15268127620220184,
      "learning_rate": 4.909039461236285e-05,
      "loss": 0.5601,
      "num_input_tokens_seen": 4423672,
      "step": 6775
    },
    {
      "epoch": 3.5534591194968552,
      "grad_norm": 0.18091660737991333,
      "learning_rate": 4.908733580827786e-05,
      "loss": 0.5005,
      "num_input_tokens_seen": 4426648,
      "step": 6780
    },
    {
      "epoch": 3.5560796645702304,
      "grad_norm": 0.30626150965690613,
      "learning_rate": 4.9084271965397014e-05,
      "loss": 0.5447,
      "num_input_tokens_seen": 4429752,
      "step": 6785
    },
    {
      "epoch": 3.558700209643606,
      "grad_norm": 0.15973016619682312,
      "learning_rate": 4.908120308436124e-05,
      "loss": 0.6402,
      "num_input_tokens_seen": 4433784,
      "step": 6790
    },
    {
      "epoch": 3.561320754716981,
      "grad_norm": 0.250694215297699,
      "learning_rate": 4.907812916581251e-05,
      "loss": 0.5498,
      "num_input_tokens_seen": 4436376,
      "step": 6795
    },
    {
      "epoch": 3.5639412997903563,
      "grad_norm": 0.24658328294754028,
      "learning_rate": 4.907505021039386e-05,
      "loss": 0.4512,
      "num_input_tokens_seen": 4439384,
      "step": 6800
    },
    {
      "epoch": 3.5665618448637315,
      "grad_norm": 0.8366602063179016,
      "learning_rate": 4.907196621874937e-05,
      "loss": 0.5516,
      "num_input_tokens_seen": 4442712,
      "step": 6805
    },
    {
      "epoch": 3.569182389937107,
      "grad_norm": 0.22067296504974365,
      "learning_rate": 4.9068877191524176e-05,
      "loss": 0.5506,
      "num_input_tokens_seen": 4446936,
      "step": 6810
    },
    {
      "epoch": 3.5718029350104823,
      "grad_norm": 0.1663796454668045,
      "learning_rate": 4.906578312936446e-05,
      "loss": 0.5736,
      "num_input_tokens_seen": 4450328,
      "step": 6815
    },
    {
      "epoch": 3.5744234800838575,
      "grad_norm": 0.20793774724006653,
      "learning_rate": 4.906268403291747e-05,
      "loss": 0.4796,
      "num_input_tokens_seen": 4453816,
      "step": 6820
    },
    {
      "epoch": 3.5770440251572326,
      "grad_norm": 0.13375434279441833,
      "learning_rate": 4.905957990283151e-05,
      "loss": 0.432,
      "num_input_tokens_seen": 4458072,
      "step": 6825
    },
    {
      "epoch": 3.579664570230608,
      "grad_norm": 0.118258036673069,
      "learning_rate": 4.905647073975592e-05,
      "loss": 0.4637,
      "num_input_tokens_seen": 4461144,
      "step": 6830
    },
    {
      "epoch": 3.582285115303983,
      "grad_norm": 0.12645678222179413,
      "learning_rate": 4.9053356544341095e-05,
      "loss": 0.4115,
      "num_input_tokens_seen": 4463928,
      "step": 6835
    },
    {
      "epoch": 3.5849056603773586,
      "grad_norm": 0.24190738797187805,
      "learning_rate": 4.90502373172385e-05,
      "loss": 0.5184,
      "num_input_tokens_seen": 4466488,
      "step": 6840
    },
    {
      "epoch": 3.5875262054507338,
      "grad_norm": 0.1054873988032341,
      "learning_rate": 4.9047113059100635e-05,
      "loss": 0.4352,
      "num_input_tokens_seen": 4470104,
      "step": 6845
    },
    {
      "epoch": 3.590146750524109,
      "grad_norm": 0.1817968785762787,
      "learning_rate": 4.904398377058106e-05,
      "loss": 0.673,
      "num_input_tokens_seen": 4472184,
      "step": 6850
    },
    {
      "epoch": 3.5927672955974845,
      "grad_norm": 0.15892833471298218,
      "learning_rate": 4.90408494523344e-05,
      "loss": 0.5467,
      "num_input_tokens_seen": 4474840,
      "step": 6855
    },
    {
      "epoch": 3.5953878406708597,
      "grad_norm": 0.3339300751686096,
      "learning_rate": 4.90377101050163e-05,
      "loss": 0.6069,
      "num_input_tokens_seen": 4477816,
      "step": 6860
    },
    {
      "epoch": 3.598008385744235,
      "grad_norm": 0.21215331554412842,
      "learning_rate": 4.903456572928348e-05,
      "loss": 0.5921,
      "num_input_tokens_seen": 4481816,
      "step": 6865
    },
    {
      "epoch": 3.60062893081761,
      "grad_norm": 0.19049356877803802,
      "learning_rate": 4.903141632579371e-05,
      "loss": 0.4181,
      "num_input_tokens_seen": 4485464,
      "step": 6870
    },
    {
      "epoch": 3.603249475890985,
      "grad_norm": 0.28355148434638977,
      "learning_rate": 4.902826189520581e-05,
      "loss": 0.4256,
      "num_input_tokens_seen": 4490392,
      "step": 6875
    },
    {
      "epoch": 3.6058700209643604,
      "grad_norm": 0.19439131021499634,
      "learning_rate": 4.9025102438179656e-05,
      "loss": 0.5432,
      "num_input_tokens_seen": 4494712,
      "step": 6880
    },
    {
      "epoch": 3.608490566037736,
      "grad_norm": 0.1632302850484848,
      "learning_rate": 4.902193795537616e-05,
      "loss": 0.577,
      "num_input_tokens_seen": 4497528,
      "step": 6885
    },
    {
      "epoch": 3.611111111111111,
      "grad_norm": 0.13515685498714447,
      "learning_rate": 4.90187684474573e-05,
      "loss": 0.4488,
      "num_input_tokens_seen": 4500536,
      "step": 6890
    },
    {
      "epoch": 3.6137316561844863,
      "grad_norm": 0.19504041969776154,
      "learning_rate": 4.901559391508611e-05,
      "loss": 0.5517,
      "num_input_tokens_seen": 4504056,
      "step": 6895
    },
    {
      "epoch": 3.6163522012578615,
      "grad_norm": 0.19265447556972504,
      "learning_rate": 4.9012414358926644e-05,
      "loss": 0.5188,
      "num_input_tokens_seen": 4506936,
      "step": 6900
    },
    {
      "epoch": 3.618972746331237,
      "grad_norm": 0.151223823428154,
      "learning_rate": 4.900922977964405e-05,
      "loss": 0.4104,
      "num_input_tokens_seen": 4509560,
      "step": 6905
    },
    {
      "epoch": 3.6215932914046123,
      "grad_norm": 0.21487511694431305,
      "learning_rate": 4.90060401779045e-05,
      "loss": 0.5241,
      "num_input_tokens_seen": 4513240,
      "step": 6910
    },
    {
      "epoch": 3.6242138364779874,
      "grad_norm": 0.17934496700763702,
      "learning_rate": 4.900284555437521e-05,
      "loss": 0.5319,
      "num_input_tokens_seen": 4515640,
      "step": 6915
    },
    {
      "epoch": 3.6268343815513626,
      "grad_norm": 0.12797154486179352,
      "learning_rate": 4.899964590972448e-05,
      "loss": 0.4524,
      "num_input_tokens_seen": 4520824,
      "step": 6920
    },
    {
      "epoch": 3.629454926624738,
      "grad_norm": 0.1584351360797882,
      "learning_rate": 4.899644124462163e-05,
      "loss": 0.5267,
      "num_input_tokens_seen": 4523896,
      "step": 6925
    },
    {
      "epoch": 3.632075471698113,
      "grad_norm": 0.1832897812128067,
      "learning_rate": 4.8993231559737026e-05,
      "loss": 0.6553,
      "num_input_tokens_seen": 4527000,
      "step": 6930
    },
    {
      "epoch": 3.6346960167714886,
      "grad_norm": 0.29078301787376404,
      "learning_rate": 4.8990016855742114e-05,
      "loss": 0.6278,
      "num_input_tokens_seen": 4529912,
      "step": 6935
    },
    {
      "epoch": 3.6373165618448637,
      "grad_norm": 0.2691931426525116,
      "learning_rate": 4.898679713330937e-05,
      "loss": 0.479,
      "num_input_tokens_seen": 4533336,
      "step": 6940
    },
    {
      "epoch": 3.639937106918239,
      "grad_norm": 0.13608726859092712,
      "learning_rate": 4.898357239311232e-05,
      "loss": 0.5341,
      "num_input_tokens_seen": 4536344,
      "step": 6945
    },
    {
      "epoch": 3.6425576519916145,
      "grad_norm": 0.17688046395778656,
      "learning_rate": 4.898034263582554e-05,
      "loss": 0.4581,
      "num_input_tokens_seen": 4540152,
      "step": 6950
    },
    {
      "epoch": 3.6451781970649897,
      "grad_norm": 0.19341319799423218,
      "learning_rate": 4.897710786212467e-05,
      "loss": 0.3761,
      "num_input_tokens_seen": 4543032,
      "step": 6955
    },
    {
      "epoch": 3.647798742138365,
      "grad_norm": 0.25433072447776794,
      "learning_rate": 4.897386807268638e-05,
      "loss": 0.5325,
      "num_input_tokens_seen": 4547064,
      "step": 6960
    },
    {
      "epoch": 3.65041928721174,
      "grad_norm": 0.1587563008069992,
      "learning_rate": 4.897062326818839e-05,
      "loss": 0.5778,
      "num_input_tokens_seen": 4551000,
      "step": 6965
    },
    {
      "epoch": 3.653039832285115,
      "grad_norm": 0.11446674168109894,
      "learning_rate": 4.8967373449309495e-05,
      "loss": 0.5794,
      "num_input_tokens_seen": 4555000,
      "step": 6970
    },
    {
      "epoch": 3.6556603773584904,
      "grad_norm": 0.1177559420466423,
      "learning_rate": 4.89641186167295e-05,
      "loss": 0.4669,
      "num_input_tokens_seen": 4558680,
      "step": 6975
    },
    {
      "epoch": 3.658280922431866,
      "grad_norm": 0.15595462918281555,
      "learning_rate": 4.8960858771129294e-05,
      "loss": 0.4893,
      "num_input_tokens_seen": 4561336,
      "step": 6980
    },
    {
      "epoch": 3.660901467505241,
      "grad_norm": 0.14751221239566803,
      "learning_rate": 4.895759391319079e-05,
      "loss": 0.5336,
      "num_input_tokens_seen": 4564568,
      "step": 6985
    },
    {
      "epoch": 3.6635220125786163,
      "grad_norm": 0.1959979087114334,
      "learning_rate": 4.8954324043596954e-05,
      "loss": 0.4667,
      "num_input_tokens_seen": 4567448,
      "step": 6990
    },
    {
      "epoch": 3.6661425576519915,
      "grad_norm": 0.11089068651199341,
      "learning_rate": 4.8951049163031826e-05,
      "loss": 0.6287,
      "num_input_tokens_seen": 4570136,
      "step": 6995
    },
    {
      "epoch": 3.668763102725367,
      "grad_norm": 0.17865322530269623,
      "learning_rate": 4.894776927218045e-05,
      "loss": 0.5239,
      "num_input_tokens_seen": 4573624,
      "step": 7000
    },
    {
      "epoch": 3.6713836477987423,
      "grad_norm": 1.460176706314087,
      "learning_rate": 4.894448437172896e-05,
      "loss": 0.507,
      "num_input_tokens_seen": 4577048,
      "step": 7005
    },
    {
      "epoch": 3.6740041928721174,
      "grad_norm": 0.15869931876659393,
      "learning_rate": 4.894119446236452e-05,
      "loss": 0.5679,
      "num_input_tokens_seen": 4580088,
      "step": 7010
    },
    {
      "epoch": 3.6766247379454926,
      "grad_norm": 0.17477408051490784,
      "learning_rate": 4.893789954477532e-05,
      "loss": 0.4544,
      "num_input_tokens_seen": 4583480,
      "step": 7015
    },
    {
      "epoch": 3.6792452830188678,
      "grad_norm": 0.17931589484214783,
      "learning_rate": 4.893459961965063e-05,
      "loss": 0.7111,
      "num_input_tokens_seen": 4586168,
      "step": 7020
    },
    {
      "epoch": 3.681865828092243,
      "grad_norm": 0.16265183687210083,
      "learning_rate": 4.8931294687680775e-05,
      "loss": 0.429,
      "num_input_tokens_seen": 4589592,
      "step": 7025
    },
    {
      "epoch": 3.6844863731656186,
      "grad_norm": 0.2628515362739563,
      "learning_rate": 4.8927984749557086e-05,
      "loss": 0.6029,
      "num_input_tokens_seen": 4592440,
      "step": 7030
    },
    {
      "epoch": 3.6871069182389937,
      "grad_norm": 0.18453630805015564,
      "learning_rate": 4.892466980597198e-05,
      "loss": 0.4105,
      "num_input_tokens_seen": 4594904,
      "step": 7035
    },
    {
      "epoch": 3.689727463312369,
      "grad_norm": 0.165715292096138,
      "learning_rate": 4.892134985761889e-05,
      "loss": 0.5953,
      "num_input_tokens_seen": 4597624,
      "step": 7040
    },
    {
      "epoch": 3.6923480083857445,
      "grad_norm": 0.3012189269065857,
      "learning_rate": 4.891802490519232e-05,
      "loss": 0.601,
      "num_input_tokens_seen": 4599928,
      "step": 7045
    },
    {
      "epoch": 3.6949685534591197,
      "grad_norm": 0.33204206824302673,
      "learning_rate": 4.891469494938781e-05,
      "loss": 0.5427,
      "num_input_tokens_seen": 4602328,
      "step": 7050
    },
    {
      "epoch": 3.697589098532495,
      "grad_norm": 0.13746879994869232,
      "learning_rate": 4.891135999090195e-05,
      "loss": 0.5446,
      "num_input_tokens_seen": 4606328,
      "step": 7055
    },
    {
      "epoch": 3.70020964360587,
      "grad_norm": 0.16492441296577454,
      "learning_rate": 4.890802003043238e-05,
      "loss": 0.4238,
      "num_input_tokens_seen": 4609176,
      "step": 7060
    },
    {
      "epoch": 3.702830188679245,
      "grad_norm": 0.13651001453399658,
      "learning_rate": 4.890467506867777e-05,
      "loss": 0.6036,
      "num_input_tokens_seen": 4612056,
      "step": 7065
    },
    {
      "epoch": 3.7054507337526204,
      "grad_norm": 0.21622426807880402,
      "learning_rate": 4.890132510633786e-05,
      "loss": 0.6439,
      "num_input_tokens_seen": 4615320,
      "step": 7070
    },
    {
      "epoch": 3.708071278825996,
      "grad_norm": 0.177177295088768,
      "learning_rate": 4.8897970144113416e-05,
      "loss": 0.5959,
      "num_input_tokens_seen": 4618200,
      "step": 7075
    },
    {
      "epoch": 3.710691823899371,
      "grad_norm": 0.2334965467453003,
      "learning_rate": 4.889461018270626e-05,
      "loss": 0.3841,
      "num_input_tokens_seen": 4621368,
      "step": 7080
    },
    {
      "epoch": 3.7133123689727463,
      "grad_norm": 0.15539178252220154,
      "learning_rate": 4.889124522281926e-05,
      "loss": 0.5673,
      "num_input_tokens_seen": 4624088,
      "step": 7085
    },
    {
      "epoch": 3.7159329140461215,
      "grad_norm": 0.15187126398086548,
      "learning_rate": 4.888787526515632e-05,
      "loss": 0.6493,
      "num_input_tokens_seen": 4628568,
      "step": 7090
    },
    {
      "epoch": 3.718553459119497,
      "grad_norm": 0.16332072019577026,
      "learning_rate": 4.8884500310422406e-05,
      "loss": 0.5752,
      "num_input_tokens_seen": 4631288,
      "step": 7095
    },
    {
      "epoch": 3.7211740041928723,
      "grad_norm": 0.1439974159002304,
      "learning_rate": 4.8881120359323516e-05,
      "loss": 0.4254,
      "num_input_tokens_seen": 4635224,
      "step": 7100
    },
    {
      "epoch": 3.7237945492662474,
      "grad_norm": 0.15671558678150177,
      "learning_rate": 4.887773541256669e-05,
      "loss": 0.7403,
      "num_input_tokens_seen": 4638648,
      "step": 7105
    },
    {
      "epoch": 3.7264150943396226,
      "grad_norm": 0.2531093955039978,
      "learning_rate": 4.887434547086004e-05,
      "loss": 0.4165,
      "num_input_tokens_seen": 4641848,
      "step": 7110
    },
    {
      "epoch": 3.7290356394129978,
      "grad_norm": 0.1757834106683731,
      "learning_rate": 4.887095053491269e-05,
      "loss": 0.537,
      "num_input_tokens_seen": 4644408,
      "step": 7115
    },
    {
      "epoch": 3.731656184486373,
      "grad_norm": 0.14330080151557922,
      "learning_rate": 4.8867550605434806e-05,
      "loss": 0.6691,
      "num_input_tokens_seen": 4647096,
      "step": 7120
    },
    {
      "epoch": 3.7342767295597485,
      "grad_norm": 0.1871563196182251,
      "learning_rate": 4.8864145683137644e-05,
      "loss": 0.5528,
      "num_input_tokens_seen": 4649176,
      "step": 7125
    },
    {
      "epoch": 3.7368972746331237,
      "grad_norm": 0.21507203578948975,
      "learning_rate": 4.886073576873346e-05,
      "loss": 0.5489,
      "num_input_tokens_seen": 4653144,
      "step": 7130
    },
    {
      "epoch": 3.739517819706499,
      "grad_norm": 0.14086538553237915,
      "learning_rate": 4.885732086293556e-05,
      "loss": 0.3977,
      "num_input_tokens_seen": 4655864,
      "step": 7135
    },
    {
      "epoch": 3.742138364779874,
      "grad_norm": 0.13338029384613037,
      "learning_rate": 4.8853900966458335e-05,
      "loss": 0.5311,
      "num_input_tokens_seen": 4659000,
      "step": 7140
    },
    {
      "epoch": 3.7447589098532497,
      "grad_norm": 0.11009198427200317,
      "learning_rate": 4.8850476080017146e-05,
      "loss": 0.5688,
      "num_input_tokens_seen": 4661720,
      "step": 7145
    },
    {
      "epoch": 3.747379454926625,
      "grad_norm": 0.32814478874206543,
      "learning_rate": 4.884704620432848e-05,
      "loss": 0.4967,
      "num_input_tokens_seen": 4666584,
      "step": 7150
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.1964779645204544,
      "learning_rate": 4.8843611340109794e-05,
      "loss": 0.5037,
      "num_input_tokens_seen": 4669272,
      "step": 7155
    },
    {
      "epoch": 3.752620545073375,
      "grad_norm": 0.23328696191310883,
      "learning_rate": 4.884017148807965e-05,
      "loss": 0.5331,
      "num_input_tokens_seen": 4672280,
      "step": 7160
    },
    {
      "epoch": 3.7552410901467503,
      "grad_norm": 0.18261465430259705,
      "learning_rate": 4.883672664895761e-05,
      "loss": 0.4035,
      "num_input_tokens_seen": 4675032,
      "step": 7165
    },
    {
      "epoch": 3.757861635220126,
      "grad_norm": 0.6005739569664001,
      "learning_rate": 4.8833276823464296e-05,
      "loss": 0.5253,
      "num_input_tokens_seen": 4678360,
      "step": 7170
    },
    {
      "epoch": 3.760482180293501,
      "grad_norm": 0.42409607768058777,
      "learning_rate": 4.882982201232138e-05,
      "loss": 0.5441,
      "num_input_tokens_seen": 4681208,
      "step": 7175
    },
    {
      "epoch": 3.7631027253668763,
      "grad_norm": 0.1368582546710968,
      "learning_rate": 4.882636221625156e-05,
      "loss": 0.5034,
      "num_input_tokens_seen": 4683768,
      "step": 7180
    },
    {
      "epoch": 3.7657232704402515,
      "grad_norm": 0.2220827043056488,
      "learning_rate": 4.8822897435978586e-05,
      "loss": 0.514,
      "num_input_tokens_seen": 4686872,
      "step": 7185
    },
    {
      "epoch": 3.768343815513627,
      "grad_norm": 0.14252004027366638,
      "learning_rate": 4.8819427672227264e-05,
      "loss": 0.5019,
      "num_input_tokens_seen": 4690424,
      "step": 7190
    },
    {
      "epoch": 3.7709643605870022,
      "grad_norm": 0.23399944603443146,
      "learning_rate": 4.881595292572341e-05,
      "loss": 0.6066,
      "num_input_tokens_seen": 4692824,
      "step": 7195
    },
    {
      "epoch": 3.7735849056603774,
      "grad_norm": 0.18168966472148895,
      "learning_rate": 4.881247319719392e-05,
      "loss": 0.5646,
      "num_input_tokens_seen": 4696440,
      "step": 7200
    },
    {
      "epoch": 3.7762054507337526,
      "grad_norm": 0.16406580805778503,
      "learning_rate": 4.88089884873667e-05,
      "loss": 0.4662,
      "num_input_tokens_seen": 4700440,
      "step": 7205
    },
    {
      "epoch": 3.7788259958071277,
      "grad_norm": 0.15097872912883759,
      "learning_rate": 4.880549879697071e-05,
      "loss": 0.4972,
      "num_input_tokens_seen": 4703320,
      "step": 7210
    },
    {
      "epoch": 3.781446540880503,
      "grad_norm": 0.16815418004989624,
      "learning_rate": 4.8802004126735965e-05,
      "loss": 0.6106,
      "num_input_tokens_seen": 4706552,
      "step": 7215
    },
    {
      "epoch": 3.7840670859538785,
      "grad_norm": 0.10384935885667801,
      "learning_rate": 4.87985044773935e-05,
      "loss": 0.4364,
      "num_input_tokens_seen": 4709592,
      "step": 7220
    },
    {
      "epoch": 3.7866876310272537,
      "grad_norm": 0.22474001348018646,
      "learning_rate": 4.879499984967541e-05,
      "loss": 0.6288,
      "num_input_tokens_seen": 4712344,
      "step": 7225
    },
    {
      "epoch": 3.789308176100629,
      "grad_norm": 0.26561570167541504,
      "learning_rate": 4.8791490244314824e-05,
      "loss": 0.5855,
      "num_input_tokens_seen": 4715000,
      "step": 7230
    },
    {
      "epoch": 3.791928721174004,
      "grad_norm": 0.13905632495880127,
      "learning_rate": 4.8787975662045904e-05,
      "loss": 0.4899,
      "num_input_tokens_seen": 4718904,
      "step": 7235
    },
    {
      "epoch": 3.7945492662473796,
      "grad_norm": 0.23579831421375275,
      "learning_rate": 4.878445610360386e-05,
      "loss": 0.6064,
      "num_input_tokens_seen": 4721656,
      "step": 7240
    },
    {
      "epoch": 3.797169811320755,
      "grad_norm": 0.4675476551055908,
      "learning_rate": 4.878093156972494e-05,
      "loss": 0.6181,
      "num_input_tokens_seen": 4726264,
      "step": 7245
    },
    {
      "epoch": 3.79979035639413,
      "grad_norm": 0.18061240017414093,
      "learning_rate": 4.877740206114645e-05,
      "loss": 0.4661,
      "num_input_tokens_seen": 4729112,
      "step": 7250
    },
    {
      "epoch": 3.802410901467505,
      "grad_norm": 0.1741410195827484,
      "learning_rate": 4.8773867578606716e-05,
      "loss": 0.5983,
      "num_input_tokens_seen": 4731832,
      "step": 7255
    },
    {
      "epoch": 3.8050314465408803,
      "grad_norm": 0.2091277837753296,
      "learning_rate": 4.877032812284511e-05,
      "loss": 0.639,
      "num_input_tokens_seen": 4738136,
      "step": 7260
    },
    {
      "epoch": 3.8076519916142555,
      "grad_norm": 0.13443362712860107,
      "learning_rate": 4.876678369460205e-05,
      "loss": 0.4659,
      "num_input_tokens_seen": 4741080,
      "step": 7265
    },
    {
      "epoch": 3.810272536687631,
      "grad_norm": 0.19349054992198944,
      "learning_rate": 4.8763234294618985e-05,
      "loss": 0.4201,
      "num_input_tokens_seen": 4744472,
      "step": 7270
    },
    {
      "epoch": 3.8128930817610063,
      "grad_norm": 0.16662479937076569,
      "learning_rate": 4.875967992363841e-05,
      "loss": 0.4286,
      "num_input_tokens_seen": 4748664,
      "step": 7275
    },
    {
      "epoch": 3.8155136268343814,
      "grad_norm": 0.3199939727783203,
      "learning_rate": 4.875612058240386e-05,
      "loss": 0.5488,
      "num_input_tokens_seen": 4751704,
      "step": 7280
    },
    {
      "epoch": 3.818134171907757,
      "grad_norm": 0.6267039775848389,
      "learning_rate": 4.875255627165991e-05,
      "loss": 0.6172,
      "num_input_tokens_seen": 4754104,
      "step": 7285
    },
    {
      "epoch": 3.8207547169811322,
      "grad_norm": 0.202953040599823,
      "learning_rate": 4.874898699215217e-05,
      "loss": 0.4816,
      "num_input_tokens_seen": 4757528,
      "step": 7290
    },
    {
      "epoch": 3.8233752620545074,
      "grad_norm": 0.10100220888853073,
      "learning_rate": 4.8745412744627296e-05,
      "loss": 0.6649,
      "num_input_tokens_seen": 4761208,
      "step": 7295
    },
    {
      "epoch": 3.8259958071278826,
      "grad_norm": 0.16438454389572144,
      "learning_rate": 4.874183352983297e-05,
      "loss": 0.4933,
      "num_input_tokens_seen": 4764088,
      "step": 7300
    },
    {
      "epoch": 3.8286163522012577,
      "grad_norm": 0.1061871275305748,
      "learning_rate": 4.873824934851794e-05,
      "loss": 0.5219,
      "num_input_tokens_seen": 4767352,
      "step": 7305
    },
    {
      "epoch": 3.831236897274633,
      "grad_norm": 0.17354880273342133,
      "learning_rate": 4.873466020143196e-05,
      "loss": 0.509,
      "num_input_tokens_seen": 4770648,
      "step": 7310
    },
    {
      "epoch": 3.8338574423480085,
      "grad_norm": 0.17704221606254578,
      "learning_rate": 4.873106608932585e-05,
      "loss": 0.5199,
      "num_input_tokens_seen": 4777624,
      "step": 7315
    },
    {
      "epoch": 3.8364779874213837,
      "grad_norm": 0.26158949732780457,
      "learning_rate": 4.872746701295145e-05,
      "loss": 0.6245,
      "num_input_tokens_seen": 4780376,
      "step": 7320
    },
    {
      "epoch": 3.839098532494759,
      "grad_norm": 0.1525159478187561,
      "learning_rate": 4.872386297306165e-05,
      "loss": 0.5553,
      "num_input_tokens_seen": 4783096,
      "step": 7325
    },
    {
      "epoch": 3.841719077568134,
      "grad_norm": 0.12908625602722168,
      "learning_rate": 4.872025397041037e-05,
      "loss": 0.3723,
      "num_input_tokens_seen": 4787000,
      "step": 7330
    },
    {
      "epoch": 3.8443396226415096,
      "grad_norm": 0.2845613360404968,
      "learning_rate": 4.871664000575258e-05,
      "loss": 0.543,
      "num_input_tokens_seen": 4790168,
      "step": 7335
    },
    {
      "epoch": 3.846960167714885,
      "grad_norm": 0.13846085965633392,
      "learning_rate": 4.871302107984427e-05,
      "loss": 0.5683,
      "num_input_tokens_seen": 4793944,
      "step": 7340
    },
    {
      "epoch": 3.84958071278826,
      "grad_norm": 0.18145637214183807,
      "learning_rate": 4.8709397193442474e-05,
      "loss": 0.4908,
      "num_input_tokens_seen": 4797016,
      "step": 7345
    },
    {
      "epoch": 3.852201257861635,
      "grad_norm": 0.11879421025514603,
      "learning_rate": 4.870576834730529e-05,
      "loss": 0.501,
      "num_input_tokens_seen": 4800216,
      "step": 7350
    },
    {
      "epoch": 3.8548218029350103,
      "grad_norm": 0.155942901968956,
      "learning_rate": 4.8702134542191816e-05,
      "loss": 0.4284,
      "num_input_tokens_seen": 4803992,
      "step": 7355
    },
    {
      "epoch": 3.8574423480083855,
      "grad_norm": 0.20163480937480927,
      "learning_rate": 4.86984957788622e-05,
      "loss": 0.5134,
      "num_input_tokens_seen": 4806872,
      "step": 7360
    },
    {
      "epoch": 3.860062893081761,
      "grad_norm": 0.18358491361141205,
      "learning_rate": 4.869485205807763e-05,
      "loss": 0.3399,
      "num_input_tokens_seen": 4809912,
      "step": 7365
    },
    {
      "epoch": 3.8626834381551363,
      "grad_norm": 0.1402786672115326,
      "learning_rate": 4.869120338060034e-05,
      "loss": 0.3629,
      "num_input_tokens_seen": 4816984,
      "step": 7370
    },
    {
      "epoch": 3.8653039832285114,
      "grad_norm": 0.13095517456531525,
      "learning_rate": 4.8687549747193585e-05,
      "loss": 0.5843,
      "num_input_tokens_seen": 4819608,
      "step": 7375
    },
    {
      "epoch": 3.867924528301887,
      "grad_norm": 0.09328384697437286,
      "learning_rate": 4.868389115862166e-05,
      "loss": 0.4182,
      "num_input_tokens_seen": 4823448,
      "step": 7380
    },
    {
      "epoch": 3.870545073375262,
      "grad_norm": 0.19095659255981445,
      "learning_rate": 4.868022761564991e-05,
      "loss": 0.5451,
      "num_input_tokens_seen": 4826328,
      "step": 7385
    },
    {
      "epoch": 3.8731656184486374,
      "grad_norm": 0.23552732169628143,
      "learning_rate": 4.8676559119044696e-05,
      "loss": 0.528,
      "num_input_tokens_seen": 4829048,
      "step": 7390
    },
    {
      "epoch": 3.8757861635220126,
      "grad_norm": 0.16711834073066711,
      "learning_rate": 4.867288566957344e-05,
      "loss": 0.4572,
      "num_input_tokens_seen": 4833048,
      "step": 7395
    },
    {
      "epoch": 3.8784067085953877,
      "grad_norm": 0.12220178544521332,
      "learning_rate": 4.866920726800457e-05,
      "loss": 0.4408,
      "num_input_tokens_seen": 4836184,
      "step": 7400
    },
    {
      "epoch": 3.881027253668763,
      "grad_norm": 0.27054715156555176,
      "learning_rate": 4.8665523915107576e-05,
      "loss": 0.5329,
      "num_input_tokens_seen": 4839896,
      "step": 7405
    },
    {
      "epoch": 3.8836477987421385,
      "grad_norm": 0.34978121519088745,
      "learning_rate": 4.866183561165296e-05,
      "loss": 0.4956,
      "num_input_tokens_seen": 4842936,
      "step": 7410
    },
    {
      "epoch": 3.8862683438155137,
      "grad_norm": 0.19046516716480255,
      "learning_rate": 4.86581423584123e-05,
      "loss": 0.6906,
      "num_input_tokens_seen": 4846200,
      "step": 7415
    },
    {
      "epoch": 3.888888888888889,
      "grad_norm": 0.17882490158081055,
      "learning_rate": 4.8654444156158154e-05,
      "loss": 0.4966,
      "num_input_tokens_seen": 4849336,
      "step": 7420
    },
    {
      "epoch": 3.891509433962264,
      "grad_norm": 0.32539454102516174,
      "learning_rate": 4.865074100566415e-05,
      "loss": 0.5368,
      "num_input_tokens_seen": 4852632,
      "step": 7425
    },
    {
      "epoch": 3.8941299790356396,
      "grad_norm": 0.19609417021274567,
      "learning_rate": 4.864703290770495e-05,
      "loss": 0.5722,
      "num_input_tokens_seen": 4855384,
      "step": 7430
    },
    {
      "epoch": 3.896750524109015,
      "grad_norm": 0.15354293584823608,
      "learning_rate": 4.864331986305626e-05,
      "loss": 0.5398,
      "num_input_tokens_seen": 4858936,
      "step": 7435
    },
    {
      "epoch": 3.89937106918239,
      "grad_norm": 0.20201966166496277,
      "learning_rate": 4.8639601872494775e-05,
      "loss": 0.5313,
      "num_input_tokens_seen": 4862488,
      "step": 7440
    },
    {
      "epoch": 3.901991614255765,
      "grad_norm": 0.16359266638755798,
      "learning_rate": 4.8635878936798287e-05,
      "loss": 0.4747,
      "num_input_tokens_seen": 4865240,
      "step": 7445
    },
    {
      "epoch": 3.9046121593291403,
      "grad_norm": 0.1357192099094391,
      "learning_rate": 4.863215105674557e-05,
      "loss": 0.491,
      "num_input_tokens_seen": 4868152,
      "step": 7450
    },
    {
      "epoch": 3.9072327044025155,
      "grad_norm": 0.20879939198493958,
      "learning_rate": 4.8628418233116457e-05,
      "loss": 0.4161,
      "num_input_tokens_seen": 4870808,
      "step": 7455
    },
    {
      "epoch": 3.909853249475891,
      "grad_norm": 0.12471804767847061,
      "learning_rate": 4.8624680466691824e-05,
      "loss": 0.5077,
      "num_input_tokens_seen": 4873944,
      "step": 7460
    },
    {
      "epoch": 3.9124737945492662,
      "grad_norm": 0.20246216654777527,
      "learning_rate": 4.862093775825356e-05,
      "loss": 0.5916,
      "num_input_tokens_seen": 4877528,
      "step": 7465
    },
    {
      "epoch": 3.9150943396226414,
      "grad_norm": 0.17784737050533295,
      "learning_rate": 4.86171901085846e-05,
      "loss": 0.4679,
      "num_input_tokens_seen": 4880504,
      "step": 7470
    },
    {
      "epoch": 3.917714884696017,
      "grad_norm": 0.1603521853685379,
      "learning_rate": 4.861343751846891e-05,
      "loss": 0.5077,
      "num_input_tokens_seen": 4884312,
      "step": 7475
    },
    {
      "epoch": 3.920335429769392,
      "grad_norm": 0.24258063733577728,
      "learning_rate": 4.860967998869148e-05,
      "loss": 0.4204,
      "num_input_tokens_seen": 4887032,
      "step": 7480
    },
    {
      "epoch": 3.9229559748427674,
      "grad_norm": 0.1915961056947708,
      "learning_rate": 4.860591752003836e-05,
      "loss": 0.5518,
      "num_input_tokens_seen": 4889976,
      "step": 7485
    },
    {
      "epoch": 3.9255765199161425,
      "grad_norm": 0.1316552609205246,
      "learning_rate": 4.8602150113296596e-05,
      "loss": 0.5881,
      "num_input_tokens_seen": 4892600,
      "step": 7490
    },
    {
      "epoch": 3.9281970649895177,
      "grad_norm": 0.31696900725364685,
      "learning_rate": 4.8598377769254305e-05,
      "loss": 0.5032,
      "num_input_tokens_seen": 4895640,
      "step": 7495
    },
    {
      "epoch": 3.930817610062893,
      "grad_norm": 0.14521050453186035,
      "learning_rate": 4.85946004887006e-05,
      "loss": 0.4361,
      "num_input_tokens_seen": 4898200,
      "step": 7500
    },
    {
      "epoch": 3.9334381551362685,
      "grad_norm": 0.25078922510147095,
      "learning_rate": 4.859081827242566e-05,
      "loss": 0.4894,
      "num_input_tokens_seen": 4901080,
      "step": 7505
    },
    {
      "epoch": 3.9360587002096437,
      "grad_norm": 0.16245388984680176,
      "learning_rate": 4.858703112122068e-05,
      "loss": 0.5596,
      "num_input_tokens_seen": 4904280,
      "step": 7510
    },
    {
      "epoch": 3.938679245283019,
      "grad_norm": 0.15413285791873932,
      "learning_rate": 4.8583239035877876e-05,
      "loss": 0.7482,
      "num_input_tokens_seen": 4907832,
      "step": 7515
    },
    {
      "epoch": 3.941299790356394,
      "grad_norm": 0.27684080600738525,
      "learning_rate": 4.8579442017190526e-05,
      "loss": 0.6743,
      "num_input_tokens_seen": 4911320,
      "step": 7520
    },
    {
      "epoch": 3.9439203354297696,
      "grad_norm": 0.11516035348176956,
      "learning_rate": 4.8575640065952905e-05,
      "loss": 0.499,
      "num_input_tokens_seen": 4914072,
      "step": 7525
    },
    {
      "epoch": 3.9465408805031448,
      "grad_norm": 0.2243247628211975,
      "learning_rate": 4.857183318296035e-05,
      "loss": 0.5566,
      "num_input_tokens_seen": 4918584,
      "step": 7530
    },
    {
      "epoch": 3.94916142557652,
      "grad_norm": 0.20161086320877075,
      "learning_rate": 4.856802136900922e-05,
      "loss": 0.4952,
      "num_input_tokens_seen": 4921496,
      "step": 7535
    },
    {
      "epoch": 3.951781970649895,
      "grad_norm": 0.1726064383983612,
      "learning_rate": 4.856420462489689e-05,
      "loss": 0.5452,
      "num_input_tokens_seen": 4924600,
      "step": 7540
    },
    {
      "epoch": 3.9544025157232703,
      "grad_norm": 0.1029500737786293,
      "learning_rate": 4.856038295142179e-05,
      "loss": 0.5008,
      "num_input_tokens_seen": 4927512,
      "step": 7545
    },
    {
      "epoch": 3.9570230607966455,
      "grad_norm": 0.23338207602500916,
      "learning_rate": 4.855655634938337e-05,
      "loss": 0.4897,
      "num_input_tokens_seen": 4930360,
      "step": 7550
    },
    {
      "epoch": 3.959643605870021,
      "grad_norm": 0.14037495851516724,
      "learning_rate": 4.8552724819582106e-05,
      "loss": 0.436,
      "num_input_tokens_seen": 4933496,
      "step": 7555
    },
    {
      "epoch": 3.9622641509433962,
      "grad_norm": 0.16859546303749084,
      "learning_rate": 4.8548888362819515e-05,
      "loss": 0.4202,
      "num_input_tokens_seen": 4936152,
      "step": 7560
    },
    {
      "epoch": 3.9648846960167714,
      "grad_norm": 0.2740533649921417,
      "learning_rate": 4.854504697989814e-05,
      "loss": 0.7357,
      "num_input_tokens_seen": 4939832,
      "step": 7565
    },
    {
      "epoch": 3.967505241090147,
      "grad_norm": 0.13508014380931854,
      "learning_rate": 4.8541200671621554e-05,
      "loss": 0.5589,
      "num_input_tokens_seen": 4943192,
      "step": 7570
    },
    {
      "epoch": 3.970125786163522,
      "grad_norm": 0.275100976228714,
      "learning_rate": 4.8537349438794354e-05,
      "loss": 0.5035,
      "num_input_tokens_seen": 4946392,
      "step": 7575
    },
    {
      "epoch": 3.9727463312368974,
      "grad_norm": 0.13629843294620514,
      "learning_rate": 4.853349328222219e-05,
      "loss": 0.4124,
      "num_input_tokens_seen": 4949464,
      "step": 7580
    },
    {
      "epoch": 3.9753668763102725,
      "grad_norm": 0.21756844222545624,
      "learning_rate": 4.8529632202711706e-05,
      "loss": 0.4371,
      "num_input_tokens_seen": 4955288,
      "step": 7585
    },
    {
      "epoch": 3.9779874213836477,
      "grad_norm": 0.16330724954605103,
      "learning_rate": 4.852576620107061e-05,
      "loss": 0.4289,
      "num_input_tokens_seen": 4958008,
      "step": 7590
    },
    {
      "epoch": 3.980607966457023,
      "grad_norm": 0.14121758937835693,
      "learning_rate": 4.8521895278107615e-05,
      "loss": 0.3918,
      "num_input_tokens_seen": 4961400,
      "step": 7595
    },
    {
      "epoch": 3.9832285115303985,
      "grad_norm": 0.215262770652771,
      "learning_rate": 4.8518019434632484e-05,
      "loss": 0.6333,
      "num_input_tokens_seen": 4964536,
      "step": 7600
    },
    {
      "epoch": 3.9858490566037736,
      "grad_norm": 0.1861032247543335,
      "learning_rate": 4.8514138671455995e-05,
      "loss": 0.6378,
      "num_input_tokens_seen": 4968280,
      "step": 7605
    },
    {
      "epoch": 3.988469601677149,
      "grad_norm": 0.11788459867238998,
      "learning_rate": 4.851025298938996e-05,
      "loss": 0.6025,
      "num_input_tokens_seen": 4971640,
      "step": 7610
    },
    {
      "epoch": 3.991090146750524,
      "grad_norm": 0.1484335958957672,
      "learning_rate": 4.850636238924722e-05,
      "loss": 0.528,
      "num_input_tokens_seen": 4975512,
      "step": 7615
    },
    {
      "epoch": 3.9937106918238996,
      "grad_norm": 0.21114809811115265,
      "learning_rate": 4.850246687184163e-05,
      "loss": 0.6185,
      "num_input_tokens_seen": 4978520,
      "step": 7620
    },
    {
      "epoch": 3.9963312368972748,
      "grad_norm": 0.1453186273574829,
      "learning_rate": 4.849856643798812e-05,
      "loss": 0.4487,
      "num_input_tokens_seen": 4981944,
      "step": 7625
    },
    {
      "epoch": 3.99895178197065,
      "grad_norm": 0.30873408913612366,
      "learning_rate": 4.849466108850259e-05,
      "loss": 0.4696,
      "num_input_tokens_seen": 4984632,
      "step": 7630
    },
    {
      "epoch": 4.0,
      "eval_loss": 0.5269851088523865,
      "eval_runtime": 10.0241,
      "eval_samples_per_second": 84.596,
      "eval_steps_per_second": 21.149,
      "num_input_tokens_seen": 4985200,
      "step": 7632
    },
    {
      "epoch": 4.001572327044025,
      "grad_norm": 0.1328001469373703,
      "learning_rate": 4.8490750824202e-05,
      "loss": 0.4953,
      "num_input_tokens_seen": 4987024,
      "step": 7635
    },
    {
      "epoch": 4.0041928721174,
      "grad_norm": 0.14263689517974854,
      "learning_rate": 4.8486835645904324e-05,
      "loss": 0.4755,
      "num_input_tokens_seen": 4989552,
      "step": 7640
    },
    {
      "epoch": 4.006813417190775,
      "grad_norm": 0.19552494585514069,
      "learning_rate": 4.84829155544286e-05,
      "loss": 0.7213,
      "num_input_tokens_seen": 4992080,
      "step": 7645
    },
    {
      "epoch": 4.009433962264151,
      "grad_norm": 0.26289471983909607,
      "learning_rate": 4.847899055059484e-05,
      "loss": 0.7218,
      "num_input_tokens_seen": 4994512,
      "step": 7650
    },
    {
      "epoch": 4.012054507337526,
      "grad_norm": 0.2321806699037552,
      "learning_rate": 4.847506063522412e-05,
      "loss": 0.3398,
      "num_input_tokens_seen": 4998000,
      "step": 7655
    },
    {
      "epoch": 4.014675052410902,
      "grad_norm": 0.20116610825061798,
      "learning_rate": 4.847112580913854e-05,
      "loss": 0.4913,
      "num_input_tokens_seen": 5002000,
      "step": 7660
    },
    {
      "epoch": 4.017295597484277,
      "grad_norm": 0.12488538771867752,
      "learning_rate": 4.846718607316121e-05,
      "loss": 0.5211,
      "num_input_tokens_seen": 5005648,
      "step": 7665
    },
    {
      "epoch": 4.019916142557652,
      "grad_norm": 0.24369074404239655,
      "learning_rate": 4.846324142811628e-05,
      "loss": 0.6501,
      "num_input_tokens_seen": 5009104,
      "step": 7670
    },
    {
      "epoch": 4.022536687631027,
      "grad_norm": 0.1331096887588501,
      "learning_rate": 4.8459291874828924e-05,
      "loss": 0.4334,
      "num_input_tokens_seen": 5012304,
      "step": 7675
    },
    {
      "epoch": 4.0251572327044025,
      "grad_norm": 0.6986499428749084,
      "learning_rate": 4.845533741412534e-05,
      "loss": 0.5568,
      "num_input_tokens_seen": 5014608,
      "step": 7680
    },
    {
      "epoch": 4.027777777777778,
      "grad_norm": 0.28843724727630615,
      "learning_rate": 4.845137804683277e-05,
      "loss": 0.5541,
      "num_input_tokens_seen": 5017776,
      "step": 7685
    },
    {
      "epoch": 4.030398322851153,
      "grad_norm": 0.17205916345119476,
      "learning_rate": 4.844741377377946e-05,
      "loss": 0.6399,
      "num_input_tokens_seen": 5020432,
      "step": 7690
    },
    {
      "epoch": 4.033018867924528,
      "grad_norm": 0.3088473379611969,
      "learning_rate": 4.844344459579469e-05,
      "loss": 0.3537,
      "num_input_tokens_seen": 5023824,
      "step": 7695
    },
    {
      "epoch": 4.035639412997903,
      "grad_norm": 0.12328032404184341,
      "learning_rate": 4.8439470513708766e-05,
      "loss": 0.5624,
      "num_input_tokens_seen": 5027344,
      "step": 7700
    },
    {
      "epoch": 4.038259958071279,
      "grad_norm": 0.2613818645477295,
      "learning_rate": 4.8435491528353026e-05,
      "loss": 0.5508,
      "num_input_tokens_seen": 5030928,
      "step": 7705
    },
    {
      "epoch": 4.040880503144654,
      "grad_norm": 0.137161523103714,
      "learning_rate": 4.843150764055982e-05,
      "loss": 0.4265,
      "num_input_tokens_seen": 5034320,
      "step": 7710
    },
    {
      "epoch": 4.04350104821803,
      "grad_norm": 0.10776714980602264,
      "learning_rate": 4.842751885116254e-05,
      "loss": 0.4894,
      "num_input_tokens_seen": 5037456,
      "step": 7715
    },
    {
      "epoch": 4.046121593291405,
      "grad_norm": 0.1235470250248909,
      "learning_rate": 4.842352516099559e-05,
      "loss": 0.463,
      "num_input_tokens_seen": 5040784,
      "step": 7720
    },
    {
      "epoch": 4.04874213836478,
      "grad_norm": 0.1694606989622116,
      "learning_rate": 4.84195265708944e-05,
      "loss": 0.6043,
      "num_input_tokens_seen": 5044176,
      "step": 7725
    },
    {
      "epoch": 4.051362683438155,
      "grad_norm": 0.17398543655872345,
      "learning_rate": 4.841552308169545e-05,
      "loss": 0.6179,
      "num_input_tokens_seen": 5047376,
      "step": 7730
    },
    {
      "epoch": 4.05398322851153,
      "grad_norm": 0.15881885588169098,
      "learning_rate": 4.841151469423619e-05,
      "loss": 0.5218,
      "num_input_tokens_seen": 5053712,
      "step": 7735
    },
    {
      "epoch": 4.056603773584905,
      "grad_norm": 0.23663181066513062,
      "learning_rate": 4.840750140935516e-05,
      "loss": 0.5159,
      "num_input_tokens_seen": 5057968,
      "step": 7740
    },
    {
      "epoch": 4.059224318658281,
      "grad_norm": 0.42382070422172546,
      "learning_rate": 4.840348322789189e-05,
      "loss": 0.7145,
      "num_input_tokens_seen": 5061328,
      "step": 7745
    },
    {
      "epoch": 4.061844863731656,
      "grad_norm": 0.19777008891105652,
      "learning_rate": 4.839946015068692e-05,
      "loss": 0.5506,
      "num_input_tokens_seen": 5064464,
      "step": 7750
    },
    {
      "epoch": 4.064465408805032,
      "grad_norm": 0.1812429428100586,
      "learning_rate": 4.839543217858184e-05,
      "loss": 0.4548,
      "num_input_tokens_seen": 5067344,
      "step": 7755
    },
    {
      "epoch": 4.067085953878407,
      "grad_norm": 0.15094338357448578,
      "learning_rate": 4.839139931241926e-05,
      "loss": 0.4861,
      "num_input_tokens_seen": 5070512,
      "step": 7760
    },
    {
      "epoch": 4.069706498951782,
      "grad_norm": 0.2994938790798187,
      "learning_rate": 4.838736155304281e-05,
      "loss": 0.4373,
      "num_input_tokens_seen": 5073488,
      "step": 7765
    },
    {
      "epoch": 4.072327044025157,
      "grad_norm": 0.20799347758293152,
      "learning_rate": 4.8383318901297135e-05,
      "loss": 0.5028,
      "num_input_tokens_seen": 5079984,
      "step": 7770
    },
    {
      "epoch": 4.0749475890985325,
      "grad_norm": 0.1282159984111786,
      "learning_rate": 4.837927135802792e-05,
      "loss": 0.5241,
      "num_input_tokens_seen": 5083056,
      "step": 7775
    },
    {
      "epoch": 4.077568134171908,
      "grad_norm": 0.25833579897880554,
      "learning_rate": 4.8375218924081856e-05,
      "loss": 0.5721,
      "num_input_tokens_seen": 5085904,
      "step": 7780
    },
    {
      "epoch": 4.080188679245283,
      "grad_norm": 0.19067327678203583,
      "learning_rate": 4.837116160030668e-05,
      "loss": 0.505,
      "num_input_tokens_seen": 5088368,
      "step": 7785
    },
    {
      "epoch": 4.082809224318658,
      "grad_norm": 0.1598152220249176,
      "learning_rate": 4.8367099387551126e-05,
      "loss": 0.4498,
      "num_input_tokens_seen": 5091088,
      "step": 7790
    },
    {
      "epoch": 4.085429769392033,
      "grad_norm": 0.13785770535469055,
      "learning_rate": 4.8363032286664964e-05,
      "loss": 0.5335,
      "num_input_tokens_seen": 5094960,
      "step": 7795
    },
    {
      "epoch": 4.088050314465409,
      "grad_norm": 0.16653403639793396,
      "learning_rate": 4.835896029849899e-05,
      "loss": 0.5322,
      "num_input_tokens_seen": 5098352,
      "step": 7800
    },
    {
      "epoch": 4.090670859538784,
      "grad_norm": 0.3688298165798187,
      "learning_rate": 4.835488342390502e-05,
      "loss": 0.5399,
      "num_input_tokens_seen": 5102416,
      "step": 7805
    },
    {
      "epoch": 4.09329140461216,
      "grad_norm": 0.12105753272771835,
      "learning_rate": 4.835080166373587e-05,
      "loss": 0.6447,
      "num_input_tokens_seen": 5105072,
      "step": 7810
    },
    {
      "epoch": 4.095911949685535,
      "grad_norm": 0.2544444501399994,
      "learning_rate": 4.834671501884542e-05,
      "loss": 0.5625,
      "num_input_tokens_seen": 5108528,
      "step": 7815
    },
    {
      "epoch": 4.09853249475891,
      "grad_norm": 0.20874294638633728,
      "learning_rate": 4.8342623490088543e-05,
      "loss": 0.6353,
      "num_input_tokens_seen": 5112176,
      "step": 7820
    },
    {
      "epoch": 4.101153039832285,
      "grad_norm": 0.10974778980016708,
      "learning_rate": 4.833852707832114e-05,
      "loss": 0.5036,
      "num_input_tokens_seen": 5116848,
      "step": 7825
    },
    {
      "epoch": 4.10377358490566,
      "grad_norm": 0.11320139467716217,
      "learning_rate": 4.833442578440013e-05,
      "loss": 0.4978,
      "num_input_tokens_seen": 5120464,
      "step": 7830
    },
    {
      "epoch": 4.106394129979035,
      "grad_norm": 0.12633338570594788,
      "learning_rate": 4.833031960918346e-05,
      "loss": 0.4428,
      "num_input_tokens_seen": 5123312,
      "step": 7835
    },
    {
      "epoch": 4.109014675052411,
      "grad_norm": 0.113125279545784,
      "learning_rate": 4.8326208553530095e-05,
      "loss": 0.4997,
      "num_input_tokens_seen": 5127088,
      "step": 7840
    },
    {
      "epoch": 4.111635220125786,
      "grad_norm": 0.15410088002681732,
      "learning_rate": 4.832209261830002e-05,
      "loss": 0.5065,
      "num_input_tokens_seen": 5130384,
      "step": 7845
    },
    {
      "epoch": 4.114255765199162,
      "grad_norm": 0.15164169669151306,
      "learning_rate": 4.831797180435423e-05,
      "loss": 0.4921,
      "num_input_tokens_seen": 5134288,
      "step": 7850
    },
    {
      "epoch": 4.116876310272537,
      "grad_norm": 0.4102325439453125,
      "learning_rate": 4.831384611255478e-05,
      "loss": 0.6712,
      "num_input_tokens_seen": 5137392,
      "step": 7855
    },
    {
      "epoch": 4.119496855345912,
      "grad_norm": 0.18581287562847137,
      "learning_rate": 4.8309715543764697e-05,
      "loss": 0.5335,
      "num_input_tokens_seen": 5139824,
      "step": 7860
    },
    {
      "epoch": 4.122117400419287,
      "grad_norm": 0.1334456503391266,
      "learning_rate": 4.8305580098848046e-05,
      "loss": 0.5963,
      "num_input_tokens_seen": 5143280,
      "step": 7865
    },
    {
      "epoch": 4.1247379454926625,
      "grad_norm": 0.15231305360794067,
      "learning_rate": 4.830143977866993e-05,
      "loss": 0.5981,
      "num_input_tokens_seen": 5145584,
      "step": 7870
    },
    {
      "epoch": 4.127358490566038,
      "grad_norm": 0.17104598879814148,
      "learning_rate": 4.829729458409645e-05,
      "loss": 0.4775,
      "num_input_tokens_seen": 5148240,
      "step": 7875
    },
    {
      "epoch": 4.129979035639413,
      "grad_norm": 0.19472065567970276,
      "learning_rate": 4.8293144515994725e-05,
      "loss": 0.6077,
      "num_input_tokens_seen": 5151632,
      "step": 7880
    },
    {
      "epoch": 4.132599580712788,
      "grad_norm": 0.47114893794059753,
      "learning_rate": 4.828898957523292e-05,
      "loss": 0.6263,
      "num_input_tokens_seen": 5154576,
      "step": 7885
    },
    {
      "epoch": 4.135220125786163,
      "grad_norm": 0.17819347977638245,
      "learning_rate": 4.8284829762680183e-05,
      "loss": 0.462,
      "num_input_tokens_seen": 5158224,
      "step": 7890
    },
    {
      "epoch": 4.137840670859539,
      "grad_norm": 0.14259512722492218,
      "learning_rate": 4.828066507920671e-05,
      "loss": 0.4239,
      "num_input_tokens_seen": 5161360,
      "step": 7895
    },
    {
      "epoch": 4.140461215932914,
      "grad_norm": 0.19179604947566986,
      "learning_rate": 4.82764955256837e-05,
      "loss": 0.4217,
      "num_input_tokens_seen": 5164656,
      "step": 7900
    },
    {
      "epoch": 4.1430817610062896,
      "grad_norm": 0.14128929376602173,
      "learning_rate": 4.827232110298338e-05,
      "loss": 0.4892,
      "num_input_tokens_seen": 5168336,
      "step": 7905
    },
    {
      "epoch": 4.145702306079665,
      "grad_norm": 0.18779884278774261,
      "learning_rate": 4.8268141811978995e-05,
      "loss": 0.5305,
      "num_input_tokens_seen": 5170736,
      "step": 7910
    },
    {
      "epoch": 4.14832285115304,
      "grad_norm": 0.1557561457157135,
      "learning_rate": 4.826395765354479e-05,
      "loss": 0.3214,
      "num_input_tokens_seen": 5173424,
      "step": 7915
    },
    {
      "epoch": 4.150943396226415,
      "grad_norm": 0.17758740484714508,
      "learning_rate": 4.825976862855606e-05,
      "loss": 0.4,
      "num_input_tokens_seen": 5175856,
      "step": 7920
    },
    {
      "epoch": 4.15356394129979,
      "grad_norm": 0.2347617894411087,
      "learning_rate": 4.8255574737889094e-05,
      "loss": 0.4709,
      "num_input_tokens_seen": 5178672,
      "step": 7925
    },
    {
      "epoch": 4.156184486373165,
      "grad_norm": 0.23363855481147766,
      "learning_rate": 4.82513759824212e-05,
      "loss": 0.4922,
      "num_input_tokens_seen": 5184496,
      "step": 7930
    },
    {
      "epoch": 4.158805031446541,
      "grad_norm": 0.10154067724943161,
      "learning_rate": 4.824717236303072e-05,
      "loss": 0.5627,
      "num_input_tokens_seen": 5188144,
      "step": 7935
    },
    {
      "epoch": 4.161425576519916,
      "grad_norm": 0.5443094372749329,
      "learning_rate": 4.8242963880597005e-05,
      "loss": 0.5059,
      "num_input_tokens_seen": 5192112,
      "step": 7940
    },
    {
      "epoch": 4.164046121593292,
      "grad_norm": 0.23250718414783478,
      "learning_rate": 4.823875053600041e-05,
      "loss": 0.5459,
      "num_input_tokens_seen": 5194320,
      "step": 7945
    },
    {
      "epoch": 4.166666666666667,
      "grad_norm": 0.32670626044273376,
      "learning_rate": 4.823453233012233e-05,
      "loss": 0.5473,
      "num_input_tokens_seen": 5197456,
      "step": 7950
    },
    {
      "epoch": 4.169287211740042,
      "grad_norm": 0.12564575672149658,
      "learning_rate": 4.8230309263845155e-05,
      "loss": 0.5438,
      "num_input_tokens_seen": 5199792,
      "step": 7955
    },
    {
      "epoch": 4.171907756813417,
      "grad_norm": 0.23281137645244598,
      "learning_rate": 4.822608133805231e-05,
      "loss": 0.6597,
      "num_input_tokens_seen": 5202672,
      "step": 7960
    },
    {
      "epoch": 4.1745283018867925,
      "grad_norm": 0.2508903741836548,
      "learning_rate": 4.8221848553628226e-05,
      "loss": 0.6262,
      "num_input_tokens_seen": 5205808,
      "step": 7965
    },
    {
      "epoch": 4.177148846960168,
      "grad_norm": 0.25709354877471924,
      "learning_rate": 4.8217610911458346e-05,
      "loss": 0.3826,
      "num_input_tokens_seen": 5208176,
      "step": 7970
    },
    {
      "epoch": 4.179769392033543,
      "grad_norm": 0.16566427052021027,
      "learning_rate": 4.821336841242915e-05,
      "loss": 0.5053,
      "num_input_tokens_seen": 5215376,
      "step": 7975
    },
    {
      "epoch": 4.182389937106918,
      "grad_norm": 0.21189017593860626,
      "learning_rate": 4.820912105742811e-05,
      "loss": 0.5531,
      "num_input_tokens_seen": 5218064,
      "step": 7980
    },
    {
      "epoch": 4.185010482180293,
      "grad_norm": 0.41069453954696655,
      "learning_rate": 4.8204868847343736e-05,
      "loss": 0.5498,
      "num_input_tokens_seen": 5221328,
      "step": 7985
    },
    {
      "epoch": 4.187631027253669,
      "grad_norm": 0.1785634607076645,
      "learning_rate": 4.820061178306553e-05,
      "loss": 0.5172,
      "num_input_tokens_seen": 5224432,
      "step": 7990
    },
    {
      "epoch": 4.190251572327044,
      "grad_norm": 0.21894392371177673,
      "learning_rate": 4.819634986548403e-05,
      "loss": 0.5432,
      "num_input_tokens_seen": 5226896,
      "step": 7995
    },
    {
      "epoch": 4.1928721174004195,
      "grad_norm": 0.17661671340465546,
      "learning_rate": 4.819208309549077e-05,
      "loss": 0.4914,
      "num_input_tokens_seen": 5231056,
      "step": 8000
    },
    {
      "epoch": 4.195492662473795,
      "grad_norm": 0.22369074821472168,
      "learning_rate": 4.8187811473978316e-05,
      "loss": 0.4166,
      "num_input_tokens_seen": 5234480,
      "step": 8005
    },
    {
      "epoch": 4.19811320754717,
      "grad_norm": 0.6924400329589844,
      "learning_rate": 4.818353500184025e-05,
      "loss": 0.6818,
      "num_input_tokens_seen": 5236816,
      "step": 8010
    },
    {
      "epoch": 4.200733752620545,
      "grad_norm": 0.6925774812698364,
      "learning_rate": 4.817925367997115e-05,
      "loss": 0.5155,
      "num_input_tokens_seen": 5239600,
      "step": 8015
    },
    {
      "epoch": 4.20335429769392,
      "grad_norm": 0.1507434844970703,
      "learning_rate": 4.8174967509266615e-05,
      "loss": 0.5245,
      "num_input_tokens_seen": 5242896,
      "step": 8020
    },
    {
      "epoch": 4.205974842767295,
      "grad_norm": 0.23755450546741486,
      "learning_rate": 4.817067649062328e-05,
      "loss": 0.59,
      "num_input_tokens_seen": 5245904,
      "step": 8025
    },
    {
      "epoch": 4.2085953878406706,
      "grad_norm": 0.25405189394950867,
      "learning_rate": 4.816638062493876e-05,
      "loss": 0.6114,
      "num_input_tokens_seen": 5249968,
      "step": 8030
    },
    {
      "epoch": 4.211215932914046,
      "grad_norm": 0.1880154013633728,
      "learning_rate": 4.8162079913111716e-05,
      "loss": 0.4828,
      "num_input_tokens_seen": 5253680,
      "step": 8035
    },
    {
      "epoch": 4.213836477987422,
      "grad_norm": 0.24362701177597046,
      "learning_rate": 4.81577743560418e-05,
      "loss": 0.7102,
      "num_input_tokens_seen": 5256144,
      "step": 8040
    },
    {
      "epoch": 4.216457023060797,
      "grad_norm": 0.15991128981113434,
      "learning_rate": 4.8153463954629684e-05,
      "loss": 0.4097,
      "num_input_tokens_seen": 5261456,
      "step": 8045
    },
    {
      "epoch": 4.219077568134172,
      "grad_norm": 0.16622094810009003,
      "learning_rate": 4.814914870977706e-05,
      "loss": 0.6078,
      "num_input_tokens_seen": 5264688,
      "step": 8050
    },
    {
      "epoch": 4.221698113207547,
      "grad_norm": 0.15036475658416748,
      "learning_rate": 4.8144828622386627e-05,
      "loss": 0.6904,
      "num_input_tokens_seen": 5267184,
      "step": 8055
    },
    {
      "epoch": 4.2243186582809225,
      "grad_norm": 0.3079656660556793,
      "learning_rate": 4.8140503693362095e-05,
      "loss": 0.5703,
      "num_input_tokens_seen": 5269776,
      "step": 8060
    },
    {
      "epoch": 4.226939203354298,
      "grad_norm": 0.16458655893802643,
      "learning_rate": 4.813617392360818e-05,
      "loss": 0.5026,
      "num_input_tokens_seen": 5273296,
      "step": 8065
    },
    {
      "epoch": 4.229559748427673,
      "grad_norm": 0.16302995383739471,
      "learning_rate": 4.813183931403065e-05,
      "loss": 0.5059,
      "num_input_tokens_seen": 5276432,
      "step": 8070
    },
    {
      "epoch": 4.232180293501048,
      "grad_norm": 0.19848062098026276,
      "learning_rate": 4.8127499865536226e-05,
      "loss": 0.4542,
      "num_input_tokens_seen": 5279696,
      "step": 8075
    },
    {
      "epoch": 4.234800838574423,
      "grad_norm": 0.1832638680934906,
      "learning_rate": 4.812315557903269e-05,
      "loss": 0.4611,
      "num_input_tokens_seen": 5283312,
      "step": 8080
    },
    {
      "epoch": 4.237421383647799,
      "grad_norm": 0.21631090342998505,
      "learning_rate": 4.8118806455428806e-05,
      "loss": 0.4084,
      "num_input_tokens_seen": 5285712,
      "step": 8085
    },
    {
      "epoch": 4.240041928721174,
      "grad_norm": 0.13080626726150513,
      "learning_rate": 4.811445249563437e-05,
      "loss": 0.5524,
      "num_input_tokens_seen": 5289136,
      "step": 8090
    },
    {
      "epoch": 4.2426624737945495,
      "grad_norm": 0.13267971575260162,
      "learning_rate": 4.811009370056017e-05,
      "loss": 0.3465,
      "num_input_tokens_seen": 5292528,
      "step": 8095
    },
    {
      "epoch": 4.245283018867925,
      "grad_norm": 0.16073928773403168,
      "learning_rate": 4.8105730071118016e-05,
      "loss": 0.57,
      "num_input_tokens_seen": 5295984,
      "step": 8100
    },
    {
      "epoch": 4.2479035639413,
      "grad_norm": 0.22472015023231506,
      "learning_rate": 4.810136160822074e-05,
      "loss": 0.4799,
      "num_input_tokens_seen": 5299120,
      "step": 8105
    },
    {
      "epoch": 4.250524109014675,
      "grad_norm": 0.1731346696615219,
      "learning_rate": 4.8096988312782174e-05,
      "loss": 0.4489,
      "num_input_tokens_seen": 5302192,
      "step": 8110
    },
    {
      "epoch": 4.25314465408805,
      "grad_norm": 0.304619699716568,
      "learning_rate": 4.809261018571715e-05,
      "loss": 0.8626,
      "num_input_tokens_seen": 5304880,
      "step": 8115
    },
    {
      "epoch": 4.255765199161425,
      "grad_norm": 0.1484355330467224,
      "learning_rate": 4.808822722794153e-05,
      "loss": 0.5893,
      "num_input_tokens_seen": 5308304,
      "step": 8120
    },
    {
      "epoch": 4.2583857442348005,
      "grad_norm": 0.2536044418811798,
      "learning_rate": 4.8083839440372194e-05,
      "loss": 0.5749,
      "num_input_tokens_seen": 5310960,
      "step": 8125
    },
    {
      "epoch": 4.261006289308176,
      "grad_norm": 0.17435207962989807,
      "learning_rate": 4.8079446823926984e-05,
      "loss": 0.4765,
      "num_input_tokens_seen": 5313264,
      "step": 8130
    },
    {
      "epoch": 4.263626834381552,
      "grad_norm": 0.22683560848236084,
      "learning_rate": 4.8075049379524805e-05,
      "loss": 0.6291,
      "num_input_tokens_seen": 5315600,
      "step": 8135
    },
    {
      "epoch": 4.266247379454927,
      "grad_norm": 0.1020987257361412,
      "learning_rate": 4.807064710808555e-05,
      "loss": 0.401,
      "num_input_tokens_seen": 5319024,
      "step": 8140
    },
    {
      "epoch": 4.268867924528302,
      "grad_norm": 0.23622983694076538,
      "learning_rate": 4.806624001053013e-05,
      "loss": 0.5107,
      "num_input_tokens_seen": 5322160,
      "step": 8145
    },
    {
      "epoch": 4.271488469601677,
      "grad_norm": 0.172508105635643,
      "learning_rate": 4.8061828087780455e-05,
      "loss": 0.4794,
      "num_input_tokens_seen": 5327152,
      "step": 8150
    },
    {
      "epoch": 4.274109014675052,
      "grad_norm": 0.23041294515132904,
      "learning_rate": 4.805741134075944e-05,
      "loss": 0.3944,
      "num_input_tokens_seen": 5330256,
      "step": 8155
    },
    {
      "epoch": 4.276729559748428,
      "grad_norm": 0.2588489055633545,
      "learning_rate": 4.8052989770391026e-05,
      "loss": 0.5697,
      "num_input_tokens_seen": 5333392,
      "step": 8160
    },
    {
      "epoch": 4.279350104821803,
      "grad_norm": 0.18680889904499054,
      "learning_rate": 4.804856337760016e-05,
      "loss": 0.4959,
      "num_input_tokens_seen": 5336720,
      "step": 8165
    },
    {
      "epoch": 4.281970649895178,
      "grad_norm": 0.11187831312417984,
      "learning_rate": 4.804413216331278e-05,
      "loss": 0.3275,
      "num_input_tokens_seen": 5339696,
      "step": 8170
    },
    {
      "epoch": 4.284591194968553,
      "grad_norm": 0.19990846514701843,
      "learning_rate": 4.8039696128455856e-05,
      "loss": 0.5975,
      "num_input_tokens_seen": 5342896,
      "step": 8175
    },
    {
      "epoch": 4.287211740041929,
      "grad_norm": 0.3144434094429016,
      "learning_rate": 4.803525527395736e-05,
      "loss": 0.5527,
      "num_input_tokens_seen": 5345392,
      "step": 8180
    },
    {
      "epoch": 4.289832285115304,
      "grad_norm": 0.20387907326221466,
      "learning_rate": 4.803080960074625e-05,
      "loss": 0.4945,
      "num_input_tokens_seen": 5349392,
      "step": 8185
    },
    {
      "epoch": 4.2924528301886795,
      "grad_norm": 0.14633704721927643,
      "learning_rate": 4.8026359109752525e-05,
      "loss": 0.4883,
      "num_input_tokens_seen": 5353008,
      "step": 8190
    },
    {
      "epoch": 4.295073375262055,
      "grad_norm": 0.2531049847602844,
      "learning_rate": 4.8021903801907175e-05,
      "loss": 0.7756,
      "num_input_tokens_seen": 5356464,
      "step": 8195
    },
    {
      "epoch": 4.29769392033543,
      "grad_norm": 0.16864211857318878,
      "learning_rate": 4.801744367814219e-05,
      "loss": 0.4376,
      "num_input_tokens_seen": 5359440,
      "step": 8200
    },
    {
      "epoch": 4.300314465408805,
      "grad_norm": 0.27046069502830505,
      "learning_rate": 4.801297873939059e-05,
      "loss": 0.6616,
      "num_input_tokens_seen": 5361968,
      "step": 8205
    },
    {
      "epoch": 4.30293501048218,
      "grad_norm": 0.1475161463022232,
      "learning_rate": 4.8008508986586384e-05,
      "loss": 0.6835,
      "num_input_tokens_seen": 5365744,
      "step": 8210
    },
    {
      "epoch": 4.305555555555555,
      "grad_norm": 0.19503197073936462,
      "learning_rate": 4.800403442066459e-05,
      "loss": 0.417,
      "num_input_tokens_seen": 5368720,
      "step": 8215
    },
    {
      "epoch": 4.3081761006289305,
      "grad_norm": 0.39651089906692505,
      "learning_rate": 4.799955504256125e-05,
      "loss": 0.5703,
      "num_input_tokens_seen": 5371408,
      "step": 8220
    },
    {
      "epoch": 4.310796645702306,
      "grad_norm": 0.26685574650764465,
      "learning_rate": 4.7995070853213376e-05,
      "loss": 0.5934,
      "num_input_tokens_seen": 5374256,
      "step": 8225
    },
    {
      "epoch": 4.313417190775682,
      "grad_norm": 0.28091001510620117,
      "learning_rate": 4.799058185355903e-05,
      "loss": 0.6158,
      "num_input_tokens_seen": 5376624,
      "step": 8230
    },
    {
      "epoch": 4.316037735849057,
      "grad_norm": 0.23547665774822235,
      "learning_rate": 4.7986088044537235e-05,
      "loss": 0.5388,
      "num_input_tokens_seen": 5380048,
      "step": 8235
    },
    {
      "epoch": 4.318658280922432,
      "grad_norm": 0.18165764212608337,
      "learning_rate": 4.798158942708807e-05,
      "loss": 0.4691,
      "num_input_tokens_seen": 5382416,
      "step": 8240
    },
    {
      "epoch": 4.321278825995807,
      "grad_norm": 0.15908931195735931,
      "learning_rate": 4.797708600215258e-05,
      "loss": 0.659,
      "num_input_tokens_seen": 5385520,
      "step": 8245
    },
    {
      "epoch": 4.323899371069182,
      "grad_norm": 0.1564486026763916,
      "learning_rate": 4.7972577770672845e-05,
      "loss": 0.7372,
      "num_input_tokens_seen": 5388368,
      "step": 8250
    },
    {
      "epoch": 4.326519916142558,
      "grad_norm": 0.1616009622812271,
      "learning_rate": 4.796806473359192e-05,
      "loss": 0.4432,
      "num_input_tokens_seen": 5391920,
      "step": 8255
    },
    {
      "epoch": 4.329140461215933,
      "grad_norm": 0.19111445546150208,
      "learning_rate": 4.796354689185389e-05,
      "loss": 0.4097,
      "num_input_tokens_seen": 5394864,
      "step": 8260
    },
    {
      "epoch": 4.331761006289308,
      "grad_norm": 0.13491980731487274,
      "learning_rate": 4.795902424640383e-05,
      "loss": 0.4879,
      "num_input_tokens_seen": 5398448,
      "step": 8265
    },
    {
      "epoch": 4.334381551362683,
      "grad_norm": 0.15057934820652008,
      "learning_rate": 4.795449679818782e-05,
      "loss": 0.4451,
      "num_input_tokens_seen": 5401904,
      "step": 8270
    },
    {
      "epoch": 4.337002096436059,
      "grad_norm": 0.1815083920955658,
      "learning_rate": 4.794996454815297e-05,
      "loss": 0.424,
      "num_input_tokens_seen": 5404432,
      "step": 8275
    },
    {
      "epoch": 4.339622641509434,
      "grad_norm": 0.2518942058086395,
      "learning_rate": 4.7945427497247355e-05,
      "loss": 0.517,
      "num_input_tokens_seen": 5407376,
      "step": 8280
    },
    {
      "epoch": 4.3422431865828095,
      "grad_norm": 0.38275355100631714,
      "learning_rate": 4.7940885646420084e-05,
      "loss": 0.5384,
      "num_input_tokens_seen": 5410768,
      "step": 8285
    },
    {
      "epoch": 4.344863731656185,
      "grad_norm": 0.1434023082256317,
      "learning_rate": 4.7936338996621264e-05,
      "loss": 0.5614,
      "num_input_tokens_seen": 5414256,
      "step": 8290
    },
    {
      "epoch": 4.34748427672956,
      "grad_norm": 0.20494520664215088,
      "learning_rate": 4.7931787548801996e-05,
      "loss": 0.4466,
      "num_input_tokens_seen": 5417296,
      "step": 8295
    },
    {
      "epoch": 4.350104821802935,
      "grad_norm": 0.23572660982608795,
      "learning_rate": 4.7927231303914394e-05,
      "loss": 0.518,
      "num_input_tokens_seen": 5420560,
      "step": 8300
    },
    {
      "epoch": 4.35272536687631,
      "grad_norm": 0.28612762689590454,
      "learning_rate": 4.7922670262911565e-05,
      "loss": 0.4752,
      "num_input_tokens_seen": 5424496,
      "step": 8305
    },
    {
      "epoch": 4.355345911949685,
      "grad_norm": 0.15548217296600342,
      "learning_rate": 4.791810442674764e-05,
      "loss": 0.5068,
      "num_input_tokens_seen": 5428944,
      "step": 8310
    },
    {
      "epoch": 4.3579664570230605,
      "grad_norm": 0.13767704367637634,
      "learning_rate": 4.791353379637773e-05,
      "loss": 0.5738,
      "num_input_tokens_seen": 5432624,
      "step": 8315
    },
    {
      "epoch": 4.360587002096436,
      "grad_norm": 0.27580299973487854,
      "learning_rate": 4.790895837275796e-05,
      "loss": 0.5399,
      "num_input_tokens_seen": 5435152,
      "step": 8320
    },
    {
      "epoch": 4.363207547169811,
      "grad_norm": 0.3172222077846527,
      "learning_rate": 4.790437815684545e-05,
      "loss": 0.5611,
      "num_input_tokens_seen": 5438640,
      "step": 8325
    },
    {
      "epoch": 4.365828092243187,
      "grad_norm": 0.17367231845855713,
      "learning_rate": 4.789979314959835e-05,
      "loss": 0.4502,
      "num_input_tokens_seen": 5441264,
      "step": 8330
    },
    {
      "epoch": 4.368448637316562,
      "grad_norm": 0.11067166924476624,
      "learning_rate": 4.789520335197577e-05,
      "loss": 0.4671,
      "num_input_tokens_seen": 5444752,
      "step": 8335
    },
    {
      "epoch": 4.371069182389937,
      "grad_norm": 0.24192354083061218,
      "learning_rate": 4.789060876493785e-05,
      "loss": 0.5371,
      "num_input_tokens_seen": 5447632,
      "step": 8340
    },
    {
      "epoch": 4.373689727463312,
      "grad_norm": 0.1840200424194336,
      "learning_rate": 4.788600938944573e-05,
      "loss": 0.6417,
      "num_input_tokens_seen": 5451344,
      "step": 8345
    },
    {
      "epoch": 4.376310272536688,
      "grad_norm": 0.22929175198078156,
      "learning_rate": 4.788140522646155e-05,
      "loss": 0.6018,
      "num_input_tokens_seen": 5454736,
      "step": 8350
    },
    {
      "epoch": 4.378930817610063,
      "grad_norm": 0.13342660665512085,
      "learning_rate": 4.787679627694843e-05,
      "loss": 0.5475,
      "num_input_tokens_seen": 5458096,
      "step": 8355
    },
    {
      "epoch": 4.381551362683438,
      "grad_norm": 0.18127883970737457,
      "learning_rate": 4.787218254187053e-05,
      "loss": 0.5055,
      "num_input_tokens_seen": 5460560,
      "step": 8360
    },
    {
      "epoch": 4.384171907756813,
      "grad_norm": 0.26909908652305603,
      "learning_rate": 4.7867564022192965e-05,
      "loss": 0.4836,
      "num_input_tokens_seen": 5463792,
      "step": 8365
    },
    {
      "epoch": 4.386792452830189,
      "grad_norm": 0.11369647085666656,
      "learning_rate": 4.786294071888191e-05,
      "loss": 0.4768,
      "num_input_tokens_seen": 5467216,
      "step": 8370
    },
    {
      "epoch": 4.389412997903564,
      "grad_norm": 0.10736338794231415,
      "learning_rate": 4.785831263290449e-05,
      "loss": 0.4873,
      "num_input_tokens_seen": 5470320,
      "step": 8375
    },
    {
      "epoch": 4.3920335429769395,
      "grad_norm": 0.11022471636533737,
      "learning_rate": 4.7853679765228845e-05,
      "loss": 0.4571,
      "num_input_tokens_seen": 5473232,
      "step": 8380
    },
    {
      "epoch": 4.394654088050315,
      "grad_norm": 0.15930873155593872,
      "learning_rate": 4.7849042116824116e-05,
      "loss": 0.6505,
      "num_input_tokens_seen": 5476432,
      "step": 8385
    },
    {
      "epoch": 4.39727463312369,
      "grad_norm": 0.3265736401081085,
      "learning_rate": 4.7844399688660456e-05,
      "loss": 0.6056,
      "num_input_tokens_seen": 5478928,
      "step": 8390
    },
    {
      "epoch": 4.399895178197065,
      "grad_norm": 0.19709043204784393,
      "learning_rate": 4.783975248170901e-05,
      "loss": 0.4159,
      "num_input_tokens_seen": 5481776,
      "step": 8395
    },
    {
      "epoch": 4.40251572327044,
      "grad_norm": 0.13647767901420593,
      "learning_rate": 4.783510049694191e-05,
      "loss": 0.4949,
      "num_input_tokens_seen": 5485008,
      "step": 8400
    },
    {
      "epoch": 4.405136268343815,
      "grad_norm": 0.09352732449769974,
      "learning_rate": 4.78304437353323e-05,
      "loss": 0.5243,
      "num_input_tokens_seen": 5492496,
      "step": 8405
    },
    {
      "epoch": 4.4077568134171905,
      "grad_norm": 0.1808728724718094,
      "learning_rate": 4.782578219785433e-05,
      "loss": 0.4993,
      "num_input_tokens_seen": 5495664,
      "step": 8410
    },
    {
      "epoch": 4.410377358490566,
      "grad_norm": 0.19169668853282928,
      "learning_rate": 4.782111588548312e-05,
      "loss": 0.4831,
      "num_input_tokens_seen": 5499792,
      "step": 8415
    },
    {
      "epoch": 4.412997903563941,
      "grad_norm": 0.13100333511829376,
      "learning_rate": 4.781644479919484e-05,
      "loss": 0.4737,
      "num_input_tokens_seen": 5502736,
      "step": 8420
    },
    {
      "epoch": 4.415618448637317,
      "grad_norm": 0.20465800166130066,
      "learning_rate": 4.7811768939966605e-05,
      "loss": 0.4859,
      "num_input_tokens_seen": 5505328,
      "step": 8425
    },
    {
      "epoch": 4.418238993710692,
      "grad_norm": 0.20419849455356598,
      "learning_rate": 4.780708830877656e-05,
      "loss": 0.5485,
      "num_input_tokens_seen": 5507952,
      "step": 8430
    },
    {
      "epoch": 4.420859538784067,
      "grad_norm": 0.26115548610687256,
      "learning_rate": 4.780240290660384e-05,
      "loss": 0.4228,
      "num_input_tokens_seen": 5510544,
      "step": 8435
    },
    {
      "epoch": 4.423480083857442,
      "grad_norm": 0.15373677015304565,
      "learning_rate": 4.7797712734428575e-05,
      "loss": 0.5932,
      "num_input_tokens_seen": 5512976,
      "step": 8440
    },
    {
      "epoch": 4.426100628930818,
      "grad_norm": 0.15377181768417358,
      "learning_rate": 4.779301779323189e-05,
      "loss": 0.5722,
      "num_input_tokens_seen": 5516496,
      "step": 8445
    },
    {
      "epoch": 4.428721174004193,
      "grad_norm": 0.14760816097259521,
      "learning_rate": 4.778831808399593e-05,
      "loss": 0.3937,
      "num_input_tokens_seen": 5520720,
      "step": 8450
    },
    {
      "epoch": 4.431341719077568,
      "grad_norm": 0.09089702367782593,
      "learning_rate": 4.77836136077038e-05,
      "loss": 0.5128,
      "num_input_tokens_seen": 5524336,
      "step": 8455
    },
    {
      "epoch": 4.433962264150943,
      "grad_norm": 0.10364136844873428,
      "learning_rate": 4.777890436533963e-05,
      "loss": 0.5025,
      "num_input_tokens_seen": 5528112,
      "step": 8460
    },
    {
      "epoch": 4.436582809224318,
      "grad_norm": 0.12203575670719147,
      "learning_rate": 4.777419035788855e-05,
      "loss": 0.5629,
      "num_input_tokens_seen": 5531664,
      "step": 8465
    },
    {
      "epoch": 4.439203354297694,
      "grad_norm": 0.2254570722579956,
      "learning_rate": 4.776947158633667e-05,
      "loss": 0.438,
      "num_input_tokens_seen": 5536368,
      "step": 8470
    },
    {
      "epoch": 4.4418238993710695,
      "grad_norm": 0.15659861266613007,
      "learning_rate": 4.7764748051671095e-05,
      "loss": 0.4817,
      "num_input_tokens_seen": 5539600,
      "step": 8475
    },
    {
      "epoch": 4.444444444444445,
      "grad_norm": 0.12367111444473267,
      "learning_rate": 4.7760019754879946e-05,
      "loss": 0.4615,
      "num_input_tokens_seen": 5542544,
      "step": 8480
    },
    {
      "epoch": 4.44706498951782,
      "grad_norm": 0.1725323498249054,
      "learning_rate": 4.7755286696952324e-05,
      "loss": 0.5167,
      "num_input_tokens_seen": 5546704,
      "step": 8485
    },
    {
      "epoch": 4.449685534591195,
      "grad_norm": 0.1500123143196106,
      "learning_rate": 4.775054887887833e-05,
      "loss": 0.4576,
      "num_input_tokens_seen": 5549520,
      "step": 8490
    },
    {
      "epoch": 4.45230607966457,
      "grad_norm": 0.2771592140197754,
      "learning_rate": 4.774580630164906e-05,
      "loss": 0.508,
      "num_input_tokens_seen": 5553040,
      "step": 8495
    },
    {
      "epoch": 4.454926624737945,
      "grad_norm": 0.15589509904384613,
      "learning_rate": 4.774105896625661e-05,
      "loss": 0.5334,
      "num_input_tokens_seen": 5556880,
      "step": 8500
    },
    {
      "epoch": 4.4575471698113205,
      "grad_norm": 0.15878327190876007,
      "learning_rate": 4.773630687369407e-05,
      "loss": 0.4249,
      "num_input_tokens_seen": 5559760,
      "step": 8505
    },
    {
      "epoch": 4.460167714884696,
      "grad_norm": 0.1355288177728653,
      "learning_rate": 4.773155002495552e-05,
      "loss": 0.4763,
      "num_input_tokens_seen": 5564688,
      "step": 8510
    },
    {
      "epoch": 4.462788259958071,
      "grad_norm": 0.19692638516426086,
      "learning_rate": 4.772678842103605e-05,
      "loss": 0.5588,
      "num_input_tokens_seen": 5568112,
      "step": 8515
    },
    {
      "epoch": 4.465408805031447,
      "grad_norm": 0.1816217005252838,
      "learning_rate": 4.772202206293172e-05,
      "loss": 0.6118,
      "num_input_tokens_seen": 5570928,
      "step": 8520
    },
    {
      "epoch": 4.468029350104822,
      "grad_norm": 0.15703155100345612,
      "learning_rate": 4.7717250951639586e-05,
      "loss": 0.5102,
      "num_input_tokens_seen": 5574480,
      "step": 8525
    },
    {
      "epoch": 4.470649895178197,
      "grad_norm": 0.24027688801288605,
      "learning_rate": 4.771247508815774e-05,
      "loss": 0.5312,
      "num_input_tokens_seen": 5577296,
      "step": 8530
    },
    {
      "epoch": 4.473270440251572,
      "grad_norm": 0.11692772060632706,
      "learning_rate": 4.770769447348522e-05,
      "loss": 0.4133,
      "num_input_tokens_seen": 5580784,
      "step": 8535
    },
    {
      "epoch": 4.475890985324948,
      "grad_norm": 0.1159553974866867,
      "learning_rate": 4.7702909108622074e-05,
      "loss": 0.4544,
      "num_input_tokens_seen": 5583856,
      "step": 8540
    },
    {
      "epoch": 4.478511530398323,
      "grad_norm": 0.15788395702838898,
      "learning_rate": 4.7698118994569355e-05,
      "loss": 0.5215,
      "num_input_tokens_seen": 5586256,
      "step": 8545
    },
    {
      "epoch": 4.481132075471698,
      "grad_norm": 0.18387554585933685,
      "learning_rate": 4.769332413232909e-05,
      "loss": 0.4874,
      "num_input_tokens_seen": 5589200,
      "step": 8550
    },
    {
      "epoch": 4.483752620545073,
      "grad_norm": 0.12115892022848129,
      "learning_rate": 4.768852452290432e-05,
      "loss": 0.5832,
      "num_input_tokens_seen": 5592976,
      "step": 8555
    },
    {
      "epoch": 4.486373165618448,
      "grad_norm": 0.2029857039451599,
      "learning_rate": 4.768372016729907e-05,
      "loss": 0.5563,
      "num_input_tokens_seen": 5595696,
      "step": 8560
    },
    {
      "epoch": 4.488993710691824,
      "grad_norm": 0.18810781836509705,
      "learning_rate": 4.767891106651834e-05,
      "loss": 0.5452,
      "num_input_tokens_seen": 5598288,
      "step": 8565
    },
    {
      "epoch": 4.4916142557651995,
      "grad_norm": 0.18095837533473969,
      "learning_rate": 4.767409722156816e-05,
      "loss": 0.5871,
      "num_input_tokens_seen": 5601296,
      "step": 8570
    },
    {
      "epoch": 4.494234800838575,
      "grad_norm": 0.2995758652687073,
      "learning_rate": 4.7669278633455515e-05,
      "loss": 0.4211,
      "num_input_tokens_seen": 5604752,
      "step": 8575
    },
    {
      "epoch": 4.49685534591195,
      "grad_norm": 0.16070158779621124,
      "learning_rate": 4.7664455303188404e-05,
      "loss": 0.5584,
      "num_input_tokens_seen": 5607824,
      "step": 8580
    },
    {
      "epoch": 4.499475890985325,
      "grad_norm": 0.20542626082897186,
      "learning_rate": 4.7659627231775815e-05,
      "loss": 0.5522,
      "num_input_tokens_seen": 5611248,
      "step": 8585
    },
    {
      "epoch": 4.5020964360587,
      "grad_norm": 0.20185352861881256,
      "learning_rate": 4.765479442022773e-05,
      "loss": 0.4696,
      "num_input_tokens_seen": 5614192,
      "step": 8590
    },
    {
      "epoch": 4.504716981132075,
      "grad_norm": 0.18128789961338043,
      "learning_rate": 4.76499568695551e-05,
      "loss": 0.5774,
      "num_input_tokens_seen": 5616944,
      "step": 8595
    },
    {
      "epoch": 4.5073375262054505,
      "grad_norm": 0.2202395647764206,
      "learning_rate": 4.764511458076991e-05,
      "loss": 0.5545,
      "num_input_tokens_seen": 5619344,
      "step": 8600
    },
    {
      "epoch": 4.509958071278826,
      "grad_norm": 0.1489349901676178,
      "learning_rate": 4.7640267554885085e-05,
      "loss": 0.4419,
      "num_input_tokens_seen": 5622608,
      "step": 8605
    },
    {
      "epoch": 4.512578616352201,
      "grad_norm": 0.26016974449157715,
      "learning_rate": 4.763541579291459e-05,
      "loss": 0.5259,
      "num_input_tokens_seen": 5625296,
      "step": 8610
    },
    {
      "epoch": 4.515199161425577,
      "grad_norm": 0.17554490268230438,
      "learning_rate": 4.763055929587336e-05,
      "loss": 0.4315,
      "num_input_tokens_seen": 5628272,
      "step": 8615
    },
    {
      "epoch": 4.517819706498952,
      "grad_norm": 0.13114464282989502,
      "learning_rate": 4.7625698064777294e-05,
      "loss": 0.4658,
      "num_input_tokens_seen": 5632944,
      "step": 8620
    },
    {
      "epoch": 4.520440251572327,
      "grad_norm": 0.23703399300575256,
      "learning_rate": 4.762083210064332e-05,
      "loss": 0.4816,
      "num_input_tokens_seen": 5636016,
      "step": 8625
    },
    {
      "epoch": 4.523060796645702,
      "grad_norm": 0.26353028416633606,
      "learning_rate": 4.761596140448934e-05,
      "loss": 0.4649,
      "num_input_tokens_seen": 5642064,
      "step": 8630
    },
    {
      "epoch": 4.5256813417190775,
      "grad_norm": 0.1575411856174469,
      "learning_rate": 4.7611085977334256e-05,
      "loss": 0.5686,
      "num_input_tokens_seen": 5645072,
      "step": 8635
    },
    {
      "epoch": 4.528301886792453,
      "grad_norm": 0.1355253905057907,
      "learning_rate": 4.7606205820197946e-05,
      "loss": 0.5563,
      "num_input_tokens_seen": 5648304,
      "step": 8640
    },
    {
      "epoch": 4.530922431865828,
      "grad_norm": 0.15970376133918762,
      "learning_rate": 4.760132093410129e-05,
      "loss": 0.6371,
      "num_input_tokens_seen": 5651120,
      "step": 8645
    },
    {
      "epoch": 4.533542976939203,
      "grad_norm": 0.30645596981048584,
      "learning_rate": 4.759643132006614e-05,
      "loss": 0.4302,
      "num_input_tokens_seen": 5653712,
      "step": 8650
    },
    {
      "epoch": 4.536163522012579,
      "grad_norm": 0.20538273453712463,
      "learning_rate": 4.759153697911535e-05,
      "loss": 0.5202,
      "num_input_tokens_seen": 5656624,
      "step": 8655
    },
    {
      "epoch": 4.538784067085954,
      "grad_norm": 0.17032364010810852,
      "learning_rate": 4.758663791227276e-05,
      "loss": 0.4696,
      "num_input_tokens_seen": 5658960,
      "step": 8660
    },
    {
      "epoch": 4.5414046121593294,
      "grad_norm": 0.21361307799816132,
      "learning_rate": 4.7581734120563206e-05,
      "loss": 0.5121,
      "num_input_tokens_seen": 5661872,
      "step": 8665
    },
    {
      "epoch": 4.544025157232705,
      "grad_norm": 0.1521407812833786,
      "learning_rate": 4.757682560501249e-05,
      "loss": 0.6214,
      "num_input_tokens_seen": 5664560,
      "step": 8670
    },
    {
      "epoch": 4.54664570230608,
      "grad_norm": 0.1431787610054016,
      "learning_rate": 4.757191236664744e-05,
      "loss": 0.4542,
      "num_input_tokens_seen": 5667664,
      "step": 8675
    },
    {
      "epoch": 4.549266247379455,
      "grad_norm": 0.21008139848709106,
      "learning_rate": 4.7566994406495825e-05,
      "loss": 0.4345,
      "num_input_tokens_seen": 5670128,
      "step": 8680
    },
    {
      "epoch": 4.55188679245283,
      "grad_norm": 0.20796748995780945,
      "learning_rate": 4.756207172558645e-05,
      "loss": 0.571,
      "num_input_tokens_seen": 5673296,
      "step": 8685
    },
    {
      "epoch": 4.554507337526205,
      "grad_norm": 0.16601869463920593,
      "learning_rate": 4.7557144324949065e-05,
      "loss": 0.5423,
      "num_input_tokens_seen": 5675920,
      "step": 8690
    },
    {
      "epoch": 4.5571278825995805,
      "grad_norm": 0.41470274329185486,
      "learning_rate": 4.755221220561443e-05,
      "loss": 0.4884,
      "num_input_tokens_seen": 5678896,
      "step": 8695
    },
    {
      "epoch": 4.559748427672956,
      "grad_norm": 0.3586844205856323,
      "learning_rate": 4.7547275368614295e-05,
      "loss": 0.5775,
      "num_input_tokens_seen": 5681488,
      "step": 8700
    },
    {
      "epoch": 4.562368972746331,
      "grad_norm": 0.2702138423919678,
      "learning_rate": 4.754233381498139e-05,
      "loss": 0.5341,
      "num_input_tokens_seen": 5684432,
      "step": 8705
    },
    {
      "epoch": 4.564989517819707,
      "grad_norm": 0.24532590806484222,
      "learning_rate": 4.753738754574942e-05,
      "loss": 0.5144,
      "num_input_tokens_seen": 5688112,
      "step": 8710
    },
    {
      "epoch": 4.567610062893082,
      "grad_norm": 0.1072121411561966,
      "learning_rate": 4.75324365619531e-05,
      "loss": 0.5717,
      "num_input_tokens_seen": 5690416,
      "step": 8715
    },
    {
      "epoch": 4.570230607966457,
      "grad_norm": 0.17699779570102692,
      "learning_rate": 4.752748086462811e-05,
      "loss": 0.5656,
      "num_input_tokens_seen": 5693744,
      "step": 8720
    },
    {
      "epoch": 4.572851153039832,
      "grad_norm": 0.11097698658704758,
      "learning_rate": 4.752252045481113e-05,
      "loss": 0.5274,
      "num_input_tokens_seen": 5697776,
      "step": 8725
    },
    {
      "epoch": 4.5754716981132075,
      "grad_norm": 0.2761296331882477,
      "learning_rate": 4.7517555333539834e-05,
      "loss": 0.809,
      "num_input_tokens_seen": 5700240,
      "step": 8730
    },
    {
      "epoch": 4.578092243186583,
      "grad_norm": 0.15607936680316925,
      "learning_rate": 4.751258550185284e-05,
      "loss": 0.4553,
      "num_input_tokens_seen": 5702992,
      "step": 8735
    },
    {
      "epoch": 4.580712788259958,
      "grad_norm": 0.350839227437973,
      "learning_rate": 4.7507610960789804e-05,
      "loss": 0.4727,
      "num_input_tokens_seen": 5706544,
      "step": 8740
    },
    {
      "epoch": 4.583333333333333,
      "grad_norm": 0.21740083396434784,
      "learning_rate": 4.7502631711391334e-05,
      "loss": 0.516,
      "num_input_tokens_seen": 5709328,
      "step": 8745
    },
    {
      "epoch": 4.585953878406709,
      "grad_norm": 0.12016762048006058,
      "learning_rate": 4.749764775469904e-05,
      "loss": 0.4362,
      "num_input_tokens_seen": 5712720,
      "step": 8750
    },
    {
      "epoch": 4.588574423480084,
      "grad_norm": 0.1370987594127655,
      "learning_rate": 4.749265909175549e-05,
      "loss": 0.6218,
      "num_input_tokens_seen": 5715536,
      "step": 8755
    },
    {
      "epoch": 4.591194968553459,
      "grad_norm": 0.19944573938846588,
      "learning_rate": 4.7487665723604275e-05,
      "loss": 0.5327,
      "num_input_tokens_seen": 5718352,
      "step": 8760
    },
    {
      "epoch": 4.593815513626835,
      "grad_norm": 0.21693143248558044,
      "learning_rate": 4.7482667651289944e-05,
      "loss": 0.4963,
      "num_input_tokens_seen": 5722096,
      "step": 8765
    },
    {
      "epoch": 4.59643605870021,
      "grad_norm": 0.265328049659729,
      "learning_rate": 4.747766487585803e-05,
      "loss": 0.4973,
      "num_input_tokens_seen": 5724816,
      "step": 8770
    },
    {
      "epoch": 4.599056603773585,
      "grad_norm": 0.13577260076999664,
      "learning_rate": 4.747265739835507e-05,
      "loss": 0.4913,
      "num_input_tokens_seen": 5729040,
      "step": 8775
    },
    {
      "epoch": 4.60167714884696,
      "grad_norm": 0.17861944437026978,
      "learning_rate": 4.746764521982856e-05,
      "loss": 0.4867,
      "num_input_tokens_seen": 5732464,
      "step": 8780
    },
    {
      "epoch": 4.604297693920335,
      "grad_norm": 0.1819932758808136,
      "learning_rate": 4.7462628341326995e-05,
      "loss": 0.3785,
      "num_input_tokens_seen": 5735472,
      "step": 8785
    },
    {
      "epoch": 4.6069182389937104,
      "grad_norm": 0.18187937140464783,
      "learning_rate": 4.7457606763899856e-05,
      "loss": 0.5002,
      "num_input_tokens_seen": 5738896,
      "step": 8790
    },
    {
      "epoch": 4.609538784067086,
      "grad_norm": 0.3177047669887543,
      "learning_rate": 4.745258048859758e-05,
      "loss": 0.4466,
      "num_input_tokens_seen": 5742032,
      "step": 8795
    },
    {
      "epoch": 4.612159329140461,
      "grad_norm": 0.1508018523454666,
      "learning_rate": 4.744754951647162e-05,
      "loss": 0.5405,
      "num_input_tokens_seen": 5745168,
      "step": 8800
    },
    {
      "epoch": 4.614779874213837,
      "grad_norm": 0.21698980033397675,
      "learning_rate": 4.74425138485744e-05,
      "loss": 0.5745,
      "num_input_tokens_seen": 5747920,
      "step": 8805
    },
    {
      "epoch": 4.617400419287212,
      "grad_norm": 0.17891071736812592,
      "learning_rate": 4.743747348595932e-05,
      "loss": 0.533,
      "num_input_tokens_seen": 5751120,
      "step": 8810
    },
    {
      "epoch": 4.620020964360587,
      "grad_norm": 0.18002447485923767,
      "learning_rate": 4.743242842968077e-05,
      "loss": 0.4719,
      "num_input_tokens_seen": 5754352,
      "step": 8815
    },
    {
      "epoch": 4.622641509433962,
      "grad_norm": 0.592458188533783,
      "learning_rate": 4.742737868079411e-05,
      "loss": 0.4732,
      "num_input_tokens_seen": 5757200,
      "step": 8820
    },
    {
      "epoch": 4.6252620545073375,
      "grad_norm": 0.19747143983840942,
      "learning_rate": 4.7422324240355696e-05,
      "loss": 0.3802,
      "num_input_tokens_seen": 5760944,
      "step": 8825
    },
    {
      "epoch": 4.627882599580713,
      "grad_norm": 0.19150897860527039,
      "learning_rate": 4.741726510942286e-05,
      "loss": 0.5373,
      "num_input_tokens_seen": 5764848,
      "step": 8830
    },
    {
      "epoch": 4.630503144654088,
      "grad_norm": 0.17511112987995148,
      "learning_rate": 4.741220128905391e-05,
      "loss": 0.4285,
      "num_input_tokens_seen": 5767984,
      "step": 8835
    },
    {
      "epoch": 4.633123689727463,
      "grad_norm": 0.16383148729801178,
      "learning_rate": 4.740713278030815e-05,
      "loss": 0.4589,
      "num_input_tokens_seen": 5771472,
      "step": 8840
    },
    {
      "epoch": 4.635744234800838,
      "grad_norm": 0.2385208159685135,
      "learning_rate": 4.740205958424584e-05,
      "loss": 0.5797,
      "num_input_tokens_seen": 5775536,
      "step": 8845
    },
    {
      "epoch": 4.638364779874214,
      "grad_norm": 0.15398965775966644,
      "learning_rate": 4.739698170192824e-05,
      "loss": 0.4558,
      "num_input_tokens_seen": 5778704,
      "step": 8850
    },
    {
      "epoch": 4.640985324947589,
      "grad_norm": 0.23238427937030792,
      "learning_rate": 4.7391899134417596e-05,
      "loss": 0.5212,
      "num_input_tokens_seen": 5781104,
      "step": 8855
    },
    {
      "epoch": 4.643605870020965,
      "grad_norm": 0.17120784521102905,
      "learning_rate": 4.73868118827771e-05,
      "loss": 0.5343,
      "num_input_tokens_seen": 5784528,
      "step": 8860
    },
    {
      "epoch": 4.64622641509434,
      "grad_norm": 0.15243002772331238,
      "learning_rate": 4.738171994807097e-05,
      "loss": 0.7743,
      "num_input_tokens_seen": 5787184,
      "step": 8865
    },
    {
      "epoch": 4.648846960167715,
      "grad_norm": 0.42950117588043213,
      "learning_rate": 4.7376623331364364e-05,
      "loss": 0.5632,
      "num_input_tokens_seen": 5790032,
      "step": 8870
    },
    {
      "epoch": 4.65146750524109,
      "grad_norm": 0.18974606692790985,
      "learning_rate": 4.737152203372345e-05,
      "loss": 0.5281,
      "num_input_tokens_seen": 5793104,
      "step": 8875
    },
    {
      "epoch": 4.654088050314465,
      "grad_norm": 0.3050162196159363,
      "learning_rate": 4.7366416056215345e-05,
      "loss": 0.6006,
      "num_input_tokens_seen": 5796368,
      "step": 8880
    },
    {
      "epoch": 4.65670859538784,
      "grad_norm": 0.24148792028427124,
      "learning_rate": 4.7361305399908185e-05,
      "loss": 0.6688,
      "num_input_tokens_seen": 5799248,
      "step": 8885
    },
    {
      "epoch": 4.659329140461216,
      "grad_norm": 0.13437247276306152,
      "learning_rate": 4.735619006587103e-05,
      "loss": 0.5915,
      "num_input_tokens_seen": 5802448,
      "step": 8890
    },
    {
      "epoch": 4.661949685534591,
      "grad_norm": 0.16281862556934357,
      "learning_rate": 4.735107005517398e-05,
      "loss": 0.5426,
      "num_input_tokens_seen": 5805936,
      "step": 8895
    },
    {
      "epoch": 4.664570230607967,
      "grad_norm": 0.22030459344387054,
      "learning_rate": 4.734594536888806e-05,
      "loss": 0.473,
      "num_input_tokens_seen": 5809744,
      "step": 8900
    },
    {
      "epoch": 4.667190775681342,
      "grad_norm": 0.17468896508216858,
      "learning_rate": 4.734081600808531e-05,
      "loss": 0.5996,
      "num_input_tokens_seen": 5812656,
      "step": 8905
    },
    {
      "epoch": 4.669811320754717,
      "grad_norm": 0.2709752917289734,
      "learning_rate": 4.733568197383872e-05,
      "loss": 0.6379,
      "num_input_tokens_seen": 5816048,
      "step": 8910
    },
    {
      "epoch": 4.672431865828092,
      "grad_norm": 0.12214405834674835,
      "learning_rate": 4.7330543267222294e-05,
      "loss": 0.6135,
      "num_input_tokens_seen": 5820080,
      "step": 8915
    },
    {
      "epoch": 4.6750524109014675,
      "grad_norm": 0.12102137506008148,
      "learning_rate": 4.732539988931097e-05,
      "loss": 0.5622,
      "num_input_tokens_seen": 5823344,
      "step": 8920
    },
    {
      "epoch": 4.677672955974843,
      "grad_norm": 0.17005978524684906,
      "learning_rate": 4.732025184118068e-05,
      "loss": 0.512,
      "num_input_tokens_seen": 5826448,
      "step": 8925
    },
    {
      "epoch": 4.680293501048218,
      "grad_norm": 0.3077332675457001,
      "learning_rate": 4.731509912390837e-05,
      "loss": 0.4822,
      "num_input_tokens_seen": 5830000,
      "step": 8930
    },
    {
      "epoch": 4.682914046121593,
      "grad_norm": 0.13204117119312286,
      "learning_rate": 4.730994173857189e-05,
      "loss": 0.3932,
      "num_input_tokens_seen": 5832464,
      "step": 8935
    },
    {
      "epoch": 4.685534591194968,
      "grad_norm": 0.2273513674736023,
      "learning_rate": 4.7304779686250126e-05,
      "loss": 0.5328,
      "num_input_tokens_seen": 5835568,
      "step": 8940
    },
    {
      "epoch": 4.688155136268344,
      "grad_norm": 0.1996448040008545,
      "learning_rate": 4.729961296802292e-05,
      "loss": 0.5104,
      "num_input_tokens_seen": 5841360,
      "step": 8945
    },
    {
      "epoch": 4.690775681341719,
      "grad_norm": 0.1759437471628189,
      "learning_rate": 4.7294441584971095e-05,
      "loss": 0.404,
      "num_input_tokens_seen": 5844240,
      "step": 8950
    },
    {
      "epoch": 4.693396226415095,
      "grad_norm": 0.3689552843570709,
      "learning_rate": 4.728926553817643e-05,
      "loss": 0.4497,
      "num_input_tokens_seen": 5848144,
      "step": 8955
    },
    {
      "epoch": 4.69601677148847,
      "grad_norm": 0.21979480981826782,
      "learning_rate": 4.728408482872171e-05,
      "loss": 0.553,
      "num_input_tokens_seen": 5850800,
      "step": 8960
    },
    {
      "epoch": 4.698637316561845,
      "grad_norm": 0.1364257037639618,
      "learning_rate": 4.727889945769067e-05,
      "loss": 0.4988,
      "num_input_tokens_seen": 5853712,
      "step": 8965
    },
    {
      "epoch": 4.70125786163522,
      "grad_norm": 0.1887691617012024,
      "learning_rate": 4.727370942616803e-05,
      "loss": 0.4452,
      "num_input_tokens_seen": 5856400,
      "step": 8970
    },
    {
      "epoch": 4.703878406708595,
      "grad_norm": 0.16733460128307343,
      "learning_rate": 4.72685147352395e-05,
      "loss": 0.5562,
      "num_input_tokens_seen": 5859760,
      "step": 8975
    },
    {
      "epoch": 4.70649895178197,
      "grad_norm": 0.12654288113117218,
      "learning_rate": 4.7263315385991735e-05,
      "loss": 0.4187,
      "num_input_tokens_seen": 5862672,
      "step": 8980
    },
    {
      "epoch": 4.709119496855346,
      "grad_norm": 0.16422803699970245,
      "learning_rate": 4.7258111379512384e-05,
      "loss": 0.5327,
      "num_input_tokens_seen": 5866288,
      "step": 8985
    },
    {
      "epoch": 4.711740041928721,
      "grad_norm": 0.13777174055576324,
      "learning_rate": 4.725290271689007e-05,
      "loss": 0.5825,
      "num_input_tokens_seen": 5869424,
      "step": 8990
    },
    {
      "epoch": 4.714360587002097,
      "grad_norm": 0.11622942984104156,
      "learning_rate": 4.724768939921438e-05,
      "loss": 0.4076,
      "num_input_tokens_seen": 5872240,
      "step": 8995
    },
    {
      "epoch": 4.716981132075472,
      "grad_norm": 0.1993066817522049,
      "learning_rate": 4.7242471427575884e-05,
      "loss": 0.4332,
      "num_input_tokens_seen": 5874672,
      "step": 9000
    },
    {
      "epoch": 4.719601677148847,
      "grad_norm": 0.14794528484344482,
      "learning_rate": 4.723724880306612e-05,
      "loss": 0.4416,
      "num_input_tokens_seen": 5878544,
      "step": 9005
    },
    {
      "epoch": 4.722222222222222,
      "grad_norm": 0.13202181458473206,
      "learning_rate": 4.72320215267776e-05,
      "loss": 0.5675,
      "num_input_tokens_seen": 5882096,
      "step": 9010
    },
    {
      "epoch": 4.7248427672955975,
      "grad_norm": 0.21809755265712738,
      "learning_rate": 4.722678959980381e-05,
      "loss": 0.5287,
      "num_input_tokens_seen": 5886512,
      "step": 9015
    },
    {
      "epoch": 4.727463312368973,
      "grad_norm": 0.19607028365135193,
      "learning_rate": 4.7221553023239216e-05,
      "loss": 0.4667,
      "num_input_tokens_seen": 5889648,
      "step": 9020
    },
    {
      "epoch": 4.730083857442348,
      "grad_norm": 0.20035099983215332,
      "learning_rate": 4.721631179817925e-05,
      "loss": 0.5099,
      "num_input_tokens_seen": 5892752,
      "step": 9025
    },
    {
      "epoch": 4.732704402515723,
      "grad_norm": 0.11108360439538956,
      "learning_rate": 4.721106592572031e-05,
      "loss": 0.6483,
      "num_input_tokens_seen": 5896112,
      "step": 9030
    },
    {
      "epoch": 4.735324947589098,
      "grad_norm": 0.1863643229007721,
      "learning_rate": 4.7205815406959774e-05,
      "loss": 0.4965,
      "num_input_tokens_seen": 5899440,
      "step": 9035
    },
    {
      "epoch": 4.737945492662474,
      "grad_norm": 0.16386502981185913,
      "learning_rate": 4.7200560242996e-05,
      "loss": 0.6184,
      "num_input_tokens_seen": 5902096,
      "step": 9040
    },
    {
      "epoch": 4.740566037735849,
      "grad_norm": 0.14567628502845764,
      "learning_rate": 4.719530043492829e-05,
      "loss": 0.4799,
      "num_input_tokens_seen": 5904880,
      "step": 9045
    },
    {
      "epoch": 4.743186582809225,
      "grad_norm": 0.1740751564502716,
      "learning_rate": 4.7190035983856954e-05,
      "loss": 0.5361,
      "num_input_tokens_seen": 5907632,
      "step": 9050
    },
    {
      "epoch": 4.7458071278826,
      "grad_norm": 0.5194262862205505,
      "learning_rate": 4.718476689088325e-05,
      "loss": 0.4903,
      "num_input_tokens_seen": 5911120,
      "step": 9055
    },
    {
      "epoch": 4.748427672955975,
      "grad_norm": 0.11609648913145065,
      "learning_rate": 4.71794931571094e-05,
      "loss": 0.4569,
      "num_input_tokens_seen": 5914928,
      "step": 9060
    },
    {
      "epoch": 4.75104821802935,
      "grad_norm": 0.16733889281749725,
      "learning_rate": 4.717421478363862e-05,
      "loss": 0.4249,
      "num_input_tokens_seen": 5918192,
      "step": 9065
    },
    {
      "epoch": 4.753668763102725,
      "grad_norm": 0.17700912058353424,
      "learning_rate": 4.7168931771575085e-05,
      "loss": 0.3924,
      "num_input_tokens_seen": 5921232,
      "step": 9070
    },
    {
      "epoch": 4.7562893081761,
      "grad_norm": 0.12958760559558868,
      "learning_rate": 4.716364412202394e-05,
      "loss": 0.4897,
      "num_input_tokens_seen": 5925040,
      "step": 9075
    },
    {
      "epoch": 4.758909853249476,
      "grad_norm": 0.27148109674453735,
      "learning_rate": 4.7158351836091306e-05,
      "loss": 0.548,
      "num_input_tokens_seen": 5927376,
      "step": 9080
    },
    {
      "epoch": 4.761530398322851,
      "grad_norm": 0.13173064589500427,
      "learning_rate": 4.7153054914884266e-05,
      "loss": 0.4339,
      "num_input_tokens_seen": 5930576,
      "step": 9085
    },
    {
      "epoch": 4.764150943396227,
      "grad_norm": 0.4172661006450653,
      "learning_rate": 4.714775335951086e-05,
      "loss": 0.4107,
      "num_input_tokens_seen": 5933328,
      "step": 9090
    },
    {
      "epoch": 4.766771488469602,
      "grad_norm": 0.13291658461093903,
      "learning_rate": 4.7142447171080137e-05,
      "loss": 0.5014,
      "num_input_tokens_seen": 5936208,
      "step": 9095
    },
    {
      "epoch": 4.769392033542977,
      "grad_norm": 0.14965291321277618,
      "learning_rate": 4.713713635070208e-05,
      "loss": 0.6715,
      "num_input_tokens_seen": 5938832,
      "step": 9100
    },
    {
      "epoch": 4.772012578616352,
      "grad_norm": 0.24147720634937286,
      "learning_rate": 4.713182089948765e-05,
      "loss": 0.4169,
      "num_input_tokens_seen": 5942128,
      "step": 9105
    },
    {
      "epoch": 4.7746331236897275,
      "grad_norm": 0.11801937967538834,
      "learning_rate": 4.712650081854878e-05,
      "loss": 0.333,
      "num_input_tokens_seen": 5945648,
      "step": 9110
    },
    {
      "epoch": 4.777253668763103,
      "grad_norm": 0.1093977689743042,
      "learning_rate": 4.712117610899837e-05,
      "loss": 0.4266,
      "num_input_tokens_seen": 5950352,
      "step": 9115
    },
    {
      "epoch": 4.779874213836478,
      "grad_norm": 0.29374390840530396,
      "learning_rate": 4.7115846771950286e-05,
      "loss": 0.644,
      "num_input_tokens_seen": 5953488,
      "step": 9120
    },
    {
      "epoch": 4.782494758909853,
      "grad_norm": 0.13677406311035156,
      "learning_rate": 4.711051280851938e-05,
      "loss": 0.4895,
      "num_input_tokens_seen": 5956784,
      "step": 9125
    },
    {
      "epoch": 4.785115303983228,
      "grad_norm": 0.23815150558948517,
      "learning_rate": 4.7105174219821435e-05,
      "loss": 0.5627,
      "num_input_tokens_seen": 5960080,
      "step": 9130
    },
    {
      "epoch": 4.787735849056604,
      "grad_norm": 0.13023631274700165,
      "learning_rate": 4.709983100697323e-05,
      "loss": 0.42,
      "num_input_tokens_seen": 5962864,
      "step": 9135
    },
    {
      "epoch": 4.790356394129979,
      "grad_norm": 0.12029954791069031,
      "learning_rate": 4.709448317109252e-05,
      "loss": 0.4891,
      "num_input_tokens_seen": 5966000,
      "step": 9140
    },
    {
      "epoch": 4.7929769392033545,
      "grad_norm": 0.1550765037536621,
      "learning_rate": 4.708913071329798e-05,
      "loss": 0.5396,
      "num_input_tokens_seen": 5969168,
      "step": 9145
    },
    {
      "epoch": 4.79559748427673,
      "grad_norm": 0.13137474656105042,
      "learning_rate": 4.708377363470932e-05,
      "loss": 0.5365,
      "num_input_tokens_seen": 5972720,
      "step": 9150
    },
    {
      "epoch": 4.798218029350105,
      "grad_norm": 0.11977386474609375,
      "learning_rate": 4.7078411936447144e-05,
      "loss": 0.3776,
      "num_input_tokens_seen": 5975760,
      "step": 9155
    },
    {
      "epoch": 4.80083857442348,
      "grad_norm": 0.17344024777412415,
      "learning_rate": 4.707304561963308e-05,
      "loss": 0.3835,
      "num_input_tokens_seen": 5979120,
      "step": 9160
    },
    {
      "epoch": 4.803459119496855,
      "grad_norm": 0.11413069069385529,
      "learning_rate": 4.70676746853897e-05,
      "loss": 0.5342,
      "num_input_tokens_seen": 5983376,
      "step": 9165
    },
    {
      "epoch": 4.80607966457023,
      "grad_norm": 0.1466069370508194,
      "learning_rate": 4.706229913484054e-05,
      "loss": 0.5917,
      "num_input_tokens_seen": 5986416,
      "step": 9170
    },
    {
      "epoch": 4.808700209643606,
      "grad_norm": 0.19918517768383026,
      "learning_rate": 4.705691896911009e-05,
      "loss": 0.4246,
      "num_input_tokens_seen": 5989136,
      "step": 9175
    },
    {
      "epoch": 4.811320754716981,
      "grad_norm": 0.25097376108169556,
      "learning_rate": 4.705153418932383e-05,
      "loss": 0.7001,
      "num_input_tokens_seen": 5992112,
      "step": 9180
    },
    {
      "epoch": 4.813941299790356,
      "grad_norm": 0.13938824832439423,
      "learning_rate": 4.7046144796608206e-05,
      "loss": 0.4663,
      "num_input_tokens_seen": 5995472,
      "step": 9185
    },
    {
      "epoch": 4.816561844863732,
      "grad_norm": 0.1988946795463562,
      "learning_rate": 4.70407507920906e-05,
      "loss": 0.4805,
      "num_input_tokens_seen": 5998736,
      "step": 9190
    },
    {
      "epoch": 4.819182389937107,
      "grad_norm": 0.15756171941757202,
      "learning_rate": 4.7035352176899384e-05,
      "loss": 0.6989,
      "num_input_tokens_seen": 6001552,
      "step": 9195
    },
    {
      "epoch": 4.821802935010482,
      "grad_norm": 0.21047107875347137,
      "learning_rate": 4.7029948952163885e-05,
      "loss": 0.5586,
      "num_input_tokens_seen": 6005520,
      "step": 9200
    },
    {
      "epoch": 4.8244234800838575,
      "grad_norm": 0.13689027726650238,
      "learning_rate": 4.7024541119014386e-05,
      "loss": 0.4578,
      "num_input_tokens_seen": 6009200,
      "step": 9205
    },
    {
      "epoch": 4.827044025157233,
      "grad_norm": 0.15584272146224976,
      "learning_rate": 4.701912867858217e-05,
      "loss": 0.4191,
      "num_input_tokens_seen": 6012912,
      "step": 9210
    },
    {
      "epoch": 4.829664570230608,
      "grad_norm": 0.15279686450958252,
      "learning_rate": 4.7013711631999434e-05,
      "loss": 0.5988,
      "num_input_tokens_seen": 6017744,
      "step": 9215
    },
    {
      "epoch": 4.832285115303983,
      "grad_norm": 0.26474249362945557,
      "learning_rate": 4.700828998039937e-05,
      "loss": 0.4668,
      "num_input_tokens_seen": 6020624,
      "step": 9220
    },
    {
      "epoch": 4.834905660377358,
      "grad_norm": 0.12291811406612396,
      "learning_rate": 4.700286372491612e-05,
      "loss": 0.484,
      "num_input_tokens_seen": 6023312,
      "step": 9225
    },
    {
      "epoch": 4.837526205450734,
      "grad_norm": 0.2693101167678833,
      "learning_rate": 4.6997432866684816e-05,
      "loss": 0.5168,
      "num_input_tokens_seen": 6026032,
      "step": 9230
    },
    {
      "epoch": 4.840146750524109,
      "grad_norm": 0.211256206035614,
      "learning_rate": 4.6991997406841504e-05,
      "loss": 0.4485,
      "num_input_tokens_seen": 6029872,
      "step": 9235
    },
    {
      "epoch": 4.8427672955974845,
      "grad_norm": 0.11137031018733978,
      "learning_rate": 4.698655734652324e-05,
      "loss": 0.5055,
      "num_input_tokens_seen": 6033232,
      "step": 9240
    },
    {
      "epoch": 4.84538784067086,
      "grad_norm": 0.5142791271209717,
      "learning_rate": 4.698111268686801e-05,
      "loss": 0.701,
      "num_input_tokens_seen": 6035888,
      "step": 9245
    },
    {
      "epoch": 4.848008385744235,
      "grad_norm": 0.10917633771896362,
      "learning_rate": 4.6975663429014785e-05,
      "loss": 0.4989,
      "num_input_tokens_seen": 6039408,
      "step": 9250
    },
    {
      "epoch": 4.85062893081761,
      "grad_norm": 0.13934636116027832,
      "learning_rate": 4.697020957410347e-05,
      "loss": 0.5775,
      "num_input_tokens_seen": 6042736,
      "step": 9255
    },
    {
      "epoch": 4.853249475890985,
      "grad_norm": 0.15495944023132324,
      "learning_rate": 4.6964751123274974e-05,
      "loss": 0.524,
      "num_input_tokens_seen": 6045296,
      "step": 9260
    },
    {
      "epoch": 4.85587002096436,
      "grad_norm": 0.24484215676784515,
      "learning_rate": 4.695928807767113e-05,
      "loss": 0.5123,
      "num_input_tokens_seen": 6047920,
      "step": 9265
    },
    {
      "epoch": 4.8584905660377355,
      "grad_norm": 0.10594244301319122,
      "learning_rate": 4.695382043843474e-05,
      "loss": 0.3862,
      "num_input_tokens_seen": 6052432,
      "step": 9270
    },
    {
      "epoch": 4.861111111111111,
      "grad_norm": 0.24061022698879242,
      "learning_rate": 4.694834820670958e-05,
      "loss": 0.4664,
      "num_input_tokens_seen": 6055472,
      "step": 9275
    },
    {
      "epoch": 4.863731656184486,
      "grad_norm": 0.23813609778881073,
      "learning_rate": 4.694287138364037e-05,
      "loss": 0.6078,
      "num_input_tokens_seen": 6059184,
      "step": 9280
    },
    {
      "epoch": 4.866352201257862,
      "grad_norm": 0.18857166171073914,
      "learning_rate": 4.6937389970372815e-05,
      "loss": 0.6356,
      "num_input_tokens_seen": 6062192,
      "step": 9285
    },
    {
      "epoch": 4.868972746331237,
      "grad_norm": 0.24348151683807373,
      "learning_rate": 4.693190396805354e-05,
      "loss": 0.5079,
      "num_input_tokens_seen": 6065264,
      "step": 9290
    },
    {
      "epoch": 4.871593291404612,
      "grad_norm": 0.14854796230793,
      "learning_rate": 4.692641337783017e-05,
      "loss": 0.5636,
      "num_input_tokens_seen": 6069008,
      "step": 9295
    },
    {
      "epoch": 4.8742138364779874,
      "grad_norm": 0.14685194194316864,
      "learning_rate": 4.6920918200851274e-05,
      "loss": 0.4462,
      "num_input_tokens_seen": 6072624,
      "step": 9300
    },
    {
      "epoch": 4.876834381551363,
      "grad_norm": 0.14054226875305176,
      "learning_rate": 4.691541843826638e-05,
      "loss": 0.6168,
      "num_input_tokens_seen": 6075280,
      "step": 9305
    },
    {
      "epoch": 4.879454926624738,
      "grad_norm": 0.2630552351474762,
      "learning_rate": 4.690991409122597e-05,
      "loss": 0.5129,
      "num_input_tokens_seen": 6078448,
      "step": 9310
    },
    {
      "epoch": 4.882075471698113,
      "grad_norm": 0.22026294469833374,
      "learning_rate": 4.690440516088149e-05,
      "loss": 0.3459,
      "num_input_tokens_seen": 6081008,
      "step": 9315
    },
    {
      "epoch": 4.884696016771488,
      "grad_norm": 0.17422820627689362,
      "learning_rate": 4.689889164838536e-05,
      "loss": 0.5226,
      "num_input_tokens_seen": 6085296,
      "step": 9320
    },
    {
      "epoch": 4.887316561844864,
      "grad_norm": 0.14989446103572845,
      "learning_rate": 4.6893373554890925e-05,
      "loss": 0.5084,
      "num_input_tokens_seen": 6087664,
      "step": 9325
    },
    {
      "epoch": 4.889937106918239,
      "grad_norm": 0.1871054619550705,
      "learning_rate": 4.688785088155251e-05,
      "loss": 0.517,
      "num_input_tokens_seen": 6090832,
      "step": 9330
    },
    {
      "epoch": 4.8925576519916145,
      "grad_norm": 0.20717886090278625,
      "learning_rate": 4.688232362952541e-05,
      "loss": 0.5706,
      "num_input_tokens_seen": 6093616,
      "step": 9335
    },
    {
      "epoch": 4.89517819706499,
      "grad_norm": 0.18745779991149902,
      "learning_rate": 4.687679179996585e-05,
      "loss": 0.3981,
      "num_input_tokens_seen": 6096592,
      "step": 9340
    },
    {
      "epoch": 4.897798742138365,
      "grad_norm": 0.09143057465553284,
      "learning_rate": 4.687125539403103e-05,
      "loss": 0.3647,
      "num_input_tokens_seen": 6100144,
      "step": 9345
    },
    {
      "epoch": 4.90041928721174,
      "grad_norm": 0.22190114855766296,
      "learning_rate": 4.68657144128791e-05,
      "loss": 0.5958,
      "num_input_tokens_seen": 6103536,
      "step": 9350
    },
    {
      "epoch": 4.903039832285115,
      "grad_norm": 0.12192828208208084,
      "learning_rate": 4.6860168857669175e-05,
      "loss": 0.5131,
      "num_input_tokens_seen": 6107408,
      "step": 9355
    },
    {
      "epoch": 4.90566037735849,
      "grad_norm": 0.1504584401845932,
      "learning_rate": 4.6854618729561315e-05,
      "loss": 0.5258,
      "num_input_tokens_seen": 6110480,
      "step": 9360
    },
    {
      "epoch": 4.9082809224318655,
      "grad_norm": 0.27398109436035156,
      "learning_rate": 4.684906402971655e-05,
      "loss": 0.4522,
      "num_input_tokens_seen": 6113488,
      "step": 9365
    },
    {
      "epoch": 4.910901467505241,
      "grad_norm": 0.15210503339767456,
      "learning_rate": 4.684350475929685e-05,
      "loss": 0.5011,
      "num_input_tokens_seen": 6117648,
      "step": 9370
    },
    {
      "epoch": 4.913522012578616,
      "grad_norm": 0.2670143246650696,
      "learning_rate": 4.683794091946517e-05,
      "loss": 0.5011,
      "num_input_tokens_seen": 6120592,
      "step": 9375
    },
    {
      "epoch": 4.916142557651992,
      "grad_norm": 0.16610638797283173,
      "learning_rate": 4.683237251138538e-05,
      "loss": 0.5596,
      "num_input_tokens_seen": 6124400,
      "step": 9380
    },
    {
      "epoch": 4.918763102725367,
      "grad_norm": 0.17931626737117767,
      "learning_rate": 4.6826799536222346e-05,
      "loss": 0.545,
      "num_input_tokens_seen": 6128400,
      "step": 9385
    },
    {
      "epoch": 4.921383647798742,
      "grad_norm": 0.11311804503202438,
      "learning_rate": 4.6821221995141865e-05,
      "loss": 0.502,
      "num_input_tokens_seen": 6131536,
      "step": 9390
    },
    {
      "epoch": 4.924004192872117,
      "grad_norm": 0.14626739919185638,
      "learning_rate": 4.681563988931068e-05,
      "loss": 0.3662,
      "num_input_tokens_seen": 6135280,
      "step": 9395
    },
    {
      "epoch": 4.926624737945493,
      "grad_norm": 0.3302150368690491,
      "learning_rate": 4.681005321989652e-05,
      "loss": 0.5479,
      "num_input_tokens_seen": 6137680,
      "step": 9400
    },
    {
      "epoch": 4.929245283018868,
      "grad_norm": 0.21378080546855927,
      "learning_rate": 4.6804461988068055e-05,
      "loss": 0.7335,
      "num_input_tokens_seen": 6141264,
      "step": 9405
    },
    {
      "epoch": 4.931865828092243,
      "grad_norm": 0.24884238839149475,
      "learning_rate": 4.6798866194994896e-05,
      "loss": 0.4554,
      "num_input_tokens_seen": 6145104,
      "step": 9410
    },
    {
      "epoch": 4.934486373165618,
      "grad_norm": 0.1225755512714386,
      "learning_rate": 4.6793265841847624e-05,
      "loss": 0.4869,
      "num_input_tokens_seen": 6148976,
      "step": 9415
    },
    {
      "epoch": 4.937106918238994,
      "grad_norm": 0.24238844215869904,
      "learning_rate": 4.6787660929797766e-05,
      "loss": 0.6703,
      "num_input_tokens_seen": 6152560,
      "step": 9420
    },
    {
      "epoch": 4.939727463312369,
      "grad_norm": 0.24659496545791626,
      "learning_rate": 4.678205146001781e-05,
      "loss": 0.6355,
      "num_input_tokens_seen": 6155024,
      "step": 9425
    },
    {
      "epoch": 4.9423480083857445,
      "grad_norm": 0.1979493498802185,
      "learning_rate": 4.67764374336812e-05,
      "loss": 0.5488,
      "num_input_tokens_seen": 6157584,
      "step": 9430
    },
    {
      "epoch": 4.94496855345912,
      "grad_norm": 0.18837112188339233,
      "learning_rate": 4.6770818851962305e-05,
      "loss": 0.5518,
      "num_input_tokens_seen": 6160240,
      "step": 9435
    },
    {
      "epoch": 4.947589098532495,
      "grad_norm": 0.1277005821466446,
      "learning_rate": 4.676519571603648e-05,
      "loss": 0.539,
      "num_input_tokens_seen": 6163696,
      "step": 9440
    },
    {
      "epoch": 4.95020964360587,
      "grad_norm": 0.13960683345794678,
      "learning_rate": 4.6759568027080034e-05,
      "loss": 0.3774,
      "num_input_tokens_seen": 6168880,
      "step": 9445
    },
    {
      "epoch": 4.952830188679245,
      "grad_norm": 0.23114456236362457,
      "learning_rate": 4.67539357862702e-05,
      "loss": 0.5213,
      "num_input_tokens_seen": 6172208,
      "step": 9450
    },
    {
      "epoch": 4.95545073375262,
      "grad_norm": 0.13261938095092773,
      "learning_rate": 4.6748298994785175e-05,
      "loss": 0.3872,
      "num_input_tokens_seen": 6174928,
      "step": 9455
    },
    {
      "epoch": 4.9580712788259955,
      "grad_norm": 0.11506618559360504,
      "learning_rate": 4.6742657653804116e-05,
      "loss": 0.499,
      "num_input_tokens_seen": 6178992,
      "step": 9460
    },
    {
      "epoch": 4.960691823899371,
      "grad_norm": 0.17603373527526855,
      "learning_rate": 4.673701176450713e-05,
      "loss": 0.4805,
      "num_input_tokens_seen": 6182128,
      "step": 9465
    },
    {
      "epoch": 4.963312368972746,
      "grad_norm": 0.2999979853630066,
      "learning_rate": 4.6731361328075286e-05,
      "loss": 0.4756,
      "num_input_tokens_seen": 6185328,
      "step": 9470
    },
    {
      "epoch": 4.965932914046122,
      "grad_norm": 0.125194251537323,
      "learning_rate": 4.672570634569056e-05,
      "loss": 0.5068,
      "num_input_tokens_seen": 6188560,
      "step": 9475
    },
    {
      "epoch": 4.968553459119497,
      "grad_norm": 0.16755349934101105,
      "learning_rate": 4.672004681853593e-05,
      "loss": 0.5998,
      "num_input_tokens_seen": 6193136,
      "step": 9480
    },
    {
      "epoch": 4.971174004192872,
      "grad_norm": 0.17056195437908173,
      "learning_rate": 4.6714382747795304e-05,
      "loss": 0.7267,
      "num_input_tokens_seen": 6196816,
      "step": 9485
    },
    {
      "epoch": 4.973794549266247,
      "grad_norm": 0.17134465277194977,
      "learning_rate": 4.670871413465354e-05,
      "loss": 0.4827,
      "num_input_tokens_seen": 6200272,
      "step": 9490
    },
    {
      "epoch": 4.976415094339623,
      "grad_norm": 0.3668573796749115,
      "learning_rate": 4.670304098029645e-05,
      "loss": 0.4628,
      "num_input_tokens_seen": 6203216,
      "step": 9495
    },
    {
      "epoch": 4.979035639412998,
      "grad_norm": 0.1413247436285019,
      "learning_rate": 4.669736328591078e-05,
      "loss": 0.4652,
      "num_input_tokens_seen": 6206736,
      "step": 9500
    },
    {
      "epoch": 4.981656184486373,
      "grad_norm": 0.13029566407203674,
      "learning_rate": 4.669168105268424e-05,
      "loss": 0.4308,
      "num_input_tokens_seen": 6209936,
      "step": 9505
    },
    {
      "epoch": 4.984276729559748,
      "grad_norm": 0.12160765379667282,
      "learning_rate": 4.6685994281805514e-05,
      "loss": 0.5116,
      "num_input_tokens_seen": 6213424,
      "step": 9510
    },
    {
      "epoch": 4.986897274633124,
      "grad_norm": 0.27701398730278015,
      "learning_rate": 4.6680302974464175e-05,
      "loss": 0.4216,
      "num_input_tokens_seen": 6217360,
      "step": 9515
    },
    {
      "epoch": 4.989517819706499,
      "grad_norm": 0.2199353277683258,
      "learning_rate": 4.66746071318508e-05,
      "loss": 0.5495,
      "num_input_tokens_seen": 6219984,
      "step": 9520
    },
    {
      "epoch": 4.9921383647798745,
      "grad_norm": 0.11715327948331833,
      "learning_rate": 4.6668906755156896e-05,
      "loss": 0.4336,
      "num_input_tokens_seen": 6223664,
      "step": 9525
    },
    {
      "epoch": 4.99475890985325,
      "grad_norm": 0.1164262592792511,
      "learning_rate": 4.6663201845574906e-05,
      "loss": 0.5118,
      "num_input_tokens_seen": 6227440,
      "step": 9530
    },
    {
      "epoch": 4.997379454926625,
      "grad_norm": 0.18450197577476501,
      "learning_rate": 4.6657492404298234e-05,
      "loss": 0.4506,
      "num_input_tokens_seen": 6230608,
      "step": 9535
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.23400825262069702,
      "learning_rate": 4.665177843252124e-05,
      "loss": 0.5107,
      "num_input_tokens_seen": 6233920,
      "step": 9540
    },
    {
      "epoch": 5.0,
      "eval_loss": 0.5173024535179138,
      "eval_runtime": 10.0102,
      "eval_samples_per_second": 84.713,
      "eval_steps_per_second": 21.178,
      "num_input_tokens_seen": 6233920,
      "step": 9540
    },
    {
      "epoch": 5.002620545073375,
      "grad_norm": 0.1398463249206543,
      "learning_rate": 4.664605993143921e-05,
      "loss": 0.4749,
      "num_input_tokens_seen": 6237312,
      "step": 9545
    },
    {
      "epoch": 5.00524109014675,
      "grad_norm": 0.1483350396156311,
      "learning_rate": 4.6640336902248396e-05,
      "loss": 0.4393,
      "num_input_tokens_seen": 6241344,
      "step": 9550
    },
    {
      "epoch": 5.0078616352201255,
      "grad_norm": 0.17506057024002075,
      "learning_rate": 4.6634609346145984e-05,
      "loss": 0.5959,
      "num_input_tokens_seen": 6244064,
      "step": 9555
    },
    {
      "epoch": 5.010482180293501,
      "grad_norm": 0.13666987419128418,
      "learning_rate": 4.6628877264330116e-05,
      "loss": 0.7044,
      "num_input_tokens_seen": 6247008,
      "step": 9560
    },
    {
      "epoch": 5.013102725366877,
      "grad_norm": 0.11857616156339645,
      "learning_rate": 4.6623140657999886e-05,
      "loss": 0.5898,
      "num_input_tokens_seen": 6250016,
      "step": 9565
    },
    {
      "epoch": 5.015723270440252,
      "grad_norm": 0.13550204038619995,
      "learning_rate": 4.661739952835532e-05,
      "loss": 0.6192,
      "num_input_tokens_seen": 6253504,
      "step": 9570
    },
    {
      "epoch": 5.018343815513627,
      "grad_norm": 0.15146604180335999,
      "learning_rate": 4.66116538765974e-05,
      "loss": 0.5366,
      "num_input_tokens_seen": 6256256,
      "step": 9575
    },
    {
      "epoch": 5.020964360587002,
      "grad_norm": 0.12623119354248047,
      "learning_rate": 4.6605903703928034e-05,
      "loss": 0.6791,
      "num_input_tokens_seen": 6260160,
      "step": 9580
    },
    {
      "epoch": 5.023584905660377,
      "grad_norm": 0.1272144764661789,
      "learning_rate": 4.6600149011550116e-05,
      "loss": 0.4817,
      "num_input_tokens_seen": 6264032,
      "step": 9585
    },
    {
      "epoch": 5.026205450733753,
      "grad_norm": 0.11457471549510956,
      "learning_rate": 4.659438980066745e-05,
      "loss": 0.5598,
      "num_input_tokens_seen": 6267552,
      "step": 9590
    },
    {
      "epoch": 5.028825995807128,
      "grad_norm": 0.11454018205404282,
      "learning_rate": 4.65886260724848e-05,
      "loss": 0.5239,
      "num_input_tokens_seen": 6270272,
      "step": 9595
    },
    {
      "epoch": 5.031446540880503,
      "grad_norm": 0.1349821239709854,
      "learning_rate": 4.658285782820787e-05,
      "loss": 0.3316,
      "num_input_tokens_seen": 6274624,
      "step": 9600
    },
    {
      "epoch": 5.034067085953878,
      "grad_norm": 0.15372054278850555,
      "learning_rate": 4.657708506904331e-05,
      "loss": 0.4864,
      "num_input_tokens_seen": 6277728,
      "step": 9605
    },
    {
      "epoch": 5.036687631027253,
      "grad_norm": 0.2592124342918396,
      "learning_rate": 4.6571307796198724e-05,
      "loss": 0.4805,
      "num_input_tokens_seen": 6281024,
      "step": 9610
    },
    {
      "epoch": 5.039308176100629,
      "grad_norm": 0.19870460033416748,
      "learning_rate": 4.656552601088265e-05,
      "loss": 0.6211,
      "num_input_tokens_seen": 6284576,
      "step": 9615
    },
    {
      "epoch": 5.0419287211740045,
      "grad_norm": 0.15290868282318115,
      "learning_rate": 4.655973971430456e-05,
      "loss": 0.5837,
      "num_input_tokens_seen": 6288096,
      "step": 9620
    },
    {
      "epoch": 5.04454926624738,
      "grad_norm": 0.23290683329105377,
      "learning_rate": 4.6553948907674895e-05,
      "loss": 0.5698,
      "num_input_tokens_seen": 6292128,
      "step": 9625
    },
    {
      "epoch": 5.047169811320755,
      "grad_norm": 0.2676048278808594,
      "learning_rate": 4.654815359220501e-05,
      "loss": 0.5287,
      "num_input_tokens_seen": 6298912,
      "step": 9630
    },
    {
      "epoch": 5.04979035639413,
      "grad_norm": 0.18829938769340515,
      "learning_rate": 4.6542353769107226e-05,
      "loss": 0.4988,
      "num_input_tokens_seen": 6301536,
      "step": 9635
    },
    {
      "epoch": 5.052410901467505,
      "grad_norm": 0.16738606989383698,
      "learning_rate": 4.653654943959481e-05,
      "loss": 0.4041,
      "num_input_tokens_seen": 6303936,
      "step": 9640
    },
    {
      "epoch": 5.05503144654088,
      "grad_norm": 0.23445887863636017,
      "learning_rate": 4.653074060488195e-05,
      "loss": 0.5358,
      "num_input_tokens_seen": 6306880,
      "step": 9645
    },
    {
      "epoch": 5.0576519916142555,
      "grad_norm": 0.16409359872341156,
      "learning_rate": 4.6524927266183796e-05,
      "loss": 0.5169,
      "num_input_tokens_seen": 6309920,
      "step": 9650
    },
    {
      "epoch": 5.060272536687631,
      "grad_norm": 0.16817130148410797,
      "learning_rate": 4.6519109424716426e-05,
      "loss": 0.4424,
      "num_input_tokens_seen": 6312480,
      "step": 9655
    },
    {
      "epoch": 5.062893081761007,
      "grad_norm": 0.11197037249803543,
      "learning_rate": 4.6513287081696855e-05,
      "loss": 0.468,
      "num_input_tokens_seen": 6316256,
      "step": 9660
    },
    {
      "epoch": 5.065513626834382,
      "grad_norm": 0.15166336297988892,
      "learning_rate": 4.6507460238343076e-05,
      "loss": 0.5666,
      "num_input_tokens_seen": 6319520,
      "step": 9665
    },
    {
      "epoch": 5.068134171907757,
      "grad_norm": 0.10841397196054459,
      "learning_rate": 4.6501628895873984e-05,
      "loss": 0.5995,
      "num_input_tokens_seen": 6323328,
      "step": 9670
    },
    {
      "epoch": 5.070754716981132,
      "grad_norm": 0.11818933486938477,
      "learning_rate": 4.649579305550942e-05,
      "loss": 0.3326,
      "num_input_tokens_seen": 6326720,
      "step": 9675
    },
    {
      "epoch": 5.073375262054507,
      "grad_norm": 0.2546842396259308,
      "learning_rate": 4.64899527184702e-05,
      "loss": 0.5147,
      "num_input_tokens_seen": 6329280,
      "step": 9680
    },
    {
      "epoch": 5.075995807127883,
      "grad_norm": 0.1175919696688652,
      "learning_rate": 4.6484107885978035e-05,
      "loss": 0.461,
      "num_input_tokens_seen": 6332384,
      "step": 9685
    },
    {
      "epoch": 5.078616352201258,
      "grad_norm": 0.15186569094657898,
      "learning_rate": 4.64782585592556e-05,
      "loss": 0.3903,
      "num_input_tokens_seen": 6336096,
      "step": 9690
    },
    {
      "epoch": 5.081236897274633,
      "grad_norm": 0.19844305515289307,
      "learning_rate": 4.647240473952652e-05,
      "loss": 0.632,
      "num_input_tokens_seen": 6338432,
      "step": 9695
    },
    {
      "epoch": 5.083857442348008,
      "grad_norm": 0.19022123515605927,
      "learning_rate": 4.6466546428015336e-05,
      "loss": 0.5091,
      "num_input_tokens_seen": 6342016,
      "step": 9700
    },
    {
      "epoch": 5.086477987421383,
      "grad_norm": 0.3264520466327667,
      "learning_rate": 4.646068362594753e-05,
      "loss": 0.6055,
      "num_input_tokens_seen": 6344896,
      "step": 9705
    },
    {
      "epoch": 5.089098532494759,
      "grad_norm": 0.1576022207736969,
      "learning_rate": 4.645481633454956e-05,
      "loss": 0.5237,
      "num_input_tokens_seen": 6348096,
      "step": 9710
    },
    {
      "epoch": 5.0917190775681345,
      "grad_norm": 0.2283770591020584,
      "learning_rate": 4.644894455504878e-05,
      "loss": 0.6342,
      "num_input_tokens_seen": 6351488,
      "step": 9715
    },
    {
      "epoch": 5.09433962264151,
      "grad_norm": 0.17739246785640717,
      "learning_rate": 4.64430682886735e-05,
      "loss": 0.5962,
      "num_input_tokens_seen": 6354464,
      "step": 9720
    },
    {
      "epoch": 5.096960167714885,
      "grad_norm": 0.13013140857219696,
      "learning_rate": 4.643718753665297e-05,
      "loss": 0.4771,
      "num_input_tokens_seen": 6358912,
      "step": 9725
    },
    {
      "epoch": 5.09958071278826,
      "grad_norm": 0.2921384871006012,
      "learning_rate": 4.643130230021737e-05,
      "loss": 0.436,
      "num_input_tokens_seen": 6362112,
      "step": 9730
    },
    {
      "epoch": 5.102201257861635,
      "grad_norm": 0.15944524109363556,
      "learning_rate": 4.6425412580597836e-05,
      "loss": 0.4816,
      "num_input_tokens_seen": 6364896,
      "step": 9735
    },
    {
      "epoch": 5.10482180293501,
      "grad_norm": 0.19615164399147034,
      "learning_rate": 4.641951837902643e-05,
      "loss": 0.6974,
      "num_input_tokens_seen": 6368160,
      "step": 9740
    },
    {
      "epoch": 5.1074423480083855,
      "grad_norm": 0.16707772016525269,
      "learning_rate": 4.641361969673614e-05,
      "loss": 0.3961,
      "num_input_tokens_seen": 6371264,
      "step": 9745
    },
    {
      "epoch": 5.110062893081761,
      "grad_norm": 0.19820302724838257,
      "learning_rate": 4.640771653496091e-05,
      "loss": 0.5652,
      "num_input_tokens_seen": 6374144,
      "step": 9750
    },
    {
      "epoch": 5.112683438155136,
      "grad_norm": 0.1422671228647232,
      "learning_rate": 4.64018088949356e-05,
      "loss": 0.4829,
      "num_input_tokens_seen": 6377600,
      "step": 9755
    },
    {
      "epoch": 5.115303983228512,
      "grad_norm": 0.276500940322876,
      "learning_rate": 4.6395896777896055e-05,
      "loss": 0.5071,
      "num_input_tokens_seen": 6380608,
      "step": 9760
    },
    {
      "epoch": 5.117924528301887,
      "grad_norm": 0.16783981025218964,
      "learning_rate": 4.638998018507899e-05,
      "loss": 0.6588,
      "num_input_tokens_seen": 6384640,
      "step": 9765
    },
    {
      "epoch": 5.120545073375262,
      "grad_norm": 0.29385194182395935,
      "learning_rate": 4.63840591177221e-05,
      "loss": 0.4422,
      "num_input_tokens_seen": 6387456,
      "step": 9770
    },
    {
      "epoch": 5.123165618448637,
      "grad_norm": 0.16568639874458313,
      "learning_rate": 4.637813357706401e-05,
      "loss": 0.4803,
      "num_input_tokens_seen": 6390880,
      "step": 9775
    },
    {
      "epoch": 5.1257861635220126,
      "grad_norm": 0.19199949502944946,
      "learning_rate": 4.6372203564344264e-05,
      "loss": 0.4149,
      "num_input_tokens_seen": 6393696,
      "step": 9780
    },
    {
      "epoch": 5.128406708595388,
      "grad_norm": 0.10991288721561432,
      "learning_rate": 4.6366269080803364e-05,
      "loss": 0.5211,
      "num_input_tokens_seen": 6397472,
      "step": 9785
    },
    {
      "epoch": 5.131027253668763,
      "grad_norm": 0.12422651052474976,
      "learning_rate": 4.636033012768273e-05,
      "loss": 0.4881,
      "num_input_tokens_seen": 6400576,
      "step": 9790
    },
    {
      "epoch": 5.133647798742138,
      "grad_norm": 0.16267085075378418,
      "learning_rate": 4.635438670622473e-05,
      "loss": 0.6256,
      "num_input_tokens_seen": 6403168,
      "step": 9795
    },
    {
      "epoch": 5.136268343815513,
      "grad_norm": 0.13594649732112885,
      "learning_rate": 4.6348438817672646e-05,
      "loss": 0.5289,
      "num_input_tokens_seen": 6406144,
      "step": 9800
    },
    {
      "epoch": 5.138888888888889,
      "grad_norm": 0.3227098286151886,
      "learning_rate": 4.634248646327072e-05,
      "loss": 0.6055,
      "num_input_tokens_seen": 6409344,
      "step": 9805
    },
    {
      "epoch": 5.1415094339622645,
      "grad_norm": 0.37387609481811523,
      "learning_rate": 4.6336529644264114e-05,
      "loss": 0.4855,
      "num_input_tokens_seen": 6411776,
      "step": 9810
    },
    {
      "epoch": 5.14412997903564,
      "grad_norm": 0.11678237468004227,
      "learning_rate": 4.633056836189893e-05,
      "loss": 0.3731,
      "num_input_tokens_seen": 6414784,
      "step": 9815
    },
    {
      "epoch": 5.146750524109015,
      "grad_norm": 0.349918395280838,
      "learning_rate": 4.632460261742219e-05,
      "loss": 0.463,
      "num_input_tokens_seen": 6417248,
      "step": 9820
    },
    {
      "epoch": 5.14937106918239,
      "grad_norm": 0.20393963158130646,
      "learning_rate": 4.631863241208187e-05,
      "loss": 0.6146,
      "num_input_tokens_seen": 6420416,
      "step": 9825
    },
    {
      "epoch": 5.151991614255765,
      "grad_norm": 0.23460832238197327,
      "learning_rate": 4.6312657747126865e-05,
      "loss": 0.4249,
      "num_input_tokens_seen": 6424288,
      "step": 9830
    },
    {
      "epoch": 5.15461215932914,
      "grad_norm": 0.13557150959968567,
      "learning_rate": 4.630667862380701e-05,
      "loss": 0.6108,
      "num_input_tokens_seen": 6426784,
      "step": 9835
    },
    {
      "epoch": 5.1572327044025155,
      "grad_norm": 0.11943891644477844,
      "learning_rate": 4.630069504337306e-05,
      "loss": 0.5592,
      "num_input_tokens_seen": 6429760,
      "step": 9840
    },
    {
      "epoch": 5.159853249475891,
      "grad_norm": 0.17237137258052826,
      "learning_rate": 4.629470700707672e-05,
      "loss": 0.4839,
      "num_input_tokens_seen": 6432480,
      "step": 9845
    },
    {
      "epoch": 5.162473794549266,
      "grad_norm": 0.14331601560115814,
      "learning_rate": 4.628871451617062e-05,
      "loss": 0.4696,
      "num_input_tokens_seen": 6437312,
      "step": 9850
    },
    {
      "epoch": 5.165094339622642,
      "grad_norm": 0.1457863450050354,
      "learning_rate": 4.628271757190832e-05,
      "loss": 0.6091,
      "num_input_tokens_seen": 6441664,
      "step": 9855
    },
    {
      "epoch": 5.167714884696017,
      "grad_norm": 0.09900735318660736,
      "learning_rate": 4.62767161755443e-05,
      "loss": 0.4184,
      "num_input_tokens_seen": 6445120,
      "step": 9860
    },
    {
      "epoch": 5.170335429769392,
      "grad_norm": 0.20787283778190613,
      "learning_rate": 4.6270710328334004e-05,
      "loss": 0.4868,
      "num_input_tokens_seen": 6447968,
      "step": 9865
    },
    {
      "epoch": 5.172955974842767,
      "grad_norm": 0.1902051419019699,
      "learning_rate": 4.626470003153377e-05,
      "loss": 0.4873,
      "num_input_tokens_seen": 6450592,
      "step": 9870
    },
    {
      "epoch": 5.1755765199161425,
      "grad_norm": 0.24964368343353271,
      "learning_rate": 4.625868528640089e-05,
      "loss": 0.4776,
      "num_input_tokens_seen": 6454496,
      "step": 9875
    },
    {
      "epoch": 5.178197064989518,
      "grad_norm": 0.18714864552021027,
      "learning_rate": 4.6252666094193586e-05,
      "loss": 0.4233,
      "num_input_tokens_seen": 6457088,
      "step": 9880
    },
    {
      "epoch": 5.180817610062893,
      "grad_norm": 0.31449517607688904,
      "learning_rate": 4.6246642456171e-05,
      "loss": 0.5902,
      "num_input_tokens_seen": 6461088,
      "step": 9885
    },
    {
      "epoch": 5.183438155136268,
      "grad_norm": 0.14486055076122284,
      "learning_rate": 4.6240614373593196e-05,
      "loss": 0.5351,
      "num_input_tokens_seen": 6463872,
      "step": 9890
    },
    {
      "epoch": 5.186058700209643,
      "grad_norm": 0.1352284550666809,
      "learning_rate": 4.623458184772119e-05,
      "loss": 0.5317,
      "num_input_tokens_seen": 6467424,
      "step": 9895
    },
    {
      "epoch": 5.188679245283019,
      "grad_norm": 0.17942363023757935,
      "learning_rate": 4.622854487981693e-05,
      "loss": 0.4694,
      "num_input_tokens_seen": 6470688,
      "step": 9900
    },
    {
      "epoch": 5.191299790356394,
      "grad_norm": 0.14414499700069427,
      "learning_rate": 4.622250347114326e-05,
      "loss": 0.478,
      "num_input_tokens_seen": 6473440,
      "step": 9905
    },
    {
      "epoch": 5.19392033542977,
      "grad_norm": 0.2739849090576172,
      "learning_rate": 4.621645762296399e-05,
      "loss": 0.5897,
      "num_input_tokens_seen": 6476352,
      "step": 9910
    },
    {
      "epoch": 5.196540880503145,
      "grad_norm": 0.13565300405025482,
      "learning_rate": 4.6210407336543835e-05,
      "loss": 0.4375,
      "num_input_tokens_seen": 6480064,
      "step": 9915
    },
    {
      "epoch": 5.19916142557652,
      "grad_norm": 0.14623145759105682,
      "learning_rate": 4.620435261314844e-05,
      "loss": 0.4685,
      "num_input_tokens_seen": 6483840,
      "step": 9920
    },
    {
      "epoch": 5.201781970649895,
      "grad_norm": 0.2142769992351532,
      "learning_rate": 4.61982934540444e-05,
      "loss": 0.7042,
      "num_input_tokens_seen": 6486368,
      "step": 9925
    },
    {
      "epoch": 5.20440251572327,
      "grad_norm": 0.10649332404136658,
      "learning_rate": 4.6192229860499206e-05,
      "loss": 0.4154,
      "num_input_tokens_seen": 6489152,
      "step": 9930
    },
    {
      "epoch": 5.2070230607966455,
      "grad_norm": 0.15688484907150269,
      "learning_rate": 4.6186161833781295e-05,
      "loss": 0.4926,
      "num_input_tokens_seen": 6493088,
      "step": 9935
    },
    {
      "epoch": 5.209643605870021,
      "grad_norm": 0.15142673254013062,
      "learning_rate": 4.618008937516003e-05,
      "loss": 0.468,
      "num_input_tokens_seen": 6496256,
      "step": 9940
    },
    {
      "epoch": 5.212264150943396,
      "grad_norm": 0.18677918612957,
      "learning_rate": 4.617401248590572e-05,
      "loss": 0.4888,
      "num_input_tokens_seen": 6499296,
      "step": 9945
    },
    {
      "epoch": 5.214884696016772,
      "grad_norm": 0.17134703695774078,
      "learning_rate": 4.616793116728955e-05,
      "loss": 0.4868,
      "num_input_tokens_seen": 6502624,
      "step": 9950
    },
    {
      "epoch": 5.217505241090147,
      "grad_norm": 0.14379531145095825,
      "learning_rate": 4.6161845420583685e-05,
      "loss": 0.7027,
      "num_input_tokens_seen": 6506272,
      "step": 9955
    },
    {
      "epoch": 5.220125786163522,
      "grad_norm": 0.20055174827575684,
      "learning_rate": 4.6155755247061174e-05,
      "loss": 0.4432,
      "num_input_tokens_seen": 6509504,
      "step": 9960
    },
    {
      "epoch": 5.222746331236897,
      "grad_norm": 0.24641694128513336,
      "learning_rate": 4.614966064799603e-05,
      "loss": 0.5866,
      "num_input_tokens_seen": 6512512,
      "step": 9965
    },
    {
      "epoch": 5.2253668763102725,
      "grad_norm": 0.0837007611989975,
      "learning_rate": 4.6143561624663166e-05,
      "loss": 0.5147,
      "num_input_tokens_seen": 6515456,
      "step": 9970
    },
    {
      "epoch": 5.227987421383648,
      "grad_norm": 0.15590429306030273,
      "learning_rate": 4.6137458178338425e-05,
      "loss": 0.4579,
      "num_input_tokens_seen": 6518880,
      "step": 9975
    },
    {
      "epoch": 5.230607966457023,
      "grad_norm": 0.21643556654453278,
      "learning_rate": 4.613135031029859e-05,
      "loss": 0.4837,
      "num_input_tokens_seen": 6521664,
      "step": 9980
    },
    {
      "epoch": 5.233228511530398,
      "grad_norm": 0.19032055139541626,
      "learning_rate": 4.612523802182134e-05,
      "loss": 0.5214,
      "num_input_tokens_seen": 6524160,
      "step": 9985
    },
    {
      "epoch": 5.235849056603773,
      "grad_norm": 0.16685576736927032,
      "learning_rate": 4.611912131418531e-05,
      "loss": 0.3984,
      "num_input_tokens_seen": 6527904,
      "step": 9990
    },
    {
      "epoch": 5.238469601677149,
      "grad_norm": 0.34227946400642395,
      "learning_rate": 4.611300018867004e-05,
      "loss": 0.7394,
      "num_input_tokens_seen": 6530560,
      "step": 9995
    },
    {
      "epoch": 5.241090146750524,
      "grad_norm": 0.17280730605125427,
      "learning_rate": 4.6106874646556e-05,
      "loss": 0.5447,
      "num_input_tokens_seen": 6534048,
      "step": 10000
    },
    {
      "epoch": 5.2437106918239,
      "grad_norm": 0.1741483360528946,
      "learning_rate": 4.6100744689124584e-05,
      "loss": 0.4492,
      "num_input_tokens_seen": 6536864,
      "step": 10005
    },
    {
      "epoch": 5.246331236897275,
      "grad_norm": 0.4126892685890198,
      "learning_rate": 4.6094610317658116e-05,
      "loss": 0.5927,
      "num_input_tokens_seen": 6540608,
      "step": 10010
    },
    {
      "epoch": 5.24895178197065,
      "grad_norm": 0.2093064934015274,
      "learning_rate": 4.608847153343982e-05,
      "loss": 0.5589,
      "num_input_tokens_seen": 6543744,
      "step": 10015
    },
    {
      "epoch": 5.251572327044025,
      "grad_norm": 0.2375878095626831,
      "learning_rate": 4.6082328337753874e-05,
      "loss": 0.4824,
      "num_input_tokens_seen": 6546656,
      "step": 10020
    },
    {
      "epoch": 5.2541928721174,
      "grad_norm": 0.20981644093990326,
      "learning_rate": 4.6076180731885353e-05,
      "loss": 0.556,
      "num_input_tokens_seen": 6549152,
      "step": 10025
    },
    {
      "epoch": 5.256813417190775,
      "grad_norm": 0.21049906313419342,
      "learning_rate": 4.6070028717120286e-05,
      "loss": 0.5759,
      "num_input_tokens_seen": 6552096,
      "step": 10030
    },
    {
      "epoch": 5.259433962264151,
      "grad_norm": 0.10719066858291626,
      "learning_rate": 4.606387229474558e-05,
      "loss": 0.5263,
      "num_input_tokens_seen": 6555232,
      "step": 10035
    },
    {
      "epoch": 5.262054507337526,
      "grad_norm": 0.1559862196445465,
      "learning_rate": 4.60577114660491e-05,
      "loss": 0.4466,
      "num_input_tokens_seen": 6557920,
      "step": 10040
    },
    {
      "epoch": 5.264675052410902,
      "grad_norm": 0.16143763065338135,
      "learning_rate": 4.605154623231962e-05,
      "loss": 0.5146,
      "num_input_tokens_seen": 6560928,
      "step": 10045
    },
    {
      "epoch": 5.267295597484277,
      "grad_norm": 0.18288692831993103,
      "learning_rate": 4.604537659484684e-05,
      "loss": 0.4597,
      "num_input_tokens_seen": 6563488,
      "step": 10050
    },
    {
      "epoch": 5.269916142557652,
      "grad_norm": 0.14872555434703827,
      "learning_rate": 4.603920255492138e-05,
      "loss": 0.4935,
      "num_input_tokens_seen": 6567520,
      "step": 10055
    },
    {
      "epoch": 5.272536687631027,
      "grad_norm": 0.10238254070281982,
      "learning_rate": 4.603302411383477e-05,
      "loss": 0.5304,
      "num_input_tokens_seen": 6570528,
      "step": 10060
    },
    {
      "epoch": 5.2751572327044025,
      "grad_norm": 0.14948873221874237,
      "learning_rate": 4.602684127287946e-05,
      "loss": 0.4291,
      "num_input_tokens_seen": 6573824,
      "step": 10065
    },
    {
      "epoch": 5.277777777777778,
      "grad_norm": 0.14362682402133942,
      "learning_rate": 4.602065403334885e-05,
      "loss": 0.5487,
      "num_input_tokens_seen": 6576896,
      "step": 10070
    },
    {
      "epoch": 5.280398322851153,
      "grad_norm": 0.3618836998939514,
      "learning_rate": 4.601446239653724e-05,
      "loss": 0.5415,
      "num_input_tokens_seen": 6579136,
      "step": 10075
    },
    {
      "epoch": 5.283018867924528,
      "grad_norm": 0.18138307332992554,
      "learning_rate": 4.600826636373984e-05,
      "loss": 0.4585,
      "num_input_tokens_seen": 6582080,
      "step": 10080
    },
    {
      "epoch": 5.285639412997903,
      "grad_norm": 0.18227246403694153,
      "learning_rate": 4.600206593625279e-05,
      "loss": 0.4822,
      "num_input_tokens_seen": 6584896,
      "step": 10085
    },
    {
      "epoch": 5.288259958071279,
      "grad_norm": 0.23692214488983154,
      "learning_rate": 4.599586111537315e-05,
      "loss": 0.515,
      "num_input_tokens_seen": 6588448,
      "step": 10090
    },
    {
      "epoch": 5.290880503144654,
      "grad_norm": 0.19631041586399078,
      "learning_rate": 4.59896519023989e-05,
      "loss": 0.423,
      "num_input_tokens_seen": 6592704,
      "step": 10095
    },
    {
      "epoch": 5.29350104821803,
      "grad_norm": 0.16373398900032043,
      "learning_rate": 4.5983438298628935e-05,
      "loss": 0.4995,
      "num_input_tokens_seen": 6596256,
      "step": 10100
    },
    {
      "epoch": 5.296121593291405,
      "grad_norm": 0.15506884455680847,
      "learning_rate": 4.597722030536306e-05,
      "loss": 0.5415,
      "num_input_tokens_seen": 6598656,
      "step": 10105
    },
    {
      "epoch": 5.29874213836478,
      "grad_norm": 0.12192460149526596,
      "learning_rate": 4.597099792390202e-05,
      "loss": 0.527,
      "num_input_tokens_seen": 6601504,
      "step": 10110
    },
    {
      "epoch": 5.301362683438155,
      "grad_norm": 0.2197960615158081,
      "learning_rate": 4.596477115554748e-05,
      "loss": 0.5479,
      "num_input_tokens_seen": 6604640,
      "step": 10115
    },
    {
      "epoch": 5.30398322851153,
      "grad_norm": 0.12405776977539062,
      "learning_rate": 4.5958540001601976e-05,
      "loss": 0.5352,
      "num_input_tokens_seen": 6607520,
      "step": 10120
    },
    {
      "epoch": 5.306603773584905,
      "grad_norm": 0.14281147718429565,
      "learning_rate": 4.595230446336902e-05,
      "loss": 0.4247,
      "num_input_tokens_seen": 6611360,
      "step": 10125
    },
    {
      "epoch": 5.309224318658281,
      "grad_norm": 0.1381833553314209,
      "learning_rate": 4.5946064542153005e-05,
      "loss": 0.4012,
      "num_input_tokens_seen": 6616672,
      "step": 10130
    },
    {
      "epoch": 5.311844863731656,
      "grad_norm": 0.18351587653160095,
      "learning_rate": 4.593982023925926e-05,
      "loss": 0.6738,
      "num_input_tokens_seen": 6619424,
      "step": 10135
    },
    {
      "epoch": 5.314465408805032,
      "grad_norm": 0.2099936306476593,
      "learning_rate": 4.5933571555994e-05,
      "loss": 0.3562,
      "num_input_tokens_seen": 6622656,
      "step": 10140
    },
    {
      "epoch": 5.317085953878407,
      "grad_norm": 0.2062096744775772,
      "learning_rate": 4.59273184936644e-05,
      "loss": 0.448,
      "num_input_tokens_seen": 6625568,
      "step": 10145
    },
    {
      "epoch": 5.319706498951782,
      "grad_norm": 0.5720161199569702,
      "learning_rate": 4.5921061053578523e-05,
      "loss": 0.6915,
      "num_input_tokens_seen": 6628896,
      "step": 10150
    },
    {
      "epoch": 5.322327044025157,
      "grad_norm": 0.21444375813007355,
      "learning_rate": 4.5914799237045355e-05,
      "loss": 0.4497,
      "num_input_tokens_seen": 6632672,
      "step": 10155
    },
    {
      "epoch": 5.3249475890985325,
      "grad_norm": 0.18019099533557892,
      "learning_rate": 4.5908533045374796e-05,
      "loss": 0.4549,
      "num_input_tokens_seen": 6637344,
      "step": 10160
    },
    {
      "epoch": 5.327568134171908,
      "grad_norm": 0.20258112251758575,
      "learning_rate": 4.590226247987766e-05,
      "loss": 0.6166,
      "num_input_tokens_seen": 6640416,
      "step": 10165
    },
    {
      "epoch": 5.330188679245283,
      "grad_norm": 0.13432931900024414,
      "learning_rate": 4.589598754186568e-05,
      "loss": 0.3591,
      "num_input_tokens_seen": 6642688,
      "step": 10170
    },
    {
      "epoch": 5.332809224318658,
      "grad_norm": 0.27848678827285767,
      "learning_rate": 4.5889708232651494e-05,
      "loss": 0.533,
      "num_input_tokens_seen": 6645952,
      "step": 10175
    },
    {
      "epoch": 5.335429769392033,
      "grad_norm": 0.17723017930984497,
      "learning_rate": 4.588342455354867e-05,
      "loss": 0.5285,
      "num_input_tokens_seen": 6648576,
      "step": 10180
    },
    {
      "epoch": 5.338050314465409,
      "grad_norm": 0.2507909834384918,
      "learning_rate": 4.5877136505871685e-05,
      "loss": 0.5948,
      "num_input_tokens_seen": 6651840,
      "step": 10185
    },
    {
      "epoch": 5.340670859538784,
      "grad_norm": 0.21049678325653076,
      "learning_rate": 4.5870844090935914e-05,
      "loss": 0.6562,
      "num_input_tokens_seen": 6654336,
      "step": 10190
    },
    {
      "epoch": 5.34329140461216,
      "grad_norm": 0.18829992413520813,
      "learning_rate": 4.586454731005768e-05,
      "loss": 0.5172,
      "num_input_tokens_seen": 6657600,
      "step": 10195
    },
    {
      "epoch": 5.345911949685535,
      "grad_norm": 0.12264261394739151,
      "learning_rate": 4.5858246164554175e-05,
      "loss": 0.4473,
      "num_input_tokens_seen": 6660768,
      "step": 10200
    },
    {
      "epoch": 5.34853249475891,
      "grad_norm": 0.3107477128505707,
      "learning_rate": 4.5851940655743536e-05,
      "loss": 0.5923,
      "num_input_tokens_seen": 6663808,
      "step": 10205
    },
    {
      "epoch": 5.351153039832285,
      "grad_norm": 0.09487741440534592,
      "learning_rate": 4.5845630784944804e-05,
      "loss": 0.528,
      "num_input_tokens_seen": 6666048,
      "step": 10210
    },
    {
      "epoch": 5.35377358490566,
      "grad_norm": 0.12491397559642792,
      "learning_rate": 4.583931655347793e-05,
      "loss": 0.4464,
      "num_input_tokens_seen": 6668672,
      "step": 10215
    },
    {
      "epoch": 5.356394129979035,
      "grad_norm": 0.3378477692604065,
      "learning_rate": 4.5832997962663785e-05,
      "loss": 0.6694,
      "num_input_tokens_seen": 6671680,
      "step": 10220
    },
    {
      "epoch": 5.359014675052411,
      "grad_norm": 0.13607174158096313,
      "learning_rate": 4.582667501382414e-05,
      "loss": 0.4239,
      "num_input_tokens_seen": 6674592,
      "step": 10225
    },
    {
      "epoch": 5.361635220125786,
      "grad_norm": 0.2032214105129242,
      "learning_rate": 4.582034770828169e-05,
      "loss": 0.4296,
      "num_input_tokens_seen": 6677632,
      "step": 10230
    },
    {
      "epoch": 5.364255765199162,
      "grad_norm": 0.23257681727409363,
      "learning_rate": 4.581401604736002e-05,
      "loss": 0.4265,
      "num_input_tokens_seen": 6680768,
      "step": 10235
    },
    {
      "epoch": 5.366876310272537,
      "grad_norm": 0.16318432986736298,
      "learning_rate": 4.580768003238365e-05,
      "loss": 0.5423,
      "num_input_tokens_seen": 6683712,
      "step": 10240
    },
    {
      "epoch": 5.369496855345912,
      "grad_norm": 0.13286568224430084,
      "learning_rate": 4.5801339664678014e-05,
      "loss": 0.5593,
      "num_input_tokens_seen": 6688512,
      "step": 10245
    },
    {
      "epoch": 5.372117400419287,
      "grad_norm": 0.16304588317871094,
      "learning_rate": 4.579499494556943e-05,
      "loss": 0.4727,
      "num_input_tokens_seen": 6691520,
      "step": 10250
    },
    {
      "epoch": 5.3747379454926625,
      "grad_norm": 0.10442323237657547,
      "learning_rate": 4.5788645876385147e-05,
      "loss": 0.5739,
      "num_input_tokens_seen": 6695360,
      "step": 10255
    },
    {
      "epoch": 5.377358490566038,
      "grad_norm": 0.12295511364936829,
      "learning_rate": 4.578229245845331e-05,
      "loss": 0.5043,
      "num_input_tokens_seen": 6698688,
      "step": 10260
    },
    {
      "epoch": 5.379979035639413,
      "grad_norm": 0.2000059336423874,
      "learning_rate": 4.5775934693102994e-05,
      "loss": 0.5757,
      "num_input_tokens_seen": 6701312,
      "step": 10265
    },
    {
      "epoch": 5.382599580712788,
      "grad_norm": 0.20587380230426788,
      "learning_rate": 4.5769572581664165e-05,
      "loss": 0.5166,
      "num_input_tokens_seen": 6703840,
      "step": 10270
    },
    {
      "epoch": 5.385220125786163,
      "grad_norm": 0.14698879420757294,
      "learning_rate": 4.57632061254677e-05,
      "loss": 0.4517,
      "num_input_tokens_seen": 6706720,
      "step": 10275
    },
    {
      "epoch": 5.387840670859539,
      "grad_norm": 0.1732882559299469,
      "learning_rate": 4.5756835325845394e-05,
      "loss": 0.4701,
      "num_input_tokens_seen": 6710240,
      "step": 10280
    },
    {
      "epoch": 5.390461215932914,
      "grad_norm": 0.1888926923274994,
      "learning_rate": 4.575046018412995e-05,
      "loss": 0.6236,
      "num_input_tokens_seen": 6713632,
      "step": 10285
    },
    {
      "epoch": 5.3930817610062896,
      "grad_norm": 0.13502047955989838,
      "learning_rate": 4.5744080701654955e-05,
      "loss": 0.46,
      "num_input_tokens_seen": 6716032,
      "step": 10290
    },
    {
      "epoch": 5.395702306079665,
      "grad_norm": 0.20870733261108398,
      "learning_rate": 4.573769687975495e-05,
      "loss": 0.4626,
      "num_input_tokens_seen": 6719136,
      "step": 10295
    },
    {
      "epoch": 5.39832285115304,
      "grad_norm": 0.28335821628570557,
      "learning_rate": 4.573130871976534e-05,
      "loss": 0.4776,
      "num_input_tokens_seen": 6721824,
      "step": 10300
    },
    {
      "epoch": 5.400943396226415,
      "grad_norm": 0.13742351531982422,
      "learning_rate": 4.572491622302247e-05,
      "loss": 0.4818,
      "num_input_tokens_seen": 6724864,
      "step": 10305
    },
    {
      "epoch": 5.40356394129979,
      "grad_norm": 0.2030615210533142,
      "learning_rate": 4.571851939086356e-05,
      "loss": 0.3953,
      "num_input_tokens_seen": 6728064,
      "step": 10310
    },
    {
      "epoch": 5.406184486373165,
      "grad_norm": 0.170846626162529,
      "learning_rate": 4.571211822462676e-05,
      "loss": 0.5755,
      "num_input_tokens_seen": 6730784,
      "step": 10315
    },
    {
      "epoch": 5.408805031446541,
      "grad_norm": 0.13193441927433014,
      "learning_rate": 4.570571272565114e-05,
      "loss": 0.5797,
      "num_input_tokens_seen": 6733408,
      "step": 10320
    },
    {
      "epoch": 5.411425576519916,
      "grad_norm": 0.025841325521469116,
      "learning_rate": 4.569930289527663e-05,
      "loss": 0.4907,
      "num_input_tokens_seen": 6738912,
      "step": 10325
    },
    {
      "epoch": 5.414046121593292,
      "grad_norm": 0.22975195944309235,
      "learning_rate": 4.56928887348441e-05,
      "loss": 0.485,
      "num_input_tokens_seen": 6742784,
      "step": 10330
    },
    {
      "epoch": 5.416666666666667,
      "grad_norm": 0.09354543685913086,
      "learning_rate": 4.568647024569533e-05,
      "loss": 0.4132,
      "num_input_tokens_seen": 6746080,
      "step": 10335
    },
    {
      "epoch": 5.419287211740042,
      "grad_norm": 0.21472515165805817,
      "learning_rate": 4.568004742917299e-05,
      "loss": 0.6232,
      "num_input_tokens_seen": 6750368,
      "step": 10340
    },
    {
      "epoch": 5.421907756813417,
      "grad_norm": 0.15544922649860382,
      "learning_rate": 4.567362028662065e-05,
      "loss": 0.4855,
      "num_input_tokens_seen": 6753984,
      "step": 10345
    },
    {
      "epoch": 5.4245283018867925,
      "grad_norm": 0.1703980714082718,
      "learning_rate": 4.566718881938281e-05,
      "loss": 0.6038,
      "num_input_tokens_seen": 6756576,
      "step": 10350
    },
    {
      "epoch": 5.427148846960168,
      "grad_norm": 0.2079327404499054,
      "learning_rate": 4.5660753028804846e-05,
      "loss": 0.495,
      "num_input_tokens_seen": 6759904,
      "step": 10355
    },
    {
      "epoch": 5.429769392033543,
      "grad_norm": 0.14476370811462402,
      "learning_rate": 4.5654312916233066e-05,
      "loss": 0.4817,
      "num_input_tokens_seen": 6762880,
      "step": 10360
    },
    {
      "epoch": 5.432389937106918,
      "grad_norm": 0.2756863534450531,
      "learning_rate": 4.564786848301466e-05,
      "loss": 0.6907,
      "num_input_tokens_seen": 6765728,
      "step": 10365
    },
    {
      "epoch": 5.435010482180293,
      "grad_norm": 0.3178292512893677,
      "learning_rate": 4.564141973049773e-05,
      "loss": 0.4693,
      "num_input_tokens_seen": 6768704,
      "step": 10370
    },
    {
      "epoch": 5.437631027253669,
      "grad_norm": 0.125722736120224,
      "learning_rate": 4.563496666003128e-05,
      "loss": 0.3785,
      "num_input_tokens_seen": 6772256,
      "step": 10375
    },
    {
      "epoch": 5.440251572327044,
      "grad_norm": 0.14321984350681305,
      "learning_rate": 4.562850927296523e-05,
      "loss": 0.5338,
      "num_input_tokens_seen": 6775552,
      "step": 10380
    },
    {
      "epoch": 5.4428721174004195,
      "grad_norm": 0.20252399146556854,
      "learning_rate": 4.5622047570650365e-05,
      "loss": 0.4166,
      "num_input_tokens_seen": 6778176,
      "step": 10385
    },
    {
      "epoch": 5.445492662473795,
      "grad_norm": 0.15648309886455536,
      "learning_rate": 4.561558155443843e-05,
      "loss": 0.5012,
      "num_input_tokens_seen": 6781888,
      "step": 10390
    },
    {
      "epoch": 5.44811320754717,
      "grad_norm": 0.13896620273590088,
      "learning_rate": 4.560911122568202e-05,
      "loss": 0.5012,
      "num_input_tokens_seen": 6785056,
      "step": 10395
    },
    {
      "epoch": 5.450733752620545,
      "grad_norm": 0.29515257477760315,
      "learning_rate": 4.560263658573467e-05,
      "loss": 0.5028,
      "num_input_tokens_seen": 6788832,
      "step": 10400
    },
    {
      "epoch": 5.45335429769392,
      "grad_norm": 0.15106314420700073,
      "learning_rate": 4.559615763595079e-05,
      "loss": 0.4812,
      "num_input_tokens_seen": 6792320,
      "step": 10405
    },
    {
      "epoch": 5.455974842767295,
      "grad_norm": 0.18209248781204224,
      "learning_rate": 4.558967437768571e-05,
      "loss": 0.6176,
      "num_input_tokens_seen": 6796000,
      "step": 10410
    },
    {
      "epoch": 5.4585953878406706,
      "grad_norm": 0.29297906160354614,
      "learning_rate": 4.558318681229565e-05,
      "loss": 0.4932,
      "num_input_tokens_seen": 6798400,
      "step": 10415
    },
    {
      "epoch": 5.461215932914046,
      "grad_norm": 0.24602067470550537,
      "learning_rate": 4.557669494113773e-05,
      "loss": 0.5469,
      "num_input_tokens_seen": 6801440,
      "step": 10420
    },
    {
      "epoch": 5.463836477987422,
      "grad_norm": 0.2054489105939865,
      "learning_rate": 4.557019876556998e-05,
      "loss": 0.4727,
      "num_input_tokens_seen": 6807072,
      "step": 10425
    },
    {
      "epoch": 5.466457023060797,
      "grad_norm": 0.21511761844158173,
      "learning_rate": 4.556369828695133e-05,
      "loss": 0.5052,
      "num_input_tokens_seen": 6810272,
      "step": 10430
    },
    {
      "epoch": 5.469077568134172,
      "grad_norm": 0.12652836740016937,
      "learning_rate": 4.5557193506641596e-05,
      "loss": 0.4327,
      "num_input_tokens_seen": 6813184,
      "step": 10435
    },
    {
      "epoch": 5.471698113207547,
      "grad_norm": 0.18874290585517883,
      "learning_rate": 4.555068442600151e-05,
      "loss": 0.4954,
      "num_input_tokens_seen": 6817824,
      "step": 10440
    },
    {
      "epoch": 5.4743186582809225,
      "grad_norm": 0.20717941224575043,
      "learning_rate": 4.5544171046392703e-05,
      "loss": 0.4499,
      "num_input_tokens_seen": 6820480,
      "step": 10445
    },
    {
      "epoch": 5.476939203354298,
      "grad_norm": 0.2263006865978241,
      "learning_rate": 4.5537653369177685e-05,
      "loss": 0.5468,
      "num_input_tokens_seen": 6823360,
      "step": 10450
    },
    {
      "epoch": 5.479559748427673,
      "grad_norm": 0.17402486503124237,
      "learning_rate": 4.5531131395719895e-05,
      "loss": 0.5512,
      "num_input_tokens_seen": 6825856,
      "step": 10455
    },
    {
      "epoch": 5.482180293501048,
      "grad_norm": 0.25295957922935486,
      "learning_rate": 4.552460512738364e-05,
      "loss": 0.5664,
      "num_input_tokens_seen": 6828896,
      "step": 10460
    },
    {
      "epoch": 5.484800838574423,
      "grad_norm": 0.13895350694656372,
      "learning_rate": 4.551807456553416e-05,
      "loss": 0.5533,
      "num_input_tokens_seen": 6832032,
      "step": 10465
    },
    {
      "epoch": 5.487421383647799,
      "grad_norm": 0.33782607316970825,
      "learning_rate": 4.551153971153755e-05,
      "loss": 0.5711,
      "num_input_tokens_seen": 6834624,
      "step": 10470
    },
    {
      "epoch": 5.490041928721174,
      "grad_norm": 0.19538797438144684,
      "learning_rate": 4.550500056676085e-05,
      "loss": 0.6727,
      "num_input_tokens_seen": 6837504,
      "step": 10475
    },
    {
      "epoch": 5.4926624737945495,
      "grad_norm": 0.17106632888317108,
      "learning_rate": 4.5498457132571966e-05,
      "loss": 0.5423,
      "num_input_tokens_seen": 6840224,
      "step": 10480
    },
    {
      "epoch": 5.495283018867925,
      "grad_norm": 0.11522742360830307,
      "learning_rate": 4.5491909410339706e-05,
      "loss": 0.389,
      "num_input_tokens_seen": 6843520,
      "step": 10485
    },
    {
      "epoch": 5.4979035639413,
      "grad_norm": 0.0970972403883934,
      "learning_rate": 4.548535740143378e-05,
      "loss": 0.4905,
      "num_input_tokens_seen": 6846464,
      "step": 10490
    },
    {
      "epoch": 5.500524109014675,
      "grad_norm": 0.22214071452617645,
      "learning_rate": 4.54788011072248e-05,
      "loss": 0.5303,
      "num_input_tokens_seen": 6849696,
      "step": 10495
    },
    {
      "epoch": 5.50314465408805,
      "grad_norm": 0.1804601401090622,
      "learning_rate": 4.5472240529084254e-05,
      "loss": 0.6722,
      "num_input_tokens_seen": 6853216,
      "step": 10500
    },
    {
      "epoch": 5.505765199161425,
      "grad_norm": 0.19461791217327118,
      "learning_rate": 4.5465675668384557e-05,
      "loss": 0.4828,
      "num_input_tokens_seen": 6855872,
      "step": 10505
    },
    {
      "epoch": 5.5083857442348005,
      "grad_norm": 0.1893475204706192,
      "learning_rate": 4.545910652649899e-05,
      "loss": 0.4296,
      "num_input_tokens_seen": 6858688,
      "step": 10510
    },
    {
      "epoch": 5.511006289308176,
      "grad_norm": 0.1749504953622818,
      "learning_rate": 4.5452533104801756e-05,
      "loss": 0.4468,
      "num_input_tokens_seen": 6861632,
      "step": 10515
    },
    {
      "epoch": 5.513626834381551,
      "grad_norm": 0.12564750015735626,
      "learning_rate": 4.544595540466792e-05,
      "loss": 0.6905,
      "num_input_tokens_seen": 6865696,
      "step": 10520
    },
    {
      "epoch": 5.516247379454927,
      "grad_norm": 0.1858426183462143,
      "learning_rate": 4.5439373427473476e-05,
      "loss": 0.5118,
      "num_input_tokens_seen": 6869056,
      "step": 10525
    },
    {
      "epoch": 5.518867924528302,
      "grad_norm": 0.31267988681793213,
      "learning_rate": 4.54327871745953e-05,
      "loss": 0.4829,
      "num_input_tokens_seen": 6871488,
      "step": 10530
    },
    {
      "epoch": 5.521488469601677,
      "grad_norm": 0.09742420166730881,
      "learning_rate": 4.542619664741116e-05,
      "loss": 0.5149,
      "num_input_tokens_seen": 6875008,
      "step": 10535
    },
    {
      "epoch": 5.524109014675052,
      "grad_norm": 0.2660734951496124,
      "learning_rate": 4.5419601847299706e-05,
      "loss": 0.7259,
      "num_input_tokens_seen": 6878336,
      "step": 10540
    },
    {
      "epoch": 5.526729559748428,
      "grad_norm": 0.11345461755990982,
      "learning_rate": 4.5413002775640503e-05,
      "loss": 0.4848,
      "num_input_tokens_seen": 6882080,
      "step": 10545
    },
    {
      "epoch": 5.529350104821803,
      "grad_norm": 0.13139666616916656,
      "learning_rate": 4.540639943381401e-05,
      "loss": 0.4133,
      "num_input_tokens_seen": 6885504,
      "step": 10550
    },
    {
      "epoch": 5.531970649895178,
      "grad_norm": 0.13657774031162262,
      "learning_rate": 4.539979182320157e-05,
      "loss": 0.5397,
      "num_input_tokens_seen": 6888704,
      "step": 10555
    },
    {
      "epoch": 5.534591194968553,
      "grad_norm": 0.20290176570415497,
      "learning_rate": 4.5393179945185404e-05,
      "loss": 0.578,
      "num_input_tokens_seen": 6891936,
      "step": 10560
    },
    {
      "epoch": 5.537211740041929,
      "grad_norm": 0.20064841210842133,
      "learning_rate": 4.538656380114865e-05,
      "loss": 0.5909,
      "num_input_tokens_seen": 6894848,
      "step": 10565
    },
    {
      "epoch": 5.539832285115304,
      "grad_norm": 0.21751298010349274,
      "learning_rate": 4.537994339247533e-05,
      "loss": 0.5106,
      "num_input_tokens_seen": 6897728,
      "step": 10570
    },
    {
      "epoch": 5.5424528301886795,
      "grad_norm": 0.2765111029148102,
      "learning_rate": 4.5373318720550366e-05,
      "loss": 0.4469,
      "num_input_tokens_seen": 6900672,
      "step": 10575
    },
    {
      "epoch": 5.545073375262055,
      "grad_norm": 0.16175203025341034,
      "learning_rate": 4.536668978675956e-05,
      "loss": 0.4414,
      "num_input_tokens_seen": 6905920,
      "step": 10580
    },
    {
      "epoch": 5.54769392033543,
      "grad_norm": 0.09040148556232452,
      "learning_rate": 4.53600565924896e-05,
      "loss": 0.4951,
      "num_input_tokens_seen": 6909408,
      "step": 10585
    },
    {
      "epoch": 5.550314465408805,
      "grad_norm": 0.16893377900123596,
      "learning_rate": 4.535341913912808e-05,
      "loss": 0.4906,
      "num_input_tokens_seen": 6912128,
      "step": 10590
    },
    {
      "epoch": 5.55293501048218,
      "grad_norm": 0.17014677822589874,
      "learning_rate": 4.534677742806348e-05,
      "loss": 0.5439,
      "num_input_tokens_seen": 6915136,
      "step": 10595
    },
    {
      "epoch": 5.555555555555555,
      "grad_norm": 0.15317772328853607,
      "learning_rate": 4.5340131460685175e-05,
      "loss": 0.5046,
      "num_input_tokens_seen": 6917856,
      "step": 10600
    },
    {
      "epoch": 5.5581761006289305,
      "grad_norm": 0.1945960372686386,
      "learning_rate": 4.533348123838341e-05,
      "loss": 0.4756,
      "num_input_tokens_seen": 6920864,
      "step": 10605
    },
    {
      "epoch": 5.560796645702306,
      "grad_norm": 0.1432117223739624,
      "learning_rate": 4.532682676254936e-05,
      "loss": 0.7035,
      "num_input_tokens_seen": 6924384,
      "step": 10610
    },
    {
      "epoch": 5.563417190775681,
      "grad_norm": 0.21314620971679688,
      "learning_rate": 4.5320168034575044e-05,
      "loss": 0.5293,
      "num_input_tokens_seen": 6926848,
      "step": 10615
    },
    {
      "epoch": 5.566037735849057,
      "grad_norm": 0.11110279709100723,
      "learning_rate": 4.53135050558534e-05,
      "loss": 0.4193,
      "num_input_tokens_seen": 6929536,
      "step": 10620
    },
    {
      "epoch": 5.568658280922432,
      "grad_norm": 0.244214728474617,
      "learning_rate": 4.530683782777825e-05,
      "loss": 0.557,
      "num_input_tokens_seen": 6932064,
      "step": 10625
    },
    {
      "epoch": 5.571278825995807,
      "grad_norm": 0.18015117943286896,
      "learning_rate": 4.53001663517443e-05,
      "loss": 0.4738,
      "num_input_tokens_seen": 6935296,
      "step": 10630
    },
    {
      "epoch": 5.573899371069182,
      "grad_norm": 0.21142606437206268,
      "learning_rate": 4.529349062914714e-05,
      "loss": 0.4625,
      "num_input_tokens_seen": 6937824,
      "step": 10635
    },
    {
      "epoch": 5.576519916142558,
      "grad_norm": 0.15589676797389984,
      "learning_rate": 4.528681066138326e-05,
      "loss": 0.5091,
      "num_input_tokens_seen": 6941600,
      "step": 10640
    },
    {
      "epoch": 5.579140461215933,
      "grad_norm": 0.11755499988794327,
      "learning_rate": 4.528012644985005e-05,
      "loss": 0.4798,
      "num_input_tokens_seen": 6944608,
      "step": 10645
    },
    {
      "epoch": 5.581761006289308,
      "grad_norm": 0.16354992985725403,
      "learning_rate": 4.527343799594574e-05,
      "loss": 0.5798,
      "num_input_tokens_seen": 6948000,
      "step": 10650
    },
    {
      "epoch": 5.584381551362683,
      "grad_norm": 0.13162054121494293,
      "learning_rate": 4.5266745301069494e-05,
      "loss": 0.3953,
      "num_input_tokens_seen": 6951424,
      "step": 10655
    },
    {
      "epoch": 5.587002096436059,
      "grad_norm": 0.15029701590538025,
      "learning_rate": 4.526004836662135e-05,
      "loss": 0.4507,
      "num_input_tokens_seen": 6956000,
      "step": 10660
    },
    {
      "epoch": 5.589622641509434,
      "grad_norm": 0.17857974767684937,
      "learning_rate": 4.525334719400223e-05,
      "loss": 0.5158,
      "num_input_tokens_seen": 6958944,
      "step": 10665
    },
    {
      "epoch": 5.5922431865828095,
      "grad_norm": 0.11156061291694641,
      "learning_rate": 4.524664178461393e-05,
      "loss": 0.4454,
      "num_input_tokens_seen": 6962304,
      "step": 10670
    },
    {
      "epoch": 5.594863731656185,
      "grad_norm": 0.20301665365695953,
      "learning_rate": 4.5239932139859154e-05,
      "loss": 0.5999,
      "num_input_tokens_seen": 6965632,
      "step": 10675
    },
    {
      "epoch": 5.59748427672956,
      "grad_norm": 0.10657626390457153,
      "learning_rate": 4.523321826114149e-05,
      "loss": 0.5827,
      "num_input_tokens_seen": 6969312,
      "step": 10680
    },
    {
      "epoch": 5.600104821802935,
      "grad_norm": 0.10464727878570557,
      "learning_rate": 4.5226500149865395e-05,
      "loss": 0.4369,
      "num_input_tokens_seen": 6972576,
      "step": 10685
    },
    {
      "epoch": 5.60272536687631,
      "grad_norm": 0.16197504103183746,
      "learning_rate": 4.5219777807436224e-05,
      "loss": 0.3759,
      "num_input_tokens_seen": 6975872,
      "step": 10690
    },
    {
      "epoch": 5.605345911949685,
      "grad_norm": 0.11687948554754257,
      "learning_rate": 4.5213051235260225e-05,
      "loss": 0.3758,
      "num_input_tokens_seen": 6978784,
      "step": 10695
    },
    {
      "epoch": 5.6079664570230605,
      "grad_norm": 0.1856047511100769,
      "learning_rate": 4.52063204347445e-05,
      "loss": 0.7108,
      "num_input_tokens_seen": 6981408,
      "step": 10700
    },
    {
      "epoch": 5.610587002096436,
      "grad_norm": 0.17328430712223053,
      "learning_rate": 4.519958540729707e-05,
      "loss": 0.4468,
      "num_input_tokens_seen": 6984800,
      "step": 10705
    },
    {
      "epoch": 5.613207547169811,
      "grad_norm": 0.1521306186914444,
      "learning_rate": 4.519284615432682e-05,
      "loss": 0.4836,
      "num_input_tokens_seen": 6988192,
      "step": 10710
    },
    {
      "epoch": 5.615828092243187,
      "grad_norm": 0.1446043699979782,
      "learning_rate": 4.518610267724352e-05,
      "loss": 0.4034,
      "num_input_tokens_seen": 6992416,
      "step": 10715
    },
    {
      "epoch": 5.618448637316562,
      "grad_norm": 0.23124082386493683,
      "learning_rate": 4.517935497745786e-05,
      "loss": 0.4563,
      "num_input_tokens_seen": 6995520,
      "step": 10720
    },
    {
      "epoch": 5.621069182389937,
      "grad_norm": 0.16374680399894714,
      "learning_rate": 4.517260305638133e-05,
      "loss": 0.467,
      "num_input_tokens_seen": 6998464,
      "step": 10725
    },
    {
      "epoch": 5.623689727463312,
      "grad_norm": 0.1622163951396942,
      "learning_rate": 4.516584691542639e-05,
      "loss": 0.4815,
      "num_input_tokens_seen": 7001600,
      "step": 10730
    },
    {
      "epoch": 5.626310272536688,
      "grad_norm": 0.19405554234981537,
      "learning_rate": 4.515908655600634e-05,
      "loss": 0.5162,
      "num_input_tokens_seen": 7004704,
      "step": 10735
    },
    {
      "epoch": 5.628930817610063,
      "grad_norm": 0.16715897619724274,
      "learning_rate": 4.515232197953537e-05,
      "loss": 0.6041,
      "num_input_tokens_seen": 7007424,
      "step": 10740
    },
    {
      "epoch": 5.631551362683438,
      "grad_norm": 0.12590739130973816,
      "learning_rate": 4.514555318742856e-05,
      "loss": 0.4352,
      "num_input_tokens_seen": 7010688,
      "step": 10745
    },
    {
      "epoch": 5.634171907756813,
      "grad_norm": 0.1122431606054306,
      "learning_rate": 4.513878018110184e-05,
      "loss": 0.3926,
      "num_input_tokens_seen": 7014048,
      "step": 10750
    },
    {
      "epoch": 5.636792452830189,
      "grad_norm": 0.2197452336549759,
      "learning_rate": 4.513200296197207e-05,
      "loss": 0.4437,
      "num_input_tokens_seen": 7017856,
      "step": 10755
    },
    {
      "epoch": 5.639412997903564,
      "grad_norm": 0.14241643249988556,
      "learning_rate": 4.512522153145696e-05,
      "loss": 0.3998,
      "num_input_tokens_seen": 7021248,
      "step": 10760
    },
    {
      "epoch": 5.6420335429769395,
      "grad_norm": 0.23826228082180023,
      "learning_rate": 4.5118435890975106e-05,
      "loss": 0.4798,
      "num_input_tokens_seen": 7024320,
      "step": 10765
    },
    {
      "epoch": 5.644654088050315,
      "grad_norm": 0.34536901116371155,
      "learning_rate": 4.511164604194598e-05,
      "loss": 0.5602,
      "num_input_tokens_seen": 7027296,
      "step": 10770
    },
    {
      "epoch": 5.64727463312369,
      "grad_norm": 0.10050990432500839,
      "learning_rate": 4.5104851985789956e-05,
      "loss": 0.5072,
      "num_input_tokens_seen": 7030784,
      "step": 10775
    },
    {
      "epoch": 5.649895178197065,
      "grad_norm": 0.10729911923408508,
      "learning_rate": 4.5098053723928254e-05,
      "loss": 0.7271,
      "num_input_tokens_seen": 7034112,
      "step": 10780
    },
    {
      "epoch": 5.65251572327044,
      "grad_norm": 0.15246786177158356,
      "learning_rate": 4.509125125778302e-05,
      "loss": 0.4092,
      "num_input_tokens_seen": 7039072,
      "step": 10785
    },
    {
      "epoch": 5.655136268343815,
      "grad_norm": 0.26122933626174927,
      "learning_rate": 4.508444458877722e-05,
      "loss": 0.4656,
      "num_input_tokens_seen": 7042304,
      "step": 10790
    },
    {
      "epoch": 5.6577568134171905,
      "grad_norm": 0.16266542673110962,
      "learning_rate": 4.5077633718334755e-05,
      "loss": 0.554,
      "num_input_tokens_seen": 7046400,
      "step": 10795
    },
    {
      "epoch": 5.660377358490566,
      "grad_norm": 0.12675204873085022,
      "learning_rate": 4.5070818647880375e-05,
      "loss": 0.4798,
      "num_input_tokens_seen": 7050656,
      "step": 10800
    },
    {
      "epoch": 5.662997903563941,
      "grad_norm": 0.20758572220802307,
      "learning_rate": 4.506399937883972e-05,
      "loss": 0.4959,
      "num_input_tokens_seen": 7053824,
      "step": 10805
    },
    {
      "epoch": 5.665618448637317,
      "grad_norm": 0.18074487149715424,
      "learning_rate": 4.5057175912639286e-05,
      "loss": 0.5644,
      "num_input_tokens_seen": 7057760,
      "step": 10810
    },
    {
      "epoch": 5.668238993710692,
      "grad_norm": 0.1697678118944168,
      "learning_rate": 4.5050348250706475e-05,
      "loss": 0.4298,
      "num_input_tokens_seen": 7061888,
      "step": 10815
    },
    {
      "epoch": 5.670859538784067,
      "grad_norm": 0.16593433916568756,
      "learning_rate": 4.5043516394469566e-05,
      "loss": 0.3872,
      "num_input_tokens_seen": 7064960,
      "step": 10820
    },
    {
      "epoch": 5.673480083857442,
      "grad_norm": 0.16209718585014343,
      "learning_rate": 4.503668034535768e-05,
      "loss": 0.4702,
      "num_input_tokens_seen": 7067968,
      "step": 10825
    },
    {
      "epoch": 5.676100628930818,
      "grad_norm": 0.10973289608955383,
      "learning_rate": 4.5029840104800876e-05,
      "loss": 0.5478,
      "num_input_tokens_seen": 7071424,
      "step": 10830
    },
    {
      "epoch": 5.678721174004193,
      "grad_norm": 0.2736682593822479,
      "learning_rate": 4.5022995674230025e-05,
      "loss": 0.5887,
      "num_input_tokens_seen": 7074656,
      "step": 10835
    },
    {
      "epoch": 5.681341719077568,
      "grad_norm": 0.1913633793592453,
      "learning_rate": 4.5016147055076914e-05,
      "loss": 0.5225,
      "num_input_tokens_seen": 7077952,
      "step": 10840
    },
    {
      "epoch": 5.683962264150943,
      "grad_norm": 0.19209012389183044,
      "learning_rate": 4.500929424877419e-05,
      "loss": 0.3808,
      "num_input_tokens_seen": 7080512,
      "step": 10845
    },
    {
      "epoch": 5.686582809224319,
      "grad_norm": 0.2836637496948242,
      "learning_rate": 4.5002437256755394e-05,
      "loss": 0.4482,
      "num_input_tokens_seen": 7083904,
      "step": 10850
    },
    {
      "epoch": 5.689203354297694,
      "grad_norm": 0.24832330644130707,
      "learning_rate": 4.4995576080454924e-05,
      "loss": 0.5178,
      "num_input_tokens_seen": 7086144,
      "step": 10855
    },
    {
      "epoch": 5.6918238993710695,
      "grad_norm": 0.14781302213668823,
      "learning_rate": 4.4988710721308056e-05,
      "loss": 0.4905,
      "num_input_tokens_seen": 7088608,
      "step": 10860
    },
    {
      "epoch": 5.694444444444445,
      "grad_norm": 0.1185210794210434,
      "learning_rate": 4.498184118075095e-05,
      "loss": 0.4318,
      "num_input_tokens_seen": 7091296,
      "step": 10865
    },
    {
      "epoch": 5.69706498951782,
      "grad_norm": 0.19680644571781158,
      "learning_rate": 4.497496746022063e-05,
      "loss": 0.4664,
      "num_input_tokens_seen": 7095008,
      "step": 10870
    },
    {
      "epoch": 5.699685534591195,
      "grad_norm": 0.24472616612911224,
      "learning_rate": 4.4968089561155015e-05,
      "loss": 0.475,
      "num_input_tokens_seen": 7098016,
      "step": 10875
    },
    {
      "epoch": 5.70230607966457,
      "grad_norm": 0.09991682320833206,
      "learning_rate": 4.496120748499286e-05,
      "loss": 0.4316,
      "num_input_tokens_seen": 7101056,
      "step": 10880
    },
    {
      "epoch": 5.704926624737945,
      "grad_norm": 0.19594179093837738,
      "learning_rate": 4.4954321233173833e-05,
      "loss": 0.391,
      "num_input_tokens_seen": 7103488,
      "step": 10885
    },
    {
      "epoch": 5.7075471698113205,
      "grad_norm": 0.15497703850269318,
      "learning_rate": 4.494743080713846e-05,
      "loss": 0.4515,
      "num_input_tokens_seen": 7107072,
      "step": 10890
    },
    {
      "epoch": 5.710167714884696,
      "grad_norm": 0.1561030000448227,
      "learning_rate": 4.494053620832812e-05,
      "loss": 0.4717,
      "num_input_tokens_seen": 7110336,
      "step": 10895
    },
    {
      "epoch": 5.712788259958071,
      "grad_norm": 0.19096073508262634,
      "learning_rate": 4.493363743818511e-05,
      "loss": 0.5899,
      "num_input_tokens_seen": 7114240,
      "step": 10900
    },
    {
      "epoch": 5.715408805031447,
      "grad_norm": 0.11271347105503082,
      "learning_rate": 4.492673449815257e-05,
      "loss": 0.5618,
      "num_input_tokens_seen": 7121216,
      "step": 10905
    },
    {
      "epoch": 5.718029350104822,
      "grad_norm": 0.2666301429271698,
      "learning_rate": 4.4919827389674496e-05,
      "loss": 0.467,
      "num_input_tokens_seen": 7124800,
      "step": 10910
    },
    {
      "epoch": 5.720649895178197,
      "grad_norm": 0.1853976845741272,
      "learning_rate": 4.491291611419579e-05,
      "loss": 0.5492,
      "num_input_tokens_seen": 7128224,
      "step": 10915
    },
    {
      "epoch": 5.723270440251572,
      "grad_norm": 0.2543230652809143,
      "learning_rate": 4.4906000673162216e-05,
      "loss": 0.4366,
      "num_input_tokens_seen": 7131456,
      "step": 10920
    },
    {
      "epoch": 5.725890985324948,
      "grad_norm": 0.2047121226787567,
      "learning_rate": 4.48990810680204e-05,
      "loss": 0.4652,
      "num_input_tokens_seen": 7134944,
      "step": 10925
    },
    {
      "epoch": 5.728511530398323,
      "grad_norm": 0.08845430612564087,
      "learning_rate": 4.4892157300217844e-05,
      "loss": 0.6174,
      "num_input_tokens_seen": 7139200,
      "step": 10930
    },
    {
      "epoch": 5.731132075471698,
      "grad_norm": 0.20477983355522156,
      "learning_rate": 4.488522937120293e-05,
      "loss": 0.4687,
      "num_input_tokens_seen": 7141792,
      "step": 10935
    },
    {
      "epoch": 5.733752620545073,
      "grad_norm": 0.08815700560808182,
      "learning_rate": 4.4878297282424886e-05,
      "loss": 0.4419,
      "num_input_tokens_seen": 7145152,
      "step": 10940
    },
    {
      "epoch": 5.736373165618449,
      "grad_norm": 0.16558800637722015,
      "learning_rate": 4.4871361035333836e-05,
      "loss": 0.6194,
      "num_input_tokens_seen": 7148192,
      "step": 10945
    },
    {
      "epoch": 5.738993710691824,
      "grad_norm": 0.12560398876667023,
      "learning_rate": 4.486442063138076e-05,
      "loss": 0.4075,
      "num_input_tokens_seen": 7151520,
      "step": 10950
    },
    {
      "epoch": 5.7416142557651995,
      "grad_norm": 0.22841639816761017,
      "learning_rate": 4.4857476072017516e-05,
      "loss": 0.418,
      "num_input_tokens_seen": 7154496,
      "step": 10955
    },
    {
      "epoch": 5.744234800838575,
      "grad_norm": 0.12149077653884888,
      "learning_rate": 4.485052735869682e-05,
      "loss": 0.5514,
      "num_input_tokens_seen": 7157600,
      "step": 10960
    },
    {
      "epoch": 5.74685534591195,
      "grad_norm": 0.22342383861541748,
      "learning_rate": 4.484357449287228e-05,
      "loss": 0.5817,
      "num_input_tokens_seen": 7160224,
      "step": 10965
    },
    {
      "epoch": 5.749475890985325,
      "grad_norm": 0.16693511605262756,
      "learning_rate": 4.483661747599833e-05,
      "loss": 0.4003,
      "num_input_tokens_seen": 7162720,
      "step": 10970
    },
    {
      "epoch": 5.7520964360587,
      "grad_norm": 0.2855328321456909,
      "learning_rate": 4.482965630953031e-05,
      "loss": 0.611,
      "num_input_tokens_seen": 7166240,
      "step": 10975
    },
    {
      "epoch": 5.754716981132075,
      "grad_norm": 0.23482602834701538,
      "learning_rate": 4.4822690994924425e-05,
      "loss": 0.5904,
      "num_input_tokens_seen": 7172480,
      "step": 10980
    },
    {
      "epoch": 5.7573375262054505,
      "grad_norm": 0.12744265794754028,
      "learning_rate": 4.481572153363773e-05,
      "loss": 0.5418,
      "num_input_tokens_seen": 7176416,
      "step": 10985
    },
    {
      "epoch": 5.759958071278826,
      "grad_norm": 0.16462461650371552,
      "learning_rate": 4.480874792712817e-05,
      "loss": 0.6166,
      "num_input_tokens_seen": 7180832,
      "step": 10990
    },
    {
      "epoch": 5.762578616352201,
      "grad_norm": 0.2498673051595688,
      "learning_rate": 4.480177017685452e-05,
      "loss": 0.5274,
      "num_input_tokens_seen": 7184416,
      "step": 10995
    },
    {
      "epoch": 5.765199161425577,
      "grad_norm": 0.1714944690465927,
      "learning_rate": 4.479478828427647e-05,
      "loss": 0.4548,
      "num_input_tokens_seen": 7186816,
      "step": 11000
    },
    {
      "epoch": 5.767819706498952,
      "grad_norm": 0.17075738310813904,
      "learning_rate": 4.478780225085454e-05,
      "loss": 0.6085,
      "num_input_tokens_seen": 7189824,
      "step": 11005
    },
    {
      "epoch": 5.770440251572327,
      "grad_norm": 0.22600896656513214,
      "learning_rate": 4.4780812078050125e-05,
      "loss": 0.5551,
      "num_input_tokens_seen": 7192640,
      "step": 11010
    },
    {
      "epoch": 5.773060796645702,
      "grad_norm": 0.20500461757183075,
      "learning_rate": 4.47738177673255e-05,
      "loss": 0.4741,
      "num_input_tokens_seen": 7195328,
      "step": 11015
    },
    {
      "epoch": 5.7756813417190775,
      "grad_norm": 0.33532360196113586,
      "learning_rate": 4.4766819320143785e-05,
      "loss": 0.718,
      "num_input_tokens_seen": 7198464,
      "step": 11020
    },
    {
      "epoch": 5.778301886792453,
      "grad_norm": 0.1317848563194275,
      "learning_rate": 4.475981673796899e-05,
      "loss": 0.4876,
      "num_input_tokens_seen": 7202080,
      "step": 11025
    },
    {
      "epoch": 5.780922431865828,
      "grad_norm": 0.161818727850914,
      "learning_rate": 4.475281002226596e-05,
      "loss": 0.5056,
      "num_input_tokens_seen": 7204832,
      "step": 11030
    },
    {
      "epoch": 5.783542976939203,
      "grad_norm": 0.21320034563541412,
      "learning_rate": 4.4745799174500414e-05,
      "loss": 0.4214,
      "num_input_tokens_seen": 7208128,
      "step": 11035
    },
    {
      "epoch": 5.786163522012579,
      "grad_norm": 0.13053114712238312,
      "learning_rate": 4.473878419613896e-05,
      "loss": 0.4735,
      "num_input_tokens_seen": 7211232,
      "step": 11040
    },
    {
      "epoch": 5.788784067085954,
      "grad_norm": 0.13094641268253326,
      "learning_rate": 4.473176508864905e-05,
      "loss": 0.4241,
      "num_input_tokens_seen": 7214720,
      "step": 11045
    },
    {
      "epoch": 5.7914046121593294,
      "grad_norm": 0.18064764142036438,
      "learning_rate": 4.4724741853498995e-05,
      "loss": 0.6038,
      "num_input_tokens_seen": 7219200,
      "step": 11050
    },
    {
      "epoch": 5.794025157232705,
      "grad_norm": 0.14690297842025757,
      "learning_rate": 4.471771449215797e-05,
      "loss": 0.5059,
      "num_input_tokens_seen": 7223808,
      "step": 11055
    },
    {
      "epoch": 5.79664570230608,
      "grad_norm": 0.24989411234855652,
      "learning_rate": 4.471068300609603e-05,
      "loss": 0.5335,
      "num_input_tokens_seen": 7227264,
      "step": 11060
    },
    {
      "epoch": 5.799266247379455,
      "grad_norm": 0.18046043813228607,
      "learning_rate": 4.4703647396784065e-05,
      "loss": 0.6002,
      "num_input_tokens_seen": 7232672,
      "step": 11065
    },
    {
      "epoch": 5.80188679245283,
      "grad_norm": 0.27495867013931274,
      "learning_rate": 4.469660766569387e-05,
      "loss": 0.473,
      "num_input_tokens_seen": 7235584,
      "step": 11070
    },
    {
      "epoch": 5.804507337526205,
      "grad_norm": 0.20242372155189514,
      "learning_rate": 4.4689563814298057e-05,
      "loss": 0.5554,
      "num_input_tokens_seen": 7238592,
      "step": 11075
    },
    {
      "epoch": 5.8071278825995805,
      "grad_norm": 0.16204021871089935,
      "learning_rate": 4.468251584407012e-05,
      "loss": 0.5331,
      "num_input_tokens_seen": 7241952,
      "step": 11080
    },
    {
      "epoch": 5.809748427672956,
      "grad_norm": 0.18473704159259796,
      "learning_rate": 4.467546375648443e-05,
      "loss": 0.4688,
      "num_input_tokens_seen": 7245056,
      "step": 11085
    },
    {
      "epoch": 5.812368972746331,
      "grad_norm": 0.17784830927848816,
      "learning_rate": 4.4668407553016176e-05,
      "loss": 0.4912,
      "num_input_tokens_seen": 7248896,
      "step": 11090
    },
    {
      "epoch": 5.814989517819707,
      "grad_norm": 0.14149576425552368,
      "learning_rate": 4.466134723514146e-05,
      "loss": 0.5653,
      "num_input_tokens_seen": 7251680,
      "step": 11095
    },
    {
      "epoch": 5.817610062893082,
      "grad_norm": 0.1494101583957672,
      "learning_rate": 4.465428280433721e-05,
      "loss": 0.6147,
      "num_input_tokens_seen": 7254432,
      "step": 11100
    },
    {
      "epoch": 5.820230607966457,
      "grad_norm": 0.49355220794677734,
      "learning_rate": 4.464721426208123e-05,
      "loss": 0.5502,
      "num_input_tokens_seen": 7257280,
      "step": 11105
    },
    {
      "epoch": 5.822851153039832,
      "grad_norm": 0.10591944307088852,
      "learning_rate": 4.464014160985217e-05,
      "loss": 0.4608,
      "num_input_tokens_seen": 7259840,
      "step": 11110
    },
    {
      "epoch": 5.8254716981132075,
      "grad_norm": 0.1638863980770111,
      "learning_rate": 4.463306484912955e-05,
      "loss": 0.4742,
      "num_input_tokens_seen": 7262624,
      "step": 11115
    },
    {
      "epoch": 5.828092243186583,
      "grad_norm": 0.1220538467168808,
      "learning_rate": 4.462598398139375e-05,
      "loss": 0.4545,
      "num_input_tokens_seen": 7265568,
      "step": 11120
    },
    {
      "epoch": 5.830712788259958,
      "grad_norm": 0.21335305273532867,
      "learning_rate": 4.4618899008126005e-05,
      "loss": 0.4169,
      "num_input_tokens_seen": 7268896,
      "step": 11125
    },
    {
      "epoch": 5.833333333333333,
      "grad_norm": 0.22140084207057953,
      "learning_rate": 4.461180993080841e-05,
      "loss": 0.3895,
      "num_input_tokens_seen": 7271328,
      "step": 11130
    },
    {
      "epoch": 5.835953878406709,
      "grad_norm": 0.23175175487995148,
      "learning_rate": 4.460471675092393e-05,
      "loss": 0.5762,
      "num_input_tokens_seen": 7274752,
      "step": 11135
    },
    {
      "epoch": 5.838574423480084,
      "grad_norm": 0.13222481310367584,
      "learning_rate": 4.4597619469956356e-05,
      "loss": 0.5862,
      "num_input_tokens_seen": 7279168,
      "step": 11140
    },
    {
      "epoch": 5.841194968553459,
      "grad_norm": 0.1699850857257843,
      "learning_rate": 4.459051808939038e-05,
      "loss": 0.5176,
      "num_input_tokens_seen": 7282944,
      "step": 11145
    },
    {
      "epoch": 5.843815513626835,
      "grad_norm": 0.1649112105369568,
      "learning_rate": 4.458341261071151e-05,
      "loss": 0.5521,
      "num_input_tokens_seen": 7285824,
      "step": 11150
    },
    {
      "epoch": 5.84643605870021,
      "grad_norm": 0.1640183925628662,
      "learning_rate": 4.457630303540615e-05,
      "loss": 0.5425,
      "num_input_tokens_seen": 7288384,
      "step": 11155
    },
    {
      "epoch": 5.849056603773585,
      "grad_norm": 0.18128643929958344,
      "learning_rate": 4.456918936496154e-05,
      "loss": 0.5814,
      "num_input_tokens_seen": 7291840,
      "step": 11160
    },
    {
      "epoch": 5.85167714884696,
      "grad_norm": 0.15538841485977173,
      "learning_rate": 4.456207160086575e-05,
      "loss": 0.4366,
      "num_input_tokens_seen": 7296032,
      "step": 11165
    },
    {
      "epoch": 5.854297693920335,
      "grad_norm": 0.1810811161994934,
      "learning_rate": 4.4554949744607775e-05,
      "loss": 0.6467,
      "num_input_tokens_seen": 7298560,
      "step": 11170
    },
    {
      "epoch": 5.8569182389937104,
      "grad_norm": 0.12559391558170319,
      "learning_rate": 4.4547823797677405e-05,
      "loss": 0.4191,
      "num_input_tokens_seen": 7301376,
      "step": 11175
    },
    {
      "epoch": 5.859538784067086,
      "grad_norm": 0.2256643921136856,
      "learning_rate": 4.45406937615653e-05,
      "loss": 0.5362,
      "num_input_tokens_seen": 7304896,
      "step": 11180
    },
    {
      "epoch": 5.862159329140461,
      "grad_norm": 0.12097077816724777,
      "learning_rate": 4.4533559637763006e-05,
      "loss": 0.5166,
      "num_input_tokens_seen": 7307968,
      "step": 11185
    },
    {
      "epoch": 5.864779874213837,
      "grad_norm": 0.19161462783813477,
      "learning_rate": 4.452642142776288e-05,
      "loss": 0.4969,
      "num_input_tokens_seen": 7311040,
      "step": 11190
    },
    {
      "epoch": 5.867400419287212,
      "grad_norm": 0.20760951936244965,
      "learning_rate": 4.451927913305816e-05,
      "loss": 0.4297,
      "num_input_tokens_seen": 7314848,
      "step": 11195
    },
    {
      "epoch": 5.870020964360587,
      "grad_norm": 0.13531188666820526,
      "learning_rate": 4.4512132755142935e-05,
      "loss": 0.5514,
      "num_input_tokens_seen": 7317440,
      "step": 11200
    },
    {
      "epoch": 5.872641509433962,
      "grad_norm": 0.15113075077533722,
      "learning_rate": 4.450498229551214e-05,
      "loss": 0.4837,
      "num_input_tokens_seen": 7320288,
      "step": 11205
    },
    {
      "epoch": 5.8752620545073375,
      "grad_norm": 0.2093614935874939,
      "learning_rate": 4.449782775566158e-05,
      "loss": 0.5396,
      "num_input_tokens_seen": 7323168,
      "step": 11210
    },
    {
      "epoch": 5.877882599580713,
      "grad_norm": 0.1705724149942398,
      "learning_rate": 4.449066913708789e-05,
      "loss": 0.4971,
      "num_input_tokens_seen": 7326048,
      "step": 11215
    },
    {
      "epoch": 5.880503144654088,
      "grad_norm": 0.19906091690063477,
      "learning_rate": 4.448350644128858e-05,
      "loss": 0.5484,
      "num_input_tokens_seen": 7328288,
      "step": 11220
    },
    {
      "epoch": 5.883123689727463,
      "grad_norm": 0.34323984384536743,
      "learning_rate": 4.4476339669762e-05,
      "loss": 0.578,
      "num_input_tokens_seen": 7331328,
      "step": 11225
    },
    {
      "epoch": 5.885744234800838,
      "grad_norm": 0.18709255754947662,
      "learning_rate": 4.446916882400737e-05,
      "loss": 0.475,
      "num_input_tokens_seen": 7334720,
      "step": 11230
    },
    {
      "epoch": 5.888364779874214,
      "grad_norm": 0.12189975380897522,
      "learning_rate": 4.446199390552472e-05,
      "loss": 0.4266,
      "num_input_tokens_seen": 7337312,
      "step": 11235
    },
    {
      "epoch": 5.890985324947589,
      "grad_norm": 0.1793793886899948,
      "learning_rate": 4.445481491581499e-05,
      "loss": 0.5433,
      "num_input_tokens_seen": 7340384,
      "step": 11240
    },
    {
      "epoch": 5.893605870020965,
      "grad_norm": 0.09814762324094772,
      "learning_rate": 4.444763185637993e-05,
      "loss": 0.5138,
      "num_input_tokens_seen": 7343936,
      "step": 11245
    },
    {
      "epoch": 5.89622641509434,
      "grad_norm": 0.1792292594909668,
      "learning_rate": 4.4440444728722166e-05,
      "loss": 0.5139,
      "num_input_tokens_seen": 7346848,
      "step": 11250
    },
    {
      "epoch": 5.898846960167715,
      "grad_norm": 0.18475528061389923,
      "learning_rate": 4.4433253534345136e-05,
      "loss": 0.5499,
      "num_input_tokens_seen": 7350112,
      "step": 11255
    },
    {
      "epoch": 5.90146750524109,
      "grad_norm": 0.1956985890865326,
      "learning_rate": 4.442605827475318e-05,
      "loss": 0.4532,
      "num_input_tokens_seen": 7353952,
      "step": 11260
    },
    {
      "epoch": 5.904088050314465,
      "grad_norm": 0.23213928937911987,
      "learning_rate": 4.4418858951451457e-05,
      "loss": 0.6132,
      "num_input_tokens_seen": 7357440,
      "step": 11265
    },
    {
      "epoch": 5.90670859538784,
      "grad_norm": 0.15166282653808594,
      "learning_rate": 4.441165556594599e-05,
      "loss": 0.6335,
      "num_input_tokens_seen": 7360800,
      "step": 11270
    },
    {
      "epoch": 5.909329140461216,
      "grad_norm": 0.22436021268367767,
      "learning_rate": 4.440444811974363e-05,
      "loss": 0.5723,
      "num_input_tokens_seen": 7364032,
      "step": 11275
    },
    {
      "epoch": 5.911949685534591,
      "grad_norm": 0.2137850672006607,
      "learning_rate": 4.43972366143521e-05,
      "loss": 0.5877,
      "num_input_tokens_seen": 7366880,
      "step": 11280
    },
    {
      "epoch": 5.914570230607967,
      "grad_norm": 0.12882071733474731,
      "learning_rate": 4.4390021051279975e-05,
      "loss": 0.4399,
      "num_input_tokens_seen": 7369664,
      "step": 11285
    },
    {
      "epoch": 5.917190775681342,
      "grad_norm": 0.18828986585140228,
      "learning_rate": 4.438280143203665e-05,
      "loss": 0.6143,
      "num_input_tokens_seen": 7372192,
      "step": 11290
    },
    {
      "epoch": 5.919811320754717,
      "grad_norm": 0.29083800315856934,
      "learning_rate": 4.43755777581324e-05,
      "loss": 0.5512,
      "num_input_tokens_seen": 7374720,
      "step": 11295
    },
    {
      "epoch": 5.922431865828092,
      "grad_norm": 0.1980191171169281,
      "learning_rate": 4.4368350031078334e-05,
      "loss": 0.5066,
      "num_input_tokens_seen": 7378112,
      "step": 11300
    },
    {
      "epoch": 5.9250524109014675,
      "grad_norm": 0.15897616744041443,
      "learning_rate": 4.43611182523864e-05,
      "loss": 0.4778,
      "num_input_tokens_seen": 7381728,
      "step": 11305
    },
    {
      "epoch": 5.927672955974843,
      "grad_norm": 0.1900622844696045,
      "learning_rate": 4.4353882423569416e-05,
      "loss": 0.5372,
      "num_input_tokens_seen": 7385504,
      "step": 11310
    },
    {
      "epoch": 5.930293501048218,
      "grad_norm": 0.1218077763915062,
      "learning_rate": 4.434664254614103e-05,
      "loss": 0.4104,
      "num_input_tokens_seen": 7389600,
      "step": 11315
    },
    {
      "epoch": 5.932914046121593,
      "grad_norm": 0.18970917165279388,
      "learning_rate": 4.4339398621615734e-05,
      "loss": 0.4884,
      "num_input_tokens_seen": 7394240,
      "step": 11320
    },
    {
      "epoch": 5.935534591194968,
      "grad_norm": 0.10761836916208267,
      "learning_rate": 4.433215065150889e-05,
      "loss": 0.5008,
      "num_input_tokens_seen": 7396704,
      "step": 11325
    },
    {
      "epoch": 5.938155136268344,
      "grad_norm": 0.1280892789363861,
      "learning_rate": 4.4324898637336674e-05,
      "loss": 0.4608,
      "num_input_tokens_seen": 7400352,
      "step": 11330
    },
    {
      "epoch": 5.940775681341719,
      "grad_norm": 0.15450318157672882,
      "learning_rate": 4.431764258061615e-05,
      "loss": 0.4079,
      "num_input_tokens_seen": 7402816,
      "step": 11335
    },
    {
      "epoch": 5.943396226415095,
      "grad_norm": 0.17605014145374298,
      "learning_rate": 4.431038248286517e-05,
      "loss": 0.4908,
      "num_input_tokens_seen": 7405568,
      "step": 11340
    },
    {
      "epoch": 5.94601677148847,
      "grad_norm": 0.14683417975902557,
      "learning_rate": 4.430311834560248e-05,
      "loss": 0.5055,
      "num_input_tokens_seen": 7408608,
      "step": 11345
    },
    {
      "epoch": 5.948637316561845,
      "grad_norm": 0.2196195125579834,
      "learning_rate": 4.4295850170347664e-05,
      "loss": 0.5699,
      "num_input_tokens_seen": 7412448,
      "step": 11350
    },
    {
      "epoch": 5.95125786163522,
      "grad_norm": 0.20641915500164032,
      "learning_rate": 4.428857795862113e-05,
      "loss": 0.4469,
      "num_input_tokens_seen": 7415104,
      "step": 11355
    },
    {
      "epoch": 5.953878406708595,
      "grad_norm": 0.14985668659210205,
      "learning_rate": 4.428130171194414e-05,
      "loss": 0.4557,
      "num_input_tokens_seen": 7418944,
      "step": 11360
    },
    {
      "epoch": 5.95649895178197,
      "grad_norm": 0.10059309750795364,
      "learning_rate": 4.427402143183879e-05,
      "loss": 0.4175,
      "num_input_tokens_seen": 7422816,
      "step": 11365
    },
    {
      "epoch": 5.959119496855346,
      "grad_norm": 0.19174177944660187,
      "learning_rate": 4.426673711982806e-05,
      "loss": 0.4278,
      "num_input_tokens_seen": 7425824,
      "step": 11370
    },
    {
      "epoch": 5.961740041928721,
      "grad_norm": 0.12766680121421814,
      "learning_rate": 4.425944877743573e-05,
      "loss": 0.5609,
      "num_input_tokens_seen": 7429088,
      "step": 11375
    },
    {
      "epoch": 5.964360587002097,
      "grad_norm": 0.14174428582191467,
      "learning_rate": 4.425215640618644e-05,
      "loss": 0.5837,
      "num_input_tokens_seen": 7432736,
      "step": 11380
    },
    {
      "epoch": 5.966981132075472,
      "grad_norm": 0.1479259878396988,
      "learning_rate": 4.424486000760567e-05,
      "loss": 0.5841,
      "num_input_tokens_seen": 7435840,
      "step": 11385
    },
    {
      "epoch": 5.969601677148847,
      "grad_norm": 0.20560652017593384,
      "learning_rate": 4.423755958321974e-05,
      "loss": 0.4379,
      "num_input_tokens_seen": 7441056,
      "step": 11390
    },
    {
      "epoch": 5.972222222222222,
      "grad_norm": 0.12632451951503754,
      "learning_rate": 4.423025513455582e-05,
      "loss": 0.5236,
      "num_input_tokens_seen": 7444640,
      "step": 11395
    },
    {
      "epoch": 5.9748427672955975,
      "grad_norm": 0.1869075745344162,
      "learning_rate": 4.422294666314192e-05,
      "loss": 0.5532,
      "num_input_tokens_seen": 7448096,
      "step": 11400
    },
    {
      "epoch": 5.977463312368973,
      "grad_norm": 0.17172567546367645,
      "learning_rate": 4.421563417050688e-05,
      "loss": 0.4817,
      "num_input_tokens_seen": 7451520,
      "step": 11405
    },
    {
      "epoch": 5.980083857442348,
      "grad_norm": 0.18648526072502136,
      "learning_rate": 4.42083176581804e-05,
      "loss": 0.4561,
      "num_input_tokens_seen": 7454048,
      "step": 11410
    },
    {
      "epoch": 5.982704402515723,
      "grad_norm": 0.461047500371933,
      "learning_rate": 4.4200997127693e-05,
      "loss": 0.4605,
      "num_input_tokens_seen": 7457216,
      "step": 11415
    },
    {
      "epoch": 5.985324947589098,
      "grad_norm": 0.15857435762882233,
      "learning_rate": 4.419367258057606e-05,
      "loss": 0.5205,
      "num_input_tokens_seen": 7460992,
      "step": 11420
    },
    {
      "epoch": 5.987945492662474,
      "grad_norm": 0.20727519690990448,
      "learning_rate": 4.418634401836179e-05,
      "loss": 0.4839,
      "num_input_tokens_seen": 7463488,
      "step": 11425
    },
    {
      "epoch": 5.990566037735849,
      "grad_norm": 0.16579808294773102,
      "learning_rate": 4.417901144258324e-05,
      "loss": 0.492,
      "num_input_tokens_seen": 7466304,
      "step": 11430
    },
    {
      "epoch": 5.993186582809225,
      "grad_norm": 0.15253110229969025,
      "learning_rate": 4.4171674854774305e-05,
      "loss": 0.5447,
      "num_input_tokens_seen": 7470176,
      "step": 11435
    },
    {
      "epoch": 5.9958071278826,
      "grad_norm": 0.24951981008052826,
      "learning_rate": 4.416433425646971e-05,
      "loss": 0.4081,
      "num_input_tokens_seen": 7473376,
      "step": 11440
    },
    {
      "epoch": 5.998427672955975,
      "grad_norm": 0.2859075963497162,
      "learning_rate": 4.415698964920503e-05,
      "loss": 0.5222,
      "num_input_tokens_seen": 7476864,
      "step": 11445
    },
    {
      "epoch": 6.0,
      "eval_loss": 0.5106295347213745,
      "eval_runtime": 9.8897,
      "eval_samples_per_second": 85.746,
      "eval_steps_per_second": 21.436,
      "num_input_tokens_seen": 7478504,
      "step": 11448
    },
    {
      "epoch": 6.00104821802935,
      "grad_norm": 0.14653846621513367,
      "learning_rate": 4.414964103451667e-05,
      "loss": 0.4527,
      "num_input_tokens_seen": 7479688,
      "step": 11450
    },
    {
      "epoch": 6.003668763102725,
      "grad_norm": 0.14359083771705627,
      "learning_rate": 4.414228841394187e-05,
      "loss": 0.4432,
      "num_input_tokens_seen": 7483176,
      "step": 11455
    },
    {
      "epoch": 6.0062893081761,
      "grad_norm": 0.14207032322883606,
      "learning_rate": 4.413493178901872e-05,
      "loss": 0.5075,
      "num_input_tokens_seen": 7486120,
      "step": 11460
    },
    {
      "epoch": 6.008909853249476,
      "grad_norm": 0.2133597731590271,
      "learning_rate": 4.412757116128614e-05,
      "loss": 0.5143,
      "num_input_tokens_seen": 7488872,
      "step": 11465
    },
    {
      "epoch": 6.011530398322851,
      "grad_norm": 0.32181817293167114,
      "learning_rate": 4.41202065322839e-05,
      "loss": 0.5441,
      "num_input_tokens_seen": 7491880,
      "step": 11470
    },
    {
      "epoch": 6.014150943396227,
      "grad_norm": 0.24826467037200928,
      "learning_rate": 4.411283790355259e-05,
      "loss": 0.4324,
      "num_input_tokens_seen": 7494952,
      "step": 11475
    },
    {
      "epoch": 6.016771488469602,
      "grad_norm": 0.10310453176498413,
      "learning_rate": 4.410546527663364e-05,
      "loss": 0.5473,
      "num_input_tokens_seen": 7498600,
      "step": 11480
    },
    {
      "epoch": 6.019392033542977,
      "grad_norm": 0.13031195104122162,
      "learning_rate": 4.409808865306931e-05,
      "loss": 0.5256,
      "num_input_tokens_seen": 7502696,
      "step": 11485
    },
    {
      "epoch": 6.022012578616352,
      "grad_norm": 0.20108938217163086,
      "learning_rate": 4.409070803440273e-05,
      "loss": 0.5072,
      "num_input_tokens_seen": 7505800,
      "step": 11490
    },
    {
      "epoch": 6.0246331236897275,
      "grad_norm": 0.24129363894462585,
      "learning_rate": 4.4083323422177824e-05,
      "loss": 0.4928,
      "num_input_tokens_seen": 7508712,
      "step": 11495
    },
    {
      "epoch": 6.027253668763103,
      "grad_norm": 0.171027272939682,
      "learning_rate": 4.407593481793937e-05,
      "loss": 0.4306,
      "num_input_tokens_seen": 7511400,
      "step": 11500
    },
    {
      "epoch": 6.029874213836478,
      "grad_norm": 0.12585048377513885,
      "learning_rate": 4.406854222323298e-05,
      "loss": 0.4814,
      "num_input_tokens_seen": 7515208,
      "step": 11505
    },
    {
      "epoch": 6.032494758909853,
      "grad_norm": 0.24779504537582397,
      "learning_rate": 4.406114563960511e-05,
      "loss": 0.3371,
      "num_input_tokens_seen": 7520712,
      "step": 11510
    },
    {
      "epoch": 6.035115303983228,
      "grad_norm": 0.10611481964588165,
      "learning_rate": 4.4053745068603026e-05,
      "loss": 0.4353,
      "num_input_tokens_seen": 7522984,
      "step": 11515
    },
    {
      "epoch": 6.037735849056604,
      "grad_norm": 0.11729887872934341,
      "learning_rate": 4.404634051177485e-05,
      "loss": 0.4818,
      "num_input_tokens_seen": 7526024,
      "step": 11520
    },
    {
      "epoch": 6.040356394129979,
      "grad_norm": 0.41550618410110474,
      "learning_rate": 4.403893197066953e-05,
      "loss": 0.5555,
      "num_input_tokens_seen": 7529288,
      "step": 11525
    },
    {
      "epoch": 6.0429769392033545,
      "grad_norm": 0.18271999061107635,
      "learning_rate": 4.4031519446836855e-05,
      "loss": 0.4757,
      "num_input_tokens_seen": 7531752,
      "step": 11530
    },
    {
      "epoch": 6.04559748427673,
      "grad_norm": 0.1973768025636673,
      "learning_rate": 4.402410294182743e-05,
      "loss": 0.4929,
      "num_input_tokens_seen": 7535112,
      "step": 11535
    },
    {
      "epoch": 6.048218029350105,
      "grad_norm": 0.2745574116706848,
      "learning_rate": 4.4016682457192714e-05,
      "loss": 0.4376,
      "num_input_tokens_seen": 7538344,
      "step": 11540
    },
    {
      "epoch": 6.05083857442348,
      "grad_norm": 0.2893034815788269,
      "learning_rate": 4.400925799448497e-05,
      "loss": 0.4698,
      "num_input_tokens_seen": 7541064,
      "step": 11545
    },
    {
      "epoch": 6.053459119496855,
      "grad_norm": 0.3760901093482971,
      "learning_rate": 4.400182955525733e-05,
      "loss": 0.4717,
      "num_input_tokens_seen": 7544808,
      "step": 11550
    },
    {
      "epoch": 6.05607966457023,
      "grad_norm": 0.1698894202709198,
      "learning_rate": 4.3994397141063734e-05,
      "loss": 0.3766,
      "num_input_tokens_seen": 7548872,
      "step": 11555
    },
    {
      "epoch": 6.058700209643606,
      "grad_norm": 0.16243337094783783,
      "learning_rate": 4.398696075345896e-05,
      "loss": 0.5657,
      "num_input_tokens_seen": 7551912,
      "step": 11560
    },
    {
      "epoch": 6.061320754716981,
      "grad_norm": 0.13499261438846588,
      "learning_rate": 4.39795203939986e-05,
      "loss": 0.4146,
      "num_input_tokens_seen": 7554856,
      "step": 11565
    },
    {
      "epoch": 6.063941299790357,
      "grad_norm": 0.5107237100601196,
      "learning_rate": 4.397207606423911e-05,
      "loss": 0.6067,
      "num_input_tokens_seen": 7558344,
      "step": 11570
    },
    {
      "epoch": 6.066561844863732,
      "grad_norm": 0.1718040555715561,
      "learning_rate": 4.396462776573775e-05,
      "loss": 0.521,
      "num_input_tokens_seen": 7561704,
      "step": 11575
    },
    {
      "epoch": 6.069182389937107,
      "grad_norm": 0.18741367757320404,
      "learning_rate": 4.395717550005262e-05,
      "loss": 0.4925,
      "num_input_tokens_seen": 7563816,
      "step": 11580
    },
    {
      "epoch": 6.071802935010482,
      "grad_norm": 0.1347350776195526,
      "learning_rate": 4.394971926874265e-05,
      "loss": 0.4251,
      "num_input_tokens_seen": 7566408,
      "step": 11585
    },
    {
      "epoch": 6.0744234800838575,
      "grad_norm": 0.17141205072402954,
      "learning_rate": 4.3942259073367604e-05,
      "loss": 0.4653,
      "num_input_tokens_seen": 7569960,
      "step": 11590
    },
    {
      "epoch": 6.077044025157233,
      "grad_norm": 0.2567731738090515,
      "learning_rate": 4.3934794915488065e-05,
      "loss": 0.4375,
      "num_input_tokens_seen": 7572872,
      "step": 11595
    },
    {
      "epoch": 6.079664570230608,
      "grad_norm": 0.22812224924564362,
      "learning_rate": 4.392732679666544e-05,
      "loss": 0.6449,
      "num_input_tokens_seen": 7576104,
      "step": 11600
    },
    {
      "epoch": 6.082285115303983,
      "grad_norm": 0.1350485384464264,
      "learning_rate": 4.3919854718461986e-05,
      "loss": 0.4396,
      "num_input_tokens_seen": 7578856,
      "step": 11605
    },
    {
      "epoch": 6.084905660377358,
      "grad_norm": 0.13440775871276855,
      "learning_rate": 4.391237868244078e-05,
      "loss": 0.4591,
      "num_input_tokens_seen": 7582408,
      "step": 11610
    },
    {
      "epoch": 6.087526205450734,
      "grad_norm": 0.13163727521896362,
      "learning_rate": 4.390489869016571e-05,
      "loss": 0.4295,
      "num_input_tokens_seen": 7585224,
      "step": 11615
    },
    {
      "epoch": 6.090146750524109,
      "grad_norm": 0.25730565190315247,
      "learning_rate": 4.389741474320152e-05,
      "loss": 0.4913,
      "num_input_tokens_seen": 7587880,
      "step": 11620
    },
    {
      "epoch": 6.0927672955974845,
      "grad_norm": 0.1510058492422104,
      "learning_rate": 4.388992684311376e-05,
      "loss": 0.5462,
      "num_input_tokens_seen": 7591784,
      "step": 11625
    },
    {
      "epoch": 6.09538784067086,
      "grad_norm": 0.19406087696552277,
      "learning_rate": 4.388243499146881e-05,
      "loss": 0.5268,
      "num_input_tokens_seen": 7594920,
      "step": 11630
    },
    {
      "epoch": 6.098008385744235,
      "grad_norm": 0.2755132019519806,
      "learning_rate": 4.3874939189833876e-05,
      "loss": 0.6357,
      "num_input_tokens_seen": 7598504,
      "step": 11635
    },
    {
      "epoch": 6.10062893081761,
      "grad_norm": 0.1263226866722107,
      "learning_rate": 4.3867439439777005e-05,
      "loss": 0.5742,
      "num_input_tokens_seen": 7601640,
      "step": 11640
    },
    {
      "epoch": 6.103249475890985,
      "grad_norm": 0.15642978250980377,
      "learning_rate": 4.385993574286705e-05,
      "loss": 0.5222,
      "num_input_tokens_seen": 7604520,
      "step": 11645
    },
    {
      "epoch": 6.10587002096436,
      "grad_norm": 0.09126173704862595,
      "learning_rate": 4.3852428100673705e-05,
      "loss": 0.3613,
      "num_input_tokens_seen": 7608136,
      "step": 11650
    },
    {
      "epoch": 6.1084905660377355,
      "grad_norm": 0.24058550596237183,
      "learning_rate": 4.3844916514767484e-05,
      "loss": 0.5666,
      "num_input_tokens_seen": 7611592,
      "step": 11655
    },
    {
      "epoch": 6.111111111111111,
      "grad_norm": 0.2996438443660736,
      "learning_rate": 4.383740098671972e-05,
      "loss": 0.5047,
      "num_input_tokens_seen": 7614184,
      "step": 11660
    },
    {
      "epoch": 6.113731656184487,
      "grad_norm": 0.19928127527236938,
      "learning_rate": 4.3829881518102576e-05,
      "loss": 0.4247,
      "num_input_tokens_seen": 7617064,
      "step": 11665
    },
    {
      "epoch": 6.116352201257862,
      "grad_norm": 0.18029452860355377,
      "learning_rate": 4.3822358110489045e-05,
      "loss": 0.4404,
      "num_input_tokens_seen": 7620680,
      "step": 11670
    },
    {
      "epoch": 6.118972746331237,
      "grad_norm": 0.17452122271060944,
      "learning_rate": 4.381483076545294e-05,
      "loss": 0.4585,
      "num_input_tokens_seen": 7623944,
      "step": 11675
    },
    {
      "epoch": 6.121593291404612,
      "grad_norm": 0.29906848073005676,
      "learning_rate": 4.380729948456888e-05,
      "loss": 0.4658,
      "num_input_tokens_seen": 7627144,
      "step": 11680
    },
    {
      "epoch": 6.1242138364779874,
      "grad_norm": 0.16436758637428284,
      "learning_rate": 4.3799764269412336e-05,
      "loss": 0.4621,
      "num_input_tokens_seen": 7630216,
      "step": 11685
    },
    {
      "epoch": 6.126834381551363,
      "grad_norm": 0.28686726093292236,
      "learning_rate": 4.379222512155958e-05,
      "loss": 0.4974,
      "num_input_tokens_seen": 7633000,
      "step": 11690
    },
    {
      "epoch": 6.129454926624738,
      "grad_norm": 0.2494363635778427,
      "learning_rate": 4.378468204258773e-05,
      "loss": 0.4204,
      "num_input_tokens_seen": 7636104,
      "step": 11695
    },
    {
      "epoch": 6.132075471698113,
      "grad_norm": 0.19061805307865143,
      "learning_rate": 4.37771350340747e-05,
      "loss": 0.6366,
      "num_input_tokens_seen": 7639208,
      "step": 11700
    },
    {
      "epoch": 6.134696016771488,
      "grad_norm": 0.12453873455524445,
      "learning_rate": 4.376958409759925e-05,
      "loss": 0.512,
      "num_input_tokens_seen": 7642184,
      "step": 11705
    },
    {
      "epoch": 6.137316561844864,
      "grad_norm": 0.2927892804145813,
      "learning_rate": 4.376202923474093e-05,
      "loss": 0.5314,
      "num_input_tokens_seen": 7645480,
      "step": 11710
    },
    {
      "epoch": 6.139937106918239,
      "grad_norm": 0.16155844926834106,
      "learning_rate": 4.3754470447080145e-05,
      "loss": 0.5169,
      "num_input_tokens_seen": 7648936,
      "step": 11715
    },
    {
      "epoch": 6.1425576519916145,
      "grad_norm": 0.1461889147758484,
      "learning_rate": 4.374690773619811e-05,
      "loss": 0.3959,
      "num_input_tokens_seen": 7651816,
      "step": 11720
    },
    {
      "epoch": 6.14517819706499,
      "grad_norm": 0.11984746903181076,
      "learning_rate": 4.373934110367685e-05,
      "loss": 0.5381,
      "num_input_tokens_seen": 7654568,
      "step": 11725
    },
    {
      "epoch": 6.147798742138365,
      "grad_norm": 0.13827669620513916,
      "learning_rate": 4.373177055109922e-05,
      "loss": 0.3316,
      "num_input_tokens_seen": 7657544,
      "step": 11730
    },
    {
      "epoch": 6.15041928721174,
      "grad_norm": 0.19999991357326508,
      "learning_rate": 4.3724196080048896e-05,
      "loss": 0.5652,
      "num_input_tokens_seen": 7660936,
      "step": 11735
    },
    {
      "epoch": 6.153039832285115,
      "grad_norm": 0.21229569613933563,
      "learning_rate": 4.3716617692110364e-05,
      "loss": 0.5579,
      "num_input_tokens_seen": 7663336,
      "step": 11740
    },
    {
      "epoch": 6.15566037735849,
      "grad_norm": 0.10654629021883011,
      "learning_rate": 4.3709035388868943e-05,
      "loss": 0.4888,
      "num_input_tokens_seen": 7666312,
      "step": 11745
    },
    {
      "epoch": 6.1582809224318655,
      "grad_norm": 0.22211478650569916,
      "learning_rate": 4.3701449171910767e-05,
      "loss": 0.558,
      "num_input_tokens_seen": 7669224,
      "step": 11750
    },
    {
      "epoch": 6.160901467505241,
      "grad_norm": 0.13592225313186646,
      "learning_rate": 4.3693859042822774e-05,
      "loss": 0.4884,
      "num_input_tokens_seen": 7672264,
      "step": 11755
    },
    {
      "epoch": 6.163522012578617,
      "grad_norm": 0.21014246344566345,
      "learning_rate": 4.368626500319275e-05,
      "loss": 0.5581,
      "num_input_tokens_seen": 7674824,
      "step": 11760
    },
    {
      "epoch": 6.166142557651992,
      "grad_norm": 0.17012080550193787,
      "learning_rate": 4.367866705460927e-05,
      "loss": 0.3666,
      "num_input_tokens_seen": 7677608,
      "step": 11765
    },
    {
      "epoch": 6.168763102725367,
      "grad_norm": 0.20846959948539734,
      "learning_rate": 4.367106519866173e-05,
      "loss": 0.5102,
      "num_input_tokens_seen": 7680680,
      "step": 11770
    },
    {
      "epoch": 6.171383647798742,
      "grad_norm": 0.18845680356025696,
      "learning_rate": 4.366345943694037e-05,
      "loss": 0.516,
      "num_input_tokens_seen": 7684072,
      "step": 11775
    },
    {
      "epoch": 6.174004192872117,
      "grad_norm": 0.11400868743658066,
      "learning_rate": 4.365584977103622e-05,
      "loss": 0.5213,
      "num_input_tokens_seen": 7687624,
      "step": 11780
    },
    {
      "epoch": 6.176624737945493,
      "grad_norm": 0.12495733052492142,
      "learning_rate": 4.364823620254113e-05,
      "loss": 0.5717,
      "num_input_tokens_seen": 7691656,
      "step": 11785
    },
    {
      "epoch": 6.179245283018868,
      "grad_norm": 0.13454601168632507,
      "learning_rate": 4.364061873304778e-05,
      "loss": 0.5372,
      "num_input_tokens_seen": 7694568,
      "step": 11790
    },
    {
      "epoch": 6.181865828092243,
      "grad_norm": 0.07382619380950928,
      "learning_rate": 4.363299736414965e-05,
      "loss": 0.4286,
      "num_input_tokens_seen": 7697544,
      "step": 11795
    },
    {
      "epoch": 6.184486373165618,
      "grad_norm": 0.15424364805221558,
      "learning_rate": 4.3625372097441045e-05,
      "loss": 0.7881,
      "num_input_tokens_seen": 7700456,
      "step": 11800
    },
    {
      "epoch": 6.187106918238993,
      "grad_norm": 0.1912245899438858,
      "learning_rate": 4.3617742934517094e-05,
      "loss": 0.4775,
      "num_input_tokens_seen": 7703112,
      "step": 11805
    },
    {
      "epoch": 6.189727463312369,
      "grad_norm": 0.15709827840328217,
      "learning_rate": 4.361010987697373e-05,
      "loss": 0.4542,
      "num_input_tokens_seen": 7705992,
      "step": 11810
    },
    {
      "epoch": 6.1923480083857445,
      "grad_norm": 0.15478253364562988,
      "learning_rate": 4.360247292640768e-05,
      "loss": 0.5023,
      "num_input_tokens_seen": 7709576,
      "step": 11815
    },
    {
      "epoch": 6.19496855345912,
      "grad_norm": 0.1743312031030655,
      "learning_rate": 4.359483208441653e-05,
      "loss": 0.4751,
      "num_input_tokens_seen": 7712712,
      "step": 11820
    },
    {
      "epoch": 6.197589098532495,
      "grad_norm": 0.20538318157196045,
      "learning_rate": 4.3587187352598646e-05,
      "loss": 0.4701,
      "num_input_tokens_seen": 7715432,
      "step": 11825
    },
    {
      "epoch": 6.20020964360587,
      "grad_norm": 0.1530299186706543,
      "learning_rate": 4.357953873255323e-05,
      "loss": 0.4278,
      "num_input_tokens_seen": 7718216,
      "step": 11830
    },
    {
      "epoch": 6.202830188679245,
      "grad_norm": 0.16776582598686218,
      "learning_rate": 4.357188622588027e-05,
      "loss": 0.4431,
      "num_input_tokens_seen": 7721096,
      "step": 11835
    },
    {
      "epoch": 6.20545073375262,
      "grad_norm": 0.32580167055130005,
      "learning_rate": 4.356422983418059e-05,
      "loss": 0.6051,
      "num_input_tokens_seen": 7724264,
      "step": 11840
    },
    {
      "epoch": 6.2080712788259955,
      "grad_norm": 0.18134555220603943,
      "learning_rate": 4.3556569559055824e-05,
      "loss": 0.4741,
      "num_input_tokens_seen": 7727176,
      "step": 11845
    },
    {
      "epoch": 6.210691823899371,
      "grad_norm": 0.17882892489433289,
      "learning_rate": 4.3548905402108406e-05,
      "loss": 0.4991,
      "num_input_tokens_seen": 7729544,
      "step": 11850
    },
    {
      "epoch": 6.213312368972747,
      "grad_norm": 0.14011739194393158,
      "learning_rate": 4.35412373649416e-05,
      "loss": 0.5795,
      "num_input_tokens_seen": 7733288,
      "step": 11855
    },
    {
      "epoch": 6.215932914046122,
      "grad_norm": 0.18848928809165955,
      "learning_rate": 4.353356544915947e-05,
      "loss": 0.4381,
      "num_input_tokens_seen": 7737096,
      "step": 11860
    },
    {
      "epoch": 6.218553459119497,
      "grad_norm": 0.28381413221359253,
      "learning_rate": 4.352588965636689e-05,
      "loss": 0.4685,
      "num_input_tokens_seen": 7741512,
      "step": 11865
    },
    {
      "epoch": 6.221174004192872,
      "grad_norm": 0.16628094017505646,
      "learning_rate": 4.351820998816954e-05,
      "loss": 0.4948,
      "num_input_tokens_seen": 7744648,
      "step": 11870
    },
    {
      "epoch": 6.223794549266247,
      "grad_norm": 0.2983183264732361,
      "learning_rate": 4.3510526446173935e-05,
      "loss": 0.3968,
      "num_input_tokens_seen": 7748392,
      "step": 11875
    },
    {
      "epoch": 6.226415094339623,
      "grad_norm": 0.20268797874450684,
      "learning_rate": 4.350283903198737e-05,
      "loss": 0.4339,
      "num_input_tokens_seen": 7752360,
      "step": 11880
    },
    {
      "epoch": 6.229035639412998,
      "grad_norm": 0.20561370253562927,
      "learning_rate": 4.349514774721797e-05,
      "loss": 0.3476,
      "num_input_tokens_seen": 7755016,
      "step": 11885
    },
    {
      "epoch": 6.231656184486373,
      "grad_norm": 0.14434035122394562,
      "learning_rate": 4.348745259347467e-05,
      "loss": 0.5518,
      "num_input_tokens_seen": 7758120,
      "step": 11890
    },
    {
      "epoch": 6.234276729559748,
      "grad_norm": 0.22928942739963531,
      "learning_rate": 4.347975357236721e-05,
      "loss": 0.6694,
      "num_input_tokens_seen": 7761032,
      "step": 11895
    },
    {
      "epoch": 6.236897274633123,
      "grad_norm": 0.18079179525375366,
      "learning_rate": 4.347205068550612e-05,
      "loss": 0.4516,
      "num_input_tokens_seen": 7763240,
      "step": 11900
    },
    {
      "epoch": 6.239517819706499,
      "grad_norm": 0.13946405053138733,
      "learning_rate": 4.3464343934502774e-05,
      "loss": 0.4832,
      "num_input_tokens_seen": 7766664,
      "step": 11905
    },
    {
      "epoch": 6.2421383647798745,
      "grad_norm": 0.20320813357830048,
      "learning_rate": 4.345663332096932e-05,
      "loss": 0.5866,
      "num_input_tokens_seen": 7769448,
      "step": 11910
    },
    {
      "epoch": 6.24475890985325,
      "grad_norm": 0.26435357332229614,
      "learning_rate": 4.344891884651875e-05,
      "loss": 0.6051,
      "num_input_tokens_seen": 7772296,
      "step": 11915
    },
    {
      "epoch": 6.247379454926625,
      "grad_norm": 0.2629462480545044,
      "learning_rate": 4.344120051276483e-05,
      "loss": 0.4242,
      "num_input_tokens_seen": 7775912,
      "step": 11920
    },
    {
      "epoch": 6.25,
      "grad_norm": 0.09680535644292831,
      "learning_rate": 4.3433478321322155e-05,
      "loss": 0.5516,
      "num_input_tokens_seen": 7779784,
      "step": 11925
    },
    {
      "epoch": 6.252620545073375,
      "grad_norm": 0.16893784701824188,
      "learning_rate": 4.3425752273806106e-05,
      "loss": 0.3716,
      "num_input_tokens_seen": 7782504,
      "step": 11930
    },
    {
      "epoch": 6.25524109014675,
      "grad_norm": 0.23350709676742554,
      "learning_rate": 4.34180223718329e-05,
      "loss": 0.3527,
      "num_input_tokens_seen": 7785864,
      "step": 11935
    },
    {
      "epoch": 6.2578616352201255,
      "grad_norm": 0.22708171606063843,
      "learning_rate": 4.341028861701953e-05,
      "loss": 0.4291,
      "num_input_tokens_seen": 7788808,
      "step": 11940
    },
    {
      "epoch": 6.260482180293501,
      "grad_norm": 0.19502030313014984,
      "learning_rate": 4.3402551010983825e-05,
      "loss": 0.4496,
      "num_input_tokens_seen": 7792232,
      "step": 11945
    },
    {
      "epoch": 6.263102725366876,
      "grad_norm": 0.1348797231912613,
      "learning_rate": 4.33948095553444e-05,
      "loss": 0.462,
      "num_input_tokens_seen": 7794856,
      "step": 11950
    },
    {
      "epoch": 6.265723270440252,
      "grad_norm": 0.15250936150550842,
      "learning_rate": 4.338706425172067e-05,
      "loss": 0.6258,
      "num_input_tokens_seen": 7797800,
      "step": 11955
    },
    {
      "epoch": 6.268343815513627,
      "grad_norm": 0.12043496966362,
      "learning_rate": 4.3379315101732866e-05,
      "loss": 0.4565,
      "num_input_tokens_seen": 7801256,
      "step": 11960
    },
    {
      "epoch": 6.270964360587002,
      "grad_norm": 0.13585029542446136,
      "learning_rate": 4.3371562107002026e-05,
      "loss": 0.4193,
      "num_input_tokens_seen": 7804296,
      "step": 11965
    },
    {
      "epoch": 6.273584905660377,
      "grad_norm": 0.14639657735824585,
      "learning_rate": 4.3363805269149984e-05,
      "loss": 0.5112,
      "num_input_tokens_seen": 7807464,
      "step": 11970
    },
    {
      "epoch": 6.276205450733753,
      "grad_norm": 0.08617033809423447,
      "learning_rate": 4.335604458979939e-05,
      "loss": 0.4652,
      "num_input_tokens_seen": 7812008,
      "step": 11975
    },
    {
      "epoch": 6.278825995807128,
      "grad_norm": 0.31053057312965393,
      "learning_rate": 4.3348280070573686e-05,
      "loss": 0.4886,
      "num_input_tokens_seen": 7815336,
      "step": 11980
    },
    {
      "epoch": 6.281446540880503,
      "grad_norm": 0.0819668099284172,
      "learning_rate": 4.334051171309712e-05,
      "loss": 0.4224,
      "num_input_tokens_seen": 7818856,
      "step": 11985
    },
    {
      "epoch": 6.284067085953878,
      "grad_norm": 0.17212842404842377,
      "learning_rate": 4.333273951899474e-05,
      "loss": 0.5862,
      "num_input_tokens_seen": 7822024,
      "step": 11990
    },
    {
      "epoch": 6.286687631027253,
      "grad_norm": 0.1767415851354599,
      "learning_rate": 4.33249634898924e-05,
      "loss": 0.6581,
      "num_input_tokens_seen": 7825640,
      "step": 11995
    },
    {
      "epoch": 6.289308176100629,
      "grad_norm": 0.11624161899089813,
      "learning_rate": 4.331718362741677e-05,
      "loss": 0.4641,
      "num_input_tokens_seen": 7827944,
      "step": 12000
    },
    {
      "epoch": 6.2919287211740045,
      "grad_norm": 0.15981866419315338,
      "learning_rate": 4.330939993319529e-05,
      "loss": 0.4831,
      "num_input_tokens_seen": 7830600,
      "step": 12005
    },
    {
      "epoch": 6.29454926624738,
      "grad_norm": 0.10737613588571548,
      "learning_rate": 4.330161240885623e-05,
      "loss": 0.3815,
      "num_input_tokens_seen": 7834056,
      "step": 12010
    },
    {
      "epoch": 6.297169811320755,
      "grad_norm": 0.2079414427280426,
      "learning_rate": 4.3293821056028654e-05,
      "loss": 0.481,
      "num_input_tokens_seen": 7836904,
      "step": 12015
    },
    {
      "epoch": 6.29979035639413,
      "grad_norm": 0.25122472643852234,
      "learning_rate": 4.3286025876342415e-05,
      "loss": 0.5015,
      "num_input_tokens_seen": 7839880,
      "step": 12020
    },
    {
      "epoch": 6.302410901467505,
      "grad_norm": 0.1374209225177765,
      "learning_rate": 4.327822687142819e-05,
      "loss": 0.4617,
      "num_input_tokens_seen": 7843432,
      "step": 12025
    },
    {
      "epoch": 6.30503144654088,
      "grad_norm": 0.21031834185123444,
      "learning_rate": 4.327042404291742e-05,
      "loss": 0.4545,
      "num_input_tokens_seen": 7846216,
      "step": 12030
    },
    {
      "epoch": 6.3076519916142555,
      "grad_norm": 0.276812881231308,
      "learning_rate": 4.326261739244239e-05,
      "loss": 0.5083,
      "num_input_tokens_seen": 7849544,
      "step": 12035
    },
    {
      "epoch": 6.310272536687631,
      "grad_norm": 0.16544657945632935,
      "learning_rate": 4.325480692163615e-05,
      "loss": 0.5313,
      "num_input_tokens_seen": 7852552,
      "step": 12040
    },
    {
      "epoch": 6.312893081761006,
      "grad_norm": 0.1776537448167801,
      "learning_rate": 4.324699263213256e-05,
      "loss": 0.5898,
      "num_input_tokens_seen": 7855336,
      "step": 12045
    },
    {
      "epoch": 6.315513626834382,
      "grad_norm": 0.16508950293064117,
      "learning_rate": 4.3239174525566285e-05,
      "loss": 0.5471,
      "num_input_tokens_seen": 7858184,
      "step": 12050
    },
    {
      "epoch": 6.318134171907757,
      "grad_norm": 0.26665350794792175,
      "learning_rate": 4.3231352603572784e-05,
      "loss": 0.4168,
      "num_input_tokens_seen": 7861192,
      "step": 12055
    },
    {
      "epoch": 6.320754716981132,
      "grad_norm": 0.14575046300888062,
      "learning_rate": 4.322352686778832e-05,
      "loss": 0.4843,
      "num_input_tokens_seen": 7864520,
      "step": 12060
    },
    {
      "epoch": 6.323375262054507,
      "grad_norm": 0.20902761816978455,
      "learning_rate": 4.3215697319849926e-05,
      "loss": 0.5784,
      "num_input_tokens_seen": 7867240,
      "step": 12065
    },
    {
      "epoch": 6.325995807127883,
      "grad_norm": 0.2401411533355713,
      "learning_rate": 4.3207863961395476e-05,
      "loss": 0.566,
      "num_input_tokens_seen": 7870120,
      "step": 12070
    },
    {
      "epoch": 6.328616352201258,
      "grad_norm": 0.2416262924671173,
      "learning_rate": 4.320002679406361e-05,
      "loss": 0.3965,
      "num_input_tokens_seen": 7873736,
      "step": 12075
    },
    {
      "epoch": 6.331236897274633,
      "grad_norm": 0.1697177141904831,
      "learning_rate": 4.319218581949378e-05,
      "loss": 0.4695,
      "num_input_tokens_seen": 7877000,
      "step": 12080
    },
    {
      "epoch": 6.333857442348008,
      "grad_norm": 0.24352717399597168,
      "learning_rate": 4.318434103932622e-05,
      "loss": 0.3721,
      "num_input_tokens_seen": 7879784,
      "step": 12085
    },
    {
      "epoch": 6.336477987421383,
      "grad_norm": 0.17159782350063324,
      "learning_rate": 4.317649245520197e-05,
      "loss": 0.5601,
      "num_input_tokens_seen": 7882792,
      "step": 12090
    },
    {
      "epoch": 6.339098532494759,
      "grad_norm": 0.20832866430282593,
      "learning_rate": 4.3168640068762875e-05,
      "loss": 0.5499,
      "num_input_tokens_seen": 7885416,
      "step": 12095
    },
    {
      "epoch": 6.3417190775681345,
      "grad_norm": 0.12498226761817932,
      "learning_rate": 4.316078388165155e-05,
      "loss": 0.5282,
      "num_input_tokens_seen": 7888936,
      "step": 12100
    },
    {
      "epoch": 6.34433962264151,
      "grad_norm": 0.11197976768016815,
      "learning_rate": 4.315292389551143e-05,
      "loss": 0.5468,
      "num_input_tokens_seen": 7891976,
      "step": 12105
    },
    {
      "epoch": 6.346960167714885,
      "grad_norm": 0.30040302872657776,
      "learning_rate": 4.3145060111986726e-05,
      "loss": 0.5919,
      "num_input_tokens_seen": 7895848,
      "step": 12110
    },
    {
      "epoch": 6.34958071278826,
      "grad_norm": 0.22309187054634094,
      "learning_rate": 4.313719253272246e-05,
      "loss": 0.7335,
      "num_input_tokens_seen": 7898760,
      "step": 12115
    },
    {
      "epoch": 6.352201257861635,
      "grad_norm": 0.17070834338665009,
      "learning_rate": 4.3129321159364436e-05,
      "loss": 0.4762,
      "num_input_tokens_seen": 7902120,
      "step": 12120
    },
    {
      "epoch": 6.35482180293501,
      "grad_norm": 0.25172245502471924,
      "learning_rate": 4.312144599355925e-05,
      "loss": 0.6874,
      "num_input_tokens_seen": 7905224,
      "step": 12125
    },
    {
      "epoch": 6.3574423480083855,
      "grad_norm": 0.17315949499607086,
      "learning_rate": 4.31135670369543e-05,
      "loss": 0.6062,
      "num_input_tokens_seen": 7908424,
      "step": 12130
    },
    {
      "epoch": 6.360062893081761,
      "grad_norm": 0.18447211384773254,
      "learning_rate": 4.310568429119779e-05,
      "loss": 0.472,
      "num_input_tokens_seen": 7911336,
      "step": 12135
    },
    {
      "epoch": 6.362683438155136,
      "grad_norm": 0.17205417156219482,
      "learning_rate": 4.309779775793868e-05,
      "loss": 0.3869,
      "num_input_tokens_seen": 7913704,
      "step": 12140
    },
    {
      "epoch": 6.365303983228512,
      "grad_norm": 0.22542323172092438,
      "learning_rate": 4.308990743882675e-05,
      "loss": 0.5516,
      "num_input_tokens_seen": 7916584,
      "step": 12145
    },
    {
      "epoch": 6.367924528301887,
      "grad_norm": 0.13191211223602295,
      "learning_rate": 4.308201333551256e-05,
      "loss": 0.6113,
      "num_input_tokens_seen": 7919880,
      "step": 12150
    },
    {
      "epoch": 6.370545073375262,
      "grad_norm": 0.19099006056785583,
      "learning_rate": 4.3074115449647476e-05,
      "loss": 0.5763,
      "num_input_tokens_seen": 7923752,
      "step": 12155
    },
    {
      "epoch": 6.373165618448637,
      "grad_norm": 0.14293061196804047,
      "learning_rate": 4.306621378288364e-05,
      "loss": 0.5816,
      "num_input_tokens_seen": 7927464,
      "step": 12160
    },
    {
      "epoch": 6.3757861635220126,
      "grad_norm": 0.17539000511169434,
      "learning_rate": 4.305830833687399e-05,
      "loss": 0.4076,
      "num_input_tokens_seen": 7930792,
      "step": 12165
    },
    {
      "epoch": 6.378406708595388,
      "grad_norm": 0.10091504454612732,
      "learning_rate": 4.305039911327225e-05,
      "loss": 0.5228,
      "num_input_tokens_seen": 7935112,
      "step": 12170
    },
    {
      "epoch": 6.381027253668763,
      "grad_norm": 0.16663490235805511,
      "learning_rate": 4.304248611373296e-05,
      "loss": 0.4668,
      "num_input_tokens_seen": 7938152,
      "step": 12175
    },
    {
      "epoch": 6.383647798742138,
      "grad_norm": 0.1174534410238266,
      "learning_rate": 4.30345693399114e-05,
      "loss": 0.5224,
      "num_input_tokens_seen": 7942536,
      "step": 12180
    },
    {
      "epoch": 6.386268343815513,
      "grad_norm": 0.1624743789434433,
      "learning_rate": 4.30266487934637e-05,
      "loss": 0.4405,
      "num_input_tokens_seen": 7945224,
      "step": 12185
    },
    {
      "epoch": 6.388888888888889,
      "grad_norm": 0.14919541776180267,
      "learning_rate": 4.3018724476046714e-05,
      "loss": 0.4822,
      "num_input_tokens_seen": 7948520,
      "step": 12190
    },
    {
      "epoch": 6.3915094339622645,
      "grad_norm": 0.1923942118883133,
      "learning_rate": 4.301079638931814e-05,
      "loss": 0.5637,
      "num_input_tokens_seen": 7951368,
      "step": 12195
    },
    {
      "epoch": 6.39412997903564,
      "grad_norm": 0.15691272914409637,
      "learning_rate": 4.3002864534936444e-05,
      "loss": 0.638,
      "num_input_tokens_seen": 7954472,
      "step": 12200
    },
    {
      "epoch": 6.396750524109015,
      "grad_norm": 0.1573014110326767,
      "learning_rate": 4.2994928914560874e-05,
      "loss": 0.5497,
      "num_input_tokens_seen": 7957000,
      "step": 12205
    },
    {
      "epoch": 6.39937106918239,
      "grad_norm": 0.15008747577667236,
      "learning_rate": 4.298698952985147e-05,
      "loss": 0.5245,
      "num_input_tokens_seen": 7959400,
      "step": 12210
    },
    {
      "epoch": 6.401991614255765,
      "grad_norm": 0.32702603936195374,
      "learning_rate": 4.297904638246906e-05,
      "loss": 0.4962,
      "num_input_tokens_seen": 7962312,
      "step": 12215
    },
    {
      "epoch": 6.40461215932914,
      "grad_norm": 0.27031034231185913,
      "learning_rate": 4.297109947407527e-05,
      "loss": 0.5803,
      "num_input_tokens_seen": 7965768,
      "step": 12220
    },
    {
      "epoch": 6.4072327044025155,
      "grad_norm": 0.1435426026582718,
      "learning_rate": 4.2963148806332486e-05,
      "loss": 0.4395,
      "num_input_tokens_seen": 7972840,
      "step": 12225
    },
    {
      "epoch": 6.409853249475891,
      "grad_norm": 0.18539367616176605,
      "learning_rate": 4.2955194380903915e-05,
      "loss": 0.5721,
      "num_input_tokens_seen": 7975304,
      "step": 12230
    },
    {
      "epoch": 6.412473794549266,
      "grad_norm": 0.16253186762332916,
      "learning_rate": 4.2947236199453524e-05,
      "loss": 0.4086,
      "num_input_tokens_seen": 7977928,
      "step": 12235
    },
    {
      "epoch": 6.415094339622642,
      "grad_norm": 0.1904004067182541,
      "learning_rate": 4.2939274263646065e-05,
      "loss": 0.4798,
      "num_input_tokens_seen": 7982120,
      "step": 12240
    },
    {
      "epoch": 6.417714884696017,
      "grad_norm": 0.16524212062358856,
      "learning_rate": 4.29313085751471e-05,
      "loss": 0.4803,
      "num_input_tokens_seen": 7985032,
      "step": 12245
    },
    {
      "epoch": 6.420335429769392,
      "grad_norm": 0.15700335800647736,
      "learning_rate": 4.292333913562294e-05,
      "loss": 0.4883,
      "num_input_tokens_seen": 7987272,
      "step": 12250
    },
    {
      "epoch": 6.422955974842767,
      "grad_norm": 0.11570843309164047,
      "learning_rate": 4.291536594674073e-05,
      "loss": 0.547,
      "num_input_tokens_seen": 7990152,
      "step": 12255
    },
    {
      "epoch": 6.4255765199161425,
      "grad_norm": 0.2128990739583969,
      "learning_rate": 4.2907389010168344e-05,
      "loss": 0.5834,
      "num_input_tokens_seen": 7992648,
      "step": 12260
    },
    {
      "epoch": 6.428197064989518,
      "grad_norm": 0.13753575086593628,
      "learning_rate": 4.2899408327574475e-05,
      "loss": 0.4603,
      "num_input_tokens_seen": 7995880,
      "step": 12265
    },
    {
      "epoch": 6.430817610062893,
      "grad_norm": 0.296333909034729,
      "learning_rate": 4.289142390062859e-05,
      "loss": 0.4678,
      "num_input_tokens_seen": 7998312,
      "step": 12270
    },
    {
      "epoch": 6.433438155136268,
      "grad_norm": 0.13863538205623627,
      "learning_rate": 4.2883435731000946e-05,
      "loss": 0.4681,
      "num_input_tokens_seen": 8001768,
      "step": 12275
    },
    {
      "epoch": 6.436058700209643,
      "grad_norm": 0.35491400957107544,
      "learning_rate": 4.287544382036256e-05,
      "loss": 0.4419,
      "num_input_tokens_seen": 8004392,
      "step": 12280
    },
    {
      "epoch": 6.438679245283019,
      "grad_norm": 0.13956794142723083,
      "learning_rate": 4.286744817038527e-05,
      "loss": 0.4868,
      "num_input_tokens_seen": 8006984,
      "step": 12285
    },
    {
      "epoch": 6.441299790356394,
      "grad_norm": 0.12908430397510529,
      "learning_rate": 4.2859448782741666e-05,
      "loss": 0.6421,
      "num_input_tokens_seen": 8010920,
      "step": 12290
    },
    {
      "epoch": 6.44392033542977,
      "grad_norm": 0.154449462890625,
      "learning_rate": 4.2851445659105126e-05,
      "loss": 0.5495,
      "num_input_tokens_seen": 8013800,
      "step": 12295
    },
    {
      "epoch": 6.446540880503145,
      "grad_norm": 0.18562449514865875,
      "learning_rate": 4.284343880114982e-05,
      "loss": 0.3998,
      "num_input_tokens_seen": 8016616,
      "step": 12300
    },
    {
      "epoch": 6.44916142557652,
      "grad_norm": 0.12320191413164139,
      "learning_rate": 4.2835428210550676e-05,
      "loss": 0.4497,
      "num_input_tokens_seen": 8020040,
      "step": 12305
    },
    {
      "epoch": 6.451781970649895,
      "grad_norm": 0.13530758023262024,
      "learning_rate": 4.282741388898344e-05,
      "loss": 0.4176,
      "num_input_tokens_seen": 8024008,
      "step": 12310
    },
    {
      "epoch": 6.45440251572327,
      "grad_norm": 0.23911285400390625,
      "learning_rate": 4.281939583812459e-05,
      "loss": 0.5211,
      "num_input_tokens_seen": 8027528,
      "step": 12315
    },
    {
      "epoch": 6.4570230607966455,
      "grad_norm": 0.3652297854423523,
      "learning_rate": 4.281137405965143e-05,
      "loss": 0.5943,
      "num_input_tokens_seen": 8030312,
      "step": 12320
    },
    {
      "epoch": 6.459643605870021,
      "grad_norm": 0.1728571504354477,
      "learning_rate": 4.280334855524203e-05,
      "loss": 0.3866,
      "num_input_tokens_seen": 8034280,
      "step": 12325
    },
    {
      "epoch": 6.462264150943396,
      "grad_norm": 0.1751277595758438,
      "learning_rate": 4.279531932657521e-05,
      "loss": 0.5237,
      "num_input_tokens_seen": 8038408,
      "step": 12330
    },
    {
      "epoch": 6.464884696016772,
      "grad_norm": 0.17428399622440338,
      "learning_rate": 4.278728637533062e-05,
      "loss": 0.4919,
      "num_input_tokens_seen": 8041544,
      "step": 12335
    },
    {
      "epoch": 6.467505241090147,
      "grad_norm": 0.14913727343082428,
      "learning_rate": 4.2779249703188626e-05,
      "loss": 0.3478,
      "num_input_tokens_seen": 8045096,
      "step": 12340
    },
    {
      "epoch": 6.470125786163522,
      "grad_norm": 0.14771801233291626,
      "learning_rate": 4.277120931183045e-05,
      "loss": 0.4401,
      "num_input_tokens_seen": 8049064,
      "step": 12345
    },
    {
      "epoch": 6.472746331236897,
      "grad_norm": 0.14940455555915833,
      "learning_rate": 4.276316520293803e-05,
      "loss": 0.5511,
      "num_input_tokens_seen": 8051816,
      "step": 12350
    },
    {
      "epoch": 6.4753668763102725,
      "grad_norm": 0.1671476811170578,
      "learning_rate": 4.27551173781941e-05,
      "loss": 0.5038,
      "num_input_tokens_seen": 8055528,
      "step": 12355
    },
    {
      "epoch": 6.477987421383648,
      "grad_norm": 0.13954474031925201,
      "learning_rate": 4.274706583928216e-05,
      "loss": 0.421,
      "num_input_tokens_seen": 8058472,
      "step": 12360
    },
    {
      "epoch": 6.480607966457023,
      "grad_norm": 0.1839284747838974,
      "learning_rate": 4.273901058788654e-05,
      "loss": 0.7252,
      "num_input_tokens_seen": 8061608,
      "step": 12365
    },
    {
      "epoch": 6.483228511530398,
      "grad_norm": 0.11252845823764801,
      "learning_rate": 4.2730951625692275e-05,
      "loss": 0.4884,
      "num_input_tokens_seen": 8064648,
      "step": 12370
    },
    {
      "epoch": 6.485849056603773,
      "grad_norm": 0.17345185577869415,
      "learning_rate": 4.27228889543852e-05,
      "loss": 0.6766,
      "num_input_tokens_seen": 8067496,
      "step": 12375
    },
    {
      "epoch": 6.488469601677149,
      "grad_norm": 0.12892837822437286,
      "learning_rate": 4.2714822575651966e-05,
      "loss": 0.4418,
      "num_input_tokens_seen": 8069864,
      "step": 12380
    },
    {
      "epoch": 6.491090146750524,
      "grad_norm": 0.19244979321956635,
      "learning_rate": 4.270675249117994e-05,
      "loss": 0.6153,
      "num_input_tokens_seen": 8072936,
      "step": 12385
    },
    {
      "epoch": 6.4937106918239,
      "grad_norm": 0.2895551919937134,
      "learning_rate": 4.269867870265731e-05,
      "loss": 0.5725,
      "num_input_tokens_seen": 8075784,
      "step": 12390
    },
    {
      "epoch": 6.496331236897275,
      "grad_norm": 0.2542673945426941,
      "learning_rate": 4.269060121177301e-05,
      "loss": 0.518,
      "num_input_tokens_seen": 8078280,
      "step": 12395
    },
    {
      "epoch": 6.49895178197065,
      "grad_norm": 0.14235711097717285,
      "learning_rate": 4.2682520020216764e-05,
      "loss": 0.3844,
      "num_input_tokens_seen": 8081992,
      "step": 12400
    },
    {
      "epoch": 6.501572327044025,
      "grad_norm": 0.16638994216918945,
      "learning_rate": 4.267443512967906e-05,
      "loss": 0.5381,
      "num_input_tokens_seen": 8085672,
      "step": 12405
    },
    {
      "epoch": 6.5041928721174,
      "grad_norm": 0.11081931740045547,
      "learning_rate": 4.2666346541851165e-05,
      "loss": 0.4271,
      "num_input_tokens_seen": 8089000,
      "step": 12410
    },
    {
      "epoch": 6.506813417190775,
      "grad_norm": 0.12221400439739227,
      "learning_rate": 4.265825425842512e-05,
      "loss": 0.421,
      "num_input_tokens_seen": 8092168,
      "step": 12415
    },
    {
      "epoch": 6.509433962264151,
      "grad_norm": 0.11912526190280914,
      "learning_rate": 4.265015828109375e-05,
      "loss": 0.5481,
      "num_input_tokens_seen": 8094728,
      "step": 12420
    },
    {
      "epoch": 6.512054507337526,
      "grad_norm": 0.32821211218833923,
      "learning_rate": 4.264205861155063e-05,
      "loss": 0.5592,
      "num_input_tokens_seen": 8098088,
      "step": 12425
    },
    {
      "epoch": 6.514675052410902,
      "grad_norm": 0.210201695561409,
      "learning_rate": 4.263395525149011e-05,
      "loss": 0.4918,
      "num_input_tokens_seen": 8100904,
      "step": 12430
    },
    {
      "epoch": 6.517295597484277,
      "grad_norm": 0.13235104084014893,
      "learning_rate": 4.262584820260734e-05,
      "loss": 0.4043,
      "num_input_tokens_seen": 8104264,
      "step": 12435
    },
    {
      "epoch": 6.519916142557652,
      "grad_norm": 0.1413646787405014,
      "learning_rate": 4.261773746659821e-05,
      "loss": 0.41,
      "num_input_tokens_seen": 8107144,
      "step": 12440
    },
    {
      "epoch": 6.522536687631027,
      "grad_norm": 0.13634786009788513,
      "learning_rate": 4.260962304515939e-05,
      "loss": 0.51,
      "num_input_tokens_seen": 8111304,
      "step": 12445
    },
    {
      "epoch": 6.5251572327044025,
      "grad_norm": 0.19557467103004456,
      "learning_rate": 4.2601504939988335e-05,
      "loss": 0.5262,
      "num_input_tokens_seen": 8114056,
      "step": 12450
    },
    {
      "epoch": 6.527777777777778,
      "grad_norm": 0.19567419588565826,
      "learning_rate": 4.259338315278325e-05,
      "loss": 0.6123,
      "num_input_tokens_seen": 8116808,
      "step": 12455
    },
    {
      "epoch": 6.530398322851153,
      "grad_norm": 0.22382239997386932,
      "learning_rate": 4.258525768524313e-05,
      "loss": 0.5115,
      "num_input_tokens_seen": 8119496,
      "step": 12460
    },
    {
      "epoch": 6.533018867924528,
      "grad_norm": 0.26898446679115295,
      "learning_rate": 4.2577128539067725e-05,
      "loss": 0.5283,
      "num_input_tokens_seen": 8122728,
      "step": 12465
    },
    {
      "epoch": 6.535639412997903,
      "grad_norm": 0.16788817942142487,
      "learning_rate": 4.256899571595756e-05,
      "loss": 0.6152,
      "num_input_tokens_seen": 8126504,
      "step": 12470
    },
    {
      "epoch": 6.538259958071279,
      "grad_norm": 0.26465174555778503,
      "learning_rate": 4.256085921761393e-05,
      "loss": 0.691,
      "num_input_tokens_seen": 8128776,
      "step": 12475
    },
    {
      "epoch": 6.540880503144654,
      "grad_norm": 0.11723022907972336,
      "learning_rate": 4.25527190457389e-05,
      "loss": 0.5892,
      "num_input_tokens_seen": 8131720,
      "step": 12480
    },
    {
      "epoch": 6.54350104821803,
      "grad_norm": 0.2235911339521408,
      "learning_rate": 4.2544575202035285e-05,
      "loss": 0.5875,
      "num_input_tokens_seen": 8135624,
      "step": 12485
    },
    {
      "epoch": 6.546121593291405,
      "grad_norm": 0.17370212078094482,
      "learning_rate": 4.253642768820671e-05,
      "loss": 0.4553,
      "num_input_tokens_seen": 8138632,
      "step": 12490
    },
    {
      "epoch": 6.54874213836478,
      "grad_norm": 0.15009701251983643,
      "learning_rate": 4.252827650595752e-05,
      "loss": 0.5145,
      "num_input_tokens_seen": 8142184,
      "step": 12495
    },
    {
      "epoch": 6.551362683438155,
      "grad_norm": 0.12996087968349457,
      "learning_rate": 4.252012165699286e-05,
      "loss": 0.568,
      "num_input_tokens_seen": 8146792,
      "step": 12500
    },
    {
      "epoch": 6.55398322851153,
      "grad_norm": 0.19210736453533173,
      "learning_rate": 4.2511963143018634e-05,
      "loss": 0.4069,
      "num_input_tokens_seen": 8150504,
      "step": 12505
    },
    {
      "epoch": 6.556603773584905,
      "grad_norm": 0.13763201236724854,
      "learning_rate": 4.25038009657415e-05,
      "loss": 0.4911,
      "num_input_tokens_seen": 8153000,
      "step": 12510
    },
    {
      "epoch": 6.559224318658281,
      "grad_norm": 0.22254621982574463,
      "learning_rate": 4.24956351268689e-05,
      "loss": 0.4148,
      "num_input_tokens_seen": 8155720,
      "step": 12515
    },
    {
      "epoch": 6.561844863731656,
      "grad_norm": 0.18853120505809784,
      "learning_rate": 4.2487465628109036e-05,
      "loss": 0.4649,
      "num_input_tokens_seen": 8159400,
      "step": 12520
    },
    {
      "epoch": 6.564465408805032,
      "grad_norm": 0.16647489368915558,
      "learning_rate": 4.247929247117087e-05,
      "loss": 0.4142,
      "num_input_tokens_seen": 8162952,
      "step": 12525
    },
    {
      "epoch": 6.567085953878407,
      "grad_norm": 0.21176503598690033,
      "learning_rate": 4.247111565776413e-05,
      "loss": 0.5143,
      "num_input_tokens_seen": 8166376,
      "step": 12530
    },
    {
      "epoch": 6.569706498951782,
      "grad_norm": 0.24661369621753693,
      "learning_rate": 4.246293518959932e-05,
      "loss": 0.5507,
      "num_input_tokens_seen": 8169064,
      "step": 12535
    },
    {
      "epoch": 6.572327044025157,
      "grad_norm": 0.13644112646579742,
      "learning_rate": 4.245475106838769e-05,
      "loss": 0.6112,
      "num_input_tokens_seen": 8171976,
      "step": 12540
    },
    {
      "epoch": 6.5749475890985325,
      "grad_norm": 0.1402328759431839,
      "learning_rate": 4.2446563295841266e-05,
      "loss": 0.4791,
      "num_input_tokens_seen": 8176200,
      "step": 12545
    },
    {
      "epoch": 6.577568134171908,
      "grad_norm": 0.1496155858039856,
      "learning_rate": 4.243837187367285e-05,
      "loss": 0.5101,
      "num_input_tokens_seen": 8179368,
      "step": 12550
    },
    {
      "epoch": 6.580188679245283,
      "grad_norm": 0.1693345457315445,
      "learning_rate": 4.2430176803595987e-05,
      "loss": 0.3887,
      "num_input_tokens_seen": 8182248,
      "step": 12555
    },
    {
      "epoch": 6.582809224318658,
      "grad_norm": 0.152715802192688,
      "learning_rate": 4.2421978087324985e-05,
      "loss": 0.6221,
      "num_input_tokens_seen": 8184776,
      "step": 12560
    },
    {
      "epoch": 6.585429769392033,
      "grad_norm": 0.1503472775220871,
      "learning_rate": 4.241377572657493e-05,
      "loss": 0.4196,
      "num_input_tokens_seen": 8187976,
      "step": 12565
    },
    {
      "epoch": 6.588050314465409,
      "grad_norm": 0.17545677721500397,
      "learning_rate": 4.2405569723061654e-05,
      "loss": 0.4818,
      "num_input_tokens_seen": 8190984,
      "step": 12570
    },
    {
      "epoch": 6.590670859538784,
      "grad_norm": 0.2584352195262909,
      "learning_rate": 4.239736007850178e-05,
      "loss": 0.4432,
      "num_input_tokens_seen": 8194568,
      "step": 12575
    },
    {
      "epoch": 6.59329140461216,
      "grad_norm": 0.19405211508274078,
      "learning_rate": 4.238914679461264e-05,
      "loss": 0.6373,
      "num_input_tokens_seen": 8197512,
      "step": 12580
    },
    {
      "epoch": 6.595911949685535,
      "grad_norm": 0.13624398410320282,
      "learning_rate": 4.2380929873112386e-05,
      "loss": 0.4664,
      "num_input_tokens_seen": 8201128,
      "step": 12585
    },
    {
      "epoch": 6.59853249475891,
      "grad_norm": 0.16405519843101501,
      "learning_rate": 4.2372709315719896e-05,
      "loss": 0.3397,
      "num_input_tokens_seen": 8204392,
      "step": 12590
    },
    {
      "epoch": 6.601153039832285,
      "grad_norm": 0.190593883395195,
      "learning_rate": 4.2364485124154806e-05,
      "loss": 0.538,
      "num_input_tokens_seen": 8206952,
      "step": 12595
    },
    {
      "epoch": 6.60377358490566,
      "grad_norm": 0.2841431796550751,
      "learning_rate": 4.235625730013754e-05,
      "loss": 0.4357,
      "num_input_tokens_seen": 8209768,
      "step": 12600
    },
    {
      "epoch": 6.606394129979035,
      "grad_norm": 0.19433674216270447,
      "learning_rate": 4.234802584538926e-05,
      "loss": 0.5879,
      "num_input_tokens_seen": 8216680,
      "step": 12605
    },
    {
      "epoch": 6.609014675052411,
      "grad_norm": 0.1632453352212906,
      "learning_rate": 4.2339790761631884e-05,
      "loss": 0.4782,
      "num_input_tokens_seen": 8220552,
      "step": 12610
    },
    {
      "epoch": 6.611635220125786,
      "grad_norm": 0.14992070198059082,
      "learning_rate": 4.233155205058811e-05,
      "loss": 0.5219,
      "num_input_tokens_seen": 8225064,
      "step": 12615
    },
    {
      "epoch": 6.614255765199162,
      "grad_norm": 0.2141435444355011,
      "learning_rate": 4.232330971398137e-05,
      "loss": 0.5602,
      "num_input_tokens_seen": 8228168,
      "step": 12620
    },
    {
      "epoch": 6.616876310272537,
      "grad_norm": 0.3797670900821686,
      "learning_rate": 4.2315063753535875e-05,
      "loss": 0.4,
      "num_input_tokens_seen": 8231944,
      "step": 12625
    },
    {
      "epoch": 6.619496855345912,
      "grad_norm": 0.19294646382331848,
      "learning_rate": 4.230681417097658e-05,
      "loss": 0.5024,
      "num_input_tokens_seen": 8235848,
      "step": 12630
    },
    {
      "epoch": 6.622117400419287,
      "grad_norm": 0.11259070783853531,
      "learning_rate": 4.229856096802921e-05,
      "loss": 0.5726,
      "num_input_tokens_seen": 8238792,
      "step": 12635
    },
    {
      "epoch": 6.6247379454926625,
      "grad_norm": 0.18688178062438965,
      "learning_rate": 4.229030414642024e-05,
      "loss": 0.4648,
      "num_input_tokens_seen": 8242088,
      "step": 12640
    },
    {
      "epoch": 6.627358490566038,
      "grad_norm": 0.170536607503891,
      "learning_rate": 4.2282043707876896e-05,
      "loss": 0.6204,
      "num_input_tokens_seen": 8244616,
      "step": 12645
    },
    {
      "epoch": 6.629979035639413,
      "grad_norm": 0.20127253234386444,
      "learning_rate": 4.2273779654127175e-05,
      "loss": 0.4266,
      "num_input_tokens_seen": 8247912,
      "step": 12650
    },
    {
      "epoch": 6.632599580712788,
      "grad_norm": 0.18093812465667725,
      "learning_rate": 4.226551198689982e-05,
      "loss": 0.5006,
      "num_input_tokens_seen": 8251816,
      "step": 12655
    },
    {
      "epoch": 6.635220125786163,
      "grad_norm": 0.13783016800880432,
      "learning_rate": 4.225724070792434e-05,
      "loss": 0.5479,
      "num_input_tokens_seen": 8255432,
      "step": 12660
    },
    {
      "epoch": 6.637840670859539,
      "grad_norm": 0.10073185712099075,
      "learning_rate": 4.2248965818930975e-05,
      "loss": 0.3555,
      "num_input_tokens_seen": 8258824,
      "step": 12665
    },
    {
      "epoch": 6.640461215932914,
      "grad_norm": 0.185990571975708,
      "learning_rate": 4.224068732165075e-05,
      "loss": 0.5395,
      "num_input_tokens_seen": 8262888,
      "step": 12670
    },
    {
      "epoch": 6.6430817610062896,
      "grad_norm": 0.32388612627983093,
      "learning_rate": 4.223240521781543e-05,
      "loss": 0.3927,
      "num_input_tokens_seen": 8266568,
      "step": 12675
    },
    {
      "epoch": 6.645702306079665,
      "grad_norm": 0.18081556260585785,
      "learning_rate": 4.2224119509157535e-05,
      "loss": 0.554,
      "num_input_tokens_seen": 8269800,
      "step": 12680
    },
    {
      "epoch": 6.64832285115304,
      "grad_norm": 0.10476398468017578,
      "learning_rate": 4.221583019741033e-05,
      "loss": 0.4141,
      "num_input_tokens_seen": 8272936,
      "step": 12685
    },
    {
      "epoch": 6.650943396226415,
      "grad_norm": 0.13513575494289398,
      "learning_rate": 4.2207537284307876e-05,
      "loss": 0.5102,
      "num_input_tokens_seen": 8276104,
      "step": 12690
    },
    {
      "epoch": 6.65356394129979,
      "grad_norm": 0.10414682328701019,
      "learning_rate": 4.219924077158493e-05,
      "loss": 0.3871,
      "num_input_tokens_seen": 8278728,
      "step": 12695
    },
    {
      "epoch": 6.656184486373165,
      "grad_norm": 0.14092327654361725,
      "learning_rate": 4.219094066097703e-05,
      "loss": 0.4538,
      "num_input_tokens_seen": 8282312,
      "step": 12700
    },
    {
      "epoch": 6.658805031446541,
      "grad_norm": 0.14747856557369232,
      "learning_rate": 4.218263695422047e-05,
      "loss": 0.5422,
      "num_input_tokens_seen": 8285800,
      "step": 12705
    },
    {
      "epoch": 6.661425576519916,
      "grad_norm": 0.1759636551141739,
      "learning_rate": 4.217432965305229e-05,
      "loss": 0.6145,
      "num_input_tokens_seen": 8288648,
      "step": 12710
    },
    {
      "epoch": 6.664046121593291,
      "grad_norm": 0.1698714643716812,
      "learning_rate": 4.2166018759210286e-05,
      "loss": 0.472,
      "num_input_tokens_seen": 8291688,
      "step": 12715
    },
    {
      "epoch": 6.666666666666667,
      "grad_norm": 0.19956651329994202,
      "learning_rate": 4.215770427443299e-05,
      "loss": 0.5795,
      "num_input_tokens_seen": 8295400,
      "step": 12720
    },
    {
      "epoch": 6.669287211740042,
      "grad_norm": 0.1312435418367386,
      "learning_rate": 4.214938620045971e-05,
      "loss": 0.4602,
      "num_input_tokens_seen": 8297704,
      "step": 12725
    },
    {
      "epoch": 6.671907756813417,
      "grad_norm": 0.18971692025661469,
      "learning_rate": 4.214106453903048e-05,
      "loss": 0.6502,
      "num_input_tokens_seen": 8300808,
      "step": 12730
    },
    {
      "epoch": 6.6745283018867925,
      "grad_norm": 0.19644416868686676,
      "learning_rate": 4.2132739291886114e-05,
      "loss": 0.6874,
      "num_input_tokens_seen": 8304104,
      "step": 12735
    },
    {
      "epoch": 6.677148846960168,
      "grad_norm": 0.19592930376529694,
      "learning_rate": 4.212441046076814e-05,
      "loss": 0.6021,
      "num_input_tokens_seen": 8307784,
      "step": 12740
    },
    {
      "epoch": 6.679769392033543,
      "grad_norm": 0.1165885254740715,
      "learning_rate": 4.211607804741887e-05,
      "loss": 0.4794,
      "num_input_tokens_seen": 8311848,
      "step": 12745
    },
    {
      "epoch": 6.682389937106918,
      "grad_norm": 0.13884328305721283,
      "learning_rate": 4.2107742053581344e-05,
      "loss": 0.5158,
      "num_input_tokens_seen": 8315272,
      "step": 12750
    },
    {
      "epoch": 6.685010482180293,
      "grad_norm": 0.18072593212127686,
      "learning_rate": 4.209940248099935e-05,
      "loss": 0.4053,
      "num_input_tokens_seen": 8318376,
      "step": 12755
    },
    {
      "epoch": 6.687631027253669,
      "grad_norm": 0.15777982771396637,
      "learning_rate": 4.2091059331417445e-05,
      "loss": 0.4929,
      "num_input_tokens_seen": 8322248,
      "step": 12760
    },
    {
      "epoch": 6.690251572327044,
      "grad_norm": 0.0924091711640358,
      "learning_rate": 4.208271260658091e-05,
      "loss": 0.3492,
      "num_input_tokens_seen": 8326056,
      "step": 12765
    },
    {
      "epoch": 6.6928721174004195,
      "grad_norm": 0.1595662534236908,
      "learning_rate": 4.2074362308235785e-05,
      "loss": 0.5921,
      "num_input_tokens_seen": 8329608,
      "step": 12770
    },
    {
      "epoch": 6.695492662473795,
      "grad_norm": 0.14195740222930908,
      "learning_rate": 4.206600843812886e-05,
      "loss": 0.5012,
      "num_input_tokens_seen": 8332072,
      "step": 12775
    },
    {
      "epoch": 6.69811320754717,
      "grad_norm": 0.1769505888223648,
      "learning_rate": 4.205765099800767e-05,
      "loss": 0.4988,
      "num_input_tokens_seen": 8334760,
      "step": 12780
    },
    {
      "epoch": 6.700733752620545,
      "grad_norm": 0.17089691758155823,
      "learning_rate": 4.204928998962049e-05,
      "loss": 0.3899,
      "num_input_tokens_seen": 8337736,
      "step": 12785
    },
    {
      "epoch": 6.70335429769392,
      "grad_norm": 0.18184494972229004,
      "learning_rate": 4.204092541471636e-05,
      "loss": 0.394,
      "num_input_tokens_seen": 8340488,
      "step": 12790
    },
    {
      "epoch": 6.705974842767295,
      "grad_norm": 0.18565821647644043,
      "learning_rate": 4.203255727504504e-05,
      "loss": 0.5158,
      "num_input_tokens_seen": 8343176,
      "step": 12795
    },
    {
      "epoch": 6.7085953878406706,
      "grad_norm": 0.14410251379013062,
      "learning_rate": 4.202418557235706e-05,
      "loss": 0.6076,
      "num_input_tokens_seen": 8346376,
      "step": 12800
    },
    {
      "epoch": 6.711215932914046,
      "grad_norm": 0.13836821913719177,
      "learning_rate": 4.201581030840368e-05,
      "loss": 0.4725,
      "num_input_tokens_seen": 8349672,
      "step": 12805
    },
    {
      "epoch": 6.713836477987421,
      "grad_norm": 0.18031899631023407,
      "learning_rate": 4.200743148493691e-05,
      "loss": 0.5223,
      "num_input_tokens_seen": 8352232,
      "step": 12810
    },
    {
      "epoch": 6.716457023060797,
      "grad_norm": 0.11521094292402267,
      "learning_rate": 4.199904910370951e-05,
      "loss": 0.3895,
      "num_input_tokens_seen": 8355592,
      "step": 12815
    },
    {
      "epoch": 6.719077568134172,
      "grad_norm": 0.12983673810958862,
      "learning_rate": 4.1990663166474964e-05,
      "loss": 0.4112,
      "num_input_tokens_seen": 8357736,
      "step": 12820
    },
    {
      "epoch": 6.721698113207547,
      "grad_norm": 0.16847966611385345,
      "learning_rate": 4.1982273674987524e-05,
      "loss": 0.4171,
      "num_input_tokens_seen": 8362568,
      "step": 12825
    },
    {
      "epoch": 6.7243186582809225,
      "grad_norm": 0.1695111095905304,
      "learning_rate": 4.197388063100218e-05,
      "loss": 0.5301,
      "num_input_tokens_seen": 8366152,
      "step": 12830
    },
    {
      "epoch": 6.726939203354298,
      "grad_norm": 0.4127589464187622,
      "learning_rate": 4.196548403627465e-05,
      "loss": 0.5101,
      "num_input_tokens_seen": 8368456,
      "step": 12835
    },
    {
      "epoch": 6.729559748427673,
      "grad_norm": 0.2233847826719284,
      "learning_rate": 4.195708389256142e-05,
      "loss": 0.5313,
      "num_input_tokens_seen": 8371560,
      "step": 12840
    },
    {
      "epoch": 6.732180293501048,
      "grad_norm": 0.19851893186569214,
      "learning_rate": 4.194868020161969e-05,
      "loss": 0.4995,
      "num_input_tokens_seen": 8374792,
      "step": 12845
    },
    {
      "epoch": 6.734800838574423,
      "grad_norm": 0.191132590174675,
      "learning_rate": 4.1940272965207425e-05,
      "loss": 0.4869,
      "num_input_tokens_seen": 8377832,
      "step": 12850
    },
    {
      "epoch": 6.737421383647799,
      "grad_norm": 0.1269596368074417,
      "learning_rate": 4.1931862185083326e-05,
      "loss": 0.5488,
      "num_input_tokens_seen": 8380872,
      "step": 12855
    },
    {
      "epoch": 6.740041928721174,
      "grad_norm": 0.12181084603071213,
      "learning_rate": 4.192344786300682e-05,
      "loss": 0.4793,
      "num_input_tokens_seen": 8384264,
      "step": 12860
    },
    {
      "epoch": 6.7426624737945495,
      "grad_norm": 0.12226337939500809,
      "learning_rate": 4.19150300007381e-05,
      "loss": 0.5164,
      "num_input_tokens_seen": 8389064,
      "step": 12865
    },
    {
      "epoch": 6.745283018867925,
      "grad_norm": 0.19904249906539917,
      "learning_rate": 4.190660860003806e-05,
      "loss": 0.5676,
      "num_input_tokens_seen": 8391976,
      "step": 12870
    },
    {
      "epoch": 6.7479035639413,
      "grad_norm": 0.14139018952846527,
      "learning_rate": 4.189818366266841e-05,
      "loss": 0.3919,
      "num_input_tokens_seen": 8395176,
      "step": 12875
    },
    {
      "epoch": 6.750524109014675,
      "grad_norm": 0.14380507171154022,
      "learning_rate": 4.188975519039151e-05,
      "loss": 0.5645,
      "num_input_tokens_seen": 8398216,
      "step": 12880
    },
    {
      "epoch": 6.75314465408805,
      "grad_norm": 0.1377839744091034,
      "learning_rate": 4.188132318497051e-05,
      "loss": 0.5446,
      "num_input_tokens_seen": 8402120,
      "step": 12885
    },
    {
      "epoch": 6.755765199161425,
      "grad_norm": 0.15175776183605194,
      "learning_rate": 4.1872887648169304e-05,
      "loss": 0.5634,
      "num_input_tokens_seen": 8405736,
      "step": 12890
    },
    {
      "epoch": 6.7583857442348005,
      "grad_norm": 0.45238080620765686,
      "learning_rate": 4.186444858175248e-05,
      "loss": 0.6382,
      "num_input_tokens_seen": 8408808,
      "step": 12895
    },
    {
      "epoch": 6.761006289308176,
      "grad_norm": 0.17001928389072418,
      "learning_rate": 4.1856005987485424e-05,
      "loss": 0.4092,
      "num_input_tokens_seen": 8411400,
      "step": 12900
    },
    {
      "epoch": 6.763626834381551,
      "grad_norm": 0.14740006625652313,
      "learning_rate": 4.1847559867134215e-05,
      "loss": 0.3636,
      "num_input_tokens_seen": 8414632,
      "step": 12905
    },
    {
      "epoch": 6.766247379454927,
      "grad_norm": 0.2350921928882599,
      "learning_rate": 4.1839110222465693e-05,
      "loss": 0.5263,
      "num_input_tokens_seen": 8417256,
      "step": 12910
    },
    {
      "epoch": 6.768867924528302,
      "grad_norm": 0.12293503433465958,
      "learning_rate": 4.183065705524742e-05,
      "loss": 0.4189,
      "num_input_tokens_seen": 8420584,
      "step": 12915
    },
    {
      "epoch": 6.771488469601677,
      "grad_norm": 0.11996056139469147,
      "learning_rate": 4.1822200367247696e-05,
      "loss": 0.6195,
      "num_input_tokens_seen": 8423912,
      "step": 12920
    },
    {
      "epoch": 6.774109014675052,
      "grad_norm": 0.15073487162590027,
      "learning_rate": 4.181374016023559e-05,
      "loss": 0.3811,
      "num_input_tokens_seen": 8427368,
      "step": 12925
    },
    {
      "epoch": 6.776729559748428,
      "grad_norm": 0.15946979820728302,
      "learning_rate": 4.1805276435980854e-05,
      "loss": 0.4953,
      "num_input_tokens_seen": 8431176,
      "step": 12930
    },
    {
      "epoch": 6.779350104821803,
      "grad_norm": 0.15108047425746918,
      "learning_rate": 4.1796809196254014e-05,
      "loss": 0.4874,
      "num_input_tokens_seen": 8433768,
      "step": 12935
    },
    {
      "epoch": 6.781970649895178,
      "grad_norm": 0.16581344604492188,
      "learning_rate": 4.178833844282632e-05,
      "loss": 0.4108,
      "num_input_tokens_seen": 8437384,
      "step": 12940
    },
    {
      "epoch": 6.784591194968553,
      "grad_norm": 0.21706141531467438,
      "learning_rate": 4.177986417746976e-05,
      "loss": 0.4077,
      "num_input_tokens_seen": 8441192,
      "step": 12945
    },
    {
      "epoch": 6.787211740041929,
      "grad_norm": 0.1441412717103958,
      "learning_rate": 4.177138640195704e-05,
      "loss": 0.6347,
      "num_input_tokens_seen": 8446504,
      "step": 12950
    },
    {
      "epoch": 6.789832285115304,
      "grad_norm": 0.1743737757205963,
      "learning_rate": 4.176290511806162e-05,
      "loss": 0.422,
      "num_input_tokens_seen": 8449384,
      "step": 12955
    },
    {
      "epoch": 6.7924528301886795,
      "grad_norm": 0.26893100142478943,
      "learning_rate": 4.1754420327557706e-05,
      "loss": 0.4489,
      "num_input_tokens_seen": 8452456,
      "step": 12960
    },
    {
      "epoch": 6.795073375262055,
      "grad_norm": 0.18738923966884613,
      "learning_rate": 4.1745932032220195e-05,
      "loss": 0.5514,
      "num_input_tokens_seen": 8457000,
      "step": 12965
    },
    {
      "epoch": 6.79769392033543,
      "grad_norm": 0.24225828051567078,
      "learning_rate": 4.173744023382474e-05,
      "loss": 0.476,
      "num_input_tokens_seen": 8461576,
      "step": 12970
    },
    {
      "epoch": 6.800314465408805,
      "grad_norm": 0.19475102424621582,
      "learning_rate": 4.172894493414776e-05,
      "loss": 0.5055,
      "num_input_tokens_seen": 8464712,
      "step": 12975
    },
    {
      "epoch": 6.80293501048218,
      "grad_norm": 0.14982694387435913,
      "learning_rate": 4.172044613496634e-05,
      "loss": 0.4134,
      "num_input_tokens_seen": 8467880,
      "step": 12980
    },
    {
      "epoch": 6.805555555555555,
      "grad_norm": 0.16071869432926178,
      "learning_rate": 4.171194383805835e-05,
      "loss": 0.4303,
      "num_input_tokens_seen": 8471208,
      "step": 12985
    },
    {
      "epoch": 6.8081761006289305,
      "grad_norm": 0.1390659213066101,
      "learning_rate": 4.170343804520237e-05,
      "loss": 0.535,
      "num_input_tokens_seen": 8474248,
      "step": 12990
    },
    {
      "epoch": 6.810796645702306,
      "grad_norm": 0.15939861536026,
      "learning_rate": 4.169492875817771e-05,
      "loss": 0.4133,
      "num_input_tokens_seen": 8477064,
      "step": 12995
    },
    {
      "epoch": 6.813417190775681,
      "grad_norm": 0.28307026624679565,
      "learning_rate": 4.1686415978764415e-05,
      "loss": 0.552,
      "num_input_tokens_seen": 8479528,
      "step": 13000
    },
    {
      "epoch": 6.816037735849057,
      "grad_norm": 0.20144134759902954,
      "learning_rate": 4.167789970874327e-05,
      "loss": 0.3973,
      "num_input_tokens_seen": 8482856,
      "step": 13005
    },
    {
      "epoch": 6.818658280922432,
      "grad_norm": 0.12725830078125,
      "learning_rate": 4.166937994989578e-05,
      "loss": 0.5099,
      "num_input_tokens_seen": 8486312,
      "step": 13010
    },
    {
      "epoch": 6.821278825995807,
      "grad_norm": 0.1659134179353714,
      "learning_rate": 4.166085670400418e-05,
      "loss": 0.5465,
      "num_input_tokens_seen": 8489096,
      "step": 13015
    },
    {
      "epoch": 6.823899371069182,
      "grad_norm": 0.10940786451101303,
      "learning_rate": 4.165232997285143e-05,
      "loss": 0.5383,
      "num_input_tokens_seen": 8493448,
      "step": 13020
    },
    {
      "epoch": 6.826519916142558,
      "grad_norm": 0.12014258652925491,
      "learning_rate": 4.164379975822124e-05,
      "loss": 0.3724,
      "num_input_tokens_seen": 8496968,
      "step": 13025
    },
    {
      "epoch": 6.829140461215933,
      "grad_norm": 0.2254563421010971,
      "learning_rate": 4.1635266061898015e-05,
      "loss": 0.445,
      "num_input_tokens_seen": 8500616,
      "step": 13030
    },
    {
      "epoch": 6.831761006289308,
      "grad_norm": 0.20461198687553406,
      "learning_rate": 4.162672888566691e-05,
      "loss": 0.5123,
      "num_input_tokens_seen": 8504136,
      "step": 13035
    },
    {
      "epoch": 6.834381551362683,
      "grad_norm": 0.150532066822052,
      "learning_rate": 4.1618188231313816e-05,
      "loss": 0.5839,
      "num_input_tokens_seen": 8507464,
      "step": 13040
    },
    {
      "epoch": 6.837002096436059,
      "grad_norm": 0.15269242227077484,
      "learning_rate": 4.1609644100625335e-05,
      "loss": 0.5814,
      "num_input_tokens_seen": 8510216,
      "step": 13045
    },
    {
      "epoch": 6.839622641509434,
      "grad_norm": 0.17922627925872803,
      "learning_rate": 4.16010964953888e-05,
      "loss": 0.5571,
      "num_input_tokens_seen": 8513960,
      "step": 13050
    },
    {
      "epoch": 6.8422431865828095,
      "grad_norm": 0.15009452402591705,
      "learning_rate": 4.159254541739227e-05,
      "loss": 0.4875,
      "num_input_tokens_seen": 8517768,
      "step": 13055
    },
    {
      "epoch": 6.844863731656185,
      "grad_norm": 0.12666770815849304,
      "learning_rate": 4.158399086842454e-05,
      "loss": 0.5844,
      "num_input_tokens_seen": 8521640,
      "step": 13060
    },
    {
      "epoch": 6.84748427672956,
      "grad_norm": 0.15448406338691711,
      "learning_rate": 4.157543285027512e-05,
      "loss": 0.4652,
      "num_input_tokens_seen": 8524968,
      "step": 13065
    },
    {
      "epoch": 6.850104821802935,
      "grad_norm": 0.246290922164917,
      "learning_rate": 4.156687136473425e-05,
      "loss": 0.7011,
      "num_input_tokens_seen": 8528200,
      "step": 13070
    },
    {
      "epoch": 6.85272536687631,
      "grad_norm": 0.19454558193683624,
      "learning_rate": 4.1558306413592894e-05,
      "loss": 0.5027,
      "num_input_tokens_seen": 8531144,
      "step": 13075
    },
    {
      "epoch": 6.855345911949685,
      "grad_norm": 0.2285558581352234,
      "learning_rate": 4.1549737998642745e-05,
      "loss": 0.5152,
      "num_input_tokens_seen": 8535240,
      "step": 13080
    },
    {
      "epoch": 6.8579664570230605,
      "grad_norm": 0.12782268226146698,
      "learning_rate": 4.154116612167622e-05,
      "loss": 0.5549,
      "num_input_tokens_seen": 8537352,
      "step": 13085
    },
    {
      "epoch": 6.860587002096436,
      "grad_norm": 0.16040568053722382,
      "learning_rate": 4.1532590784486444e-05,
      "loss": 0.4377,
      "num_input_tokens_seen": 8540776,
      "step": 13090
    },
    {
      "epoch": 6.863207547169811,
      "grad_norm": 0.1764349788427353,
      "learning_rate": 4.1524011988867294e-05,
      "loss": 0.5486,
      "num_input_tokens_seen": 8543944,
      "step": 13095
    },
    {
      "epoch": 6.865828092243187,
      "grad_norm": 0.2078891396522522,
      "learning_rate": 4.151542973661336e-05,
      "loss": 0.5117,
      "num_input_tokens_seen": 8546984,
      "step": 13100
    },
    {
      "epoch": 6.868448637316562,
      "grad_norm": 0.241987407207489,
      "learning_rate": 4.150684402951994e-05,
      "loss": 0.8371,
      "num_input_tokens_seen": 8550376,
      "step": 13105
    },
    {
      "epoch": 6.871069182389937,
      "grad_norm": 0.20469245314598083,
      "learning_rate": 4.149825486938306e-05,
      "loss": 0.5376,
      "num_input_tokens_seen": 8553512,
      "step": 13110
    },
    {
      "epoch": 6.873689727463312,
      "grad_norm": 0.12512046098709106,
      "learning_rate": 4.148966225799949e-05,
      "loss": 0.4676,
      "num_input_tokens_seen": 8557320,
      "step": 13115
    },
    {
      "epoch": 6.876310272536688,
      "grad_norm": 0.19731883704662323,
      "learning_rate": 4.14810661971667e-05,
      "loss": 0.4551,
      "num_input_tokens_seen": 8559848,
      "step": 13120
    },
    {
      "epoch": 6.878930817610063,
      "grad_norm": 0.10995720326900482,
      "learning_rate": 4.147246668868289e-05,
      "loss": 0.3301,
      "num_input_tokens_seen": 8563400,
      "step": 13125
    },
    {
      "epoch": 6.881551362683438,
      "grad_norm": 0.14202900230884552,
      "learning_rate": 4.1463863734346964e-05,
      "loss": 0.4732,
      "num_input_tokens_seen": 8566728,
      "step": 13130
    },
    {
      "epoch": 6.884171907756813,
      "grad_norm": 0.20559698343276978,
      "learning_rate": 4.145525733595859e-05,
      "loss": 0.4494,
      "num_input_tokens_seen": 8570056,
      "step": 13135
    },
    {
      "epoch": 6.886792452830189,
      "grad_norm": 0.18005584180355072,
      "learning_rate": 4.14466474953181e-05,
      "loss": 0.4605,
      "num_input_tokens_seen": 8573064,
      "step": 13140
    },
    {
      "epoch": 6.889412997903564,
      "grad_norm": 0.15832652151584625,
      "learning_rate": 4.14380342142266e-05,
      "loss": 0.6063,
      "num_input_tokens_seen": 8577640,
      "step": 13145
    },
    {
      "epoch": 6.8920335429769395,
      "grad_norm": 0.18445985019207,
      "learning_rate": 4.142941749448587e-05,
      "loss": 0.624,
      "num_input_tokens_seen": 8581256,
      "step": 13150
    },
    {
      "epoch": 6.894654088050315,
      "grad_norm": 0.17532506585121155,
      "learning_rate": 4.142079733789843e-05,
      "loss": 0.4304,
      "num_input_tokens_seen": 8584168,
      "step": 13155
    },
    {
      "epoch": 6.89727463312369,
      "grad_norm": 0.12329734861850739,
      "learning_rate": 4.141217374626753e-05,
      "loss": 0.5209,
      "num_input_tokens_seen": 8587432,
      "step": 13160
    },
    {
      "epoch": 6.899895178197065,
      "grad_norm": 0.1272217482328415,
      "learning_rate": 4.140354672139713e-05,
      "loss": 0.5407,
      "num_input_tokens_seen": 8590728,
      "step": 13165
    },
    {
      "epoch": 6.90251572327044,
      "grad_norm": 0.1758262813091278,
      "learning_rate": 4.1394916265091886e-05,
      "loss": 0.4237,
      "num_input_tokens_seen": 8594120,
      "step": 13170
    },
    {
      "epoch": 6.905136268343815,
      "grad_norm": 0.13892902433872223,
      "learning_rate": 4.1386282379157214e-05,
      "loss": 0.5139,
      "num_input_tokens_seen": 8596904,
      "step": 13175
    },
    {
      "epoch": 6.9077568134171905,
      "grad_norm": 0.12482967227697372,
      "learning_rate": 4.137764506539921e-05,
      "loss": 0.4386,
      "num_input_tokens_seen": 8599848,
      "step": 13180
    },
    {
      "epoch": 6.910377358490566,
      "grad_norm": 0.11666932702064514,
      "learning_rate": 4.136900432562471e-05,
      "loss": 0.4536,
      "num_input_tokens_seen": 8603272,
      "step": 13185
    },
    {
      "epoch": 6.912997903563941,
      "grad_norm": 0.12284085154533386,
      "learning_rate": 4.136036016164123e-05,
      "loss": 0.4686,
      "num_input_tokens_seen": 8607240,
      "step": 13190
    },
    {
      "epoch": 6.915618448637317,
      "grad_norm": 0.39728331565856934,
      "learning_rate": 4.135171257525707e-05,
      "loss": 0.6697,
      "num_input_tokens_seen": 8613160,
      "step": 13195
    },
    {
      "epoch": 6.918238993710692,
      "grad_norm": 0.1440892517566681,
      "learning_rate": 4.1343061568281195e-05,
      "loss": 0.5138,
      "num_input_tokens_seen": 8617480,
      "step": 13200
    },
    {
      "epoch": 6.920859538784067,
      "grad_norm": 0.1452832669019699,
      "learning_rate": 4.133440714252328e-05,
      "loss": 0.4932,
      "num_input_tokens_seen": 8620520,
      "step": 13205
    },
    {
      "epoch": 6.923480083857442,
      "grad_norm": 0.15498147904872894,
      "learning_rate": 4.132574929979375e-05,
      "loss": 0.3593,
      "num_input_tokens_seen": 8626568,
      "step": 13210
    },
    {
      "epoch": 6.926100628930818,
      "grad_norm": 0.1269749104976654,
      "learning_rate": 4.1317088041903714e-05,
      "loss": 0.5234,
      "num_input_tokens_seen": 8630184,
      "step": 13215
    },
    {
      "epoch": 6.928721174004193,
      "grad_norm": 0.16238611936569214,
      "learning_rate": 4.1308423370665015e-05,
      "loss": 0.4574,
      "num_input_tokens_seen": 8633576,
      "step": 13220
    },
    {
      "epoch": 6.931341719077568,
      "grad_norm": 0.10683578997850418,
      "learning_rate": 4.1299755287890206e-05,
      "loss": 0.4907,
      "num_input_tokens_seen": 8636520,
      "step": 13225
    },
    {
      "epoch": 6.933962264150943,
      "grad_norm": 0.1395188719034195,
      "learning_rate": 4.129108379539255e-05,
      "loss": 0.3638,
      "num_input_tokens_seen": 8639816,
      "step": 13230
    },
    {
      "epoch": 6.936582809224319,
      "grad_norm": 0.20531854033470154,
      "learning_rate": 4.128240889498602e-05,
      "loss": 0.5459,
      "num_input_tokens_seen": 8642888,
      "step": 13235
    },
    {
      "epoch": 6.939203354297694,
      "grad_norm": 0.1719614863395691,
      "learning_rate": 4.127373058848531e-05,
      "loss": 0.5396,
      "num_input_tokens_seen": 8645704,
      "step": 13240
    },
    {
      "epoch": 6.9418238993710695,
      "grad_norm": 0.12915363907814026,
      "learning_rate": 4.126504887770582e-05,
      "loss": 0.4408,
      "num_input_tokens_seen": 8648648,
      "step": 13245
    },
    {
      "epoch": 6.944444444444445,
      "grad_norm": 0.18407829105854034,
      "learning_rate": 4.1256363764463666e-05,
      "loss": 0.4447,
      "num_input_tokens_seen": 8651560,
      "step": 13250
    },
    {
      "epoch": 6.94706498951782,
      "grad_norm": 0.11382641643285751,
      "learning_rate": 4.1247675250575676e-05,
      "loss": 0.5712,
      "num_input_tokens_seen": 8655208,
      "step": 13255
    },
    {
      "epoch": 6.949685534591195,
      "grad_norm": 0.1549016237258911,
      "learning_rate": 4.1238983337859396e-05,
      "loss": 0.3544,
      "num_input_tokens_seen": 8657576,
      "step": 13260
    },
    {
      "epoch": 6.95230607966457,
      "grad_norm": 0.22368991374969482,
      "learning_rate": 4.123028802813305e-05,
      "loss": 0.7364,
      "num_input_tokens_seen": 8660872,
      "step": 13265
    },
    {
      "epoch": 6.954926624737945,
      "grad_norm": 0.12002988159656525,
      "learning_rate": 4.1221589323215625e-05,
      "loss": 0.4864,
      "num_input_tokens_seen": 8665192,
      "step": 13270
    },
    {
      "epoch": 6.9575471698113205,
      "grad_norm": 0.08476640284061432,
      "learning_rate": 4.121288722492678e-05,
      "loss": 0.4481,
      "num_input_tokens_seen": 8670088,
      "step": 13275
    },
    {
      "epoch": 6.960167714884696,
      "grad_norm": 0.27841246128082275,
      "learning_rate": 4.120418173508688e-05,
      "loss": 0.5242,
      "num_input_tokens_seen": 8673576,
      "step": 13280
    },
    {
      "epoch": 6.962788259958071,
      "grad_norm": 0.16594508290290833,
      "learning_rate": 4.1195472855517045e-05,
      "loss": 0.6987,
      "num_input_tokens_seen": 8676424,
      "step": 13285
    },
    {
      "epoch": 6.965408805031447,
      "grad_norm": 0.16418541967868805,
      "learning_rate": 4.1186760588039045e-05,
      "loss": 0.4652,
      "num_input_tokens_seen": 8679912,
      "step": 13290
    },
    {
      "epoch": 6.968029350104822,
      "grad_norm": 0.10832539945840836,
      "learning_rate": 4.11780449344754e-05,
      "loss": 0.7433,
      "num_input_tokens_seen": 8683880,
      "step": 13295
    },
    {
      "epoch": 6.970649895178197,
      "grad_norm": 0.12712763249874115,
      "learning_rate": 4.1169325896649326e-05,
      "loss": 0.5733,
      "num_input_tokens_seen": 8687784,
      "step": 13300
    },
    {
      "epoch": 6.973270440251572,
      "grad_norm": 0.14156852662563324,
      "learning_rate": 4.116060347638474e-05,
      "loss": 0.4723,
      "num_input_tokens_seen": 8690920,
      "step": 13305
    },
    {
      "epoch": 6.975890985324948,
      "grad_norm": 0.2412731647491455,
      "learning_rate": 4.115187767550628e-05,
      "loss": 0.4096,
      "num_input_tokens_seen": 8693960,
      "step": 13310
    },
    {
      "epoch": 6.978511530398323,
      "grad_norm": 0.10989008843898773,
      "learning_rate": 4.114314849583928e-05,
      "loss": 0.5503,
      "num_input_tokens_seen": 8697640,
      "step": 13315
    },
    {
      "epoch": 6.981132075471698,
      "grad_norm": 0.11412730813026428,
      "learning_rate": 4.113441593920979e-05,
      "loss": 0.7179,
      "num_input_tokens_seen": 8699912,
      "step": 13320
    },
    {
      "epoch": 6.983752620545073,
      "grad_norm": 0.1672830581665039,
      "learning_rate": 4.112568000744454e-05,
      "loss": 0.6421,
      "num_input_tokens_seen": 8703240,
      "step": 13325
    },
    {
      "epoch": 6.986373165618449,
      "grad_norm": 0.15019944310188293,
      "learning_rate": 4.1116940702371025e-05,
      "loss": 0.3973,
      "num_input_tokens_seen": 8706376,
      "step": 13330
    },
    {
      "epoch": 6.988993710691824,
      "grad_norm": 0.16977445781230927,
      "learning_rate": 4.1108198025817376e-05,
      "loss": 0.6151,
      "num_input_tokens_seen": 8709224,
      "step": 13335
    },
    {
      "epoch": 6.9916142557651995,
      "grad_norm": 0.3348049223423004,
      "learning_rate": 4.1099451979612476e-05,
      "loss": 0.5747,
      "num_input_tokens_seen": 8711816,
      "step": 13340
    },
    {
      "epoch": 6.994234800838575,
      "grad_norm": 0.22382502257823944,
      "learning_rate": 4.109070256558589e-05,
      "loss": 0.5077,
      "num_input_tokens_seen": 8715496,
      "step": 13345
    },
    {
      "epoch": 6.99685534591195,
      "grad_norm": 0.14362528920173645,
      "learning_rate": 4.108194978556791e-05,
      "loss": 0.4126,
      "num_input_tokens_seen": 8718312,
      "step": 13350
    },
    {
      "epoch": 6.999475890985325,
      "grad_norm": 0.22272919118404388,
      "learning_rate": 4.1073193641389495e-05,
      "loss": 0.4821,
      "num_input_tokens_seen": 8722632,
      "step": 13355
    },
    {
      "epoch": 7.0,
      "eval_loss": 0.5057855248451233,
      "eval_runtime": 9.99,
      "eval_samples_per_second": 84.885,
      "eval_steps_per_second": 21.221,
      "num_input_tokens_seen": 8722744,
      "step": 13356
    },
    {
      "epoch": 7.0020964360587,
      "grad_norm": 0.2369641810655594,
      "learning_rate": 4.1064434134882345e-05,
      "loss": 0.5077,
      "num_input_tokens_seen": 8725944,
      "step": 13360
    },
    {
      "epoch": 7.004716981132075,
      "grad_norm": 0.137380450963974,
      "learning_rate": 4.1055671267878847e-05,
      "loss": 0.4932,
      "num_input_tokens_seen": 8729592,
      "step": 13365
    },
    {
      "epoch": 7.0073375262054505,
      "grad_norm": 0.164423868060112,
      "learning_rate": 4.1046905042212106e-05,
      "loss": 0.4429,
      "num_input_tokens_seen": 8734264,
      "step": 13370
    },
    {
      "epoch": 7.009958071278826,
      "grad_norm": 0.11640036851167679,
      "learning_rate": 4.103813545971589e-05,
      "loss": 0.4279,
      "num_input_tokens_seen": 8738744,
      "step": 13375
    },
    {
      "epoch": 7.012578616352202,
      "grad_norm": 0.1501612812280655,
      "learning_rate": 4.10293625222247e-05,
      "loss": 0.4304,
      "num_input_tokens_seen": 8741720,
      "step": 13380
    },
    {
      "epoch": 7.015199161425577,
      "grad_norm": 0.14716166257858276,
      "learning_rate": 4.102058623157376e-05,
      "loss": 0.4877,
      "num_input_tokens_seen": 8745112,
      "step": 13385
    },
    {
      "epoch": 7.017819706498952,
      "grad_norm": 0.37862083315849304,
      "learning_rate": 4.101180658959894e-05,
      "loss": 0.4602,
      "num_input_tokens_seen": 8747960,
      "step": 13390
    },
    {
      "epoch": 7.020440251572327,
      "grad_norm": 0.1840529441833496,
      "learning_rate": 4.1003023598136866e-05,
      "loss": 0.5256,
      "num_input_tokens_seen": 8750808,
      "step": 13395
    },
    {
      "epoch": 7.023060796645702,
      "grad_norm": 0.18153488636016846,
      "learning_rate": 4.099423725902482e-05,
      "loss": 0.5079,
      "num_input_tokens_seen": 8753368,
      "step": 13400
    },
    {
      "epoch": 7.0256813417190775,
      "grad_norm": 0.29958948493003845,
      "learning_rate": 4.098544757410081e-05,
      "loss": 0.5462,
      "num_input_tokens_seen": 8756056,
      "step": 13405
    },
    {
      "epoch": 7.028301886792453,
      "grad_norm": 0.175140842795372,
      "learning_rate": 4.097665454520354e-05,
      "loss": 0.4744,
      "num_input_tokens_seen": 8759256,
      "step": 13410
    },
    {
      "epoch": 7.030922431865828,
      "grad_norm": 0.30073580145835876,
      "learning_rate": 4.096785817417241e-05,
      "loss": 0.4321,
      "num_input_tokens_seen": 8763032,
      "step": 13415
    },
    {
      "epoch": 7.033542976939203,
      "grad_norm": 0.2729150950908661,
      "learning_rate": 4.0959058462847524e-05,
      "loss": 0.3735,
      "num_input_tokens_seen": 8766072,
      "step": 13420
    },
    {
      "epoch": 7.036163522012578,
      "grad_norm": 0.3270089030265808,
      "learning_rate": 4.095025541306968e-05,
      "loss": 0.5046,
      "num_input_tokens_seen": 8769976,
      "step": 13425
    },
    {
      "epoch": 7.038784067085954,
      "grad_norm": 0.16096433997154236,
      "learning_rate": 4.0941449026680374e-05,
      "loss": 0.5571,
      "num_input_tokens_seen": 8773528,
      "step": 13430
    },
    {
      "epoch": 7.0414046121593294,
      "grad_norm": 0.3318123519420624,
      "learning_rate": 4.09326393055218e-05,
      "loss": 0.5347,
      "num_input_tokens_seen": 8776440,
      "step": 13435
    },
    {
      "epoch": 7.044025157232705,
      "grad_norm": 0.12555162608623505,
      "learning_rate": 4.0923826251436845e-05,
      "loss": 0.369,
      "num_input_tokens_seen": 8780408,
      "step": 13440
    },
    {
      "epoch": 7.04664570230608,
      "grad_norm": 0.14893433451652527,
      "learning_rate": 4.0915009866269115e-05,
      "loss": 0.4358,
      "num_input_tokens_seen": 8782872,
      "step": 13445
    },
    {
      "epoch": 7.049266247379455,
      "grad_norm": 0.12033006548881531,
      "learning_rate": 4.090619015186288e-05,
      "loss": 0.4712,
      "num_input_tokens_seen": 8786776,
      "step": 13450
    },
    {
      "epoch": 7.05188679245283,
      "grad_norm": 0.15413697063922882,
      "learning_rate": 4.0897367110063145e-05,
      "loss": 0.5537,
      "num_input_tokens_seen": 8791576,
      "step": 13455
    },
    {
      "epoch": 7.054507337526205,
      "grad_norm": 0.18639330565929413,
      "learning_rate": 4.088854074271557e-05,
      "loss": 0.4544,
      "num_input_tokens_seen": 8794072,
      "step": 13460
    },
    {
      "epoch": 7.0571278825995805,
      "grad_norm": 0.10928129404783249,
      "learning_rate": 4.0879711051666534e-05,
      "loss": 0.5044,
      "num_input_tokens_seen": 8798424,
      "step": 13465
    },
    {
      "epoch": 7.059748427672956,
      "grad_norm": 0.2773078978061676,
      "learning_rate": 4.087087803876313e-05,
      "loss": 0.3927,
      "num_input_tokens_seen": 8801208,
      "step": 13470
    },
    {
      "epoch": 7.062368972746331,
      "grad_norm": 0.18343870341777802,
      "learning_rate": 4.086204170585308e-05,
      "loss": 0.5203,
      "num_input_tokens_seen": 8804472,
      "step": 13475
    },
    {
      "epoch": 7.064989517819707,
      "grad_norm": 0.13096268475055695,
      "learning_rate": 4.085320205478488e-05,
      "loss": 0.5511,
      "num_input_tokens_seen": 8808152,
      "step": 13480
    },
    {
      "epoch": 7.067610062893082,
      "grad_norm": 0.12536989152431488,
      "learning_rate": 4.084435908740768e-05,
      "loss": 0.4508,
      "num_input_tokens_seen": 8811352,
      "step": 13485
    },
    {
      "epoch": 7.070230607966457,
      "grad_norm": 0.43682366609573364,
      "learning_rate": 4.0835512805571306e-05,
      "loss": 0.508,
      "num_input_tokens_seen": 8814328,
      "step": 13490
    },
    {
      "epoch": 7.072851153039832,
      "grad_norm": 0.09625066071748734,
      "learning_rate": 4.082666321112633e-05,
      "loss": 0.5299,
      "num_input_tokens_seen": 8817976,
      "step": 13495
    },
    {
      "epoch": 7.0754716981132075,
      "grad_norm": 0.19525152444839478,
      "learning_rate": 4.081781030592396e-05,
      "loss": 0.4127,
      "num_input_tokens_seen": 8820856,
      "step": 13500
    },
    {
      "epoch": 7.078092243186583,
      "grad_norm": 0.16918151080608368,
      "learning_rate": 4.080895409181613e-05,
      "loss": 0.5984,
      "num_input_tokens_seen": 8824120,
      "step": 13505
    },
    {
      "epoch": 7.080712788259958,
      "grad_norm": 0.14439554512500763,
      "learning_rate": 4.080009457065547e-05,
      "loss": 0.5207,
      "num_input_tokens_seen": 8827192,
      "step": 13510
    },
    {
      "epoch": 7.083333333333333,
      "grad_norm": 0.24653105437755585,
      "learning_rate": 4.0791231744295286e-05,
      "loss": 0.4161,
      "num_input_tokens_seen": 8831064,
      "step": 13515
    },
    {
      "epoch": 7.085953878406708,
      "grad_norm": 0.12946996092796326,
      "learning_rate": 4.078236561458957e-05,
      "loss": 0.5719,
      "num_input_tokens_seen": 8834200,
      "step": 13520
    },
    {
      "epoch": 7.088574423480084,
      "grad_norm": 0.17009997367858887,
      "learning_rate": 4.077349618339302e-05,
      "loss": 0.4074,
      "num_input_tokens_seen": 8837240,
      "step": 13525
    },
    {
      "epoch": 7.091194968553459,
      "grad_norm": 0.1192779541015625,
      "learning_rate": 4.076462345256103e-05,
      "loss": 0.6462,
      "num_input_tokens_seen": 8841976,
      "step": 13530
    },
    {
      "epoch": 7.093815513626835,
      "grad_norm": 0.2083342969417572,
      "learning_rate": 4.0755747423949655e-05,
      "loss": 0.4233,
      "num_input_tokens_seen": 8845336,
      "step": 13535
    },
    {
      "epoch": 7.09643605870021,
      "grad_norm": 0.1644333451986313,
      "learning_rate": 4.074686809941568e-05,
      "loss": 0.4661,
      "num_input_tokens_seen": 8848184,
      "step": 13540
    },
    {
      "epoch": 7.099056603773585,
      "grad_norm": 0.4134347438812256,
      "learning_rate": 4.0737985480816546e-05,
      "loss": 0.549,
      "num_input_tokens_seen": 8850840,
      "step": 13545
    },
    {
      "epoch": 7.10167714884696,
      "grad_norm": 0.3265760838985443,
      "learning_rate": 4.0729099570010395e-05,
      "loss": 0.624,
      "num_input_tokens_seen": 8854296,
      "step": 13550
    },
    {
      "epoch": 7.104297693920335,
      "grad_norm": 0.16467256844043732,
      "learning_rate": 4.072021036885607e-05,
      "loss": 0.4498,
      "num_input_tokens_seen": 8857560,
      "step": 13555
    },
    {
      "epoch": 7.1069182389937104,
      "grad_norm": 0.09552530199289322,
      "learning_rate": 4.0711317879213075e-05,
      "loss": 0.4666,
      "num_input_tokens_seen": 8861496,
      "step": 13560
    },
    {
      "epoch": 7.109538784067086,
      "grad_norm": 0.1503831446170807,
      "learning_rate": 4.070242210294162e-05,
      "loss": 0.4907,
      "num_input_tokens_seen": 8864344,
      "step": 13565
    },
    {
      "epoch": 7.112159329140461,
      "grad_norm": 0.10474760830402374,
      "learning_rate": 4.069352304190261e-05,
      "loss": 0.5633,
      "num_input_tokens_seen": 8868152,
      "step": 13570
    },
    {
      "epoch": 7.114779874213837,
      "grad_norm": 0.15460126101970673,
      "learning_rate": 4.068462069795761e-05,
      "loss": 0.3548,
      "num_input_tokens_seen": 8870776,
      "step": 13575
    },
    {
      "epoch": 7.117400419287212,
      "grad_norm": 0.23513440787792206,
      "learning_rate": 4.067571507296891e-05,
      "loss": 0.59,
      "num_input_tokens_seen": 8873944,
      "step": 13580
    },
    {
      "epoch": 7.120020964360587,
      "grad_norm": 0.36747387051582336,
      "learning_rate": 4.0666806168799455e-05,
      "loss": 0.5389,
      "num_input_tokens_seen": 8876792,
      "step": 13585
    },
    {
      "epoch": 7.122641509433962,
      "grad_norm": 0.3417227864265442,
      "learning_rate": 4.0657893987312876e-05,
      "loss": 0.6153,
      "num_input_tokens_seen": 8879672,
      "step": 13590
    },
    {
      "epoch": 7.1252620545073375,
      "grad_norm": 0.2507050335407257,
      "learning_rate": 4.064897853037352e-05,
      "loss": 0.386,
      "num_input_tokens_seen": 8882904,
      "step": 13595
    },
    {
      "epoch": 7.127882599580713,
      "grad_norm": 0.18929414451122284,
      "learning_rate": 4.064005979984639e-05,
      "loss": 0.4031,
      "num_input_tokens_seen": 8886328,
      "step": 13600
    },
    {
      "epoch": 7.130503144654088,
      "grad_norm": 0.3465249538421631,
      "learning_rate": 4.0631137797597176e-05,
      "loss": 0.482,
      "num_input_tokens_seen": 8888888,
      "step": 13605
    },
    {
      "epoch": 7.133123689727463,
      "grad_norm": 0.16177856922149658,
      "learning_rate": 4.0622212525492266e-05,
      "loss": 0.4866,
      "num_input_tokens_seen": 8892536,
      "step": 13610
    },
    {
      "epoch": 7.135744234800838,
      "grad_norm": 0.12678876519203186,
      "learning_rate": 4.061328398539872e-05,
      "loss": 0.4467,
      "num_input_tokens_seen": 8895768,
      "step": 13615
    },
    {
      "epoch": 7.138364779874214,
      "grad_norm": 0.15033555030822754,
      "learning_rate": 4.060435217918429e-05,
      "loss": 0.4762,
      "num_input_tokens_seen": 8899320,
      "step": 13620
    },
    {
      "epoch": 7.140985324947589,
      "grad_norm": 0.22290277481079102,
      "learning_rate": 4.0595417108717405e-05,
      "loss": 0.4775,
      "num_input_tokens_seen": 8901816,
      "step": 13625
    },
    {
      "epoch": 7.143605870020965,
      "grad_norm": 0.11305152624845505,
      "learning_rate": 4.0586478775867185e-05,
      "loss": 0.396,
      "num_input_tokens_seen": 8904984,
      "step": 13630
    },
    {
      "epoch": 7.14622641509434,
      "grad_norm": 0.14555801451206207,
      "learning_rate": 4.057753718250343e-05,
      "loss": 0.46,
      "num_input_tokens_seen": 8909272,
      "step": 13635
    },
    {
      "epoch": 7.148846960167715,
      "grad_norm": 0.13638164103031158,
      "learning_rate": 4.056859233049661e-05,
      "loss": 0.5506,
      "num_input_tokens_seen": 8912088,
      "step": 13640
    },
    {
      "epoch": 7.15146750524109,
      "grad_norm": 0.14862920343875885,
      "learning_rate": 4.05596442217179e-05,
      "loss": 0.5663,
      "num_input_tokens_seen": 8915160,
      "step": 13645
    },
    {
      "epoch": 7.154088050314465,
      "grad_norm": 0.11300987750291824,
      "learning_rate": 4.055069285803912e-05,
      "loss": 0.4358,
      "num_input_tokens_seen": 8918200,
      "step": 13650
    },
    {
      "epoch": 7.15670859538784,
      "grad_norm": 0.11787806451320648,
      "learning_rate": 4.0541738241332815e-05,
      "loss": 0.3946,
      "num_input_tokens_seen": 8921464,
      "step": 13655
    },
    {
      "epoch": 7.159329140461216,
      "grad_norm": 0.17802219092845917,
      "learning_rate": 4.053278037347217e-05,
      "loss": 0.5003,
      "num_input_tokens_seen": 8924344,
      "step": 13660
    },
    {
      "epoch": 7.161949685534591,
      "grad_norm": 0.1937159150838852,
      "learning_rate": 4.052381925633109e-05,
      "loss": 0.471,
      "num_input_tokens_seen": 8927352,
      "step": 13665
    },
    {
      "epoch": 7.164570230607967,
      "grad_norm": 0.15598060190677643,
      "learning_rate": 4.051485489178412e-05,
      "loss": 0.5415,
      "num_input_tokens_seen": 8930936,
      "step": 13670
    },
    {
      "epoch": 7.167190775681342,
      "grad_norm": 0.2954428493976593,
      "learning_rate": 4.05058872817065e-05,
      "loss": 0.5569,
      "num_input_tokens_seen": 8934296,
      "step": 13675
    },
    {
      "epoch": 7.169811320754717,
      "grad_norm": 0.11587727069854736,
      "learning_rate": 4.0496916427974177e-05,
      "loss": 0.3615,
      "num_input_tokens_seen": 8937016,
      "step": 13680
    },
    {
      "epoch": 7.172431865828092,
      "grad_norm": 0.15589061379432678,
      "learning_rate": 4.048794233246373e-05,
      "loss": 0.4024,
      "num_input_tokens_seen": 8940792,
      "step": 13685
    },
    {
      "epoch": 7.1750524109014675,
      "grad_norm": 0.13163655996322632,
      "learning_rate": 4.0478964997052435e-05,
      "loss": 0.5416,
      "num_input_tokens_seen": 8944024,
      "step": 13690
    },
    {
      "epoch": 7.177672955974843,
      "grad_norm": 0.20546801388263702,
      "learning_rate": 4.046998442361827e-05,
      "loss": 0.4939,
      "num_input_tokens_seen": 8946872,
      "step": 13695
    },
    {
      "epoch": 7.180293501048218,
      "grad_norm": 0.13815580308437347,
      "learning_rate": 4.046100061403984e-05,
      "loss": 0.4687,
      "num_input_tokens_seen": 8950168,
      "step": 13700
    },
    {
      "epoch": 7.182914046121593,
      "grad_norm": 0.2436678260564804,
      "learning_rate": 4.045201357019648e-05,
      "loss": 0.4888,
      "num_input_tokens_seen": 8953112,
      "step": 13705
    },
    {
      "epoch": 7.185534591194968,
      "grad_norm": 0.22633856534957886,
      "learning_rate": 4.044302329396816e-05,
      "loss": 0.4095,
      "num_input_tokens_seen": 8956312,
      "step": 13710
    },
    {
      "epoch": 7.188155136268344,
      "grad_norm": 0.18366019427776337,
      "learning_rate": 4.0434029787235564e-05,
      "loss": 0.436,
      "num_input_tokens_seen": 8959032,
      "step": 13715
    },
    {
      "epoch": 7.190775681341719,
      "grad_norm": 0.20838242769241333,
      "learning_rate": 4.042503305188e-05,
      "loss": 0.5995,
      "num_input_tokens_seen": 8962744,
      "step": 13720
    },
    {
      "epoch": 7.193396226415095,
      "grad_norm": 0.12322830408811569,
      "learning_rate": 4.041603308978351e-05,
      "loss": 0.506,
      "num_input_tokens_seen": 8966520,
      "step": 13725
    },
    {
      "epoch": 7.19601677148847,
      "grad_norm": 0.1750705987215042,
      "learning_rate": 4.040702990282878e-05,
      "loss": 0.5706,
      "num_input_tokens_seen": 8969304,
      "step": 13730
    },
    {
      "epoch": 7.198637316561845,
      "grad_norm": 0.09935373812913895,
      "learning_rate": 4.039802349289914e-05,
      "loss": 0.5466,
      "num_input_tokens_seen": 8973432,
      "step": 13735
    },
    {
      "epoch": 7.20125786163522,
      "grad_norm": 0.1816687285900116,
      "learning_rate": 4.038901386187868e-05,
      "loss": 0.539,
      "num_input_tokens_seen": 8976248,
      "step": 13740
    },
    {
      "epoch": 7.203878406708595,
      "grad_norm": 0.15258489549160004,
      "learning_rate": 4.038000101165208e-05,
      "loss": 0.4812,
      "num_input_tokens_seen": 8980792,
      "step": 13745
    },
    {
      "epoch": 7.20649895178197,
      "grad_norm": 0.1778855174779892,
      "learning_rate": 4.0370984944104725e-05,
      "loss": 0.6111,
      "num_input_tokens_seen": 8983480,
      "step": 13750
    },
    {
      "epoch": 7.209119496855346,
      "grad_norm": 0.2675851583480835,
      "learning_rate": 4.0361965661122695e-05,
      "loss": 0.5393,
      "num_input_tokens_seen": 8986200,
      "step": 13755
    },
    {
      "epoch": 7.211740041928721,
      "grad_norm": 0.17385247349739075,
      "learning_rate": 4.035294316459271e-05,
      "loss": 0.5061,
      "num_input_tokens_seen": 8989816,
      "step": 13760
    },
    {
      "epoch": 7.214360587002097,
      "grad_norm": 0.11995985358953476,
      "learning_rate": 4.034391745640215e-05,
      "loss": 0.32,
      "num_input_tokens_seen": 8992600,
      "step": 13765
    },
    {
      "epoch": 7.216981132075472,
      "grad_norm": 0.16276727616786957,
      "learning_rate": 4.033488853843913e-05,
      "loss": 0.5053,
      "num_input_tokens_seen": 8997432,
      "step": 13770
    },
    {
      "epoch": 7.219601677148847,
      "grad_norm": 0.27375009655952454,
      "learning_rate": 4.0325856412592363e-05,
      "loss": 0.5497,
      "num_input_tokens_seen": 8999864,
      "step": 13775
    },
    {
      "epoch": 7.222222222222222,
      "grad_norm": 0.09908919036388397,
      "learning_rate": 4.031682108075129e-05,
      "loss": 0.4251,
      "num_input_tokens_seen": 9003160,
      "step": 13780
    },
    {
      "epoch": 7.2248427672955975,
      "grad_norm": 0.2169523686170578,
      "learning_rate": 4.030778254480599e-05,
      "loss": 0.4749,
      "num_input_tokens_seen": 9006040,
      "step": 13785
    },
    {
      "epoch": 7.227463312368973,
      "grad_norm": 0.1471545249223709,
      "learning_rate": 4.029874080664722e-05,
      "loss": 0.4617,
      "num_input_tokens_seen": 9009624,
      "step": 13790
    },
    {
      "epoch": 7.230083857442348,
      "grad_norm": 0.13261619210243225,
      "learning_rate": 4.0289695868166414e-05,
      "loss": 0.4824,
      "num_input_tokens_seen": 9013912,
      "step": 13795
    },
    {
      "epoch": 7.232704402515723,
      "grad_norm": 0.1905713826417923,
      "learning_rate": 4.028064773125566e-05,
      "loss": 0.4647,
      "num_input_tokens_seen": 9017016,
      "step": 13800
    },
    {
      "epoch": 7.235324947589098,
      "grad_norm": 0.2743663489818573,
      "learning_rate": 4.0271596397807734e-05,
      "loss": 0.4504,
      "num_input_tokens_seen": 9019000,
      "step": 13805
    },
    {
      "epoch": 7.237945492662474,
      "grad_norm": 0.20848765969276428,
      "learning_rate": 4.026254186971607e-05,
      "loss": 0.5762,
      "num_input_tokens_seen": 9022520,
      "step": 13810
    },
    {
      "epoch": 7.240566037735849,
      "grad_norm": 0.19204892218112946,
      "learning_rate": 4.025348414887477e-05,
      "loss": 0.5143,
      "num_input_tokens_seen": 9025496,
      "step": 13815
    },
    {
      "epoch": 7.243186582809225,
      "grad_norm": 0.15397082269191742,
      "learning_rate": 4.024442323717861e-05,
      "loss": 0.4765,
      "num_input_tokens_seen": 9029464,
      "step": 13820
    },
    {
      "epoch": 7.2458071278826,
      "grad_norm": 0.2187599390745163,
      "learning_rate": 4.023535913652302e-05,
      "loss": 0.5667,
      "num_input_tokens_seen": 9032088,
      "step": 13825
    },
    {
      "epoch": 7.248427672955975,
      "grad_norm": 0.10332349687814713,
      "learning_rate": 4.022629184880412e-05,
      "loss": 0.4424,
      "num_input_tokens_seen": 9035768,
      "step": 13830
    },
    {
      "epoch": 7.25104821802935,
      "grad_norm": 0.16857126355171204,
      "learning_rate": 4.021722137591867e-05,
      "loss": 0.4905,
      "num_input_tokens_seen": 9038840,
      "step": 13835
    },
    {
      "epoch": 7.253668763102725,
      "grad_norm": 0.17230692505836487,
      "learning_rate": 4.020814771976412e-05,
      "loss": 0.3692,
      "num_input_tokens_seen": 9042328,
      "step": 13840
    },
    {
      "epoch": 7.2562893081761,
      "grad_norm": 0.2880427837371826,
      "learning_rate": 4.019907088223857e-05,
      "loss": 0.4994,
      "num_input_tokens_seen": 9045336,
      "step": 13845
    },
    {
      "epoch": 7.258909853249476,
      "grad_norm": 0.1968982070684433,
      "learning_rate": 4.018999086524079e-05,
      "loss": 0.5675,
      "num_input_tokens_seen": 9048312,
      "step": 13850
    },
    {
      "epoch": 7.261530398322851,
      "grad_norm": 0.13985410332679749,
      "learning_rate": 4.018090767067022e-05,
      "loss": 0.6104,
      "num_input_tokens_seen": 9052760,
      "step": 13855
    },
    {
      "epoch": 7.264150943396227,
      "grad_norm": 0.34456494450569153,
      "learning_rate": 4.017182130042696e-05,
      "loss": 0.7234,
      "num_input_tokens_seen": 9055864,
      "step": 13860
    },
    {
      "epoch": 7.266771488469602,
      "grad_norm": 0.2543192505836487,
      "learning_rate": 4.0162731756411764e-05,
      "loss": 0.5065,
      "num_input_tokens_seen": 9058776,
      "step": 13865
    },
    {
      "epoch": 7.269392033542977,
      "grad_norm": 0.1751331239938736,
      "learning_rate": 4.015363904052608e-05,
      "loss": 0.6337,
      "num_input_tokens_seen": 9062360,
      "step": 13870
    },
    {
      "epoch": 7.272012578616352,
      "grad_norm": 0.12095390260219574,
      "learning_rate": 4.014454315467199e-05,
      "loss": 0.6101,
      "num_input_tokens_seen": 9065016,
      "step": 13875
    },
    {
      "epoch": 7.2746331236897275,
      "grad_norm": 0.1986192911863327,
      "learning_rate": 4.013544410075224e-05,
      "loss": 0.5125,
      "num_input_tokens_seen": 9071992,
      "step": 13880
    },
    {
      "epoch": 7.277253668763103,
      "grad_norm": 0.12365567684173584,
      "learning_rate": 4.0126341880670273e-05,
      "loss": 0.5255,
      "num_input_tokens_seen": 9075672,
      "step": 13885
    },
    {
      "epoch": 7.279874213836478,
      "grad_norm": 0.16266365349292755,
      "learning_rate": 4.011723649633014e-05,
      "loss": 0.4483,
      "num_input_tokens_seen": 9078936,
      "step": 13890
    },
    {
      "epoch": 7.282494758909853,
      "grad_norm": 0.24378827214241028,
      "learning_rate": 4.0108127949636606e-05,
      "loss": 0.6798,
      "num_input_tokens_seen": 9082168,
      "step": 13895
    },
    {
      "epoch": 7.285115303983228,
      "grad_norm": 0.2705378830432892,
      "learning_rate": 4.009901624249507e-05,
      "loss": 0.6402,
      "num_input_tokens_seen": 9084664,
      "step": 13900
    },
    {
      "epoch": 7.287735849056604,
      "grad_norm": 0.10479357838630676,
      "learning_rate": 4.008990137681159e-05,
      "loss": 0.4636,
      "num_input_tokens_seen": 9087864,
      "step": 13905
    },
    {
      "epoch": 7.290356394129979,
      "grad_norm": 0.14520077407360077,
      "learning_rate": 4.00807833544929e-05,
      "loss": 0.5879,
      "num_input_tokens_seen": 9091416,
      "step": 13910
    },
    {
      "epoch": 7.2929769392033545,
      "grad_norm": 0.2229585498571396,
      "learning_rate": 4.0071662177446376e-05,
      "loss": 0.4709,
      "num_input_tokens_seen": 9094296,
      "step": 13915
    },
    {
      "epoch": 7.29559748427673,
      "grad_norm": 0.20791618525981903,
      "learning_rate": 4.006253784758008e-05,
      "loss": 0.5562,
      "num_input_tokens_seen": 9097112,
      "step": 13920
    },
    {
      "epoch": 7.298218029350105,
      "grad_norm": 0.161110520362854,
      "learning_rate": 4.0053410366802704e-05,
      "loss": 0.4762,
      "num_input_tokens_seen": 9100632,
      "step": 13925
    },
    {
      "epoch": 7.30083857442348,
      "grad_norm": 0.22143281996250153,
      "learning_rate": 4.004427973702362e-05,
      "loss": 0.4389,
      "num_input_tokens_seen": 9103864,
      "step": 13930
    },
    {
      "epoch": 7.303459119496855,
      "grad_norm": 0.1610436737537384,
      "learning_rate": 4.003514596015285e-05,
      "loss": 0.4624,
      "num_input_tokens_seen": 9107928,
      "step": 13935
    },
    {
      "epoch": 7.30607966457023,
      "grad_norm": 0.17244653403759003,
      "learning_rate": 4.002600903810108e-05,
      "loss": 0.4087,
      "num_input_tokens_seen": 9111384,
      "step": 13940
    },
    {
      "epoch": 7.308700209643606,
      "grad_norm": 0.18173056840896606,
      "learning_rate": 4.001686897277963e-05,
      "loss": 0.4665,
      "num_input_tokens_seen": 9114488,
      "step": 13945
    },
    {
      "epoch": 7.311320754716981,
      "grad_norm": 0.11087478697299957,
      "learning_rate": 4.000772576610052e-05,
      "loss": 0.5232,
      "num_input_tokens_seen": 9117208,
      "step": 13950
    },
    {
      "epoch": 7.313941299790357,
      "grad_norm": 0.16549961268901825,
      "learning_rate": 3.99985794199764e-05,
      "loss": 0.5071,
      "num_input_tokens_seen": 9121176,
      "step": 13955
    },
    {
      "epoch": 7.316561844863732,
      "grad_norm": 0.22925050556659698,
      "learning_rate": 3.998942993632058e-05,
      "loss": 0.5995,
      "num_input_tokens_seen": 9124216,
      "step": 13960
    },
    {
      "epoch": 7.319182389937107,
      "grad_norm": 0.19052913784980774,
      "learning_rate": 3.998027731704702e-05,
      "loss": 0.5356,
      "num_input_tokens_seen": 9126936,
      "step": 13965
    },
    {
      "epoch": 7.321802935010482,
      "grad_norm": 0.14161501824855804,
      "learning_rate": 3.9971121564070354e-05,
      "loss": 0.5572,
      "num_input_tokens_seen": 9129688,
      "step": 13970
    },
    {
      "epoch": 7.3244234800838575,
      "grad_norm": 0.1963827759027481,
      "learning_rate": 3.996196267930585e-05,
      "loss": 0.4449,
      "num_input_tokens_seen": 9132152,
      "step": 13975
    },
    {
      "epoch": 7.327044025157233,
      "grad_norm": 0.20386679470539093,
      "learning_rate": 3.995280066466945e-05,
      "loss": 0.5749,
      "num_input_tokens_seen": 9135352,
      "step": 13980
    },
    {
      "epoch": 7.329664570230608,
      "grad_norm": 0.12230800837278366,
      "learning_rate": 3.9943635522077746e-05,
      "loss": 0.6306,
      "num_input_tokens_seen": 9139064,
      "step": 13985
    },
    {
      "epoch": 7.332285115303983,
      "grad_norm": 0.17453762888908386,
      "learning_rate": 3.993446725344797e-05,
      "loss": 0.4417,
      "num_input_tokens_seen": 9141624,
      "step": 13990
    },
    {
      "epoch": 7.334905660377358,
      "grad_norm": 0.16215865314006805,
      "learning_rate": 3.992529586069803e-05,
      "loss": 0.4148,
      "num_input_tokens_seen": 9144120,
      "step": 13995
    },
    {
      "epoch": 7.337526205450734,
      "grad_norm": 0.23297211527824402,
      "learning_rate": 3.991612134574648e-05,
      "loss": 0.4315,
      "num_input_tokens_seen": 9147384,
      "step": 14000
    },
    {
      "epoch": 7.340146750524109,
      "grad_norm": 0.10537437349557877,
      "learning_rate": 3.99069437105125e-05,
      "loss": 0.4533,
      "num_input_tokens_seen": 9150744,
      "step": 14005
    },
    {
      "epoch": 7.3427672955974845,
      "grad_norm": 0.1351461410522461,
      "learning_rate": 3.9897762956915976e-05,
      "loss": 0.4851,
      "num_input_tokens_seen": 9153560,
      "step": 14010
    },
    {
      "epoch": 7.34538784067086,
      "grad_norm": 0.15932606160640717,
      "learning_rate": 3.9888579086877403e-05,
      "loss": 0.3498,
      "num_input_tokens_seen": 9156696,
      "step": 14015
    },
    {
      "epoch": 7.348008385744235,
      "grad_norm": 0.19201664626598358,
      "learning_rate": 3.987939210231794e-05,
      "loss": 0.4273,
      "num_input_tokens_seen": 9159416,
      "step": 14020
    },
    {
      "epoch": 7.35062893081761,
      "grad_norm": 0.16438254714012146,
      "learning_rate": 3.987020200515941e-05,
      "loss": 0.3892,
      "num_input_tokens_seen": 9162584,
      "step": 14025
    },
    {
      "epoch": 7.353249475890985,
      "grad_norm": 0.1652875691652298,
      "learning_rate": 3.9861008797324255e-05,
      "loss": 0.4575,
      "num_input_tokens_seen": 9165848,
      "step": 14030
    },
    {
      "epoch": 7.35587002096436,
      "grad_norm": 0.18034614622592926,
      "learning_rate": 3.9851812480735605e-05,
      "loss": 0.4353,
      "num_input_tokens_seen": 9168600,
      "step": 14035
    },
    {
      "epoch": 7.3584905660377355,
      "grad_norm": 0.15733616054058075,
      "learning_rate": 3.9842613057317235e-05,
      "loss": 0.4017,
      "num_input_tokens_seen": 9173336,
      "step": 14040
    },
    {
      "epoch": 7.361111111111111,
      "grad_norm": 0.18251726031303406,
      "learning_rate": 3.9833410528993534e-05,
      "loss": 0.5062,
      "num_input_tokens_seen": 9176376,
      "step": 14045
    },
    {
      "epoch": 7.363731656184487,
      "grad_norm": 0.1751348376274109,
      "learning_rate": 3.9824204897689585e-05,
      "loss": 0.5145,
      "num_input_tokens_seen": 9180088,
      "step": 14050
    },
    {
      "epoch": 7.366352201257862,
      "grad_norm": 0.38061049580574036,
      "learning_rate": 3.98149961653311e-05,
      "loss": 0.4548,
      "num_input_tokens_seen": 9183064,
      "step": 14055
    },
    {
      "epoch": 7.368972746331237,
      "grad_norm": 0.20884519815444946,
      "learning_rate": 3.980578433384442e-05,
      "loss": 0.4813,
      "num_input_tokens_seen": 9186040,
      "step": 14060
    },
    {
      "epoch": 7.371593291404612,
      "grad_norm": 0.15717215836048126,
      "learning_rate": 3.979656940515659e-05,
      "loss": 0.475,
      "num_input_tokens_seen": 9189080,
      "step": 14065
    },
    {
      "epoch": 7.3742138364779874,
      "grad_norm": 0.12382405251264572,
      "learning_rate": 3.9787351381195235e-05,
      "loss": 0.6154,
      "num_input_tokens_seen": 9191992,
      "step": 14070
    },
    {
      "epoch": 7.376834381551363,
      "grad_norm": 0.1918151080608368,
      "learning_rate": 3.977813026388868e-05,
      "loss": 0.5068,
      "num_input_tokens_seen": 9194520,
      "step": 14075
    },
    {
      "epoch": 7.379454926624738,
      "grad_norm": 0.2503295838832855,
      "learning_rate": 3.976890605516587e-05,
      "loss": 0.5095,
      "num_input_tokens_seen": 9196632,
      "step": 14080
    },
    {
      "epoch": 7.382075471698113,
      "grad_norm": 0.19071894884109497,
      "learning_rate": 3.97596787569564e-05,
      "loss": 0.5585,
      "num_input_tokens_seen": 9199832,
      "step": 14085
    },
    {
      "epoch": 7.384696016771488,
      "grad_norm": 0.15438112616539001,
      "learning_rate": 3.975044837119053e-05,
      "loss": 0.4915,
      "num_input_tokens_seen": 9203384,
      "step": 14090
    },
    {
      "epoch": 7.387316561844864,
      "grad_norm": 0.38030245900154114,
      "learning_rate": 3.974121489979914e-05,
      "loss": 0.553,
      "num_input_tokens_seen": 9206840,
      "step": 14095
    },
    {
      "epoch": 7.389937106918239,
      "grad_norm": 0.2712450921535492,
      "learning_rate": 3.973197834471377e-05,
      "loss": 0.6562,
      "num_input_tokens_seen": 9209432,
      "step": 14100
    },
    {
      "epoch": 7.3925576519916145,
      "grad_norm": 0.13886003196239471,
      "learning_rate": 3.9722738707866595e-05,
      "loss": 0.4648,
      "num_input_tokens_seen": 9212056,
      "step": 14105
    },
    {
      "epoch": 7.39517819706499,
      "grad_norm": 0.13540644943714142,
      "learning_rate": 3.971349599119046e-05,
      "loss": 0.3971,
      "num_input_tokens_seen": 9215224,
      "step": 14110
    },
    {
      "epoch": 7.397798742138365,
      "grad_norm": 0.09997647255659103,
      "learning_rate": 3.970425019661881e-05,
      "loss": 0.4023,
      "num_input_tokens_seen": 9219672,
      "step": 14115
    },
    {
      "epoch": 7.40041928721174,
      "grad_norm": 0.26534661650657654,
      "learning_rate": 3.9695001326085785e-05,
      "loss": 0.4967,
      "num_input_tokens_seen": 9222616,
      "step": 14120
    },
    {
      "epoch": 7.403039832285115,
      "grad_norm": 0.15475668013095856,
      "learning_rate": 3.968574938152613e-05,
      "loss": 0.7322,
      "num_input_tokens_seen": 9227032,
      "step": 14125
    },
    {
      "epoch": 7.40566037735849,
      "grad_norm": 0.38589203357696533,
      "learning_rate": 3.9676494364875247e-05,
      "loss": 0.557,
      "num_input_tokens_seen": 9229688,
      "step": 14130
    },
    {
      "epoch": 7.4082809224318655,
      "grad_norm": 0.1516973078250885,
      "learning_rate": 3.966723627806918e-05,
      "loss": 0.5251,
      "num_input_tokens_seen": 9233016,
      "step": 14135
    },
    {
      "epoch": 7.410901467505241,
      "grad_norm": 0.14896747469902039,
      "learning_rate": 3.9657975123044624e-05,
      "loss": 0.4818,
      "num_input_tokens_seen": 9236504,
      "step": 14140
    },
    {
      "epoch": 7.413522012578617,
      "grad_norm": 0.16520848870277405,
      "learning_rate": 3.964871090173888e-05,
      "loss": 0.4685,
      "num_input_tokens_seen": 9240344,
      "step": 14145
    },
    {
      "epoch": 7.416142557651992,
      "grad_norm": 0.15939462184906006,
      "learning_rate": 3.963944361608995e-05,
      "loss": 0.4849,
      "num_input_tokens_seen": 9242904,
      "step": 14150
    },
    {
      "epoch": 7.418763102725367,
      "grad_norm": 0.25582125782966614,
      "learning_rate": 3.963017326803644e-05,
      "loss": 0.5938,
      "num_input_tokens_seen": 9246072,
      "step": 14155
    },
    {
      "epoch": 7.421383647798742,
      "grad_norm": 0.4033554792404175,
      "learning_rate": 3.962089985951758e-05,
      "loss": 0.5127,
      "num_input_tokens_seen": 9248696,
      "step": 14160
    },
    {
      "epoch": 7.424004192872117,
      "grad_norm": 0.14948010444641113,
      "learning_rate": 3.961162339247327e-05,
      "loss": 0.6761,
      "num_input_tokens_seen": 9251160,
      "step": 14165
    },
    {
      "epoch": 7.426624737945493,
      "grad_norm": 0.1352875530719757,
      "learning_rate": 3.9602343868844055e-05,
      "loss": 0.5367,
      "num_input_tokens_seen": 9254040,
      "step": 14170
    },
    {
      "epoch": 7.429245283018868,
      "grad_norm": 0.10703811049461365,
      "learning_rate": 3.959306129057109e-05,
      "loss": 0.5134,
      "num_input_tokens_seen": 9257816,
      "step": 14175
    },
    {
      "epoch": 7.431865828092243,
      "grad_norm": 0.16646799445152283,
      "learning_rate": 3.958377565959619e-05,
      "loss": 0.4194,
      "num_input_tokens_seen": 9260184,
      "step": 14180
    },
    {
      "epoch": 7.434486373165618,
      "grad_norm": 0.26212647557258606,
      "learning_rate": 3.95744869778618e-05,
      "loss": 0.6016,
      "num_input_tokens_seen": 9263032,
      "step": 14185
    },
    {
      "epoch": 7.437106918238994,
      "grad_norm": 0.14796128869056702,
      "learning_rate": 3.956519524731101e-05,
      "loss": 0.5707,
      "num_input_tokens_seen": 9267448,
      "step": 14190
    },
    {
      "epoch": 7.439727463312369,
      "grad_norm": 0.1853507161140442,
      "learning_rate": 3.9555900469887555e-05,
      "loss": 0.4536,
      "num_input_tokens_seen": 9270200,
      "step": 14195
    },
    {
      "epoch": 7.4423480083857445,
      "grad_norm": 0.17234648764133453,
      "learning_rate": 3.954660264753577e-05,
      "loss": 0.4785,
      "num_input_tokens_seen": 9272856,
      "step": 14200
    },
    {
      "epoch": 7.44496855345912,
      "grad_norm": 0.11929885298013687,
      "learning_rate": 3.953730178220067e-05,
      "loss": 0.4643,
      "num_input_tokens_seen": 9276504,
      "step": 14205
    },
    {
      "epoch": 7.447589098532495,
      "grad_norm": 0.1422407180070877,
      "learning_rate": 3.952799787582789e-05,
      "loss": 0.5436,
      "num_input_tokens_seen": 9279736,
      "step": 14210
    },
    {
      "epoch": 7.45020964360587,
      "grad_norm": 0.23915161192417145,
      "learning_rate": 3.9518690930363694e-05,
      "loss": 0.4353,
      "num_input_tokens_seen": 9282680,
      "step": 14215
    },
    {
      "epoch": 7.452830188679245,
      "grad_norm": 0.21135158836841583,
      "learning_rate": 3.9509380947754994e-05,
      "loss": 0.4828,
      "num_input_tokens_seen": 9285912,
      "step": 14220
    },
    {
      "epoch": 7.45545073375262,
      "grad_norm": 0.18740375339984894,
      "learning_rate": 3.9500067929949336e-05,
      "loss": 0.3936,
      "num_input_tokens_seen": 9289176,
      "step": 14225
    },
    {
      "epoch": 7.4580712788259955,
      "grad_norm": 0.2223196178674698,
      "learning_rate": 3.9490751878894895e-05,
      "loss": 0.4587,
      "num_input_tokens_seen": 9291736,
      "step": 14230
    },
    {
      "epoch": 7.460691823899371,
      "grad_norm": 0.20769789814949036,
      "learning_rate": 3.948143279654047e-05,
      "loss": 0.4799,
      "num_input_tokens_seen": 9296312,
      "step": 14235
    },
    {
      "epoch": 7.463312368972747,
      "grad_norm": 0.24582630395889282,
      "learning_rate": 3.947211068483553e-05,
      "loss": 0.4456,
      "num_input_tokens_seen": 9298904,
      "step": 14240
    },
    {
      "epoch": 7.465932914046122,
      "grad_norm": 0.13890546560287476,
      "learning_rate": 3.946278554573013e-05,
      "loss": 0.4975,
      "num_input_tokens_seen": 9301432,
      "step": 14245
    },
    {
      "epoch": 7.468553459119497,
      "grad_norm": 0.15515267848968506,
      "learning_rate": 3.9453457381175e-05,
      "loss": 0.5197,
      "num_input_tokens_seen": 9304376,
      "step": 14250
    },
    {
      "epoch": 7.471174004192872,
      "grad_norm": 0.16672436892986298,
      "learning_rate": 3.9444126193121474e-05,
      "loss": 0.4741,
      "num_input_tokens_seen": 9307864,
      "step": 14255
    },
    {
      "epoch": 7.473794549266247,
      "grad_norm": 0.1951926201581955,
      "learning_rate": 3.943479198352153e-05,
      "loss": 0.5474,
      "num_input_tokens_seen": 9311032,
      "step": 14260
    },
    {
      "epoch": 7.476415094339623,
      "grad_norm": 0.11328805983066559,
      "learning_rate": 3.94254547543278e-05,
      "loss": 0.3704,
      "num_input_tokens_seen": 9314264,
      "step": 14265
    },
    {
      "epoch": 7.479035639412998,
      "grad_norm": 0.1329527348279953,
      "learning_rate": 3.9416114507493504e-05,
      "loss": 0.4074,
      "num_input_tokens_seen": 9317272,
      "step": 14270
    },
    {
      "epoch": 7.481656184486373,
      "grad_norm": 0.2939611077308655,
      "learning_rate": 3.940677124497252e-05,
      "loss": 0.5079,
      "num_input_tokens_seen": 9320504,
      "step": 14275
    },
    {
      "epoch": 7.484276729559748,
      "grad_norm": 0.15981575846672058,
      "learning_rate": 3.939742496871935e-05,
      "loss": 0.4806,
      "num_input_tokens_seen": 9324792,
      "step": 14280
    },
    {
      "epoch": 7.486897274633124,
      "grad_norm": 0.10011264681816101,
      "learning_rate": 3.9388075680689136e-05,
      "loss": 0.4121,
      "num_input_tokens_seen": 9327544,
      "step": 14285
    },
    {
      "epoch": 7.489517819706499,
      "grad_norm": 0.1852010041475296,
      "learning_rate": 3.9378723382837634e-05,
      "loss": 0.543,
      "num_input_tokens_seen": 9330488,
      "step": 14290
    },
    {
      "epoch": 7.4921383647798745,
      "grad_norm": 0.13864444196224213,
      "learning_rate": 3.936936807712124e-05,
      "loss": 0.5519,
      "num_input_tokens_seen": 9335288,
      "step": 14295
    },
    {
      "epoch": 7.49475890985325,
      "grad_norm": 0.1163550391793251,
      "learning_rate": 3.936000976549698e-05,
      "loss": 0.4626,
      "num_input_tokens_seen": 9339384,
      "step": 14300
    },
    {
      "epoch": 7.497379454926625,
      "grad_norm": 0.14219306409358978,
      "learning_rate": 3.935064844992251e-05,
      "loss": 0.613,
      "num_input_tokens_seen": 9342872,
      "step": 14305
    },
    {
      "epoch": 7.5,
      "grad_norm": 0.13920888304710388,
      "learning_rate": 3.93412841323561e-05,
      "loss": 0.4508,
      "num_input_tokens_seen": 9345976,
      "step": 14310
    },
    {
      "epoch": 7.502620545073375,
      "grad_norm": 0.22267094254493713,
      "learning_rate": 3.933191681475666e-05,
      "loss": 0.4371,
      "num_input_tokens_seen": 9349208,
      "step": 14315
    },
    {
      "epoch": 7.50524109014675,
      "grad_norm": 0.15520869195461273,
      "learning_rate": 3.932254649908373e-05,
      "loss": 0.4667,
      "num_input_tokens_seen": 9352184,
      "step": 14320
    },
    {
      "epoch": 7.5078616352201255,
      "grad_norm": 0.16940587759017944,
      "learning_rate": 3.931317318729748e-05,
      "loss": 0.5231,
      "num_input_tokens_seen": 9356216,
      "step": 14325
    },
    {
      "epoch": 7.510482180293501,
      "grad_norm": 0.2452559620141983,
      "learning_rate": 3.930379688135869e-05,
      "loss": 0.4938,
      "num_input_tokens_seen": 9359544,
      "step": 14330
    },
    {
      "epoch": 7.513102725366876,
      "grad_norm": 0.13405098021030426,
      "learning_rate": 3.929441758322878e-05,
      "loss": 0.4554,
      "num_input_tokens_seen": 9362200,
      "step": 14335
    },
    {
      "epoch": 7.515723270440252,
      "grad_norm": 0.12553317844867706,
      "learning_rate": 3.928503529486979e-05,
      "loss": 0.4335,
      "num_input_tokens_seen": 9366520,
      "step": 14340
    },
    {
      "epoch": 7.518343815513627,
      "grad_norm": 0.12029028683900833,
      "learning_rate": 3.927565001824438e-05,
      "loss": 0.4131,
      "num_input_tokens_seen": 9372440,
      "step": 14345
    },
    {
      "epoch": 7.520964360587002,
      "grad_norm": 0.18426483869552612,
      "learning_rate": 3.926626175531586e-05,
      "loss": 0.3861,
      "num_input_tokens_seen": 9375064,
      "step": 14350
    },
    {
      "epoch": 7.523584905660377,
      "grad_norm": 0.1768079698085785,
      "learning_rate": 3.925687050804814e-05,
      "loss": 0.3944,
      "num_input_tokens_seen": 9377720,
      "step": 14355
    },
    {
      "epoch": 7.526205450733753,
      "grad_norm": 0.1028532013297081,
      "learning_rate": 3.924747627840575e-05,
      "loss": 0.4959,
      "num_input_tokens_seen": 9380984,
      "step": 14360
    },
    {
      "epoch": 7.528825995807128,
      "grad_norm": 0.2413407266139984,
      "learning_rate": 3.923807906835388e-05,
      "loss": 0.4122,
      "num_input_tokens_seen": 9383992,
      "step": 14365
    },
    {
      "epoch": 7.531446540880503,
      "grad_norm": 0.37556561827659607,
      "learning_rate": 3.92286788798583e-05,
      "loss": 0.5149,
      "num_input_tokens_seen": 9386360,
      "step": 14370
    },
    {
      "epoch": 7.534067085953878,
      "grad_norm": 0.1714393049478531,
      "learning_rate": 3.921927571488543e-05,
      "loss": 0.5596,
      "num_input_tokens_seen": 9389144,
      "step": 14375
    },
    {
      "epoch": 7.536687631027254,
      "grad_norm": 0.28539273142814636,
      "learning_rate": 3.9209869575402296e-05,
      "loss": 0.3685,
      "num_input_tokens_seen": 9392024,
      "step": 14380
    },
    {
      "epoch": 7.539308176100629,
      "grad_norm": 0.21632763743400574,
      "learning_rate": 3.9200460463376566e-05,
      "loss": 0.4702,
      "num_input_tokens_seen": 9394872,
      "step": 14385
    },
    {
      "epoch": 7.5419287211740045,
      "grad_norm": 0.1824803203344345,
      "learning_rate": 3.9191048380776515e-05,
      "loss": 0.6591,
      "num_input_tokens_seen": 9397304,
      "step": 14390
    },
    {
      "epoch": 7.54454926624738,
      "grad_norm": 0.16866914927959442,
      "learning_rate": 3.918163332957104e-05,
      "loss": 0.5415,
      "num_input_tokens_seen": 9400664,
      "step": 14395
    },
    {
      "epoch": 7.547169811320755,
      "grad_norm": 0.10764313489198685,
      "learning_rate": 3.9172215311729665e-05,
      "loss": 0.5522,
      "num_input_tokens_seen": 9403864,
      "step": 14400
    },
    {
      "epoch": 7.54979035639413,
      "grad_norm": 0.2679060697555542,
      "learning_rate": 3.916279432922254e-05,
      "loss": 0.4888,
      "num_input_tokens_seen": 9407544,
      "step": 14405
    },
    {
      "epoch": 7.552410901467505,
      "grad_norm": 0.10206279158592224,
      "learning_rate": 3.915337038402041e-05,
      "loss": 0.3943,
      "num_input_tokens_seen": 9410968,
      "step": 14410
    },
    {
      "epoch": 7.55503144654088,
      "grad_norm": 0.26685526967048645,
      "learning_rate": 3.914394347809467e-05,
      "loss": 0.5464,
      "num_input_tokens_seen": 9413400,
      "step": 14415
    },
    {
      "epoch": 7.5576519916142555,
      "grad_norm": 0.2641712427139282,
      "learning_rate": 3.9134513613417315e-05,
      "loss": 0.3948,
      "num_input_tokens_seen": 9415832,
      "step": 14420
    },
    {
      "epoch": 7.560272536687631,
      "grad_norm": 0.2012123018503189,
      "learning_rate": 3.912508079196097e-05,
      "loss": 0.5806,
      "num_input_tokens_seen": 9419064,
      "step": 14425
    },
    {
      "epoch": 7.562893081761006,
      "grad_norm": 0.13622990250587463,
      "learning_rate": 3.9115645015698866e-05,
      "loss": 0.5227,
      "num_input_tokens_seen": 9421752,
      "step": 14430
    },
    {
      "epoch": 7.565513626834382,
      "grad_norm": 0.18987759947776794,
      "learning_rate": 3.9106206286604865e-05,
      "loss": 0.5573,
      "num_input_tokens_seen": 9424984,
      "step": 14435
    },
    {
      "epoch": 7.568134171907757,
      "grad_norm": 0.139963760972023,
      "learning_rate": 3.909676460665345e-05,
      "loss": 0.4343,
      "num_input_tokens_seen": 9428216,
      "step": 14440
    },
    {
      "epoch": 7.570754716981132,
      "grad_norm": 0.21532203257083893,
      "learning_rate": 3.908731997781969e-05,
      "loss": 0.502,
      "num_input_tokens_seen": 9431576,
      "step": 14445
    },
    {
      "epoch": 7.573375262054507,
      "grad_norm": 0.27320876717567444,
      "learning_rate": 3.907787240207932e-05,
      "loss": 0.6192,
      "num_input_tokens_seen": 9434680,
      "step": 14450
    },
    {
      "epoch": 7.575995807127883,
      "grad_norm": 0.23002973198890686,
      "learning_rate": 3.9068421881408645e-05,
      "loss": 0.3158,
      "num_input_tokens_seen": 9437432,
      "step": 14455
    },
    {
      "epoch": 7.578616352201258,
      "grad_norm": 0.13273748755455017,
      "learning_rate": 3.9058968417784614e-05,
      "loss": 0.4134,
      "num_input_tokens_seen": 9441208,
      "step": 14460
    },
    {
      "epoch": 7.581236897274633,
      "grad_norm": 0.1727103292942047,
      "learning_rate": 3.9049512013184785e-05,
      "loss": 0.5067,
      "num_input_tokens_seen": 9444056,
      "step": 14465
    },
    {
      "epoch": 7.583857442348008,
      "grad_norm": 0.5892871022224426,
      "learning_rate": 3.9040052669587325e-05,
      "loss": 0.6265,
      "num_input_tokens_seen": 9446968,
      "step": 14470
    },
    {
      "epoch": 7.586477987421384,
      "grad_norm": 0.1298278421163559,
      "learning_rate": 3.9030590388971026e-05,
      "loss": 0.4999,
      "num_input_tokens_seen": 9450712,
      "step": 14475
    },
    {
      "epoch": 7.589098532494759,
      "grad_norm": 0.1181408241391182,
      "learning_rate": 3.902112517331529e-05,
      "loss": 0.4642,
      "num_input_tokens_seen": 9453240,
      "step": 14480
    },
    {
      "epoch": 7.5917190775681345,
      "grad_norm": 0.14714540541172028,
      "learning_rate": 3.901165702460012e-05,
      "loss": 0.4204,
      "num_input_tokens_seen": 9456728,
      "step": 14485
    },
    {
      "epoch": 7.59433962264151,
      "grad_norm": 0.17267128825187683,
      "learning_rate": 3.900218594480616e-05,
      "loss": 0.4821,
      "num_input_tokens_seen": 9459800,
      "step": 14490
    },
    {
      "epoch": 7.596960167714885,
      "grad_norm": 0.1090492233633995,
      "learning_rate": 3.899271193591465e-05,
      "loss": 0.4644,
      "num_input_tokens_seen": 9463384,
      "step": 14495
    },
    {
      "epoch": 7.59958071278826,
      "grad_norm": 0.17878051102161407,
      "learning_rate": 3.8983234999907435e-05,
      "loss": 0.4865,
      "num_input_tokens_seen": 9466520,
      "step": 14500
    },
    {
      "epoch": 7.602201257861635,
      "grad_norm": 0.18542629480361938,
      "learning_rate": 3.897375513876699e-05,
      "loss": 0.4,
      "num_input_tokens_seen": 9469944,
      "step": 14505
    },
    {
      "epoch": 7.60482180293501,
      "grad_norm": 1.2440954446792603,
      "learning_rate": 3.8964272354476396e-05,
      "loss": 0.5727,
      "num_input_tokens_seen": 9472280,
      "step": 14510
    },
    {
      "epoch": 7.6074423480083855,
      "grad_norm": 0.20260906219482422,
      "learning_rate": 3.895478664901933e-05,
      "loss": 0.4913,
      "num_input_tokens_seen": 9474808,
      "step": 14515
    },
    {
      "epoch": 7.610062893081761,
      "grad_norm": 0.14242053031921387,
      "learning_rate": 3.89452980243801e-05,
      "loss": 0.4913,
      "num_input_tokens_seen": 9479064,
      "step": 14520
    },
    {
      "epoch": 7.612683438155136,
      "grad_norm": 0.13093413412570953,
      "learning_rate": 3.893580648254362e-05,
      "loss": 0.5083,
      "num_input_tokens_seen": 9483128,
      "step": 14525
    },
    {
      "epoch": 7.615303983228512,
      "grad_norm": 0.1915416568517685,
      "learning_rate": 3.8926312025495414e-05,
      "loss": 0.4723,
      "num_input_tokens_seen": 9486136,
      "step": 14530
    },
    {
      "epoch": 7.617924528301887,
      "grad_norm": 0.32830575108528137,
      "learning_rate": 3.891681465522161e-05,
      "loss": 0.838,
      "num_input_tokens_seen": 9488568,
      "step": 14535
    },
    {
      "epoch": 7.620545073375262,
      "grad_norm": 0.19083169102668762,
      "learning_rate": 3.890731437370896e-05,
      "loss": 0.5082,
      "num_input_tokens_seen": 9492440,
      "step": 14540
    },
    {
      "epoch": 7.623165618448637,
      "grad_norm": 0.16593684256076813,
      "learning_rate": 3.88978111829448e-05,
      "loss": 0.4382,
      "num_input_tokens_seen": 9495192,
      "step": 14545
    },
    {
      "epoch": 7.6257861635220126,
      "grad_norm": 0.36214572191238403,
      "learning_rate": 3.88883050849171e-05,
      "loss": 0.5293,
      "num_input_tokens_seen": 9497688,
      "step": 14550
    },
    {
      "epoch": 7.628406708595388,
      "grad_norm": 0.14330947399139404,
      "learning_rate": 3.8878796081614417e-05,
      "loss": 0.4796,
      "num_input_tokens_seen": 9500984,
      "step": 14555
    },
    {
      "epoch": 7.631027253668763,
      "grad_norm": 0.11421045660972595,
      "learning_rate": 3.8869284175025935e-05,
      "loss": 0.4767,
      "num_input_tokens_seen": 9504152,
      "step": 14560
    },
    {
      "epoch": 7.633647798742138,
      "grad_norm": 0.11702099442481995,
      "learning_rate": 3.885976936714143e-05,
      "loss": 0.4861,
      "num_input_tokens_seen": 9506840,
      "step": 14565
    },
    {
      "epoch": 7.636268343815514,
      "grad_norm": 0.11602992564439774,
      "learning_rate": 3.885025165995129e-05,
      "loss": 0.5179,
      "num_input_tokens_seen": 9509880,
      "step": 14570
    },
    {
      "epoch": 7.638888888888889,
      "grad_norm": 0.264838844537735,
      "learning_rate": 3.884073105544652e-05,
      "loss": 0.4814,
      "num_input_tokens_seen": 9513464,
      "step": 14575
    },
    {
      "epoch": 7.6415094339622645,
      "grad_norm": 0.31222113966941833,
      "learning_rate": 3.8831207555618723e-05,
      "loss": 0.5492,
      "num_input_tokens_seen": 9516216,
      "step": 14580
    },
    {
      "epoch": 7.64412997903564,
      "grad_norm": 0.14767862856388092,
      "learning_rate": 3.882168116246009e-05,
      "loss": 0.3987,
      "num_input_tokens_seen": 9519032,
      "step": 14585
    },
    {
      "epoch": 7.646750524109015,
      "grad_norm": 0.08593698590993881,
      "learning_rate": 3.881215187796344e-05,
      "loss": 0.4062,
      "num_input_tokens_seen": 9523640,
      "step": 14590
    },
    {
      "epoch": 7.64937106918239,
      "grad_norm": 0.17450553178787231,
      "learning_rate": 3.8802619704122204e-05,
      "loss": 0.457,
      "num_input_tokens_seen": 9526616,
      "step": 14595
    },
    {
      "epoch": 7.651991614255765,
      "grad_norm": 0.14106734097003937,
      "learning_rate": 3.8793084642930387e-05,
      "loss": 0.5277,
      "num_input_tokens_seen": 9529656,
      "step": 14600
    },
    {
      "epoch": 7.65461215932914,
      "grad_norm": 0.13567566871643066,
      "learning_rate": 3.878354669638263e-05,
      "loss": 0.4411,
      "num_input_tokens_seen": 9532984,
      "step": 14605
    },
    {
      "epoch": 7.6572327044025155,
      "grad_norm": 0.351300984621048,
      "learning_rate": 3.8774005866474155e-05,
      "loss": 0.4674,
      "num_input_tokens_seen": 9535928,
      "step": 14610
    },
    {
      "epoch": 7.659853249475891,
      "grad_norm": 0.13071110844612122,
      "learning_rate": 3.876446215520079e-05,
      "loss": 0.5073,
      "num_input_tokens_seen": 9539256,
      "step": 14615
    },
    {
      "epoch": 7.662473794549266,
      "grad_norm": 0.11245118081569672,
      "learning_rate": 3.875491556455898e-05,
      "loss": 0.4727,
      "num_input_tokens_seen": 9542296,
      "step": 14620
    },
    {
      "epoch": 7.665094339622642,
      "grad_norm": 0.16074411571025848,
      "learning_rate": 3.8745366096545754e-05,
      "loss": 0.3973,
      "num_input_tokens_seen": 9545368,
      "step": 14625
    },
    {
      "epoch": 7.667714884696017,
      "grad_norm": 0.10209383070468903,
      "learning_rate": 3.873581375315876e-05,
      "loss": 0.461,
      "num_input_tokens_seen": 9548472,
      "step": 14630
    },
    {
      "epoch": 7.670335429769392,
      "grad_norm": 0.0989476889371872,
      "learning_rate": 3.872625853639623e-05,
      "loss": 0.4135,
      "num_input_tokens_seen": 9552248,
      "step": 14635
    },
    {
      "epoch": 7.672955974842767,
      "grad_norm": 0.13951142132282257,
      "learning_rate": 3.871670044825701e-05,
      "loss": 0.7258,
      "num_input_tokens_seen": 9555320,
      "step": 14640
    },
    {
      "epoch": 7.6755765199161425,
      "grad_norm": 0.2184307873249054,
      "learning_rate": 3.870713949074056e-05,
      "loss": 0.4976,
      "num_input_tokens_seen": 9558904,
      "step": 14645
    },
    {
      "epoch": 7.678197064989518,
      "grad_norm": 0.2978760004043579,
      "learning_rate": 3.86975756658469e-05,
      "loss": 0.553,
      "num_input_tokens_seen": 9561624,
      "step": 14650
    },
    {
      "epoch": 7.680817610062893,
      "grad_norm": 0.2721463441848755,
      "learning_rate": 3.868800897557667e-05,
      "loss": 0.6298,
      "num_input_tokens_seen": 9564568,
      "step": 14655
    },
    {
      "epoch": 7.683438155136268,
      "grad_norm": 0.17341560125350952,
      "learning_rate": 3.867843942193113e-05,
      "loss": 0.5713,
      "num_input_tokens_seen": 9567704,
      "step": 14660
    },
    {
      "epoch": 7.686058700209644,
      "grad_norm": 0.14522112905979156,
      "learning_rate": 3.866886700691212e-05,
      "loss": 0.4857,
      "num_input_tokens_seen": 9571512,
      "step": 14665
    },
    {
      "epoch": 7.688679245283019,
      "grad_norm": 0.1605389565229416,
      "learning_rate": 3.8659291732522075e-05,
      "loss": 0.5751,
      "num_input_tokens_seen": 9575448,
      "step": 14670
    },
    {
      "epoch": 7.691299790356394,
      "grad_norm": 0.19303956627845764,
      "learning_rate": 3.8649713600764026e-05,
      "loss": 0.2994,
      "num_input_tokens_seen": 9583704,
      "step": 14675
    },
    {
      "epoch": 7.69392033542977,
      "grad_norm": 0.17278699576854706,
      "learning_rate": 3.864013261364162e-05,
      "loss": 0.4747,
      "num_input_tokens_seen": 9586392,
      "step": 14680
    },
    {
      "epoch": 7.696540880503145,
      "grad_norm": 0.2019912600517273,
      "learning_rate": 3.8630548773159084e-05,
      "loss": 0.6262,
      "num_input_tokens_seen": 9589336,
      "step": 14685
    },
    {
      "epoch": 7.69916142557652,
      "grad_norm": 0.10949312895536423,
      "learning_rate": 3.862096208132125e-05,
      "loss": 0.4494,
      "num_input_tokens_seen": 9593016,
      "step": 14690
    },
    {
      "epoch": 7.701781970649895,
      "grad_norm": 0.2839570939540863,
      "learning_rate": 3.861137254013355e-05,
      "loss": 0.4498,
      "num_input_tokens_seen": 9595512,
      "step": 14695
    },
    {
      "epoch": 7.70440251572327,
      "grad_norm": 0.24105790257453918,
      "learning_rate": 3.8601780151601996e-05,
      "loss": 0.4973,
      "num_input_tokens_seen": 9598904,
      "step": 14700
    },
    {
      "epoch": 7.7070230607966455,
      "grad_norm": 0.12326765060424805,
      "learning_rate": 3.859218491773321e-05,
      "loss": 0.3254,
      "num_input_tokens_seen": 9602712,
      "step": 14705
    },
    {
      "epoch": 7.709643605870021,
      "grad_norm": 0.14395438134670258,
      "learning_rate": 3.858258684053441e-05,
      "loss": 0.489,
      "num_input_tokens_seen": 9605560,
      "step": 14710
    },
    {
      "epoch": 7.712264150943396,
      "grad_norm": 0.13472506403923035,
      "learning_rate": 3.8572985922013405e-05,
      "loss": 0.6097,
      "num_input_tokens_seen": 9609048,
      "step": 14715
    },
    {
      "epoch": 7.714884696016772,
      "grad_norm": 0.1934245526790619,
      "learning_rate": 3.856338216417859e-05,
      "loss": 0.4757,
      "num_input_tokens_seen": 9612824,
      "step": 14720
    },
    {
      "epoch": 7.717505241090147,
      "grad_norm": 0.09117812663316727,
      "learning_rate": 3.855377556903897e-05,
      "loss": 0.43,
      "num_input_tokens_seen": 9615704,
      "step": 14725
    },
    {
      "epoch": 7.720125786163522,
      "grad_norm": 0.1880381554365158,
      "learning_rate": 3.8544166138604124e-05,
      "loss": 0.5443,
      "num_input_tokens_seen": 9618840,
      "step": 14730
    },
    {
      "epoch": 7.722746331236897,
      "grad_norm": 0.12499641627073288,
      "learning_rate": 3.8534553874884244e-05,
      "loss": 0.4325,
      "num_input_tokens_seen": 9622552,
      "step": 14735
    },
    {
      "epoch": 7.7253668763102725,
      "grad_norm": 0.13251005113124847,
      "learning_rate": 3.852493877989011e-05,
      "loss": 0.6295,
      "num_input_tokens_seen": 9625720,
      "step": 14740
    },
    {
      "epoch": 7.727987421383648,
      "grad_norm": 0.16410739719867706,
      "learning_rate": 3.851532085563308e-05,
      "loss": 0.4241,
      "num_input_tokens_seen": 9628664,
      "step": 14745
    },
    {
      "epoch": 7.730607966457023,
      "grad_norm": 0.14712536334991455,
      "learning_rate": 3.850570010412511e-05,
      "loss": 0.451,
      "num_input_tokens_seen": 9632088,
      "step": 14750
    },
    {
      "epoch": 7.733228511530398,
      "grad_norm": 0.17426319420337677,
      "learning_rate": 3.8496076527378766e-05,
      "loss": 0.5269,
      "num_input_tokens_seen": 9635160,
      "step": 14755
    },
    {
      "epoch": 7.735849056603773,
      "grad_norm": 0.13824664056301117,
      "learning_rate": 3.848645012740718e-05,
      "loss": 0.4994,
      "num_input_tokens_seen": 9638776,
      "step": 14760
    },
    {
      "epoch": 7.738469601677149,
      "grad_norm": 0.22307994961738586,
      "learning_rate": 3.8476820906224096e-05,
      "loss": 0.5128,
      "num_input_tokens_seen": 9641624,
      "step": 14765
    },
    {
      "epoch": 7.741090146750524,
      "grad_norm": 0.13436448574066162,
      "learning_rate": 3.846718886584382e-05,
      "loss": 0.5636,
      "num_input_tokens_seen": 9649176,
      "step": 14770
    },
    {
      "epoch": 7.7437106918239,
      "grad_norm": 0.18427331745624542,
      "learning_rate": 3.845755400828128e-05,
      "loss": 0.7026,
      "num_input_tokens_seen": 9651768,
      "step": 14775
    },
    {
      "epoch": 7.746331236897275,
      "grad_norm": 0.14879272878170013,
      "learning_rate": 3.844791633555198e-05,
      "loss": 0.3227,
      "num_input_tokens_seen": 9654552,
      "step": 14780
    },
    {
      "epoch": 7.74895178197065,
      "grad_norm": 0.14643728733062744,
      "learning_rate": 3.8438275849671996e-05,
      "loss": 0.4836,
      "num_input_tokens_seen": 9657112,
      "step": 14785
    },
    {
      "epoch": 7.751572327044025,
      "grad_norm": 0.09858390688896179,
      "learning_rate": 3.842863255265802e-05,
      "loss": 0.4101,
      "num_input_tokens_seen": 9661176,
      "step": 14790
    },
    {
      "epoch": 7.7541928721174,
      "grad_norm": 0.17601850628852844,
      "learning_rate": 3.841898644652732e-05,
      "loss": 0.4666,
      "num_input_tokens_seen": 9663480,
      "step": 14795
    },
    {
      "epoch": 7.756813417190775,
      "grad_norm": 0.12263753265142441,
      "learning_rate": 3.840933753329774e-05,
      "loss": 0.482,
      "num_input_tokens_seen": 9666808,
      "step": 14800
    },
    {
      "epoch": 7.759433962264151,
      "grad_norm": 0.2696529030799866,
      "learning_rate": 3.839968581498773e-05,
      "loss": 0.6355,
      "num_input_tokens_seen": 9669336,
      "step": 14805
    },
    {
      "epoch": 7.762054507337526,
      "grad_norm": 0.18968410789966583,
      "learning_rate": 3.839003129361632e-05,
      "loss": 0.5933,
      "num_input_tokens_seen": 9671928,
      "step": 14810
    },
    {
      "epoch": 7.764675052410902,
      "grad_norm": 0.1080281212925911,
      "learning_rate": 3.8380373971203134e-05,
      "loss": 0.5668,
      "num_input_tokens_seen": 9676120,
      "step": 14815
    },
    {
      "epoch": 7.767295597484277,
      "grad_norm": 0.1770203858613968,
      "learning_rate": 3.8370713849768346e-05,
      "loss": 0.4042,
      "num_input_tokens_seen": 9679032,
      "step": 14820
    },
    {
      "epoch": 7.769916142557652,
      "grad_norm": 0.16581431031227112,
      "learning_rate": 3.8361050931332774e-05,
      "loss": 0.4678,
      "num_input_tokens_seen": 9682744,
      "step": 14825
    },
    {
      "epoch": 7.772536687631027,
      "grad_norm": 0.17430970072746277,
      "learning_rate": 3.835138521791778e-05,
      "loss": 0.6752,
      "num_input_tokens_seen": 9686040,
      "step": 14830
    },
    {
      "epoch": 7.7751572327044025,
      "grad_norm": 0.25070443749427795,
      "learning_rate": 3.8341716711545314e-05,
      "loss": 0.5042,
      "num_input_tokens_seen": 9690168,
      "step": 14835
    },
    {
      "epoch": 7.777777777777778,
      "grad_norm": 0.2624719440937042,
      "learning_rate": 3.833204541423793e-05,
      "loss": 0.5379,
      "num_input_tokens_seen": 9693496,
      "step": 14840
    },
    {
      "epoch": 7.780398322851153,
      "grad_norm": 0.1992136389017105,
      "learning_rate": 3.832237132801874e-05,
      "loss": 0.4757,
      "num_input_tokens_seen": 9696760,
      "step": 14845
    },
    {
      "epoch": 7.783018867924528,
      "grad_norm": 0.1713799685239792,
      "learning_rate": 3.831269445491146e-05,
      "loss": 0.5618,
      "num_input_tokens_seen": 9699672,
      "step": 14850
    },
    {
      "epoch": 7.785639412997903,
      "grad_norm": 0.12621672451496124,
      "learning_rate": 3.8303014796940375e-05,
      "loss": 0.4238,
      "num_input_tokens_seen": 9703384,
      "step": 14855
    },
    {
      "epoch": 7.788259958071279,
      "grad_norm": 0.10366907715797424,
      "learning_rate": 3.829333235613036e-05,
      "loss": 0.5176,
      "num_input_tokens_seen": 9706328,
      "step": 14860
    },
    {
      "epoch": 7.790880503144654,
      "grad_norm": 0.24439725279808044,
      "learning_rate": 3.828364713450688e-05,
      "loss": 0.5743,
      "num_input_tokens_seen": 9709432,
      "step": 14865
    },
    {
      "epoch": 7.79350104821803,
      "grad_norm": 0.2702085077762604,
      "learning_rate": 3.827395913409597e-05,
      "loss": 0.6031,
      "num_input_tokens_seen": 9712696,
      "step": 14870
    },
    {
      "epoch": 7.796121593291405,
      "grad_norm": 0.2509147822856903,
      "learning_rate": 3.826426835692424e-05,
      "loss": 0.5681,
      "num_input_tokens_seen": 9715704,
      "step": 14875
    },
    {
      "epoch": 7.79874213836478,
      "grad_norm": 0.16947922110557556,
      "learning_rate": 3.825457480501891e-05,
      "loss": 0.7239,
      "num_input_tokens_seen": 9720344,
      "step": 14880
    },
    {
      "epoch": 7.801362683438155,
      "grad_norm": 0.10318566113710403,
      "learning_rate": 3.824487848040773e-05,
      "loss": 0.4441,
      "num_input_tokens_seen": 9723544,
      "step": 14885
    },
    {
      "epoch": 7.80398322851153,
      "grad_norm": 0.1894834041595459,
      "learning_rate": 3.823517938511908e-05,
      "loss": 0.7327,
      "num_input_tokens_seen": 9726424,
      "step": 14890
    },
    {
      "epoch": 7.806603773584905,
      "grad_norm": 0.20336273312568665,
      "learning_rate": 3.8225477521181906e-05,
      "loss": 0.4734,
      "num_input_tokens_seen": 9729144,
      "step": 14895
    },
    {
      "epoch": 7.809224318658281,
      "grad_norm": 0.1252165585756302,
      "learning_rate": 3.821577289062571e-05,
      "loss": 0.4785,
      "num_input_tokens_seen": 9732504,
      "step": 14900
    },
    {
      "epoch": 7.811844863731656,
      "grad_norm": 0.15306644141674042,
      "learning_rate": 3.82060654954806e-05,
      "loss": 0.4655,
      "num_input_tokens_seen": 9737048,
      "step": 14905
    },
    {
      "epoch": 7.814465408805032,
      "grad_norm": 0.21409708261489868,
      "learning_rate": 3.819635533777726e-05,
      "loss": 0.5315,
      "num_input_tokens_seen": 9739960,
      "step": 14910
    },
    {
      "epoch": 7.817085953878407,
      "grad_norm": 0.16926659643650055,
      "learning_rate": 3.8186642419546924e-05,
      "loss": 0.4096,
      "num_input_tokens_seen": 9743288,
      "step": 14915
    },
    {
      "epoch": 7.819706498951782,
      "grad_norm": 0.20856575667858124,
      "learning_rate": 3.817692674282145e-05,
      "loss": 0.7115,
      "num_input_tokens_seen": 9746072,
      "step": 14920
    },
    {
      "epoch": 7.822327044025157,
      "grad_norm": 0.16318799555301666,
      "learning_rate": 3.816720830963322e-05,
      "loss": 0.5529,
      "num_input_tokens_seen": 9748920,
      "step": 14925
    },
    {
      "epoch": 7.8249475890985325,
      "grad_norm": 0.12212925404310226,
      "learning_rate": 3.815748712201523e-05,
      "loss": 0.3558,
      "num_input_tokens_seen": 9751864,
      "step": 14930
    },
    {
      "epoch": 7.827568134171908,
      "grad_norm": 0.13031019270420074,
      "learning_rate": 3.814776318200105e-05,
      "loss": 0.4547,
      "num_input_tokens_seen": 9755128,
      "step": 14935
    },
    {
      "epoch": 7.830188679245283,
      "grad_norm": 0.30202701687812805,
      "learning_rate": 3.8138036491624805e-05,
      "loss": 0.4019,
      "num_input_tokens_seen": 9757976,
      "step": 14940
    },
    {
      "epoch": 7.832809224318658,
      "grad_norm": 0.1084076315164566,
      "learning_rate": 3.812830705292122e-05,
      "loss": 0.4296,
      "num_input_tokens_seen": 9761464,
      "step": 14945
    },
    {
      "epoch": 7.835429769392033,
      "grad_norm": 0.23447051644325256,
      "learning_rate": 3.8118574867925564e-05,
      "loss": 0.4976,
      "num_input_tokens_seen": 9768408,
      "step": 14950
    },
    {
      "epoch": 7.838050314465409,
      "grad_norm": 0.13662725687026978,
      "learning_rate": 3.810883993867371e-05,
      "loss": 0.5464,
      "num_input_tokens_seen": 9771832,
      "step": 14955
    },
    {
      "epoch": 7.840670859538784,
      "grad_norm": 0.16294816136360168,
      "learning_rate": 3.80991022672021e-05,
      "loss": 0.5952,
      "num_input_tokens_seen": 9775736,
      "step": 14960
    },
    {
      "epoch": 7.84329140461216,
      "grad_norm": 0.17291755974292755,
      "learning_rate": 3.8089361855547744e-05,
      "loss": 0.5337,
      "num_input_tokens_seen": 9779096,
      "step": 14965
    },
    {
      "epoch": 7.845911949685535,
      "grad_norm": 0.25240465998649597,
      "learning_rate": 3.8079618705748214e-05,
      "loss": 0.7538,
      "num_input_tokens_seen": 9782264,
      "step": 14970
    },
    {
      "epoch": 7.84853249475891,
      "grad_norm": 0.15083010494709015,
      "learning_rate": 3.806987281984167e-05,
      "loss": 0.5183,
      "num_input_tokens_seen": 9785368,
      "step": 14975
    },
    {
      "epoch": 7.851153039832285,
      "grad_norm": 0.25374361872673035,
      "learning_rate": 3.806012419986684e-05,
      "loss": 0.4107,
      "num_input_tokens_seen": 9789304,
      "step": 14980
    },
    {
      "epoch": 7.85377358490566,
      "grad_norm": 0.1740136742591858,
      "learning_rate": 3.805037284786303e-05,
      "loss": 0.6232,
      "num_input_tokens_seen": 9792184,
      "step": 14985
    },
    {
      "epoch": 7.856394129979035,
      "grad_norm": 0.16014987230300903,
      "learning_rate": 3.804061876587009e-05,
      "loss": 0.5731,
      "num_input_tokens_seen": 9795928,
      "step": 14990
    },
    {
      "epoch": 7.859014675052411,
      "grad_norm": 0.14539669454097748,
      "learning_rate": 3.80308619559285e-05,
      "loss": 0.5172,
      "num_input_tokens_seen": 9798776,
      "step": 14995
    },
    {
      "epoch": 7.861635220125786,
      "grad_norm": 0.16334837675094604,
      "learning_rate": 3.802110242007925e-05,
      "loss": 0.4199,
      "num_input_tokens_seen": 9802008,
      "step": 15000
    },
    {
      "epoch": 7.864255765199162,
      "grad_norm": 0.27070340514183044,
      "learning_rate": 3.801134016036392e-05,
      "loss": 0.6405,
      "num_input_tokens_seen": 9805912,
      "step": 15005
    },
    {
      "epoch": 7.866876310272537,
      "grad_norm": 0.17004276812076569,
      "learning_rate": 3.8001575178824665e-05,
      "loss": 0.4953,
      "num_input_tokens_seen": 9808664,
      "step": 15010
    },
    {
      "epoch": 7.869496855345912,
      "grad_norm": 0.08964405208826065,
      "learning_rate": 3.799180747750422e-05,
      "loss": 0.4276,
      "num_input_tokens_seen": 9811256,
      "step": 15015
    },
    {
      "epoch": 7.872117400419287,
      "grad_norm": 0.15097062289714813,
      "learning_rate": 3.798203705844586e-05,
      "loss": 0.5538,
      "num_input_tokens_seen": 9816440,
      "step": 15020
    },
    {
      "epoch": 7.8747379454926625,
      "grad_norm": 0.18384787440299988,
      "learning_rate": 3.7972263923693476e-05,
      "loss": 0.4677,
      "num_input_tokens_seen": 9818776,
      "step": 15025
    },
    {
      "epoch": 7.877358490566038,
      "grad_norm": 0.13971500098705292,
      "learning_rate": 3.796248807529145e-05,
      "loss": 0.5179,
      "num_input_tokens_seen": 9822520,
      "step": 15030
    },
    {
      "epoch": 7.879979035639413,
      "grad_norm": 0.1071922555565834,
      "learning_rate": 3.795270951528482e-05,
      "loss": 0.3964,
      "num_input_tokens_seen": 9825496,
      "step": 15035
    },
    {
      "epoch": 7.882599580712788,
      "grad_norm": 0.14393438398838043,
      "learning_rate": 3.7942928245719124e-05,
      "loss": 0.6459,
      "num_input_tokens_seen": 9829080,
      "step": 15040
    },
    {
      "epoch": 7.885220125786163,
      "grad_norm": 0.22517621517181396,
      "learning_rate": 3.7933144268640495e-05,
      "loss": 0.4488,
      "num_input_tokens_seen": 9832056,
      "step": 15045
    },
    {
      "epoch": 7.887840670859539,
      "grad_norm": 0.1569695621728897,
      "learning_rate": 3.7923357586095645e-05,
      "loss": 0.5085,
      "num_input_tokens_seen": 9834680,
      "step": 15050
    },
    {
      "epoch": 7.890461215932914,
      "grad_norm": 0.12956680357456207,
      "learning_rate": 3.7913568200131823e-05,
      "loss": 0.5804,
      "num_input_tokens_seen": 9838008,
      "step": 15055
    },
    {
      "epoch": 7.8930817610062896,
      "grad_norm": 0.2155853509902954,
      "learning_rate": 3.790377611279685e-05,
      "loss": 0.605,
      "num_input_tokens_seen": 9841528,
      "step": 15060
    },
    {
      "epoch": 7.895702306079665,
      "grad_norm": 0.2328968346118927,
      "learning_rate": 3.789398132613914e-05,
      "loss": 0.6347,
      "num_input_tokens_seen": 9844984,
      "step": 15065
    },
    {
      "epoch": 7.89832285115304,
      "grad_norm": 0.11303059756755829,
      "learning_rate": 3.788418384220763e-05,
      "loss": 0.4983,
      "num_input_tokens_seen": 9848568,
      "step": 15070
    },
    {
      "epoch": 7.900943396226415,
      "grad_norm": 0.16948720812797546,
      "learning_rate": 3.7874383663051846e-05,
      "loss": 0.3346,
      "num_input_tokens_seen": 9851640,
      "step": 15075
    },
    {
      "epoch": 7.90356394129979,
      "grad_norm": 0.18417340517044067,
      "learning_rate": 3.7864580790721894e-05,
      "loss": 0.3968,
      "num_input_tokens_seen": 9855192,
      "step": 15080
    },
    {
      "epoch": 7.906184486373165,
      "grad_norm": 0.13470759987831116,
      "learning_rate": 3.785477522726839e-05,
      "loss": 0.5016,
      "num_input_tokens_seen": 9858072,
      "step": 15085
    },
    {
      "epoch": 7.908805031446541,
      "grad_norm": 0.16908471286296844,
      "learning_rate": 3.784496697474258e-05,
      "loss": 0.4491,
      "num_input_tokens_seen": 9861784,
      "step": 15090
    },
    {
      "epoch": 7.911425576519916,
      "grad_norm": 0.2463563233613968,
      "learning_rate": 3.783515603519622e-05,
      "loss": 0.5487,
      "num_input_tokens_seen": 9866104,
      "step": 15095
    },
    {
      "epoch": 7.914046121593291,
      "grad_norm": 0.28367719054222107,
      "learning_rate": 3.782534241068164e-05,
      "loss": 0.625,
      "num_input_tokens_seen": 9869816,
      "step": 15100
    },
    {
      "epoch": 7.916666666666667,
      "grad_norm": 0.152965247631073,
      "learning_rate": 3.781552610325175e-05,
      "loss": 0.3242,
      "num_input_tokens_seen": 9872504,
      "step": 15105
    },
    {
      "epoch": 7.919287211740042,
      "grad_norm": 0.14772839844226837,
      "learning_rate": 3.780570711496001e-05,
      "loss": 0.5007,
      "num_input_tokens_seen": 9875320,
      "step": 15110
    },
    {
      "epoch": 7.921907756813417,
      "grad_norm": 0.10840928554534912,
      "learning_rate": 3.779588544786044e-05,
      "loss": 0.389,
      "num_input_tokens_seen": 9878136,
      "step": 15115
    },
    {
      "epoch": 7.9245283018867925,
      "grad_norm": 0.18067726492881775,
      "learning_rate": 3.778606110400761e-05,
      "loss": 0.4729,
      "num_input_tokens_seen": 9880792,
      "step": 15120
    },
    {
      "epoch": 7.927148846960168,
      "grad_norm": 0.13486094772815704,
      "learning_rate": 3.7776234085456665e-05,
      "loss": 0.4256,
      "num_input_tokens_seen": 9884312,
      "step": 15125
    },
    {
      "epoch": 7.929769392033543,
      "grad_norm": 0.1810661405324936,
      "learning_rate": 3.776640439426332e-05,
      "loss": 0.4798,
      "num_input_tokens_seen": 9887416,
      "step": 15130
    },
    {
      "epoch": 7.932389937106918,
      "grad_norm": 0.1514590084552765,
      "learning_rate": 3.775657203248381e-05,
      "loss": 0.3809,
      "num_input_tokens_seen": 9890200,
      "step": 15135
    },
    {
      "epoch": 7.935010482180293,
      "grad_norm": 0.23107558488845825,
      "learning_rate": 3.774673700217497e-05,
      "loss": 0.4178,
      "num_input_tokens_seen": 9893368,
      "step": 15140
    },
    {
      "epoch": 7.937631027253669,
      "grad_norm": 0.20347434282302856,
      "learning_rate": 3.773689930539418e-05,
      "loss": 0.5313,
      "num_input_tokens_seen": 9899224,
      "step": 15145
    },
    {
      "epoch": 7.940251572327044,
      "grad_norm": 0.23948611319065094,
      "learning_rate": 3.7727058944199354e-05,
      "loss": 0.5316,
      "num_input_tokens_seen": 9901944,
      "step": 15150
    },
    {
      "epoch": 7.9428721174004195,
      "grad_norm": 0.10384946316480637,
      "learning_rate": 3.7717215920648994e-05,
      "loss": 0.6118,
      "num_input_tokens_seen": 9905720,
      "step": 15155
    },
    {
      "epoch": 7.945492662473795,
      "grad_norm": 0.1613946259021759,
      "learning_rate": 3.770737023680215e-05,
      "loss": 0.6454,
      "num_input_tokens_seen": 9908440,
      "step": 15160
    },
    {
      "epoch": 7.94811320754717,
      "grad_norm": 0.1776055097579956,
      "learning_rate": 3.7697521894718436e-05,
      "loss": 0.5468,
      "num_input_tokens_seen": 9911256,
      "step": 15165
    },
    {
      "epoch": 7.950733752620545,
      "grad_norm": 0.18778087198734283,
      "learning_rate": 3.7687670896458e-05,
      "loss": 0.5303,
      "num_input_tokens_seen": 9913336,
      "step": 15170
    },
    {
      "epoch": 7.95335429769392,
      "grad_norm": 0.13326703011989594,
      "learning_rate": 3.7677817244081556e-05,
      "loss": 0.3874,
      "num_input_tokens_seen": 9917048,
      "step": 15175
    },
    {
      "epoch": 7.955974842767295,
      "grad_norm": 0.3043297231197357,
      "learning_rate": 3.766796093965038e-05,
      "loss": 0.4846,
      "num_input_tokens_seen": 9920952,
      "step": 15180
    },
    {
      "epoch": 7.9585953878406706,
      "grad_norm": 0.1651763617992401,
      "learning_rate": 3.76581019852263e-05,
      "loss": 0.4308,
      "num_input_tokens_seen": 9924376,
      "step": 15185
    },
    {
      "epoch": 7.961215932914046,
      "grad_norm": 0.12430520355701447,
      "learning_rate": 3.7648240382871705e-05,
      "loss": 0.6254,
      "num_input_tokens_seen": 9927160,
      "step": 15190
    },
    {
      "epoch": 7.963836477987421,
      "grad_norm": 0.12757617235183716,
      "learning_rate": 3.763837613464951e-05,
      "loss": 0.5586,
      "num_input_tokens_seen": 9930840,
      "step": 15195
    },
    {
      "epoch": 7.966457023060797,
      "grad_norm": 0.13527220487594604,
      "learning_rate": 3.7628509242623215e-05,
      "loss": 0.6869,
      "num_input_tokens_seen": 9934104,
      "step": 15200
    },
    {
      "epoch": 7.969077568134172,
      "grad_norm": 0.17302480340003967,
      "learning_rate": 3.761863970885686e-05,
      "loss": 0.4291,
      "num_input_tokens_seen": 9937016,
      "step": 15205
    },
    {
      "epoch": 7.971698113207547,
      "grad_norm": 0.2921009957790375,
      "learning_rate": 3.760876753541504e-05,
      "loss": 0.5322,
      "num_input_tokens_seen": 9939736,
      "step": 15210
    },
    {
      "epoch": 7.9743186582809225,
      "grad_norm": 0.14657901227474213,
      "learning_rate": 3.7598892724362907e-05,
      "loss": 0.5316,
      "num_input_tokens_seen": 9942744,
      "step": 15215
    },
    {
      "epoch": 7.976939203354298,
      "grad_norm": 0.3160702586174011,
      "learning_rate": 3.758901527776615e-05,
      "loss": 0.6406,
      "num_input_tokens_seen": 9946776,
      "step": 15220
    },
    {
      "epoch": 7.979559748427673,
      "grad_norm": 0.22521963715553284,
      "learning_rate": 3.7579135197691015e-05,
      "loss": 0.5164,
      "num_input_tokens_seen": 9949304,
      "step": 15225
    },
    {
      "epoch": 7.982180293501048,
      "grad_norm": 0.1794375330209732,
      "learning_rate": 3.7569252486204306e-05,
      "loss": 0.4954,
      "num_input_tokens_seen": 9953112,
      "step": 15230
    },
    {
      "epoch": 7.984800838574423,
      "grad_norm": 0.1721857488155365,
      "learning_rate": 3.755936714537338e-05,
      "loss": 0.5055,
      "num_input_tokens_seen": 9956056,
      "step": 15235
    },
    {
      "epoch": 7.987421383647799,
      "grad_norm": 0.22215861082077026,
      "learning_rate": 3.754947917726612e-05,
      "loss": 0.3828,
      "num_input_tokens_seen": 9959416,
      "step": 15240
    },
    {
      "epoch": 7.990041928721174,
      "grad_norm": 0.17284847795963287,
      "learning_rate": 3.7539588583951e-05,
      "loss": 0.4311,
      "num_input_tokens_seen": 9962680,
      "step": 15245
    },
    {
      "epoch": 7.9926624737945495,
      "grad_norm": 0.11985237896442413,
      "learning_rate": 3.752969536749701e-05,
      "loss": 0.617,
      "num_input_tokens_seen": 9966616,
      "step": 15250
    },
    {
      "epoch": 7.995283018867925,
      "grad_norm": 0.11288110911846161,
      "learning_rate": 3.7519799529973685e-05,
      "loss": 0.4692,
      "num_input_tokens_seen": 9971864,
      "step": 15255
    },
    {
      "epoch": 7.9979035639413,
      "grad_norm": 0.13363637030124664,
      "learning_rate": 3.7509901073451134e-05,
      "loss": 0.8077,
      "num_input_tokens_seen": 9975864,
      "step": 15260
    },
    {
      "epoch": 8.0,
      "eval_loss": 0.5022030472755432,
      "eval_runtime": 9.8867,
      "eval_samples_per_second": 85.772,
      "eval_steps_per_second": 21.443,
      "num_input_tokens_seen": 9977520,
      "step": 15264
    },
    {
      "epoch": 8.000524109014675,
      "grad_norm": 0.2053503543138504,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.5801,
      "num_input_tokens_seen": 9978032,
      "step": 15265
    },
    {
      "epoch": 8.00314465408805,
      "grad_norm": 0.1917293667793274,
      "learning_rate": 3.749009631169147e-05,
      "loss": 0.6701,
      "num_input_tokens_seen": 9981616,
      "step": 15270
    },
    {
      "epoch": 8.005765199161425,
      "grad_norm": 0.19064012169837952,
      "learning_rate": 3.74801900105973e-05,
      "loss": 0.4246,
      "num_input_tokens_seen": 9984624,
      "step": 15275
    },
    {
      "epoch": 8.0083857442348,
      "grad_norm": 0.19414712488651276,
      "learning_rate": 3.747028109878975e-05,
      "loss": 0.4585,
      "num_input_tokens_seen": 9987568,
      "step": 15280
    },
    {
      "epoch": 8.011006289308176,
      "grad_norm": 0.1835051327943802,
      "learning_rate": 3.746036957834167e-05,
      "loss": 0.4591,
      "num_input_tokens_seen": 9991280,
      "step": 15285
    },
    {
      "epoch": 8.01362683438155,
      "grad_norm": 0.139021635055542,
      "learning_rate": 3.7450455451326415e-05,
      "loss": 0.458,
      "num_input_tokens_seen": 9994288,
      "step": 15290
    },
    {
      "epoch": 8.016247379454926,
      "grad_norm": 0.17400440573692322,
      "learning_rate": 3.7440538719817935e-05,
      "loss": 0.4418,
      "num_input_tokens_seen": 9997200,
      "step": 15295
    },
    {
      "epoch": 8.018867924528301,
      "grad_norm": 0.18660840392112732,
      "learning_rate": 3.743061938589068e-05,
      "loss": 0.6095,
      "num_input_tokens_seen": 10000176,
      "step": 15300
    },
    {
      "epoch": 8.021488469601676,
      "grad_norm": 0.15704508125782013,
      "learning_rate": 3.742069745161966e-05,
      "loss": 0.5284,
      "num_input_tokens_seen": 10003504,
      "step": 15305
    },
    {
      "epoch": 8.024109014675052,
      "grad_norm": 0.23818475008010864,
      "learning_rate": 3.741077291908045e-05,
      "loss": 0.4918,
      "num_input_tokens_seen": 10008688,
      "step": 15310
    },
    {
      "epoch": 8.026729559748428,
      "grad_norm": 0.1876126378774643,
      "learning_rate": 3.740084579034913e-05,
      "loss": 0.6447,
      "num_input_tokens_seen": 10011536,
      "step": 15315
    },
    {
      "epoch": 8.029350104821804,
      "grad_norm": 0.22220507264137268,
      "learning_rate": 3.739091606750234e-05,
      "loss": 0.4804,
      "num_input_tokens_seen": 10014096,
      "step": 15320
    },
    {
      "epoch": 8.031970649895179,
      "grad_norm": 0.11576581001281738,
      "learning_rate": 3.7380983752617275e-05,
      "loss": 0.4708,
      "num_input_tokens_seen": 10017360,
      "step": 15325
    },
    {
      "epoch": 8.034591194968554,
      "grad_norm": 0.1563175469636917,
      "learning_rate": 3.737104884777166e-05,
      "loss": 0.6193,
      "num_input_tokens_seen": 10020560,
      "step": 15330
    },
    {
      "epoch": 8.03721174004193,
      "grad_norm": 0.20191240310668945,
      "learning_rate": 3.736111135504376e-05,
      "loss": 0.5557,
      "num_input_tokens_seen": 10024176,
      "step": 15335
    },
    {
      "epoch": 8.039832285115304,
      "grad_norm": 0.15378685295581818,
      "learning_rate": 3.735117127651239e-05,
      "loss": 0.3735,
      "num_input_tokens_seen": 10028336,
      "step": 15340
    },
    {
      "epoch": 8.04245283018868,
      "grad_norm": 0.16237467527389526,
      "learning_rate": 3.734122861425689e-05,
      "loss": 0.473,
      "num_input_tokens_seen": 10031376,
      "step": 15345
    },
    {
      "epoch": 8.045073375262055,
      "grad_norm": 0.2503816485404968,
      "learning_rate": 3.7331283370357173e-05,
      "loss": 0.4984,
      "num_input_tokens_seen": 10033904,
      "step": 15350
    },
    {
      "epoch": 8.04769392033543,
      "grad_norm": 0.15565520524978638,
      "learning_rate": 3.732133554689365e-05,
      "loss": 0.4984,
      "num_input_tokens_seen": 10036720,
      "step": 15355
    },
    {
      "epoch": 8.050314465408805,
      "grad_norm": 0.20040857791900635,
      "learning_rate": 3.73113851459473e-05,
      "loss": 0.5509,
      "num_input_tokens_seen": 10040016,
      "step": 15360
    },
    {
      "epoch": 8.05293501048218,
      "grad_norm": 0.14143791794776917,
      "learning_rate": 3.730143216959964e-05,
      "loss": 0.5439,
      "num_input_tokens_seen": 10043088,
      "step": 15365
    },
    {
      "epoch": 8.055555555555555,
      "grad_norm": 0.1678479015827179,
      "learning_rate": 3.729147661993271e-05,
      "loss": 0.5808,
      "num_input_tokens_seen": 10046160,
      "step": 15370
    },
    {
      "epoch": 8.05817610062893,
      "grad_norm": 0.16101588308811188,
      "learning_rate": 3.728151849902909e-05,
      "loss": 0.4893,
      "num_input_tokens_seen": 10048592,
      "step": 15375
    },
    {
      "epoch": 8.060796645702306,
      "grad_norm": 0.0903841108083725,
      "learning_rate": 3.727155780897193e-05,
      "loss": 0.5221,
      "num_input_tokens_seen": 10051536,
      "step": 15380
    },
    {
      "epoch": 8.06341719077568,
      "grad_norm": 0.16146060824394226,
      "learning_rate": 3.7261594551844865e-05,
      "loss": 0.5716,
      "num_input_tokens_seen": 10055632,
      "step": 15385
    },
    {
      "epoch": 8.066037735849056,
      "grad_norm": 0.11669182777404785,
      "learning_rate": 3.7251628729732116e-05,
      "loss": 0.554,
      "num_input_tokens_seen": 10060528,
      "step": 15390
    },
    {
      "epoch": 8.068658280922431,
      "grad_norm": 0.20486299693584442,
      "learning_rate": 3.724166034471841e-05,
      "loss": 0.3732,
      "num_input_tokens_seen": 10063664,
      "step": 15395
    },
    {
      "epoch": 8.071278825995806,
      "grad_norm": 0.2835920751094818,
      "learning_rate": 3.723168939888901e-05,
      "loss": 0.5098,
      "num_input_tokens_seen": 10066288,
      "step": 15400
    },
    {
      "epoch": 8.073899371069182,
      "grad_norm": 0.3114957809448242,
      "learning_rate": 3.7221715894329747e-05,
      "loss": 0.6763,
      "num_input_tokens_seen": 10070064,
      "step": 15405
    },
    {
      "epoch": 8.076519916142558,
      "grad_norm": 0.19402512907981873,
      "learning_rate": 3.7211739833126945e-05,
      "loss": 0.4442,
      "num_input_tokens_seen": 10074128,
      "step": 15410
    },
    {
      "epoch": 8.079140461215934,
      "grad_norm": 0.1515740007162094,
      "learning_rate": 3.720176121736748e-05,
      "loss": 0.3103,
      "num_input_tokens_seen": 10077936,
      "step": 15415
    },
    {
      "epoch": 8.081761006289309,
      "grad_norm": 0.1768779456615448,
      "learning_rate": 3.719178004913879e-05,
      "loss": 0.5285,
      "num_input_tokens_seen": 10080976,
      "step": 15420
    },
    {
      "epoch": 8.084381551362684,
      "grad_norm": 0.18326307833194733,
      "learning_rate": 3.7181796330528796e-05,
      "loss": 0.4962,
      "num_input_tokens_seen": 10083984,
      "step": 15425
    },
    {
      "epoch": 8.08700209643606,
      "grad_norm": 0.2225414216518402,
      "learning_rate": 3.717181006362599e-05,
      "loss": 0.6453,
      "num_input_tokens_seen": 10086800,
      "step": 15430
    },
    {
      "epoch": 8.089622641509434,
      "grad_norm": 0.10204736888408661,
      "learning_rate": 3.7161821250519384e-05,
      "loss": 0.4343,
      "num_input_tokens_seen": 10090544,
      "step": 15435
    },
    {
      "epoch": 8.09224318658281,
      "grad_norm": 0.39148661494255066,
      "learning_rate": 3.7151829893298515e-05,
      "loss": 0.4482,
      "num_input_tokens_seen": 10093360,
      "step": 15440
    },
    {
      "epoch": 8.094863731656185,
      "grad_norm": 0.1672953963279724,
      "learning_rate": 3.714183599405347e-05,
      "loss": 0.6283,
      "num_input_tokens_seen": 10097520,
      "step": 15445
    },
    {
      "epoch": 8.09748427672956,
      "grad_norm": 0.11677629500627518,
      "learning_rate": 3.713183955487487e-05,
      "loss": 0.4416,
      "num_input_tokens_seen": 10100240,
      "step": 15450
    },
    {
      "epoch": 8.100104821802935,
      "grad_norm": 0.1441928595304489,
      "learning_rate": 3.7121840577853815e-05,
      "loss": 0.462,
      "num_input_tokens_seen": 10103280,
      "step": 15455
    },
    {
      "epoch": 8.10272536687631,
      "grad_norm": 0.11704909056425095,
      "learning_rate": 3.7111839065082027e-05,
      "loss": 0.5283,
      "num_input_tokens_seen": 10106288,
      "step": 15460
    },
    {
      "epoch": 8.105345911949685,
      "grad_norm": 0.2521974742412567,
      "learning_rate": 3.710183501865168e-05,
      "loss": 0.4341,
      "num_input_tokens_seen": 10110064,
      "step": 15465
    },
    {
      "epoch": 8.10796645702306,
      "grad_norm": 0.13467028737068176,
      "learning_rate": 3.7091828440655514e-05,
      "loss": 0.3924,
      "num_input_tokens_seen": 10113360,
      "step": 15470
    },
    {
      "epoch": 8.110587002096436,
      "grad_norm": 0.21202407777309418,
      "learning_rate": 3.708181933318678e-05,
      "loss": 0.5516,
      "num_input_tokens_seen": 10119760,
      "step": 15475
    },
    {
      "epoch": 8.11320754716981,
      "grad_norm": 0.1263396292924881,
      "learning_rate": 3.707180769833929e-05,
      "loss": 0.4701,
      "num_input_tokens_seen": 10126864,
      "step": 15480
    },
    {
      "epoch": 8.115828092243186,
      "grad_norm": 0.1768403947353363,
      "learning_rate": 3.706179353820734e-05,
      "loss": 0.4125,
      "num_input_tokens_seen": 10129776,
      "step": 15485
    },
    {
      "epoch": 8.118448637316561,
      "grad_norm": 0.16376937925815582,
      "learning_rate": 3.7051776854885794e-05,
      "loss": 0.5653,
      "num_input_tokens_seen": 10132816,
      "step": 15490
    },
    {
      "epoch": 8.121069182389936,
      "grad_norm": 0.17332549393177032,
      "learning_rate": 3.7041757650470025e-05,
      "loss": 0.3503,
      "num_input_tokens_seen": 10136112,
      "step": 15495
    },
    {
      "epoch": 8.123689727463312,
      "grad_norm": 0.10851074755191803,
      "learning_rate": 3.703173592705594e-05,
      "loss": 0.481,
      "num_input_tokens_seen": 10138896,
      "step": 15500
    },
    {
      "epoch": 8.126310272536688,
      "grad_norm": 0.127703458070755,
      "learning_rate": 3.702171168673996e-05,
      "loss": 0.478,
      "num_input_tokens_seen": 10141456,
      "step": 15505
    },
    {
      "epoch": 8.128930817610064,
      "grad_norm": 0.1621476262807846,
      "learning_rate": 3.701168493161905e-05,
      "loss": 0.5707,
      "num_input_tokens_seen": 10144944,
      "step": 15510
    },
    {
      "epoch": 8.131551362683439,
      "grad_norm": 0.1363028585910797,
      "learning_rate": 3.7001655663790676e-05,
      "loss": 0.4161,
      "num_input_tokens_seen": 10148016,
      "step": 15515
    },
    {
      "epoch": 8.134171907756814,
      "grad_norm": 0.11154991388320923,
      "learning_rate": 3.699162388535288e-05,
      "loss": 0.5471,
      "num_input_tokens_seen": 10152048,
      "step": 15520
    },
    {
      "epoch": 8.13679245283019,
      "grad_norm": 0.13346083462238312,
      "learning_rate": 3.698158959840416e-05,
      "loss": 0.5501,
      "num_input_tokens_seen": 10155600,
      "step": 15525
    },
    {
      "epoch": 8.139412997903564,
      "grad_norm": 0.2153785675764084,
      "learning_rate": 3.6971552805043585e-05,
      "loss": 0.5242,
      "num_input_tokens_seen": 10158160,
      "step": 15530
    },
    {
      "epoch": 8.14203354297694,
      "grad_norm": 0.13118530809879303,
      "learning_rate": 3.6961513507370754e-05,
      "loss": 0.5152,
      "num_input_tokens_seen": 10161296,
      "step": 15535
    },
    {
      "epoch": 8.144654088050315,
      "grad_norm": 0.10593048483133316,
      "learning_rate": 3.695147170748576e-05,
      "loss": 0.5153,
      "num_input_tokens_seen": 10164112,
      "step": 15540
    },
    {
      "epoch": 8.14727463312369,
      "grad_norm": 0.198774516582489,
      "learning_rate": 3.694142740748923e-05,
      "loss": 0.485,
      "num_input_tokens_seen": 10167248,
      "step": 15545
    },
    {
      "epoch": 8.149895178197065,
      "grad_norm": 0.10969062149524689,
      "learning_rate": 3.693138060948232e-05,
      "loss": 0.383,
      "num_input_tokens_seen": 10171088,
      "step": 15550
    },
    {
      "epoch": 8.15251572327044,
      "grad_norm": 0.1452464461326599,
      "learning_rate": 3.692133131556671e-05,
      "loss": 0.5321,
      "num_input_tokens_seen": 10174192,
      "step": 15555
    },
    {
      "epoch": 8.155136268343815,
      "grad_norm": 0.16138401627540588,
      "learning_rate": 3.6911279527844596e-05,
      "loss": 0.5411,
      "num_input_tokens_seen": 10177264,
      "step": 15560
    },
    {
      "epoch": 8.15775681341719,
      "grad_norm": 0.22372187674045563,
      "learning_rate": 3.69012252484187e-05,
      "loss": 0.6189,
      "num_input_tokens_seen": 10179824,
      "step": 15565
    },
    {
      "epoch": 8.160377358490566,
      "grad_norm": 0.15556205809116364,
      "learning_rate": 3.6891168479392245e-05,
      "loss": 0.4577,
      "num_input_tokens_seen": 10183024,
      "step": 15570
    },
    {
      "epoch": 8.16299790356394,
      "grad_norm": 0.11553327739238739,
      "learning_rate": 3.688110922286901e-05,
      "loss": 0.4612,
      "num_input_tokens_seen": 10186832,
      "step": 15575
    },
    {
      "epoch": 8.165618448637316,
      "grad_norm": 0.15743526816368103,
      "learning_rate": 3.687104748095328e-05,
      "loss": 0.5421,
      "num_input_tokens_seen": 10189360,
      "step": 15580
    },
    {
      "epoch": 8.168238993710691,
      "grad_norm": 0.13826389610767365,
      "learning_rate": 3.686098325574984e-05,
      "loss": 0.5184,
      "num_input_tokens_seen": 10193360,
      "step": 15585
    },
    {
      "epoch": 8.170859538784066,
      "grad_norm": 0.18768270313739777,
      "learning_rate": 3.685091654936403e-05,
      "loss": 0.4954,
      "num_input_tokens_seen": 10197136,
      "step": 15590
    },
    {
      "epoch": 8.173480083857442,
      "grad_norm": 0.13486653566360474,
      "learning_rate": 3.6840847363901675e-05,
      "loss": 0.4792,
      "num_input_tokens_seen": 10200592,
      "step": 15595
    },
    {
      "epoch": 8.176100628930818,
      "grad_norm": 0.1809166520833969,
      "learning_rate": 3.683077570146914e-05,
      "loss": 0.4952,
      "num_input_tokens_seen": 10203088,
      "step": 15600
    },
    {
      "epoch": 8.178721174004194,
      "grad_norm": 0.18603764474391937,
      "learning_rate": 3.68207015641733e-05,
      "loss": 0.6313,
      "num_input_tokens_seen": 10205840,
      "step": 15605
    },
    {
      "epoch": 8.181341719077569,
      "grad_norm": 0.14470355212688446,
      "learning_rate": 3.681062495412155e-05,
      "loss": 0.3783,
      "num_input_tokens_seen": 10209040,
      "step": 15610
    },
    {
      "epoch": 8.183962264150944,
      "grad_norm": 0.17330826818943024,
      "learning_rate": 3.68005458734218e-05,
      "loss": 0.4767,
      "num_input_tokens_seen": 10212048,
      "step": 15615
    },
    {
      "epoch": 8.18658280922432,
      "grad_norm": 0.17179428040981293,
      "learning_rate": 3.679046432418248e-05,
      "loss": 0.5449,
      "num_input_tokens_seen": 10214896,
      "step": 15620
    },
    {
      "epoch": 8.189203354297694,
      "grad_norm": 0.21347098052501678,
      "learning_rate": 3.6780380308512543e-05,
      "loss": 0.5055,
      "num_input_tokens_seen": 10217872,
      "step": 15625
    },
    {
      "epoch": 8.19182389937107,
      "grad_norm": 0.18440550565719604,
      "learning_rate": 3.677029382852144e-05,
      "loss": 0.484,
      "num_input_tokens_seen": 10221488,
      "step": 15630
    },
    {
      "epoch": 8.194444444444445,
      "grad_norm": 0.19204963743686676,
      "learning_rate": 3.676020488631915e-05,
      "loss": 0.4056,
      "num_input_tokens_seen": 10224048,
      "step": 15635
    },
    {
      "epoch": 8.19706498951782,
      "grad_norm": 0.13452106714248657,
      "learning_rate": 3.675011348401617e-05,
      "loss": 0.4559,
      "num_input_tokens_seen": 10228368,
      "step": 15640
    },
    {
      "epoch": 8.199685534591195,
      "grad_norm": 0.1319764107465744,
      "learning_rate": 3.6740019623723494e-05,
      "loss": 0.4732,
      "num_input_tokens_seen": 10231728,
      "step": 15645
    },
    {
      "epoch": 8.20230607966457,
      "grad_norm": 0.19675534963607788,
      "learning_rate": 3.672992330755265e-05,
      "loss": 0.4455,
      "num_input_tokens_seen": 10234960,
      "step": 15650
    },
    {
      "epoch": 8.204926624737945,
      "grad_norm": 0.10887814313173294,
      "learning_rate": 3.671982453761568e-05,
      "loss": 0.4638,
      "num_input_tokens_seen": 10238448,
      "step": 15655
    },
    {
      "epoch": 8.20754716981132,
      "grad_norm": 0.19728805124759674,
      "learning_rate": 3.6709723316025114e-05,
      "loss": 0.4124,
      "num_input_tokens_seen": 10240912,
      "step": 15660
    },
    {
      "epoch": 8.210167714884696,
      "grad_norm": 0.14855864644050598,
      "learning_rate": 3.669961964489403e-05,
      "loss": 0.4902,
      "num_input_tokens_seen": 10244816,
      "step": 15665
    },
    {
      "epoch": 8.21278825995807,
      "grad_norm": 0.20162586867809296,
      "learning_rate": 3.668951352633599e-05,
      "loss": 0.535,
      "num_input_tokens_seen": 10247728,
      "step": 15670
    },
    {
      "epoch": 8.215408805031446,
      "grad_norm": 0.12297359108924866,
      "learning_rate": 3.6679404962465084e-05,
      "loss": 0.4211,
      "num_input_tokens_seen": 10251248,
      "step": 15675
    },
    {
      "epoch": 8.218029350104821,
      "grad_norm": 0.12823595106601715,
      "learning_rate": 3.66692939553959e-05,
      "loss": 0.51,
      "num_input_tokens_seen": 10254160,
      "step": 15680
    },
    {
      "epoch": 8.220649895178196,
      "grad_norm": 0.15277662873268127,
      "learning_rate": 3.6659180507243554e-05,
      "loss": 0.4321,
      "num_input_tokens_seen": 10256944,
      "step": 15685
    },
    {
      "epoch": 8.223270440251572,
      "grad_norm": 0.1739761084318161,
      "learning_rate": 3.664906462012366e-05,
      "loss": 0.5407,
      "num_input_tokens_seen": 10259888,
      "step": 15690
    },
    {
      "epoch": 8.225890985324948,
      "grad_norm": 0.19386698305606842,
      "learning_rate": 3.663894629615234e-05,
      "loss": 0.6173,
      "num_input_tokens_seen": 10262640,
      "step": 15695
    },
    {
      "epoch": 8.228511530398324,
      "grad_norm": 0.18381951749324799,
      "learning_rate": 3.6628825537446246e-05,
      "loss": 0.5407,
      "num_input_tokens_seen": 10265424,
      "step": 15700
    },
    {
      "epoch": 8.231132075471699,
      "grad_norm": 0.13968774676322937,
      "learning_rate": 3.6618702346122525e-05,
      "loss": 0.53,
      "num_input_tokens_seen": 10267888,
      "step": 15705
    },
    {
      "epoch": 8.233752620545074,
      "grad_norm": 0.07891759276390076,
      "learning_rate": 3.6608576724298817e-05,
      "loss": 0.3553,
      "num_input_tokens_seen": 10270992,
      "step": 15710
    },
    {
      "epoch": 8.23637316561845,
      "grad_norm": 0.23455460369586945,
      "learning_rate": 3.65984486740933e-05,
      "loss": 0.5565,
      "num_input_tokens_seen": 10274128,
      "step": 15715
    },
    {
      "epoch": 8.238993710691824,
      "grad_norm": 0.19122721254825592,
      "learning_rate": 3.6588318197624644e-05,
      "loss": 0.4342,
      "num_input_tokens_seen": 10278128,
      "step": 15720
    },
    {
      "epoch": 8.2416142557652,
      "grad_norm": 0.16197705268859863,
      "learning_rate": 3.6578185297012015e-05,
      "loss": 0.4801,
      "num_input_tokens_seen": 10280720,
      "step": 15725
    },
    {
      "epoch": 8.244234800838575,
      "grad_norm": 0.16618570685386658,
      "learning_rate": 3.656804997437512e-05,
      "loss": 0.4881,
      "num_input_tokens_seen": 10283280,
      "step": 15730
    },
    {
      "epoch": 8.24685534591195,
      "grad_norm": 0.10323794186115265,
      "learning_rate": 3.6557912231834155e-05,
      "loss": 0.5287,
      "num_input_tokens_seen": 10287600,
      "step": 15735
    },
    {
      "epoch": 8.249475890985325,
      "grad_norm": 0.20854675769805908,
      "learning_rate": 3.65477720715098e-05,
      "loss": 0.4574,
      "num_input_tokens_seen": 10289808,
      "step": 15740
    },
    {
      "epoch": 8.2520964360587,
      "grad_norm": 0.3028404414653778,
      "learning_rate": 3.653762949552328e-05,
      "loss": 0.5219,
      "num_input_tokens_seen": 10292656,
      "step": 15745
    },
    {
      "epoch": 8.254716981132075,
      "grad_norm": 0.11581404507160187,
      "learning_rate": 3.6527484505996295e-05,
      "loss": 0.4384,
      "num_input_tokens_seen": 10295664,
      "step": 15750
    },
    {
      "epoch": 8.25733752620545,
      "grad_norm": 0.2347566783428192,
      "learning_rate": 3.651733710505106e-05,
      "loss": 0.542,
      "num_input_tokens_seen": 10299248,
      "step": 15755
    },
    {
      "epoch": 8.259958071278826,
      "grad_norm": 0.1564943492412567,
      "learning_rate": 3.6507187294810305e-05,
      "loss": 0.5425,
      "num_input_tokens_seen": 10302832,
      "step": 15760
    },
    {
      "epoch": 8.2625786163522,
      "grad_norm": 0.15952247381210327,
      "learning_rate": 3.649703507739724e-05,
      "loss": 0.5074,
      "num_input_tokens_seen": 10307888,
      "step": 15765
    },
    {
      "epoch": 8.265199161425576,
      "grad_norm": 0.2032589614391327,
      "learning_rate": 3.6486880454935606e-05,
      "loss": 0.4864,
      "num_input_tokens_seen": 10311248,
      "step": 15770
    },
    {
      "epoch": 8.267819706498951,
      "grad_norm": 0.20951493084430695,
      "learning_rate": 3.647672342954963e-05,
      "loss": 0.5083,
      "num_input_tokens_seen": 10314320,
      "step": 15775
    },
    {
      "epoch": 8.270440251572326,
      "grad_norm": 0.15408289432525635,
      "learning_rate": 3.646656400336405e-05,
      "loss": 0.5399,
      "num_input_tokens_seen": 10317936,
      "step": 15780
    },
    {
      "epoch": 8.273060796645701,
      "grad_norm": 0.1545063555240631,
      "learning_rate": 3.6456402178504094e-05,
      "loss": 0.5752,
      "num_input_tokens_seen": 10321136,
      "step": 15785
    },
    {
      "epoch": 8.275681341719078,
      "grad_norm": 0.17883019149303436,
      "learning_rate": 3.6446237957095505e-05,
      "loss": 0.5662,
      "num_input_tokens_seen": 10323760,
      "step": 15790
    },
    {
      "epoch": 8.278301886792454,
      "grad_norm": 0.3034683167934418,
      "learning_rate": 3.643607134126452e-05,
      "loss": 0.5164,
      "num_input_tokens_seen": 10327088,
      "step": 15795
    },
    {
      "epoch": 8.280922431865829,
      "grad_norm": 0.1512373983860016,
      "learning_rate": 3.642590233313789e-05,
      "loss": 0.6109,
      "num_input_tokens_seen": 10330448,
      "step": 15800
    },
    {
      "epoch": 8.283542976939204,
      "grad_norm": 0.22496742010116577,
      "learning_rate": 3.6415730934842827e-05,
      "loss": 0.4136,
      "num_input_tokens_seen": 10334672,
      "step": 15805
    },
    {
      "epoch": 8.286163522012579,
      "grad_norm": 0.17958934605121613,
      "learning_rate": 3.64055571485071e-05,
      "loss": 0.5375,
      "num_input_tokens_seen": 10337904,
      "step": 15810
    },
    {
      "epoch": 8.288784067085954,
      "grad_norm": 0.20117492973804474,
      "learning_rate": 3.639538097625894e-05,
      "loss": 0.444,
      "num_input_tokens_seen": 10341488,
      "step": 15815
    },
    {
      "epoch": 8.29140461215933,
      "grad_norm": 0.19377894699573517,
      "learning_rate": 3.638520242022708e-05,
      "loss": 0.4822,
      "num_input_tokens_seen": 10344944,
      "step": 15820
    },
    {
      "epoch": 8.294025157232705,
      "grad_norm": 0.22725988924503326,
      "learning_rate": 3.637502148254077e-05,
      "loss": 0.5726,
      "num_input_tokens_seen": 10347248,
      "step": 15825
    },
    {
      "epoch": 8.29664570230608,
      "grad_norm": 0.19063705205917358,
      "learning_rate": 3.636483816532974e-05,
      "loss": 0.6158,
      "num_input_tokens_seen": 10349712,
      "step": 15830
    },
    {
      "epoch": 8.299266247379455,
      "grad_norm": 0.1552148461341858,
      "learning_rate": 3.6354652470724217e-05,
      "loss": 0.4782,
      "num_input_tokens_seen": 10353936,
      "step": 15835
    },
    {
      "epoch": 8.30188679245283,
      "grad_norm": 0.3030988276004791,
      "learning_rate": 3.634446440085494e-05,
      "loss": 0.4605,
      "num_input_tokens_seen": 10356432,
      "step": 15840
    },
    {
      "epoch": 8.304507337526205,
      "grad_norm": 0.19216430187225342,
      "learning_rate": 3.633427395785313e-05,
      "loss": 0.7372,
      "num_input_tokens_seen": 10360624,
      "step": 15845
    },
    {
      "epoch": 8.30712788259958,
      "grad_norm": 0.18537810444831848,
      "learning_rate": 3.6324081143850516e-05,
      "loss": 0.5132,
      "num_input_tokens_seen": 10364432,
      "step": 15850
    },
    {
      "epoch": 8.309748427672956,
      "grad_norm": 0.16377833485603333,
      "learning_rate": 3.6313885960979326e-05,
      "loss": 0.4611,
      "num_input_tokens_seen": 10367184,
      "step": 15855
    },
    {
      "epoch": 8.31236897274633,
      "grad_norm": 0.17855152487754822,
      "learning_rate": 3.630368841137226e-05,
      "loss": 0.4772,
      "num_input_tokens_seen": 10370192,
      "step": 15860
    },
    {
      "epoch": 8.314989517819706,
      "grad_norm": 0.1382598727941513,
      "learning_rate": 3.6293488497162554e-05,
      "loss": 0.5702,
      "num_input_tokens_seen": 10373744,
      "step": 15865
    },
    {
      "epoch": 8.317610062893081,
      "grad_norm": 0.1442812979221344,
      "learning_rate": 3.628328622048388e-05,
      "loss": 0.5407,
      "num_input_tokens_seen": 10376400,
      "step": 15870
    },
    {
      "epoch": 8.320230607966456,
      "grad_norm": 0.11824765056371689,
      "learning_rate": 3.627308158347046e-05,
      "loss": 0.391,
      "num_input_tokens_seen": 10379600,
      "step": 15875
    },
    {
      "epoch": 8.322851153039831,
      "grad_norm": 0.21216389536857605,
      "learning_rate": 3.626287458825699e-05,
      "loss": 0.4742,
      "num_input_tokens_seen": 10383088,
      "step": 15880
    },
    {
      "epoch": 8.325471698113208,
      "grad_norm": 0.1954452395439148,
      "learning_rate": 3.625266523697864e-05,
      "loss": 0.5037,
      "num_input_tokens_seen": 10386288,
      "step": 15885
    },
    {
      "epoch": 8.328092243186584,
      "grad_norm": 0.13795538246631622,
      "learning_rate": 3.62424535317711e-05,
      "loss": 0.6563,
      "num_input_tokens_seen": 10389744,
      "step": 15890
    },
    {
      "epoch": 8.330712788259959,
      "grad_norm": 0.1448611468076706,
      "learning_rate": 3.623223947477055e-05,
      "loss": 0.4278,
      "num_input_tokens_seen": 10392880,
      "step": 15895
    },
    {
      "epoch": 8.333333333333334,
      "grad_norm": 0.19296446442604065,
      "learning_rate": 3.6222023068113635e-05,
      "loss": 0.6056,
      "num_input_tokens_seen": 10397168,
      "step": 15900
    },
    {
      "epoch": 8.335953878406709,
      "grad_norm": 0.20161838829517365,
      "learning_rate": 3.621180431393754e-05,
      "loss": 0.5079,
      "num_input_tokens_seen": 10400784,
      "step": 15905
    },
    {
      "epoch": 8.338574423480084,
      "grad_norm": 0.16627736389636993,
      "learning_rate": 3.6201583214379874e-05,
      "loss": 0.3082,
      "num_input_tokens_seen": 10403568,
      "step": 15910
    },
    {
      "epoch": 8.34119496855346,
      "grad_norm": 0.2072869837284088,
      "learning_rate": 3.61913597715788e-05,
      "loss": 0.4795,
      "num_input_tokens_seen": 10406800,
      "step": 15915
    },
    {
      "epoch": 8.343815513626835,
      "grad_norm": 0.10208550840616226,
      "learning_rate": 3.6181133987672944e-05,
      "loss": 0.4136,
      "num_input_tokens_seen": 10409744,
      "step": 15920
    },
    {
      "epoch": 8.34643605870021,
      "grad_norm": 0.23498384654521942,
      "learning_rate": 3.617090586480142e-05,
      "loss": 0.5094,
      "num_input_tokens_seen": 10412880,
      "step": 15925
    },
    {
      "epoch": 8.349056603773585,
      "grad_norm": 0.24244830012321472,
      "learning_rate": 3.6160675405103826e-05,
      "loss": 0.5045,
      "num_input_tokens_seen": 10416432,
      "step": 15930
    },
    {
      "epoch": 8.35167714884696,
      "grad_norm": 0.28177303075790405,
      "learning_rate": 3.615044261072027e-05,
      "loss": 0.4823,
      "num_input_tokens_seen": 10419824,
      "step": 15935
    },
    {
      "epoch": 8.354297693920335,
      "grad_norm": 0.1263771802186966,
      "learning_rate": 3.614020748379133e-05,
      "loss": 0.4811,
      "num_input_tokens_seen": 10422800,
      "step": 15940
    },
    {
      "epoch": 8.35691823899371,
      "grad_norm": 0.18906717002391815,
      "learning_rate": 3.6129970026458076e-05,
      "loss": 0.4134,
      "num_input_tokens_seen": 10425808,
      "step": 15945
    },
    {
      "epoch": 8.359538784067086,
      "grad_norm": 0.1211095005273819,
      "learning_rate": 3.611973024086207e-05,
      "loss": 0.4828,
      "num_input_tokens_seen": 10428944,
      "step": 15950
    },
    {
      "epoch": 8.36215932914046,
      "grad_norm": 0.144003227353096,
      "learning_rate": 3.610948812914536e-05,
      "loss": 0.533,
      "num_input_tokens_seen": 10433648,
      "step": 15955
    },
    {
      "epoch": 8.364779874213836,
      "grad_norm": 0.12987564504146576,
      "learning_rate": 3.609924369345048e-05,
      "loss": 0.4563,
      "num_input_tokens_seen": 10436784,
      "step": 15960
    },
    {
      "epoch": 8.367400419287211,
      "grad_norm": 0.2141723930835724,
      "learning_rate": 3.6088996935920444e-05,
      "loss": 0.4722,
      "num_input_tokens_seen": 10439280,
      "step": 15965
    },
    {
      "epoch": 8.370020964360586,
      "grad_norm": 0.13537536561489105,
      "learning_rate": 3.607874785869876e-05,
      "loss": 0.3951,
      "num_input_tokens_seen": 10441744,
      "step": 15970
    },
    {
      "epoch": 8.372641509433961,
      "grad_norm": 0.16130521893501282,
      "learning_rate": 3.6068496463929416e-05,
      "loss": 0.4626,
      "num_input_tokens_seen": 10445424,
      "step": 15975
    },
    {
      "epoch": 8.375262054507338,
      "grad_norm": 0.18309557437896729,
      "learning_rate": 3.605824275375689e-05,
      "loss": 0.7032,
      "num_input_tokens_seen": 10448688,
      "step": 15980
    },
    {
      "epoch": 8.377882599580714,
      "grad_norm": 0.14772088825702667,
      "learning_rate": 3.604798673032613e-05,
      "loss": 0.517,
      "num_input_tokens_seen": 10453040,
      "step": 15985
    },
    {
      "epoch": 8.380503144654089,
      "grad_norm": 0.1392832100391388,
      "learning_rate": 3.60377283957826e-05,
      "loss": 0.5373,
      "num_input_tokens_seen": 10457072,
      "step": 15990
    },
    {
      "epoch": 8.383123689727464,
      "grad_norm": 0.5470751523971558,
      "learning_rate": 3.6027467752272203e-05,
      "loss": 0.3749,
      "num_input_tokens_seen": 10459280,
      "step": 15995
    },
    {
      "epoch": 8.385744234800839,
      "grad_norm": 0.24992576241493225,
      "learning_rate": 3.601720480194137e-05,
      "loss": 0.4566,
      "num_input_tokens_seen": 10462192,
      "step": 16000
    },
    {
      "epoch": 8.388364779874214,
      "grad_norm": 0.1629134863615036,
      "learning_rate": 3.600693954693698e-05,
      "loss": 0.4946,
      "num_input_tokens_seen": 10466672,
      "step": 16005
    },
    {
      "epoch": 8.39098532494759,
      "grad_norm": 0.14057357609272003,
      "learning_rate": 3.5996671989406414e-05,
      "loss": 0.4847,
      "num_input_tokens_seen": 10469776,
      "step": 16010
    },
    {
      "epoch": 8.393605870020965,
      "grad_norm": 0.24802453815937042,
      "learning_rate": 3.598640213149752e-05,
      "loss": 0.406,
      "num_input_tokens_seen": 10472848,
      "step": 16015
    },
    {
      "epoch": 8.39622641509434,
      "grad_norm": 0.5616933107376099,
      "learning_rate": 3.597612997535864e-05,
      "loss": 0.5791,
      "num_input_tokens_seen": 10476368,
      "step": 16020
    },
    {
      "epoch": 8.398846960167715,
      "grad_norm": 0.14671394228935242,
      "learning_rate": 3.596585552313859e-05,
      "loss": 0.5069,
      "num_input_tokens_seen": 10479472,
      "step": 16025
    },
    {
      "epoch": 8.40146750524109,
      "grad_norm": 0.21832002699375153,
      "learning_rate": 3.595557877698666e-05,
      "loss": 0.5609,
      "num_input_tokens_seen": 10482000,
      "step": 16030
    },
    {
      "epoch": 8.404088050314465,
      "grad_norm": 0.24715301394462585,
      "learning_rate": 3.594529973905266e-05,
      "loss": 0.4998,
      "num_input_tokens_seen": 10484592,
      "step": 16035
    },
    {
      "epoch": 8.40670859538784,
      "grad_norm": 0.41140687465667725,
      "learning_rate": 3.593501841148681e-05,
      "loss": 0.4795,
      "num_input_tokens_seen": 10487216,
      "step": 16040
    },
    {
      "epoch": 8.409329140461216,
      "grad_norm": 0.1944611519575119,
      "learning_rate": 3.5924734796439864e-05,
      "loss": 0.5013,
      "num_input_tokens_seen": 10490512,
      "step": 16045
    },
    {
      "epoch": 8.41194968553459,
      "grad_norm": 0.14257055521011353,
      "learning_rate": 3.5914448896063044e-05,
      "loss": 0.4056,
      "num_input_tokens_seen": 10493552,
      "step": 16050
    },
    {
      "epoch": 8.414570230607966,
      "grad_norm": 0.20303930342197418,
      "learning_rate": 3.590416071250802e-05,
      "loss": 0.5619,
      "num_input_tokens_seen": 10497264,
      "step": 16055
    },
    {
      "epoch": 8.417190775681341,
      "grad_norm": 0.21463006734848022,
      "learning_rate": 3.589387024792699e-05,
      "loss": 0.3804,
      "num_input_tokens_seen": 10499984,
      "step": 16060
    },
    {
      "epoch": 8.419811320754716,
      "grad_norm": 0.1661885380744934,
      "learning_rate": 3.5883577504472576e-05,
      "loss": 0.4461,
      "num_input_tokens_seen": 10503248,
      "step": 16065
    },
    {
      "epoch": 8.422431865828091,
      "grad_norm": 0.11123379319906235,
      "learning_rate": 3.587328248429792e-05,
      "loss": 0.5029,
      "num_input_tokens_seen": 10507088,
      "step": 16070
    },
    {
      "epoch": 8.425052410901468,
      "grad_norm": 0.1353880763053894,
      "learning_rate": 3.586298518955661e-05,
      "loss": 0.4457,
      "num_input_tokens_seen": 10510416,
      "step": 16075
    },
    {
      "epoch": 8.427672955974844,
      "grad_norm": 0.17165528237819672,
      "learning_rate": 3.5852685622402746e-05,
      "loss": 0.5542,
      "num_input_tokens_seen": 10514384,
      "step": 16080
    },
    {
      "epoch": 8.430293501048219,
      "grad_norm": 0.09782934933900833,
      "learning_rate": 3.584238378499085e-05,
      "loss": 0.5887,
      "num_input_tokens_seen": 10517872,
      "step": 16085
    },
    {
      "epoch": 8.432914046121594,
      "grad_norm": 0.16528111696243286,
      "learning_rate": 3.5832079679475974e-05,
      "loss": 0.5005,
      "num_input_tokens_seen": 10521200,
      "step": 16090
    },
    {
      "epoch": 8.435534591194969,
      "grad_norm": 0.10892472416162491,
      "learning_rate": 3.582177330801359e-05,
      "loss": 0.4899,
      "num_input_tokens_seen": 10523856,
      "step": 16095
    },
    {
      "epoch": 8.438155136268344,
      "grad_norm": 0.12413088977336884,
      "learning_rate": 3.58114646727597e-05,
      "loss": 0.5333,
      "num_input_tokens_seen": 10527728,
      "step": 16100
    },
    {
      "epoch": 8.44077568134172,
      "grad_norm": 0.17800016701221466,
      "learning_rate": 3.580115377587075e-05,
      "loss": 0.473,
      "num_input_tokens_seen": 10531536,
      "step": 16105
    },
    {
      "epoch": 8.443396226415095,
      "grad_norm": 0.16437461972236633,
      "learning_rate": 3.5790840619503654e-05,
      "loss": 0.4236,
      "num_input_tokens_seen": 10534352,
      "step": 16110
    },
    {
      "epoch": 8.44601677148847,
      "grad_norm": 0.16046595573425293,
      "learning_rate": 3.57805252058158e-05,
      "loss": 0.4886,
      "num_input_tokens_seen": 10537552,
      "step": 16115
    },
    {
      "epoch": 8.448637316561845,
      "grad_norm": 0.17331430315971375,
      "learning_rate": 3.577020753696507e-05,
      "loss": 0.5086,
      "num_input_tokens_seen": 10539952,
      "step": 16120
    },
    {
      "epoch": 8.45125786163522,
      "grad_norm": 0.16226010024547577,
      "learning_rate": 3.57598876151098e-05,
      "loss": 0.4689,
      "num_input_tokens_seen": 10543408,
      "step": 16125
    },
    {
      "epoch": 8.453878406708595,
      "grad_norm": 0.096927210688591,
      "learning_rate": 3.5749565442408796e-05,
      "loss": 0.392,
      "num_input_tokens_seen": 10546352,
      "step": 16130
    },
    {
      "epoch": 8.45649895178197,
      "grad_norm": 0.11097201704978943,
      "learning_rate": 3.573924102102133e-05,
      "loss": 0.4315,
      "num_input_tokens_seen": 10550352,
      "step": 16135
    },
    {
      "epoch": 8.459119496855346,
      "grad_norm": 0.11601868271827698,
      "learning_rate": 3.572891435310717e-05,
      "loss": 0.6256,
      "num_input_tokens_seen": 10554864,
      "step": 16140
    },
    {
      "epoch": 8.46174004192872,
      "grad_norm": 0.15461117029190063,
      "learning_rate": 3.5718585440826527e-05,
      "loss": 0.4117,
      "num_input_tokens_seen": 10558960,
      "step": 16145
    },
    {
      "epoch": 8.464360587002096,
      "grad_norm": 0.13390198349952698,
      "learning_rate": 3.570825428634009e-05,
      "loss": 0.4915,
      "num_input_tokens_seen": 10563600,
      "step": 16150
    },
    {
      "epoch": 8.466981132075471,
      "grad_norm": 0.2054513543844223,
      "learning_rate": 3.569792089180903e-05,
      "loss": 0.4844,
      "num_input_tokens_seen": 10566480,
      "step": 16155
    },
    {
      "epoch": 8.469601677148846,
      "grad_norm": 0.14556485414505005,
      "learning_rate": 3.568758525939497e-05,
      "loss": 0.3979,
      "num_input_tokens_seen": 10569616,
      "step": 16160
    },
    {
      "epoch": 8.472222222222221,
      "grad_norm": 0.14444677531719208,
      "learning_rate": 3.5677247391259995e-05,
      "loss": 0.5548,
      "num_input_tokens_seen": 10572336,
      "step": 16165
    },
    {
      "epoch": 8.474842767295598,
      "grad_norm": 0.23285461962223053,
      "learning_rate": 3.566690728956668e-05,
      "loss": 0.6144,
      "num_input_tokens_seen": 10574576,
      "step": 16170
    },
    {
      "epoch": 8.477463312368974,
      "grad_norm": 0.10744938999414444,
      "learning_rate": 3.5656564956478056e-05,
      "loss": 0.434,
      "num_input_tokens_seen": 10578640,
      "step": 16175
    },
    {
      "epoch": 8.480083857442349,
      "grad_norm": 0.13479292392730713,
      "learning_rate": 3.5646220394157626e-05,
      "loss": 0.3956,
      "num_input_tokens_seen": 10581392,
      "step": 16180
    },
    {
      "epoch": 8.482704402515724,
      "grad_norm": 0.20329895615577698,
      "learning_rate": 3.563587360476933e-05,
      "loss": 0.714,
      "num_input_tokens_seen": 10584144,
      "step": 16185
    },
    {
      "epoch": 8.485324947589099,
      "grad_norm": 0.32417649030685425,
      "learning_rate": 3.562552459047763e-05,
      "loss": 0.519,
      "num_input_tokens_seen": 10587600,
      "step": 16190
    },
    {
      "epoch": 8.487945492662474,
      "grad_norm": 0.24230065941810608,
      "learning_rate": 3.561517335344741e-05,
      "loss": 0.5084,
      "num_input_tokens_seen": 10590832,
      "step": 16195
    },
    {
      "epoch": 8.49056603773585,
      "grad_norm": 0.1777334064245224,
      "learning_rate": 3.560481989584402e-05,
      "loss": 0.5426,
      "num_input_tokens_seen": 10593712,
      "step": 16200
    },
    {
      "epoch": 8.493186582809225,
      "grad_norm": 0.1189345195889473,
      "learning_rate": 3.55944642198333e-05,
      "loss": 0.4482,
      "num_input_tokens_seen": 10596336,
      "step": 16205
    },
    {
      "epoch": 8.4958071278826,
      "grad_norm": 0.12890280783176422,
      "learning_rate": 3.5584106327581536e-05,
      "loss": 0.4731,
      "num_input_tokens_seen": 10599536,
      "step": 16210
    },
    {
      "epoch": 8.498427672955975,
      "grad_norm": 0.23539921641349792,
      "learning_rate": 3.557374622125547e-05,
      "loss": 0.2709,
      "num_input_tokens_seen": 10602576,
      "step": 16215
    },
    {
      "epoch": 8.50104821802935,
      "grad_norm": 0.18078073859214783,
      "learning_rate": 3.556338390302234e-05,
      "loss": 0.419,
      "num_input_tokens_seen": 10605616,
      "step": 16220
    },
    {
      "epoch": 8.503668763102725,
      "grad_norm": 0.24226514995098114,
      "learning_rate": 3.55530193750498e-05,
      "loss": 0.6614,
      "num_input_tokens_seen": 10609040,
      "step": 16225
    },
    {
      "epoch": 8.5062893081761,
      "grad_norm": 0.1250118762254715,
      "learning_rate": 3.5542652639505996e-05,
      "loss": 0.5803,
      "num_input_tokens_seen": 10612304,
      "step": 16230
    },
    {
      "epoch": 8.508909853249476,
      "grad_norm": 0.15247629582881927,
      "learning_rate": 3.5532283698559556e-05,
      "loss": 0.419,
      "num_input_tokens_seen": 10614928,
      "step": 16235
    },
    {
      "epoch": 8.51153039832285,
      "grad_norm": 0.13414108753204346,
      "learning_rate": 3.55219125543795e-05,
      "loss": 0.6081,
      "num_input_tokens_seen": 10618320,
      "step": 16240
    },
    {
      "epoch": 8.514150943396226,
      "grad_norm": 0.2575390934944153,
      "learning_rate": 3.5511539209135396e-05,
      "loss": 0.5075,
      "num_input_tokens_seen": 10621296,
      "step": 16245
    },
    {
      "epoch": 8.516771488469601,
      "grad_norm": 0.14297901093959808,
      "learning_rate": 3.550116366499721e-05,
      "loss": 0.5411,
      "num_input_tokens_seen": 10623984,
      "step": 16250
    },
    {
      "epoch": 8.519392033542976,
      "grad_norm": 0.19575834274291992,
      "learning_rate": 3.549078592413538e-05,
      "loss": 0.4638,
      "num_input_tokens_seen": 10628560,
      "step": 16255
    },
    {
      "epoch": 8.522012578616351,
      "grad_norm": 0.2511533200740814,
      "learning_rate": 3.548040598872083e-05,
      "loss": 0.4432,
      "num_input_tokens_seen": 10631856,
      "step": 16260
    },
    {
      "epoch": 8.524633123689728,
      "grad_norm": 0.14475637674331665,
      "learning_rate": 3.5470023860924905e-05,
      "loss": 0.4696,
      "num_input_tokens_seen": 10635696,
      "step": 16265
    },
    {
      "epoch": 8.527253668763104,
      "grad_norm": 0.2869107127189636,
      "learning_rate": 3.5459639542919444e-05,
      "loss": 0.5172,
      "num_input_tokens_seen": 10638384,
      "step": 16270
    },
    {
      "epoch": 8.529874213836479,
      "grad_norm": 0.18690098822116852,
      "learning_rate": 3.5449253036876715e-05,
      "loss": 0.5154,
      "num_input_tokens_seen": 10641456,
      "step": 16275
    },
    {
      "epoch": 8.532494758909854,
      "grad_norm": 0.32674163579940796,
      "learning_rate": 3.543886434496947e-05,
      "loss": 0.4216,
      "num_input_tokens_seen": 10644336,
      "step": 16280
    },
    {
      "epoch": 8.535115303983229,
      "grad_norm": 0.15268854796886444,
      "learning_rate": 3.542847346937089e-05,
      "loss": 0.4209,
      "num_input_tokens_seen": 10647440,
      "step": 16285
    },
    {
      "epoch": 8.537735849056604,
      "grad_norm": 0.13819049298763275,
      "learning_rate": 3.541808041225464e-05,
      "loss": 0.467,
      "num_input_tokens_seen": 10649872,
      "step": 16290
    },
    {
      "epoch": 8.54035639412998,
      "grad_norm": 0.11388176679611206,
      "learning_rate": 3.540768517579481e-05,
      "loss": 0.5381,
      "num_input_tokens_seen": 10652976,
      "step": 16295
    },
    {
      "epoch": 8.542976939203355,
      "grad_norm": 0.16769404709339142,
      "learning_rate": 3.539728776216599e-05,
      "loss": 0.4511,
      "num_input_tokens_seen": 10656016,
      "step": 16300
    },
    {
      "epoch": 8.54559748427673,
      "grad_norm": 0.1997755765914917,
      "learning_rate": 3.5386888173543185e-05,
      "loss": 0.5325,
      "num_input_tokens_seen": 10659536,
      "step": 16305
    },
    {
      "epoch": 8.548218029350105,
      "grad_norm": 0.21682070195674896,
      "learning_rate": 3.5376486412101875e-05,
      "loss": 0.3889,
      "num_input_tokens_seen": 10662416,
      "step": 16310
    },
    {
      "epoch": 8.55083857442348,
      "grad_norm": 0.09498199820518494,
      "learning_rate": 3.5366082480017994e-05,
      "loss": 0.4337,
      "num_input_tokens_seen": 10666320,
      "step": 16315
    },
    {
      "epoch": 8.553459119496855,
      "grad_norm": 0.2158992439508438,
      "learning_rate": 3.535567637946792e-05,
      "loss": 0.6666,
      "num_input_tokens_seen": 10669136,
      "step": 16320
    },
    {
      "epoch": 8.55607966457023,
      "grad_norm": 0.18772639334201813,
      "learning_rate": 3.534526811262848e-05,
      "loss": 0.4734,
      "num_input_tokens_seen": 10672240,
      "step": 16325
    },
    {
      "epoch": 8.558700209643606,
      "grad_norm": 0.17574481666088104,
      "learning_rate": 3.533485768167698e-05,
      "loss": 0.4989,
      "num_input_tokens_seen": 10674992,
      "step": 16330
    },
    {
      "epoch": 8.56132075471698,
      "grad_norm": 0.1937495470046997,
      "learning_rate": 3.532444508879117e-05,
      "loss": 0.5332,
      "num_input_tokens_seen": 10677616,
      "step": 16335
    },
    {
      "epoch": 8.563941299790356,
      "grad_norm": 0.16123315691947937,
      "learning_rate": 3.531403033614922e-05,
      "loss": 0.4918,
      "num_input_tokens_seen": 10681136,
      "step": 16340
    },
    {
      "epoch": 8.566561844863731,
      "grad_norm": 0.20404431223869324,
      "learning_rate": 3.530361342592981e-05,
      "loss": 0.6412,
      "num_input_tokens_seen": 10684368,
      "step": 16345
    },
    {
      "epoch": 8.569182389937106,
      "grad_norm": 0.12984082102775574,
      "learning_rate": 3.5293194360312006e-05,
      "loss": 0.5599,
      "num_input_tokens_seen": 10687632,
      "step": 16350
    },
    {
      "epoch": 8.571802935010481,
      "grad_norm": 0.16187255084514618,
      "learning_rate": 3.528277314147538e-05,
      "loss": 0.5084,
      "num_input_tokens_seen": 10692336,
      "step": 16355
    },
    {
      "epoch": 8.574423480083858,
      "grad_norm": 0.10501859337091446,
      "learning_rate": 3.5272349771599925e-05,
      "loss": 0.5417,
      "num_input_tokens_seen": 10696048,
      "step": 16360
    },
    {
      "epoch": 8.577044025157234,
      "grad_norm": 0.4274337887763977,
      "learning_rate": 3.526192425286609e-05,
      "loss": 0.4674,
      "num_input_tokens_seen": 10700304,
      "step": 16365
    },
    {
      "epoch": 8.579664570230609,
      "grad_norm": 0.18405888974666595,
      "learning_rate": 3.525149658745476e-05,
      "loss": 0.5077,
      "num_input_tokens_seen": 10703120,
      "step": 16370
    },
    {
      "epoch": 8.582285115303984,
      "grad_norm": 0.08797333389520645,
      "learning_rate": 3.5241066777547315e-05,
      "loss": 0.5027,
      "num_input_tokens_seen": 10707536,
      "step": 16375
    },
    {
      "epoch": 8.584905660377359,
      "grad_norm": 0.13657347857952118,
      "learning_rate": 3.523063482532554e-05,
      "loss": 0.575,
      "num_input_tokens_seen": 10710384,
      "step": 16380
    },
    {
      "epoch": 8.587526205450734,
      "grad_norm": 0.22482828795909882,
      "learning_rate": 3.5220200732971665e-05,
      "loss": 0.483,
      "num_input_tokens_seen": 10713840,
      "step": 16385
    },
    {
      "epoch": 8.59014675052411,
      "grad_norm": 0.13642726838588715,
      "learning_rate": 3.52097645026684e-05,
      "loss": 0.4364,
      "num_input_tokens_seen": 10716720,
      "step": 16390
    },
    {
      "epoch": 8.592767295597485,
      "grad_norm": 0.17681005597114563,
      "learning_rate": 3.519932613659889e-05,
      "loss": 0.5394,
      "num_input_tokens_seen": 10720720,
      "step": 16395
    },
    {
      "epoch": 8.59538784067086,
      "grad_norm": 0.17145729064941406,
      "learning_rate": 3.5188885636946696e-05,
      "loss": 0.4656,
      "num_input_tokens_seen": 10723952,
      "step": 16400
    },
    {
      "epoch": 8.598008385744235,
      "grad_norm": 0.21995288133621216,
      "learning_rate": 3.5178443005895884e-05,
      "loss": 0.5844,
      "num_input_tokens_seen": 10726096,
      "step": 16405
    },
    {
      "epoch": 8.60062893081761,
      "grad_norm": 0.10801391303539276,
      "learning_rate": 3.51679982456309e-05,
      "loss": 0.4063,
      "num_input_tokens_seen": 10729456,
      "step": 16410
    },
    {
      "epoch": 8.603249475890985,
      "grad_norm": 0.11996271461248398,
      "learning_rate": 3.51575513583367e-05,
      "loss": 0.3782,
      "num_input_tokens_seen": 10731824,
      "step": 16415
    },
    {
      "epoch": 8.60587002096436,
      "grad_norm": 0.12905818223953247,
      "learning_rate": 3.5147102346198635e-05,
      "loss": 0.3748,
      "num_input_tokens_seen": 10734864,
      "step": 16420
    },
    {
      "epoch": 8.608490566037736,
      "grad_norm": 0.18255086243152618,
      "learning_rate": 3.5136651211402515e-05,
      "loss": 0.4901,
      "num_input_tokens_seen": 10737744,
      "step": 16425
    },
    {
      "epoch": 8.61111111111111,
      "grad_norm": 0.17542758584022522,
      "learning_rate": 3.5126197956134614e-05,
      "loss": 0.3495,
      "num_input_tokens_seen": 10740656,
      "step": 16430
    },
    {
      "epoch": 8.613731656184486,
      "grad_norm": 0.24186812341213226,
      "learning_rate": 3.5115742582581626e-05,
      "loss": 0.7436,
      "num_input_tokens_seen": 10744688,
      "step": 16435
    },
    {
      "epoch": 8.616352201257861,
      "grad_norm": 0.1601296067237854,
      "learning_rate": 3.51052850929307e-05,
      "loss": 0.5553,
      "num_input_tokens_seen": 10747184,
      "step": 16440
    },
    {
      "epoch": 8.618972746331236,
      "grad_norm": 0.1923363357782364,
      "learning_rate": 3.509482548936941e-05,
      "loss": 0.6285,
      "num_input_tokens_seen": 10750320,
      "step": 16445
    },
    {
      "epoch": 8.621593291404611,
      "grad_norm": 0.15598824620246887,
      "learning_rate": 3.5084363774085804e-05,
      "loss": 0.409,
      "num_input_tokens_seen": 10753328,
      "step": 16450
    },
    {
      "epoch": 8.624213836477988,
      "grad_norm": 0.15142881870269775,
      "learning_rate": 3.507389994926834e-05,
      "loss": 0.5094,
      "num_input_tokens_seen": 10756208,
      "step": 16455
    },
    {
      "epoch": 8.626834381551364,
      "grad_norm": 0.21682582795619965,
      "learning_rate": 3.5063434017105946e-05,
      "loss": 0.4583,
      "num_input_tokens_seen": 10759888,
      "step": 16460
    },
    {
      "epoch": 8.629454926624739,
      "grad_norm": 0.24904772639274597,
      "learning_rate": 3.505296597978795e-05,
      "loss": 0.4559,
      "num_input_tokens_seen": 10763152,
      "step": 16465
    },
    {
      "epoch": 8.632075471698114,
      "grad_norm": 0.20025841891765594,
      "learning_rate": 3.504249583950417e-05,
      "loss": 0.4992,
      "num_input_tokens_seen": 10767024,
      "step": 16470
    },
    {
      "epoch": 8.634696016771489,
      "grad_norm": 0.14593705534934998,
      "learning_rate": 3.503202359844483e-05,
      "loss": 0.381,
      "num_input_tokens_seen": 10770096,
      "step": 16475
    },
    {
      "epoch": 8.637316561844864,
      "grad_norm": 0.19756081700325012,
      "learning_rate": 3.50215492588006e-05,
      "loss": 0.5587,
      "num_input_tokens_seen": 10774576,
      "step": 16480
    },
    {
      "epoch": 8.63993710691824,
      "grad_norm": 0.12068936973810196,
      "learning_rate": 3.5011072822762605e-05,
      "loss": 0.3734,
      "num_input_tokens_seen": 10777296,
      "step": 16485
    },
    {
      "epoch": 8.642557651991615,
      "grad_norm": 0.12480912357568741,
      "learning_rate": 3.500059429252238e-05,
      "loss": 0.4355,
      "num_input_tokens_seen": 10779696,
      "step": 16490
    },
    {
      "epoch": 8.64517819706499,
      "grad_norm": 0.2043144851922989,
      "learning_rate": 3.499011367027193e-05,
      "loss": 0.4313,
      "num_input_tokens_seen": 10783184,
      "step": 16495
    },
    {
      "epoch": 8.647798742138365,
      "grad_norm": 0.13129819929599762,
      "learning_rate": 3.4979630958203665e-05,
      "loss": 0.6479,
      "num_input_tokens_seen": 10786512,
      "step": 16500
    },
    {
      "epoch": 8.65041928721174,
      "grad_norm": 0.1589341163635254,
      "learning_rate": 3.496914615851046e-05,
      "loss": 0.4477,
      "num_input_tokens_seen": 10789872,
      "step": 16505
    },
    {
      "epoch": 8.653039832285115,
      "grad_norm": 0.13432809710502625,
      "learning_rate": 3.495865927338562e-05,
      "loss": 0.4342,
      "num_input_tokens_seen": 10792944,
      "step": 16510
    },
    {
      "epoch": 8.65566037735849,
      "grad_norm": 0.20240139961242676,
      "learning_rate": 3.494817030502286e-05,
      "loss": 0.5098,
      "num_input_tokens_seen": 10796624,
      "step": 16515
    },
    {
      "epoch": 8.658280922431866,
      "grad_norm": 0.13596144318580627,
      "learning_rate": 3.493767925561638e-05,
      "loss": 0.4872,
      "num_input_tokens_seen": 10800144,
      "step": 16520
    },
    {
      "epoch": 8.66090146750524,
      "grad_norm": 0.16646413505077362,
      "learning_rate": 3.492718612736077e-05,
      "loss": 0.442,
      "num_input_tokens_seen": 10803152,
      "step": 16525
    },
    {
      "epoch": 8.663522012578616,
      "grad_norm": 0.2757636308670044,
      "learning_rate": 3.491669092245108e-05,
      "loss": 0.5292,
      "num_input_tokens_seen": 10805552,
      "step": 16530
    },
    {
      "epoch": 8.666142557651991,
      "grad_norm": 0.1675666868686676,
      "learning_rate": 3.4906193643082784e-05,
      "loss": 0.4752,
      "num_input_tokens_seen": 10808880,
      "step": 16535
    },
    {
      "epoch": 8.668763102725366,
      "grad_norm": 0.17688873410224915,
      "learning_rate": 3.4895694291451795e-05,
      "loss": 0.461,
      "num_input_tokens_seen": 10812240,
      "step": 16540
    },
    {
      "epoch": 8.671383647798741,
      "grad_norm": 0.27651679515838623,
      "learning_rate": 3.4885192869754444e-05,
      "loss": 0.4471,
      "num_input_tokens_seen": 10815600,
      "step": 16545
    },
    {
      "epoch": 8.674004192872118,
      "grad_norm": 0.15440334379673004,
      "learning_rate": 3.4874689380187537e-05,
      "loss": 0.4708,
      "num_input_tokens_seen": 10819120,
      "step": 16550
    },
    {
      "epoch": 8.676624737945493,
      "grad_norm": 0.20403526723384857,
      "learning_rate": 3.486418382494826e-05,
      "loss": 0.4232,
      "num_input_tokens_seen": 10821744,
      "step": 16555
    },
    {
      "epoch": 8.679245283018869,
      "grad_norm": 0.30493488907814026,
      "learning_rate": 3.485367620623426e-05,
      "loss": 0.4582,
      "num_input_tokens_seen": 10824336,
      "step": 16560
    },
    {
      "epoch": 8.681865828092244,
      "grad_norm": 0.21671976149082184,
      "learning_rate": 3.484316652624362e-05,
      "loss": 0.6096,
      "num_input_tokens_seen": 10826672,
      "step": 16565
    },
    {
      "epoch": 8.684486373165619,
      "grad_norm": 0.13318860530853271,
      "learning_rate": 3.483265478717484e-05,
      "loss": 0.4182,
      "num_input_tokens_seen": 10830192,
      "step": 16570
    },
    {
      "epoch": 8.687106918238994,
      "grad_norm": 0.2867143154144287,
      "learning_rate": 3.482214099122686e-05,
      "loss": 0.6684,
      "num_input_tokens_seen": 10834096,
      "step": 16575
    },
    {
      "epoch": 8.68972746331237,
      "grad_norm": 0.21335697174072266,
      "learning_rate": 3.481162514059903e-05,
      "loss": 0.4905,
      "num_input_tokens_seen": 10837360,
      "step": 16580
    },
    {
      "epoch": 8.692348008385745,
      "grad_norm": 0.15922965109348297,
      "learning_rate": 3.480110723749116e-05,
      "loss": 0.5836,
      "num_input_tokens_seen": 10841168,
      "step": 16585
    },
    {
      "epoch": 8.69496855345912,
      "grad_norm": 0.20324312150478363,
      "learning_rate": 3.479058728410348e-05,
      "loss": 0.3634,
      "num_input_tokens_seen": 10844784,
      "step": 16590
    },
    {
      "epoch": 8.697589098532495,
      "grad_norm": 0.1972815841436386,
      "learning_rate": 3.478006528263663e-05,
      "loss": 0.482,
      "num_input_tokens_seen": 10847376,
      "step": 16595
    },
    {
      "epoch": 8.70020964360587,
      "grad_norm": 0.16956497728824615,
      "learning_rate": 3.4769541235291704e-05,
      "loss": 0.4305,
      "num_input_tokens_seen": 10850000,
      "step": 16600
    },
    {
      "epoch": 8.702830188679245,
      "grad_norm": 0.24145916104316711,
      "learning_rate": 3.475901514427021e-05,
      "loss": 0.3927,
      "num_input_tokens_seen": 10852528,
      "step": 16605
    },
    {
      "epoch": 8.70545073375262,
      "grad_norm": 0.2650786340236664,
      "learning_rate": 3.4748487011774075e-05,
      "loss": 0.4603,
      "num_input_tokens_seen": 10855312,
      "step": 16610
    },
    {
      "epoch": 8.708071278825996,
      "grad_norm": 0.11061473935842514,
      "learning_rate": 3.473795684000569e-05,
      "loss": 0.6244,
      "num_input_tokens_seen": 10859376,
      "step": 16615
    },
    {
      "epoch": 8.71069182389937,
      "grad_norm": 0.1875632256269455,
      "learning_rate": 3.4727424631167815e-05,
      "loss": 0.5293,
      "num_input_tokens_seen": 10862224,
      "step": 16620
    },
    {
      "epoch": 8.713312368972746,
      "grad_norm": 0.2578648328781128,
      "learning_rate": 3.471689038746369e-05,
      "loss": 0.6446,
      "num_input_tokens_seen": 10865104,
      "step": 16625
    },
    {
      "epoch": 8.715932914046121,
      "grad_norm": 0.174208864569664,
      "learning_rate": 3.470635411109695e-05,
      "loss": 0.4154,
      "num_input_tokens_seen": 10867792,
      "step": 16630
    },
    {
      "epoch": 8.718553459119496,
      "grad_norm": 0.24583186209201813,
      "learning_rate": 3.469581580427167e-05,
      "loss": 0.529,
      "num_input_tokens_seen": 10870768,
      "step": 16635
    },
    {
      "epoch": 8.721174004192871,
      "grad_norm": 0.12855452299118042,
      "learning_rate": 3.4685275469192335e-05,
      "loss": 0.358,
      "num_input_tokens_seen": 10875056,
      "step": 16640
    },
    {
      "epoch": 8.723794549266248,
      "grad_norm": 0.21881558001041412,
      "learning_rate": 3.467473310806387e-05,
      "loss": 0.5459,
      "num_input_tokens_seen": 10878288,
      "step": 16645
    },
    {
      "epoch": 8.726415094339622,
      "grad_norm": 0.21593829989433289,
      "learning_rate": 3.466418872309161e-05,
      "loss": 0.5267,
      "num_input_tokens_seen": 10880720,
      "step": 16650
    },
    {
      "epoch": 8.729035639412999,
      "grad_norm": 0.18044479191303253,
      "learning_rate": 3.465364231648133e-05,
      "loss": 0.3194,
      "num_input_tokens_seen": 10884528,
      "step": 16655
    },
    {
      "epoch": 8.731656184486374,
      "grad_norm": 0.18525166809558868,
      "learning_rate": 3.464309389043921e-05,
      "loss": 0.5939,
      "num_input_tokens_seen": 10886960,
      "step": 16660
    },
    {
      "epoch": 8.734276729559749,
      "grad_norm": 0.12196366488933563,
      "learning_rate": 3.4632543447171856e-05,
      "loss": 0.5038,
      "num_input_tokens_seen": 10889904,
      "step": 16665
    },
    {
      "epoch": 8.736897274633124,
      "grad_norm": 0.158524751663208,
      "learning_rate": 3.4621990988886315e-05,
      "loss": 0.6071,
      "num_input_tokens_seen": 10893072,
      "step": 16670
    },
    {
      "epoch": 8.7395178197065,
      "grad_norm": 0.14804241061210632,
      "learning_rate": 3.461143651779003e-05,
      "loss": 0.5192,
      "num_input_tokens_seen": 10897392,
      "step": 16675
    },
    {
      "epoch": 8.742138364779874,
      "grad_norm": 0.11649242788553238,
      "learning_rate": 3.4600880036090876e-05,
      "loss": 0.5827,
      "num_input_tokens_seen": 10901008,
      "step": 16680
    },
    {
      "epoch": 8.74475890985325,
      "grad_norm": 0.1562197059392929,
      "learning_rate": 3.459032154599715e-05,
      "loss": 0.5548,
      "num_input_tokens_seen": 10904848,
      "step": 16685
    },
    {
      "epoch": 8.747379454926625,
      "grad_norm": 0.23452891409397125,
      "learning_rate": 3.457976104971757e-05,
      "loss": 0.5011,
      "num_input_tokens_seen": 10908528,
      "step": 16690
    },
    {
      "epoch": 8.75,
      "grad_norm": 0.1804642230272293,
      "learning_rate": 3.4569198549461276e-05,
      "loss": 0.5854,
      "num_input_tokens_seen": 10912080,
      "step": 16695
    },
    {
      "epoch": 8.752620545073375,
      "grad_norm": 0.12914183735847473,
      "learning_rate": 3.455863404743781e-05,
      "loss": 0.4969,
      "num_input_tokens_seen": 10916528,
      "step": 16700
    },
    {
      "epoch": 8.75524109014675,
      "grad_norm": 0.18157972395420074,
      "learning_rate": 3.454806754585716e-05,
      "loss": 0.4618,
      "num_input_tokens_seen": 10920688,
      "step": 16705
    },
    {
      "epoch": 8.757861635220126,
      "grad_norm": 0.22277849912643433,
      "learning_rate": 3.453749904692971e-05,
      "loss": 0.4679,
      "num_input_tokens_seen": 10923600,
      "step": 16710
    },
    {
      "epoch": 8.7604821802935,
      "grad_norm": 0.12191981077194214,
      "learning_rate": 3.452692855286626e-05,
      "loss": 0.4117,
      "num_input_tokens_seen": 10926544,
      "step": 16715
    },
    {
      "epoch": 8.763102725366876,
      "grad_norm": 0.12648673355579376,
      "learning_rate": 3.4516356065878055e-05,
      "loss": 0.4233,
      "num_input_tokens_seen": 10929392,
      "step": 16720
    },
    {
      "epoch": 8.765723270440251,
      "grad_norm": 0.19302968680858612,
      "learning_rate": 3.450578158817671e-05,
      "loss": 0.4448,
      "num_input_tokens_seen": 10933456,
      "step": 16725
    },
    {
      "epoch": 8.768343815513626,
      "grad_norm": 0.191706582903862,
      "learning_rate": 3.4495205121974325e-05,
      "loss": 0.6336,
      "num_input_tokens_seen": 10936336,
      "step": 16730
    },
    {
      "epoch": 8.770964360587001,
      "grad_norm": 0.17411883175373077,
      "learning_rate": 3.448462666948335e-05,
      "loss": 0.5147,
      "num_input_tokens_seen": 10939696,
      "step": 16735
    },
    {
      "epoch": 8.773584905660378,
      "grad_norm": 0.17510469257831573,
      "learning_rate": 3.4474046232916674e-05,
      "loss": 0.4037,
      "num_input_tokens_seen": 10942192,
      "step": 16740
    },
    {
      "epoch": 8.776205450733752,
      "grad_norm": 0.2264733612537384,
      "learning_rate": 3.4463463814487616e-05,
      "loss": 0.3105,
      "num_input_tokens_seen": 10944624,
      "step": 16745
    },
    {
      "epoch": 8.778825995807129,
      "grad_norm": 0.09984707087278366,
      "learning_rate": 3.44528794164099e-05,
      "loss": 0.421,
      "num_input_tokens_seen": 10947952,
      "step": 16750
    },
    {
      "epoch": 8.781446540880504,
      "grad_norm": 0.25856950879096985,
      "learning_rate": 3.4442293040897646e-05,
      "loss": 0.4585,
      "num_input_tokens_seen": 10951440,
      "step": 16755
    },
    {
      "epoch": 8.784067085953879,
      "grad_norm": 0.15685325860977173,
      "learning_rate": 3.443170469016542e-05,
      "loss": 0.425,
      "num_input_tokens_seen": 10954224,
      "step": 16760
    },
    {
      "epoch": 8.786687631027254,
      "grad_norm": 0.41606998443603516,
      "learning_rate": 3.442111436642817e-05,
      "loss": 0.5027,
      "num_input_tokens_seen": 10957392,
      "step": 16765
    },
    {
      "epoch": 8.78930817610063,
      "grad_norm": 0.09781327098608017,
      "learning_rate": 3.441052207190127e-05,
      "loss": 0.5369,
      "num_input_tokens_seen": 10961136,
      "step": 16770
    },
    {
      "epoch": 8.791928721174004,
      "grad_norm": 0.1491025984287262,
      "learning_rate": 3.4399927808800526e-05,
      "loss": 0.5235,
      "num_input_tokens_seen": 10964272,
      "step": 16775
    },
    {
      "epoch": 8.79454926624738,
      "grad_norm": 0.15143737196922302,
      "learning_rate": 3.438933157934211e-05,
      "loss": 0.3937,
      "num_input_tokens_seen": 10967344,
      "step": 16780
    },
    {
      "epoch": 8.797169811320755,
      "grad_norm": 0.16212455928325653,
      "learning_rate": 3.4378733385742653e-05,
      "loss": 0.4341,
      "num_input_tokens_seen": 10970800,
      "step": 16785
    },
    {
      "epoch": 8.79979035639413,
      "grad_norm": 0.10377275943756104,
      "learning_rate": 3.436813323021918e-05,
      "loss": 0.5052,
      "num_input_tokens_seen": 10974544,
      "step": 16790
    },
    {
      "epoch": 8.802410901467505,
      "grad_norm": 0.4216497242450714,
      "learning_rate": 3.43575311149891e-05,
      "loss": 0.5476,
      "num_input_tokens_seen": 10976784,
      "step": 16795
    },
    {
      "epoch": 8.80503144654088,
      "grad_norm": 0.1362956017255783,
      "learning_rate": 3.434692704227028e-05,
      "loss": 0.5597,
      "num_input_tokens_seen": 10980400,
      "step": 16800
    },
    {
      "epoch": 8.807651991614255,
      "grad_norm": 0.16704051196575165,
      "learning_rate": 3.4336321014280944e-05,
      "loss": 0.3655,
      "num_input_tokens_seen": 10986160,
      "step": 16805
    },
    {
      "epoch": 8.81027253668763,
      "grad_norm": 0.18315689265727997,
      "learning_rate": 3.432571303323977e-05,
      "loss": 0.5276,
      "num_input_tokens_seen": 10989584,
      "step": 16810
    },
    {
      "epoch": 8.812893081761006,
      "grad_norm": 0.20867633819580078,
      "learning_rate": 3.431510310136583e-05,
      "loss": 0.5755,
      "num_input_tokens_seen": 10992464,
      "step": 16815
    },
    {
      "epoch": 8.815513626834381,
      "grad_norm": 0.16216357052326202,
      "learning_rate": 3.4304491220878585e-05,
      "loss": 0.5843,
      "num_input_tokens_seen": 10995472,
      "step": 16820
    },
    {
      "epoch": 8.818134171907756,
      "grad_norm": 0.3114825487136841,
      "learning_rate": 3.429387739399793e-05,
      "loss": 0.4542,
      "num_input_tokens_seen": 10999088,
      "step": 16825
    },
    {
      "epoch": 8.820754716981131,
      "grad_norm": 0.15657185018062592,
      "learning_rate": 3.428326162294416e-05,
      "loss": 0.4881,
      "num_input_tokens_seen": 11002000,
      "step": 16830
    },
    {
      "epoch": 8.823375262054507,
      "grad_norm": 0.2877059280872345,
      "learning_rate": 3.427264390993795e-05,
      "loss": 0.6612,
      "num_input_tokens_seen": 11005360,
      "step": 16835
    },
    {
      "epoch": 8.825995807127882,
      "grad_norm": 0.13634544610977173,
      "learning_rate": 3.426202425720044e-05,
      "loss": 0.4151,
      "num_input_tokens_seen": 11008272,
      "step": 16840
    },
    {
      "epoch": 8.828616352201259,
      "grad_norm": 0.15599559247493744,
      "learning_rate": 3.425140266695312e-05,
      "loss": 0.4647,
      "num_input_tokens_seen": 11011760,
      "step": 16845
    },
    {
      "epoch": 8.831236897274634,
      "grad_norm": 0.16441063582897186,
      "learning_rate": 3.4240779141417905e-05,
      "loss": 0.5346,
      "num_input_tokens_seen": 11014736,
      "step": 16850
    },
    {
      "epoch": 8.833857442348009,
      "grad_norm": 0.18347816169261932,
      "learning_rate": 3.423015368281711e-05,
      "loss": 0.4194,
      "num_input_tokens_seen": 11018128,
      "step": 16855
    },
    {
      "epoch": 8.836477987421384,
      "grad_norm": 0.2021685540676117,
      "learning_rate": 3.421952629337348e-05,
      "loss": 0.4214,
      "num_input_tokens_seen": 11022032,
      "step": 16860
    },
    {
      "epoch": 8.83909853249476,
      "grad_norm": 0.18003667891025543,
      "learning_rate": 3.420889697531013e-05,
      "loss": 0.4749,
      "num_input_tokens_seen": 11024944,
      "step": 16865
    },
    {
      "epoch": 8.841719077568134,
      "grad_norm": 0.14485317468643188,
      "learning_rate": 3.419826573085058e-05,
      "loss": 0.4566,
      "num_input_tokens_seen": 11028272,
      "step": 16870
    },
    {
      "epoch": 8.84433962264151,
      "grad_norm": 0.3317781388759613,
      "learning_rate": 3.418763256221879e-05,
      "loss": 0.5139,
      "num_input_tokens_seen": 11031088,
      "step": 16875
    },
    {
      "epoch": 8.846960167714885,
      "grad_norm": 0.20718568563461304,
      "learning_rate": 3.4176997471639085e-05,
      "loss": 0.5435,
      "num_input_tokens_seen": 11033456,
      "step": 16880
    },
    {
      "epoch": 8.84958071278826,
      "grad_norm": 0.1290135383605957,
      "learning_rate": 3.4166360461336206e-05,
      "loss": 0.5135,
      "num_input_tokens_seen": 11036976,
      "step": 16885
    },
    {
      "epoch": 8.852201257861635,
      "grad_norm": 0.12149849534034729,
      "learning_rate": 3.41557215335353e-05,
      "loss": 0.4856,
      "num_input_tokens_seen": 11039472,
      "step": 16890
    },
    {
      "epoch": 8.85482180293501,
      "grad_norm": 0.15236108005046844,
      "learning_rate": 3.4145080690461887e-05,
      "loss": 0.4571,
      "num_input_tokens_seen": 11042064,
      "step": 16895
    },
    {
      "epoch": 8.857442348008385,
      "grad_norm": 0.10929001867771149,
      "learning_rate": 3.4134437934341936e-05,
      "loss": 0.5297,
      "num_input_tokens_seen": 11045232,
      "step": 16900
    },
    {
      "epoch": 8.86006289308176,
      "grad_norm": 0.148976668715477,
      "learning_rate": 3.4123793267401784e-05,
      "loss": 0.4412,
      "num_input_tokens_seen": 11048784,
      "step": 16905
    },
    {
      "epoch": 8.862683438155136,
      "grad_norm": 0.14461228251457214,
      "learning_rate": 3.411314669186816e-05,
      "loss": 0.5274,
      "num_input_tokens_seen": 11052528,
      "step": 16910
    },
    {
      "epoch": 8.865303983228511,
      "grad_norm": 0.3804866075515747,
      "learning_rate": 3.410249820996823e-05,
      "loss": 0.7611,
      "num_input_tokens_seen": 11055824,
      "step": 16915
    },
    {
      "epoch": 8.867924528301886,
      "grad_norm": 0.15461388230323792,
      "learning_rate": 3.4091847823929513e-05,
      "loss": 0.3401,
      "num_input_tokens_seen": 11059504,
      "step": 16920
    },
    {
      "epoch": 8.870545073375261,
      "grad_norm": 0.13339926302433014,
      "learning_rate": 3.4081195535979946e-05,
      "loss": 0.4739,
      "num_input_tokens_seen": 11062544,
      "step": 16925
    },
    {
      "epoch": 8.873165618448636,
      "grad_norm": 0.12548039853572845,
      "learning_rate": 3.40705413483479e-05,
      "loss": 0.4607,
      "num_input_tokens_seen": 11065584,
      "step": 16930
    },
    {
      "epoch": 8.875786163522012,
      "grad_norm": 0.2359761744737625,
      "learning_rate": 3.405988526326207e-05,
      "loss": 0.4043,
      "num_input_tokens_seen": 11068336,
      "step": 16935
    },
    {
      "epoch": 8.878406708595389,
      "grad_norm": 0.2041347771883011,
      "learning_rate": 3.40492272829516e-05,
      "loss": 0.5285,
      "num_input_tokens_seen": 11071184,
      "step": 16940
    },
    {
      "epoch": 8.881027253668764,
      "grad_norm": 0.3759242296218872,
      "learning_rate": 3.403856740964604e-05,
      "loss": 0.3189,
      "num_input_tokens_seen": 11074320,
      "step": 16945
    },
    {
      "epoch": 8.883647798742139,
      "grad_norm": 0.1325937956571579,
      "learning_rate": 3.4027905645575275e-05,
      "loss": 0.4458,
      "num_input_tokens_seen": 11077168,
      "step": 16950
    },
    {
      "epoch": 8.886268343815514,
      "grad_norm": 0.3215600848197937,
      "learning_rate": 3.401724199296966e-05,
      "loss": 0.4722,
      "num_input_tokens_seen": 11080080,
      "step": 16955
    },
    {
      "epoch": 8.88888888888889,
      "grad_norm": 0.13802139461040497,
      "learning_rate": 3.400657645405989e-05,
      "loss": 0.3848,
      "num_input_tokens_seen": 11083408,
      "step": 16960
    },
    {
      "epoch": 8.891509433962264,
      "grad_norm": 0.16379199922084808,
      "learning_rate": 3.3995909031077065e-05,
      "loss": 0.5871,
      "num_input_tokens_seen": 11087184,
      "step": 16965
    },
    {
      "epoch": 8.89412997903564,
      "grad_norm": 0.14888381958007812,
      "learning_rate": 3.398523972625272e-05,
      "loss": 0.4241,
      "num_input_tokens_seen": 11090864,
      "step": 16970
    },
    {
      "epoch": 8.896750524109015,
      "grad_norm": 0.16051813960075378,
      "learning_rate": 3.3974568541818723e-05,
      "loss": 0.4401,
      "num_input_tokens_seen": 11094288,
      "step": 16975
    },
    {
      "epoch": 8.89937106918239,
      "grad_norm": 0.11469803005456924,
      "learning_rate": 3.396389548000738e-05,
      "loss": 0.311,
      "num_input_tokens_seen": 11096656,
      "step": 16980
    },
    {
      "epoch": 8.901991614255765,
      "grad_norm": 0.11118430644273758,
      "learning_rate": 3.395322054305137e-05,
      "loss": 0.4609,
      "num_input_tokens_seen": 11100144,
      "step": 16985
    },
    {
      "epoch": 8.90461215932914,
      "grad_norm": 0.1148969754576683,
      "learning_rate": 3.394254373318376e-05,
      "loss": 0.4198,
      "num_input_tokens_seen": 11103440,
      "step": 16990
    },
    {
      "epoch": 8.907232704402515,
      "grad_norm": 0.16275745630264282,
      "learning_rate": 3.3931865052638026e-05,
      "loss": 0.4333,
      "num_input_tokens_seen": 11107056,
      "step": 16995
    },
    {
      "epoch": 8.90985324947589,
      "grad_norm": 0.3924511671066284,
      "learning_rate": 3.392118450364802e-05,
      "loss": 0.4607,
      "num_input_tokens_seen": 11110640,
      "step": 17000
    },
    {
      "epoch": 8.912473794549266,
      "grad_norm": 0.03252582252025604,
      "learning_rate": 3.391050208844799e-05,
      "loss": 0.3387,
      "num_input_tokens_seen": 11117008,
      "step": 17005
    },
    {
      "epoch": 8.915094339622641,
      "grad_norm": 0.18637146055698395,
      "learning_rate": 3.389981780927259e-05,
      "loss": 0.4342,
      "num_input_tokens_seen": 11120176,
      "step": 17010
    },
    {
      "epoch": 8.917714884696016,
      "grad_norm": 0.21585886180400848,
      "learning_rate": 3.3889131668356834e-05,
      "loss": 0.5062,
      "num_input_tokens_seen": 11123504,
      "step": 17015
    },
    {
      "epoch": 8.920335429769391,
      "grad_norm": 0.21431078016757965,
      "learning_rate": 3.387844366793614e-05,
      "loss": 0.532,
      "num_input_tokens_seen": 11126000,
      "step": 17020
    },
    {
      "epoch": 8.922955974842766,
      "grad_norm": 0.10308961570262909,
      "learning_rate": 3.3867753810246326e-05,
      "loss": 0.46,
      "num_input_tokens_seen": 11129584,
      "step": 17025
    },
    {
      "epoch": 8.925576519916142,
      "grad_norm": 0.17425796389579773,
      "learning_rate": 3.3857062097523586e-05,
      "loss": 0.5131,
      "num_input_tokens_seen": 11132240,
      "step": 17030
    },
    {
      "epoch": 8.928197064989519,
      "grad_norm": 0.1940600425004959,
      "learning_rate": 3.38463685320045e-05,
      "loss": 0.5181,
      "num_input_tokens_seen": 11135120,
      "step": 17035
    },
    {
      "epoch": 8.930817610062894,
      "grad_norm": 0.3052026927471161,
      "learning_rate": 3.3835673115926045e-05,
      "loss": 0.3604,
      "num_input_tokens_seen": 11137968,
      "step": 17040
    },
    {
      "epoch": 8.933438155136269,
      "grad_norm": 0.12499068677425385,
      "learning_rate": 3.382497585152557e-05,
      "loss": 0.3912,
      "num_input_tokens_seen": 11140368,
      "step": 17045
    },
    {
      "epoch": 8.936058700209644,
      "grad_norm": 0.16869841516017914,
      "learning_rate": 3.3814276741040835e-05,
      "loss": 0.5625,
      "num_input_tokens_seen": 11145168,
      "step": 17050
    },
    {
      "epoch": 8.93867924528302,
      "grad_norm": 0.15786735713481903,
      "learning_rate": 3.380357578670996e-05,
      "loss": 0.5113,
      "num_input_tokens_seen": 11149008,
      "step": 17055
    },
    {
      "epoch": 8.941299790356394,
      "grad_norm": 0.22580747306346893,
      "learning_rate": 3.379287299077146e-05,
      "loss": 0.4769,
      "num_input_tokens_seen": 11156336,
      "step": 17060
    },
    {
      "epoch": 8.94392033542977,
      "grad_norm": 0.19751808047294617,
      "learning_rate": 3.3782168355464263e-05,
      "loss": 0.5908,
      "num_input_tokens_seen": 11159152,
      "step": 17065
    },
    {
      "epoch": 8.946540880503145,
      "grad_norm": 0.14663411676883698,
      "learning_rate": 3.377146188302762e-05,
      "loss": 0.42,
      "num_input_tokens_seen": 11162096,
      "step": 17070
    },
    {
      "epoch": 8.94916142557652,
      "grad_norm": 0.21823903918266296,
      "learning_rate": 3.376075357570124e-05,
      "loss": 0.4763,
      "num_input_tokens_seen": 11165200,
      "step": 17075
    },
    {
      "epoch": 8.951781970649895,
      "grad_norm": 0.262013703584671,
      "learning_rate": 3.3750043435725144e-05,
      "loss": 0.6231,
      "num_input_tokens_seen": 11168368,
      "step": 17080
    },
    {
      "epoch": 8.95440251572327,
      "grad_norm": 0.10788179934024811,
      "learning_rate": 3.373933146533979e-05,
      "loss": 0.5651,
      "num_input_tokens_seen": 11172048,
      "step": 17085
    },
    {
      "epoch": 8.957023060796645,
      "grad_norm": 0.11482631415128708,
      "learning_rate": 3.3728617666786e-05,
      "loss": 0.486,
      "num_input_tokens_seen": 11175600,
      "step": 17090
    },
    {
      "epoch": 8.95964360587002,
      "grad_norm": 0.26714038848876953,
      "learning_rate": 3.371790204230497e-05,
      "loss": 0.4554,
      "num_input_tokens_seen": 11178576,
      "step": 17095
    },
    {
      "epoch": 8.962264150943396,
      "grad_norm": 0.16777403652668,
      "learning_rate": 3.3707184594138295e-05,
      "loss": 0.5501,
      "num_input_tokens_seen": 11181168,
      "step": 17100
    },
    {
      "epoch": 8.964884696016771,
      "grad_norm": 0.24020417034626007,
      "learning_rate": 3.369646532452795e-05,
      "loss": 0.4247,
      "num_input_tokens_seen": 11184400,
      "step": 17105
    },
    {
      "epoch": 8.967505241090146,
      "grad_norm": 0.1661829650402069,
      "learning_rate": 3.3685744235716254e-05,
      "loss": 0.5214,
      "num_input_tokens_seen": 11187536,
      "step": 17110
    },
    {
      "epoch": 8.970125786163521,
      "grad_norm": 0.16050678491592407,
      "learning_rate": 3.367502132994598e-05,
      "loss": 0.4497,
      "num_input_tokens_seen": 11191568,
      "step": 17115
    },
    {
      "epoch": 8.972746331236896,
      "grad_norm": 0.39527395367622375,
      "learning_rate": 3.366429660946019e-05,
      "loss": 0.416,
      "num_input_tokens_seen": 11195024,
      "step": 17120
    },
    {
      "epoch": 8.975366876310272,
      "grad_norm": 0.2293180525302887,
      "learning_rate": 3.365357007650241e-05,
      "loss": 0.5329,
      "num_input_tokens_seen": 11198480,
      "step": 17125
    },
    {
      "epoch": 8.977987421383649,
      "grad_norm": 0.0904393270611763,
      "learning_rate": 3.364284173331649e-05,
      "loss": 0.5103,
      "num_input_tokens_seen": 11201712,
      "step": 17130
    },
    {
      "epoch": 8.980607966457024,
      "grad_norm": 0.09996692836284637,
      "learning_rate": 3.3632111582146686e-05,
      "loss": 0.4199,
      "num_input_tokens_seen": 11204720,
      "step": 17135
    },
    {
      "epoch": 8.983228511530399,
      "grad_norm": 0.209699809551239,
      "learning_rate": 3.362137962523761e-05,
      "loss": 0.4578,
      "num_input_tokens_seen": 11207440,
      "step": 17140
    },
    {
      "epoch": 8.985849056603774,
      "grad_norm": 0.16584846377372742,
      "learning_rate": 3.361064586483429e-05,
      "loss": 0.4654,
      "num_input_tokens_seen": 11210384,
      "step": 17145
    },
    {
      "epoch": 8.98846960167715,
      "grad_norm": 0.21530073881149292,
      "learning_rate": 3.359991030318208e-05,
      "loss": 0.4748,
      "num_input_tokens_seen": 11213392,
      "step": 17150
    },
    {
      "epoch": 8.991090146750524,
      "grad_norm": 0.18671788275241852,
      "learning_rate": 3.358917294252675e-05,
      "loss": 0.5631,
      "num_input_tokens_seen": 11216016,
      "step": 17155
    },
    {
      "epoch": 8.9937106918239,
      "grad_norm": 0.15900592505931854,
      "learning_rate": 3.357843378511443e-05,
      "loss": 0.5425,
      "num_input_tokens_seen": 11218160,
      "step": 17160
    },
    {
      "epoch": 8.996331236897275,
      "grad_norm": 0.13843724131584167,
      "learning_rate": 3.3567692833191635e-05,
      "loss": 0.7291,
      "num_input_tokens_seen": 11222000,
      "step": 17165
    },
    {
      "epoch": 8.99895178197065,
      "grad_norm": 0.1858283430337906,
      "learning_rate": 3.355695008900525e-05,
      "loss": 0.5135,
      "num_input_tokens_seen": 11224816,
      "step": 17170
    },
    {
      "epoch": 9.0,
      "eval_loss": 0.499555379152298,
      "eval_runtime": 9.9577,
      "eval_samples_per_second": 85.161,
      "eval_steps_per_second": 21.29,
      "num_input_tokens_seen": 11225416,
      "step": 17172
    },
    {
      "epoch": 9.001572327044025,
      "grad_norm": 0.18673266470432281,
      "learning_rate": 3.354620555480251e-05,
      "loss": 0.6034,
      "num_input_tokens_seen": 11227112,
      "step": 17175
    },
    {
      "epoch": 9.0041928721174,
      "grad_norm": 0.12699857354164124,
      "learning_rate": 3.353545923283108e-05,
      "loss": 0.4356,
      "num_input_tokens_seen": 11229544,
      "step": 17180
    },
    {
      "epoch": 9.006813417190775,
      "grad_norm": 0.13198339939117432,
      "learning_rate": 3.352471112533895e-05,
      "loss": 0.5212,
      "num_input_tokens_seen": 11233640,
      "step": 17185
    },
    {
      "epoch": 9.00943396226415,
      "grad_norm": 0.19841104745864868,
      "learning_rate": 3.351396123457451e-05,
      "loss": 0.4053,
      "num_input_tokens_seen": 11236552,
      "step": 17190
    },
    {
      "epoch": 9.012054507337526,
      "grad_norm": 0.25109755992889404,
      "learning_rate": 3.350320956278652e-05,
      "loss": 0.4364,
      "num_input_tokens_seen": 11240008,
      "step": 17195
    },
    {
      "epoch": 9.014675052410901,
      "grad_norm": 0.2417827546596527,
      "learning_rate": 3.349245611222409e-05,
      "loss": 0.4955,
      "num_input_tokens_seen": 11243176,
      "step": 17200
    },
    {
      "epoch": 9.017295597484276,
      "grad_norm": 0.1639915555715561,
      "learning_rate": 3.348170088513673e-05,
      "loss": 0.468,
      "num_input_tokens_seen": 11246568,
      "step": 17205
    },
    {
      "epoch": 9.019916142557651,
      "grad_norm": 0.13635645806789398,
      "learning_rate": 3.347094388377431e-05,
      "loss": 0.4499,
      "num_input_tokens_seen": 11251080,
      "step": 17210
    },
    {
      "epoch": 9.022536687631026,
      "grad_norm": 0.23396164178848267,
      "learning_rate": 3.3460185110387076e-05,
      "loss": 0.6914,
      "num_input_tokens_seen": 11253256,
      "step": 17215
    },
    {
      "epoch": 9.025157232704403,
      "grad_norm": 0.13156326115131378,
      "learning_rate": 3.344942456722563e-05,
      "loss": 0.4756,
      "num_input_tokens_seen": 11256936,
      "step": 17220
    },
    {
      "epoch": 9.027777777777779,
      "grad_norm": 0.12413366138935089,
      "learning_rate": 3.343866225654096e-05,
      "loss": 0.4968,
      "num_input_tokens_seen": 11259784,
      "step": 17225
    },
    {
      "epoch": 9.030398322851154,
      "grad_norm": 0.15126465260982513,
      "learning_rate": 3.342789818058443e-05,
      "loss": 0.4776,
      "num_input_tokens_seen": 11262728,
      "step": 17230
    },
    {
      "epoch": 9.033018867924529,
      "grad_norm": 0.13563445210456848,
      "learning_rate": 3.341713234160774e-05,
      "loss": 0.4572,
      "num_input_tokens_seen": 11265640,
      "step": 17235
    },
    {
      "epoch": 9.035639412997904,
      "grad_norm": 0.1211775690317154,
      "learning_rate": 3.3406364741862986e-05,
      "loss": 0.4237,
      "num_input_tokens_seen": 11270344,
      "step": 17240
    },
    {
      "epoch": 9.03825995807128,
      "grad_norm": 0.5843890905380249,
      "learning_rate": 3.3395595383602644e-05,
      "loss": 0.627,
      "num_input_tokens_seen": 11272584,
      "step": 17245
    },
    {
      "epoch": 9.040880503144654,
      "grad_norm": 0.16062459349632263,
      "learning_rate": 3.338482426907953e-05,
      "loss": 0.5258,
      "num_input_tokens_seen": 11276872,
      "step": 17250
    },
    {
      "epoch": 9.04350104821803,
      "grad_norm": 0.15116052329540253,
      "learning_rate": 3.337405140054683e-05,
      "loss": 0.4878,
      "num_input_tokens_seen": 11279688,
      "step": 17255
    },
    {
      "epoch": 9.046121593291405,
      "grad_norm": 0.14594623446464539,
      "learning_rate": 3.336327678025813e-05,
      "loss": 0.4371,
      "num_input_tokens_seen": 11282888,
      "step": 17260
    },
    {
      "epoch": 9.04874213836478,
      "grad_norm": 0.15036801993846893,
      "learning_rate": 3.3352500410467324e-05,
      "loss": 0.5737,
      "num_input_tokens_seen": 11285928,
      "step": 17265
    },
    {
      "epoch": 9.051362683438155,
      "grad_norm": 0.1925780177116394,
      "learning_rate": 3.3341722293428726e-05,
      "loss": 0.4564,
      "num_input_tokens_seen": 11289288,
      "step": 17270
    },
    {
      "epoch": 9.05398322851153,
      "grad_norm": 0.1205994039773941,
      "learning_rate": 3.3330942431397e-05,
      "loss": 0.5975,
      "num_input_tokens_seen": 11295976,
      "step": 17275
    },
    {
      "epoch": 9.056603773584905,
      "grad_norm": 0.1384219080209732,
      "learning_rate": 3.332016082662715e-05,
      "loss": 0.4954,
      "num_input_tokens_seen": 11299816,
      "step": 17280
    },
    {
      "epoch": 9.05922431865828,
      "grad_norm": 0.11130956560373306,
      "learning_rate": 3.3309377481374584e-05,
      "loss": 0.5621,
      "num_input_tokens_seen": 11303176,
      "step": 17285
    },
    {
      "epoch": 9.061844863731656,
      "grad_norm": 0.18837323784828186,
      "learning_rate": 3.3298592397895046e-05,
      "loss": 0.3727,
      "num_input_tokens_seen": 11306376,
      "step": 17290
    },
    {
      "epoch": 9.064465408805031,
      "grad_norm": 0.15273983776569366,
      "learning_rate": 3.3287805578444645e-05,
      "loss": 0.3866,
      "num_input_tokens_seen": 11310696,
      "step": 17295
    },
    {
      "epoch": 9.067085953878406,
      "grad_norm": 0.2863749861717224,
      "learning_rate": 3.327701702527989e-05,
      "loss": 0.4129,
      "num_input_tokens_seen": 11314888,
      "step": 17300
    },
    {
      "epoch": 9.069706498951781,
      "grad_norm": 0.20384709537029266,
      "learning_rate": 3.3266226740657584e-05,
      "loss": 0.5332,
      "num_input_tokens_seen": 11317576,
      "step": 17305
    },
    {
      "epoch": 9.072327044025156,
      "grad_norm": 0.16098056733608246,
      "learning_rate": 3.3255434726834954e-05,
      "loss": 0.5035,
      "num_input_tokens_seen": 11320840,
      "step": 17310
    },
    {
      "epoch": 9.074947589098532,
      "grad_norm": 0.15532973408699036,
      "learning_rate": 3.324464098606957e-05,
      "loss": 0.3216,
      "num_input_tokens_seen": 11324040,
      "step": 17315
    },
    {
      "epoch": 9.077568134171909,
      "grad_norm": 0.2416616976261139,
      "learning_rate": 3.323384552061934e-05,
      "loss": 0.5608,
      "num_input_tokens_seen": 11327784,
      "step": 17320
    },
    {
      "epoch": 9.080188679245284,
      "grad_norm": 0.1073940321803093,
      "learning_rate": 3.322304833274257e-05,
      "loss": 0.4463,
      "num_input_tokens_seen": 11331304,
      "step": 17325
    },
    {
      "epoch": 9.082809224318659,
      "grad_norm": 0.16137218475341797,
      "learning_rate": 3.32122494246979e-05,
      "loss": 0.5891,
      "num_input_tokens_seen": 11334312,
      "step": 17330
    },
    {
      "epoch": 9.085429769392034,
      "grad_norm": 0.12114764004945755,
      "learning_rate": 3.320144879874434e-05,
      "loss": 0.504,
      "num_input_tokens_seen": 11339144,
      "step": 17335
    },
    {
      "epoch": 9.08805031446541,
      "grad_norm": 0.19588419795036316,
      "learning_rate": 3.319064645714126e-05,
      "loss": 0.3417,
      "num_input_tokens_seen": 11342664,
      "step": 17340
    },
    {
      "epoch": 9.090670859538784,
      "grad_norm": 0.17714448273181915,
      "learning_rate": 3.3179842402148386e-05,
      "loss": 0.4233,
      "num_input_tokens_seen": 11345768,
      "step": 17345
    },
    {
      "epoch": 9.09329140461216,
      "grad_norm": 0.12833896279335022,
      "learning_rate": 3.316903663602581e-05,
      "loss": 0.6274,
      "num_input_tokens_seen": 11349448,
      "step": 17350
    },
    {
      "epoch": 9.095911949685535,
      "grad_norm": 0.17209745943546295,
      "learning_rate": 3.315822916103395e-05,
      "loss": 0.4152,
      "num_input_tokens_seen": 11352744,
      "step": 17355
    },
    {
      "epoch": 9.09853249475891,
      "grad_norm": 0.5821444988250732,
      "learning_rate": 3.3147419979433635e-05,
      "loss": 0.5438,
      "num_input_tokens_seen": 11354920,
      "step": 17360
    },
    {
      "epoch": 9.101153039832285,
      "grad_norm": 0.21220895648002625,
      "learning_rate": 3.313660909348601e-05,
      "loss": 0.4162,
      "num_input_tokens_seen": 11359240,
      "step": 17365
    },
    {
      "epoch": 9.10377358490566,
      "grad_norm": 0.1572081595659256,
      "learning_rate": 3.312579650545259e-05,
      "loss": 0.5447,
      "num_input_tokens_seen": 11362056,
      "step": 17370
    },
    {
      "epoch": 9.106394129979035,
      "grad_norm": 0.2646786868572235,
      "learning_rate": 3.311498221759524e-05,
      "loss": 0.3918,
      "num_input_tokens_seen": 11364648,
      "step": 17375
    },
    {
      "epoch": 9.10901467505241,
      "grad_norm": 0.12011672556400299,
      "learning_rate": 3.310416623217619e-05,
      "loss": 0.4094,
      "num_input_tokens_seen": 11368168,
      "step": 17380
    },
    {
      "epoch": 9.111635220125786,
      "grad_norm": 0.17152754962444305,
      "learning_rate": 3.309334855145803e-05,
      "loss": 0.5224,
      "num_input_tokens_seen": 11370856,
      "step": 17385
    },
    {
      "epoch": 9.114255765199161,
      "grad_norm": 0.1811707317829132,
      "learning_rate": 3.3082529177703684e-05,
      "loss": 0.4772,
      "num_input_tokens_seen": 11373768,
      "step": 17390
    },
    {
      "epoch": 9.116876310272536,
      "grad_norm": 0.22288577258586884,
      "learning_rate": 3.3071708113176435e-05,
      "loss": 0.5212,
      "num_input_tokens_seen": 11377192,
      "step": 17395
    },
    {
      "epoch": 9.119496855345911,
      "grad_norm": 0.14258116483688354,
      "learning_rate": 3.306088536013995e-05,
      "loss": 0.4599,
      "num_input_tokens_seen": 11379912,
      "step": 17400
    },
    {
      "epoch": 9.122117400419286,
      "grad_norm": 0.13021507859230042,
      "learning_rate": 3.3050060920858206e-05,
      "loss": 0.496,
      "num_input_tokens_seen": 11383496,
      "step": 17405
    },
    {
      "epoch": 9.124737945492662,
      "grad_norm": 0.21312887966632843,
      "learning_rate": 3.3039234797595544e-05,
      "loss": 0.6247,
      "num_input_tokens_seen": 11386504,
      "step": 17410
    },
    {
      "epoch": 9.127358490566039,
      "grad_norm": 0.7032839059829712,
      "learning_rate": 3.3028406992616695e-05,
      "loss": 0.4343,
      "num_input_tokens_seen": 11388776,
      "step": 17415
    },
    {
      "epoch": 9.129979035639414,
      "grad_norm": 0.1540578305721283,
      "learning_rate": 3.301757750818669e-05,
      "loss": 0.6256,
      "num_input_tokens_seen": 11391976,
      "step": 17420
    },
    {
      "epoch": 9.132599580712789,
      "grad_norm": 0.12846116721630096,
      "learning_rate": 3.300674634657094e-05,
      "loss": 0.4601,
      "num_input_tokens_seen": 11394856,
      "step": 17425
    },
    {
      "epoch": 9.135220125786164,
      "grad_norm": 0.27986523509025574,
      "learning_rate": 3.2995913510035206e-05,
      "loss": 0.5338,
      "num_input_tokens_seen": 11397896,
      "step": 17430
    },
    {
      "epoch": 9.13784067085954,
      "grad_norm": 0.2670206129550934,
      "learning_rate": 3.298507900084558e-05,
      "loss": 0.5764,
      "num_input_tokens_seen": 11401096,
      "step": 17435
    },
    {
      "epoch": 9.140461215932914,
      "grad_norm": 0.2625999450683594,
      "learning_rate": 3.2974242821268524e-05,
      "loss": 0.4479,
      "num_input_tokens_seen": 11404616,
      "step": 17440
    },
    {
      "epoch": 9.14308176100629,
      "grad_norm": 0.12220294028520584,
      "learning_rate": 3.2963404973570855e-05,
      "loss": 0.4715,
      "num_input_tokens_seen": 11407848,
      "step": 17445
    },
    {
      "epoch": 9.145702306079665,
      "grad_norm": 0.1696995496749878,
      "learning_rate": 3.2952565460019715e-05,
      "loss": 0.4591,
      "num_input_tokens_seen": 11410760,
      "step": 17450
    },
    {
      "epoch": 9.14832285115304,
      "grad_norm": 0.22967718541622162,
      "learning_rate": 3.29417242828826e-05,
      "loss": 0.5493,
      "num_input_tokens_seen": 11413864,
      "step": 17455
    },
    {
      "epoch": 9.150943396226415,
      "grad_norm": 0.1571381688117981,
      "learning_rate": 3.2930881444427374e-05,
      "loss": 0.4134,
      "num_input_tokens_seen": 11416808,
      "step": 17460
    },
    {
      "epoch": 9.15356394129979,
      "grad_norm": 0.1698523461818695,
      "learning_rate": 3.292003694692223e-05,
      "loss": 0.5526,
      "num_input_tokens_seen": 11419944,
      "step": 17465
    },
    {
      "epoch": 9.156184486373165,
      "grad_norm": 0.20826804637908936,
      "learning_rate": 3.290919079263572e-05,
      "loss": 0.4268,
      "num_input_tokens_seen": 11422728,
      "step": 17470
    },
    {
      "epoch": 9.15880503144654,
      "grad_norm": 0.1286371499300003,
      "learning_rate": 3.2898342983836715e-05,
      "loss": 0.5107,
      "num_input_tokens_seen": 11430984,
      "step": 17475
    },
    {
      "epoch": 9.161425576519916,
      "grad_norm": 0.21618929505348206,
      "learning_rate": 3.288749352279447e-05,
      "loss": 0.4813,
      "num_input_tokens_seen": 11433672,
      "step": 17480
    },
    {
      "epoch": 9.164046121593291,
      "grad_norm": 0.20281681418418884,
      "learning_rate": 3.2876642411778573e-05,
      "loss": 0.5174,
      "num_input_tokens_seen": 11436168,
      "step": 17485
    },
    {
      "epoch": 9.166666666666666,
      "grad_norm": 0.07008562237024307,
      "learning_rate": 3.286578965305894e-05,
      "loss": 0.3426,
      "num_input_tokens_seen": 11439528,
      "step": 17490
    },
    {
      "epoch": 9.169287211740041,
      "grad_norm": 0.23928670585155487,
      "learning_rate": 3.285493524890585e-05,
      "loss": 0.5359,
      "num_input_tokens_seen": 11442312,
      "step": 17495
    },
    {
      "epoch": 9.171907756813416,
      "grad_norm": 0.22982166707515717,
      "learning_rate": 3.284407920158992e-05,
      "loss": 0.4317,
      "num_input_tokens_seen": 11445416,
      "step": 17500
    },
    {
      "epoch": 9.174528301886792,
      "grad_norm": 0.2127855271100998,
      "learning_rate": 3.283322151338211e-05,
      "loss": 0.6387,
      "num_input_tokens_seen": 11448712,
      "step": 17505
    },
    {
      "epoch": 9.177148846960169,
      "grad_norm": 0.19134452939033508,
      "learning_rate": 3.282236218655373e-05,
      "loss": 0.5382,
      "num_input_tokens_seen": 11452104,
      "step": 17510
    },
    {
      "epoch": 9.179769392033544,
      "grad_norm": 0.14202438294887543,
      "learning_rate": 3.281150122337642e-05,
      "loss": 0.3437,
      "num_input_tokens_seen": 11456136,
      "step": 17515
    },
    {
      "epoch": 9.182389937106919,
      "grad_norm": 0.14200322329998016,
      "learning_rate": 3.280063862612216e-05,
      "loss": 0.413,
      "num_input_tokens_seen": 11459272,
      "step": 17520
    },
    {
      "epoch": 9.185010482180294,
      "grad_norm": 0.1524047553539276,
      "learning_rate": 3.2789774397063314e-05,
      "loss": 0.4548,
      "num_input_tokens_seen": 11463208,
      "step": 17525
    },
    {
      "epoch": 9.18763102725367,
      "grad_norm": 0.18855604529380798,
      "learning_rate": 3.277890853847253e-05,
      "loss": 0.4536,
      "num_input_tokens_seen": 11466472,
      "step": 17530
    },
    {
      "epoch": 9.190251572327044,
      "grad_norm": 0.21728144586086273,
      "learning_rate": 3.2768041052622824e-05,
      "loss": 0.5763,
      "num_input_tokens_seen": 11469192,
      "step": 17535
    },
    {
      "epoch": 9.19287211740042,
      "grad_norm": 0.16075053811073303,
      "learning_rate": 3.275717194178755e-05,
      "loss": 0.4417,
      "num_input_tokens_seen": 11472232,
      "step": 17540
    },
    {
      "epoch": 9.195492662473795,
      "grad_norm": 0.13717587292194366,
      "learning_rate": 3.27463012082404e-05,
      "loss": 0.5221,
      "num_input_tokens_seen": 11476040,
      "step": 17545
    },
    {
      "epoch": 9.19811320754717,
      "grad_norm": 0.20515874028205872,
      "learning_rate": 3.2735428854255416e-05,
      "loss": 0.4716,
      "num_input_tokens_seen": 11478728,
      "step": 17550
    },
    {
      "epoch": 9.200733752620545,
      "grad_norm": 0.19897878170013428,
      "learning_rate": 3.272455488210697e-05,
      "loss": 0.5631,
      "num_input_tokens_seen": 11481448,
      "step": 17555
    },
    {
      "epoch": 9.20335429769392,
      "grad_norm": 0.17933663725852966,
      "learning_rate": 3.271367929406977e-05,
      "loss": 0.4815,
      "num_input_tokens_seen": 11484424,
      "step": 17560
    },
    {
      "epoch": 9.205974842767295,
      "grad_norm": 0.1756221055984497,
      "learning_rate": 3.270280209241886e-05,
      "loss": 0.4566,
      "num_input_tokens_seen": 11487560,
      "step": 17565
    },
    {
      "epoch": 9.20859538784067,
      "grad_norm": 0.29432255029678345,
      "learning_rate": 3.2691923279429624e-05,
      "loss": 0.4673,
      "num_input_tokens_seen": 11490632,
      "step": 17570
    },
    {
      "epoch": 9.211215932914046,
      "grad_norm": 0.25568830966949463,
      "learning_rate": 3.2681042857377804e-05,
      "loss": 0.4493,
      "num_input_tokens_seen": 11493032,
      "step": 17575
    },
    {
      "epoch": 9.213836477987421,
      "grad_norm": 0.15445634722709656,
      "learning_rate": 3.2670160828539435e-05,
      "loss": 0.4136,
      "num_input_tokens_seen": 11496200,
      "step": 17580
    },
    {
      "epoch": 9.216457023060796,
      "grad_norm": 0.19820159673690796,
      "learning_rate": 3.2659277195190934e-05,
      "loss": 0.5109,
      "num_input_tokens_seen": 11498696,
      "step": 17585
    },
    {
      "epoch": 9.219077568134171,
      "grad_norm": 0.14676755666732788,
      "learning_rate": 3.2648391959609026e-05,
      "loss": 0.4089,
      "num_input_tokens_seen": 11502312,
      "step": 17590
    },
    {
      "epoch": 9.221698113207546,
      "grad_norm": 0.13765153288841248,
      "learning_rate": 3.263750512407078e-05,
      "loss": 0.4775,
      "num_input_tokens_seen": 11505160,
      "step": 17595
    },
    {
      "epoch": 9.224318658280922,
      "grad_norm": 0.13013991713523865,
      "learning_rate": 3.262661669085359e-05,
      "loss": 0.3818,
      "num_input_tokens_seen": 11508456,
      "step": 17600
    },
    {
      "epoch": 9.226939203354299,
      "grad_norm": 0.16734535992145538,
      "learning_rate": 3.2615726662235205e-05,
      "loss": 0.5632,
      "num_input_tokens_seen": 11511816,
      "step": 17605
    },
    {
      "epoch": 9.229559748427674,
      "grad_norm": 0.1672126203775406,
      "learning_rate": 3.2604835040493676e-05,
      "loss": 0.537,
      "num_input_tokens_seen": 11514088,
      "step": 17610
    },
    {
      "epoch": 9.232180293501049,
      "grad_norm": 0.1697588562965393,
      "learning_rate": 3.259394182790742e-05,
      "loss": 0.4282,
      "num_input_tokens_seen": 11517096,
      "step": 17615
    },
    {
      "epoch": 9.234800838574424,
      "grad_norm": 0.1322755068540573,
      "learning_rate": 3.2583047026755174e-05,
      "loss": 0.427,
      "num_input_tokens_seen": 11520168,
      "step": 17620
    },
    {
      "epoch": 9.2374213836478,
      "grad_norm": 0.1466263383626938,
      "learning_rate": 3.2572150639316e-05,
      "loss": 0.6032,
      "num_input_tokens_seen": 11523944,
      "step": 17625
    },
    {
      "epoch": 9.240041928721174,
      "grad_norm": 0.08419789373874664,
      "learning_rate": 3.256125266786931e-05,
      "loss": 0.6307,
      "num_input_tokens_seen": 11527720,
      "step": 17630
    },
    {
      "epoch": 9.24266247379455,
      "grad_norm": 0.14635245501995087,
      "learning_rate": 3.255035311469481e-05,
      "loss": 0.5501,
      "num_input_tokens_seen": 11531304,
      "step": 17635
    },
    {
      "epoch": 9.245283018867925,
      "grad_norm": 0.17037317156791687,
      "learning_rate": 3.2539451982072586e-05,
      "loss": 0.4415,
      "num_input_tokens_seen": 11535176,
      "step": 17640
    },
    {
      "epoch": 9.2479035639413,
      "grad_norm": 0.16240628063678741,
      "learning_rate": 3.2528549272283016e-05,
      "loss": 0.5372,
      "num_input_tokens_seen": 11539016,
      "step": 17645
    },
    {
      "epoch": 9.250524109014675,
      "grad_norm": 0.13941870629787445,
      "learning_rate": 3.251764498760683e-05,
      "loss": 0.4158,
      "num_input_tokens_seen": 11543752,
      "step": 17650
    },
    {
      "epoch": 9.25314465408805,
      "grad_norm": 0.19691359996795654,
      "learning_rate": 3.2506739130325084e-05,
      "loss": 0.4912,
      "num_input_tokens_seen": 11546664,
      "step": 17655
    },
    {
      "epoch": 9.255765199161425,
      "grad_norm": 0.27460694313049316,
      "learning_rate": 3.2495831702719146e-05,
      "loss": 0.5173,
      "num_input_tokens_seen": 11549032,
      "step": 17660
    },
    {
      "epoch": 9.2583857442348,
      "grad_norm": 0.13772115111351013,
      "learning_rate": 3.248492270707074e-05,
      "loss": 0.5871,
      "num_input_tokens_seen": 11552328,
      "step": 17665
    },
    {
      "epoch": 9.261006289308176,
      "grad_norm": 0.19956009089946747,
      "learning_rate": 3.247401214566189e-05,
      "loss": 0.5945,
      "num_input_tokens_seen": 11555464,
      "step": 17670
    },
    {
      "epoch": 9.26362683438155,
      "grad_norm": 0.17862193286418915,
      "learning_rate": 3.2463100020774965e-05,
      "loss": 0.491,
      "num_input_tokens_seen": 11558088,
      "step": 17675
    },
    {
      "epoch": 9.266247379454926,
      "grad_norm": 0.15213610231876373,
      "learning_rate": 3.245218633469267e-05,
      "loss": 0.5307,
      "num_input_tokens_seen": 11560872,
      "step": 17680
    },
    {
      "epoch": 9.268867924528301,
      "grad_norm": 0.18136872351169586,
      "learning_rate": 3.244127108969801e-05,
      "loss": 0.4445,
      "num_input_tokens_seen": 11564264,
      "step": 17685
    },
    {
      "epoch": 9.271488469601676,
      "grad_norm": 0.14963790774345398,
      "learning_rate": 3.2430354288074314e-05,
      "loss": 0.4644,
      "num_input_tokens_seen": 11567656,
      "step": 17690
    },
    {
      "epoch": 9.274109014675052,
      "grad_norm": 0.1555885225534439,
      "learning_rate": 3.241943593210529e-05,
      "loss": 0.5371,
      "num_input_tokens_seen": 11571176,
      "step": 17695
    },
    {
      "epoch": 9.276729559748428,
      "grad_norm": 0.12548086047172546,
      "learning_rate": 3.2408516024074916e-05,
      "loss": 0.5572,
      "num_input_tokens_seen": 11574216,
      "step": 17700
    },
    {
      "epoch": 9.279350104821804,
      "grad_norm": 0.17085380852222443,
      "learning_rate": 3.2397594566267505e-05,
      "loss": 0.3788,
      "num_input_tokens_seen": 11577608,
      "step": 17705
    },
    {
      "epoch": 9.281970649895179,
      "grad_norm": 0.13973408937454224,
      "learning_rate": 3.23866715609677e-05,
      "loss": 0.5193,
      "num_input_tokens_seen": 11581000,
      "step": 17710
    },
    {
      "epoch": 9.284591194968554,
      "grad_norm": 0.14901433885097504,
      "learning_rate": 3.237574701046049e-05,
      "loss": 0.5067,
      "num_input_tokens_seen": 11584840,
      "step": 17715
    },
    {
      "epoch": 9.28721174004193,
      "grad_norm": 0.1715969741344452,
      "learning_rate": 3.236482091703115e-05,
      "loss": 0.4984,
      "num_input_tokens_seen": 11587912,
      "step": 17720
    },
    {
      "epoch": 9.289832285115304,
      "grad_norm": 0.18357504904270172,
      "learning_rate": 3.235389328296528e-05,
      "loss": 0.502,
      "num_input_tokens_seen": 11591752,
      "step": 17725
    },
    {
      "epoch": 9.29245283018868,
      "grad_norm": 0.20311838388442993,
      "learning_rate": 3.2342964110548845e-05,
      "loss": 0.5661,
      "num_input_tokens_seen": 11594248,
      "step": 17730
    },
    {
      "epoch": 9.295073375262055,
      "grad_norm": 0.22822779417037964,
      "learning_rate": 3.2332033402068093e-05,
      "loss": 0.4825,
      "num_input_tokens_seen": 11597448,
      "step": 17735
    },
    {
      "epoch": 9.29769392033543,
      "grad_norm": 0.1846347600221634,
      "learning_rate": 3.2321101159809605e-05,
      "loss": 0.7227,
      "num_input_tokens_seen": 11600360,
      "step": 17740
    },
    {
      "epoch": 9.300314465408805,
      "grad_norm": 0.21479177474975586,
      "learning_rate": 3.231016738606027e-05,
      "loss": 0.5446,
      "num_input_tokens_seen": 11603528,
      "step": 17745
    },
    {
      "epoch": 9.30293501048218,
      "grad_norm": 0.18584942817687988,
      "learning_rate": 3.2299232083107313e-05,
      "loss": 0.5251,
      "num_input_tokens_seen": 11606600,
      "step": 17750
    },
    {
      "epoch": 9.305555555555555,
      "grad_norm": 0.14694617688655853,
      "learning_rate": 3.228829525323829e-05,
      "loss": 0.4471,
      "num_input_tokens_seen": 11609064,
      "step": 17755
    },
    {
      "epoch": 9.30817610062893,
      "grad_norm": 0.1314953714609146,
      "learning_rate": 3.227735689874104e-05,
      "loss": 0.4423,
      "num_input_tokens_seen": 11612328,
      "step": 17760
    },
    {
      "epoch": 9.310796645702306,
      "grad_norm": 0.18781793117523193,
      "learning_rate": 3.226641702190375e-05,
      "loss": 0.5021,
      "num_input_tokens_seen": 11615912,
      "step": 17765
    },
    {
      "epoch": 9.31341719077568,
      "grad_norm": 0.1597997546195984,
      "learning_rate": 3.225547562501493e-05,
      "loss": 0.5404,
      "num_input_tokens_seen": 11618888,
      "step": 17770
    },
    {
      "epoch": 9.316037735849056,
      "grad_norm": 0.1665494441986084,
      "learning_rate": 3.2244532710363375e-05,
      "loss": 0.7289,
      "num_input_tokens_seen": 11621544,
      "step": 17775
    },
    {
      "epoch": 9.318658280922431,
      "grad_norm": 0.15710864961147308,
      "learning_rate": 3.2233588280238224e-05,
      "loss": 0.4256,
      "num_input_tokens_seen": 11624040,
      "step": 17780
    },
    {
      "epoch": 9.321278825995806,
      "grad_norm": 0.20829817652702332,
      "learning_rate": 3.2222642336928944e-05,
      "loss": 0.501,
      "num_input_tokens_seen": 11626728,
      "step": 17785
    },
    {
      "epoch": 9.323899371069182,
      "grad_norm": 0.22632820904254913,
      "learning_rate": 3.2211694882725275e-05,
      "loss": 0.3739,
      "num_input_tokens_seen": 11629832,
      "step": 17790
    },
    {
      "epoch": 9.326519916142558,
      "grad_norm": 0.1953151822090149,
      "learning_rate": 3.220074591991731e-05,
      "loss": 0.3907,
      "num_input_tokens_seen": 11633320,
      "step": 17795
    },
    {
      "epoch": 9.329140461215934,
      "grad_norm": 0.1702290177345276,
      "learning_rate": 3.218979545079546e-05,
      "loss": 0.4507,
      "num_input_tokens_seen": 11636104,
      "step": 17800
    },
    {
      "epoch": 9.331761006289309,
      "grad_norm": 0.14468370378017426,
      "learning_rate": 3.217884347765042e-05,
      "loss": 0.4441,
      "num_input_tokens_seen": 11639752,
      "step": 17805
    },
    {
      "epoch": 9.334381551362684,
      "grad_norm": 0.36230823397636414,
      "learning_rate": 3.216789000277323e-05,
      "loss": 0.4642,
      "num_input_tokens_seen": 11642664,
      "step": 17810
    },
    {
      "epoch": 9.33700209643606,
      "grad_norm": 0.17404048144817352,
      "learning_rate": 3.2156935028455224e-05,
      "loss": 0.5244,
      "num_input_tokens_seen": 11646024,
      "step": 17815
    },
    {
      "epoch": 9.339622641509434,
      "grad_norm": 0.10852047055959702,
      "learning_rate": 3.2145978556988056e-05,
      "loss": 0.4697,
      "num_input_tokens_seen": 11649192,
      "step": 17820
    },
    {
      "epoch": 9.34224318658281,
      "grad_norm": 0.13520720601081848,
      "learning_rate": 3.213502059066371e-05,
      "loss": 0.582,
      "num_input_tokens_seen": 11653192,
      "step": 17825
    },
    {
      "epoch": 9.344863731656185,
      "grad_norm": 0.2398144006729126,
      "learning_rate": 3.212406113177445e-05,
      "loss": 0.4695,
      "num_input_tokens_seen": 11657224,
      "step": 17830
    },
    {
      "epoch": 9.34748427672956,
      "grad_norm": 0.15263542532920837,
      "learning_rate": 3.211310018261287e-05,
      "loss": 0.6103,
      "num_input_tokens_seen": 11660328,
      "step": 17835
    },
    {
      "epoch": 9.350104821802935,
      "grad_norm": 0.11168210208415985,
      "learning_rate": 3.21021377454719e-05,
      "loss": 0.5598,
      "num_input_tokens_seen": 11663400,
      "step": 17840
    },
    {
      "epoch": 9.35272536687631,
      "grad_norm": 0.19451870024204254,
      "learning_rate": 3.209117382264473e-05,
      "loss": 0.5359,
      "num_input_tokens_seen": 11665992,
      "step": 17845
    },
    {
      "epoch": 9.355345911949685,
      "grad_norm": 0.3815658390522003,
      "learning_rate": 3.208020841642489e-05,
      "loss": 0.4457,
      "num_input_tokens_seen": 11669000,
      "step": 17850
    },
    {
      "epoch": 9.35796645702306,
      "grad_norm": 0.17315223813056946,
      "learning_rate": 3.206924152910623e-05,
      "loss": 0.4807,
      "num_input_tokens_seen": 11671976,
      "step": 17855
    },
    {
      "epoch": 9.360587002096436,
      "grad_norm": 0.12208079546689987,
      "learning_rate": 3.205827316298289e-05,
      "loss": 0.4318,
      "num_input_tokens_seen": 11674984,
      "step": 17860
    },
    {
      "epoch": 9.36320754716981,
      "grad_norm": 0.2390955090522766,
      "learning_rate": 3.204730332034934e-05,
      "loss": 0.4552,
      "num_input_tokens_seen": 11677864,
      "step": 17865
    },
    {
      "epoch": 9.365828092243186,
      "grad_norm": 0.09086273610591888,
      "learning_rate": 3.2036332003500324e-05,
      "loss": 0.4288,
      "num_input_tokens_seen": 11681128,
      "step": 17870
    },
    {
      "epoch": 9.368448637316561,
      "grad_norm": 0.17865242063999176,
      "learning_rate": 3.2025359214730924e-05,
      "loss": 0.4842,
      "num_input_tokens_seen": 11683880,
      "step": 17875
    },
    {
      "epoch": 9.371069182389936,
      "grad_norm": 0.21900725364685059,
      "learning_rate": 3.201438495633654e-05,
      "loss": 0.4893,
      "num_input_tokens_seen": 11687496,
      "step": 17880
    },
    {
      "epoch": 9.373689727463312,
      "grad_norm": 0.23637308180332184,
      "learning_rate": 3.200340923061284e-05,
      "loss": 0.5018,
      "num_input_tokens_seen": 11690696,
      "step": 17885
    },
    {
      "epoch": 9.376310272536688,
      "grad_norm": 0.14997448027133942,
      "learning_rate": 3.199243203985583e-05,
      "loss": 0.4453,
      "num_input_tokens_seen": 11693544,
      "step": 17890
    },
    {
      "epoch": 9.378930817610064,
      "grad_norm": 0.1891237497329712,
      "learning_rate": 3.1981453386361804e-05,
      "loss": 0.55,
      "num_input_tokens_seen": 11696840,
      "step": 17895
    },
    {
      "epoch": 9.381551362683439,
      "grad_norm": 0.14682511985301971,
      "learning_rate": 3.1970473272427386e-05,
      "loss": 0.5006,
      "num_input_tokens_seen": 11700264,
      "step": 17900
    },
    {
      "epoch": 9.384171907756814,
      "grad_norm": 0.22819316387176514,
      "learning_rate": 3.195949170034948e-05,
      "loss": 0.3553,
      "num_input_tokens_seen": 11703496,
      "step": 17905
    },
    {
      "epoch": 9.38679245283019,
      "grad_norm": 0.32798805832862854,
      "learning_rate": 3.19485086724253e-05,
      "loss": 0.6021,
      "num_input_tokens_seen": 11706664,
      "step": 17910
    },
    {
      "epoch": 9.389412997903564,
      "grad_norm": 0.2560352087020874,
      "learning_rate": 3.193752419095239e-05,
      "loss": 0.5165,
      "num_input_tokens_seen": 11709480,
      "step": 17915
    },
    {
      "epoch": 9.39203354297694,
      "grad_norm": 0.10938461124897003,
      "learning_rate": 3.192653825822857e-05,
      "loss": 0.3979,
      "num_input_tokens_seen": 11712712,
      "step": 17920
    },
    {
      "epoch": 9.394654088050315,
      "grad_norm": 0.22729185223579407,
      "learning_rate": 3.1915550876551964e-05,
      "loss": 0.5851,
      "num_input_tokens_seen": 11715464,
      "step": 17925
    },
    {
      "epoch": 9.39727463312369,
      "grad_norm": 0.1552584171295166,
      "learning_rate": 3.190456204822101e-05,
      "loss": 0.5511,
      "num_input_tokens_seen": 11718120,
      "step": 17930
    },
    {
      "epoch": 9.399895178197065,
      "grad_norm": 0.14896003901958466,
      "learning_rate": 3.1893571775534445e-05,
      "loss": 0.5168,
      "num_input_tokens_seen": 11721512,
      "step": 17935
    },
    {
      "epoch": 9.40251572327044,
      "grad_norm": 0.1530793011188507,
      "learning_rate": 3.1882580060791315e-05,
      "loss": 0.3535,
      "num_input_tokens_seen": 11724008,
      "step": 17940
    },
    {
      "epoch": 9.405136268343815,
      "grad_norm": 0.14932698011398315,
      "learning_rate": 3.187158690629095e-05,
      "loss": 0.4872,
      "num_input_tokens_seen": 11726696,
      "step": 17945
    },
    {
      "epoch": 9.40775681341719,
      "grad_norm": 0.09928027540445328,
      "learning_rate": 3.1860592314333e-05,
      "loss": 0.3991,
      "num_input_tokens_seen": 11730408,
      "step": 17950
    },
    {
      "epoch": 9.410377358490566,
      "grad_norm": 0.20407281816005707,
      "learning_rate": 3.184959628721742e-05,
      "loss": 0.5689,
      "num_input_tokens_seen": 11733320,
      "step": 17955
    },
    {
      "epoch": 9.41299790356394,
      "grad_norm": 0.2085757702589035,
      "learning_rate": 3.183859882724443e-05,
      "loss": 0.4715,
      "num_input_tokens_seen": 11738472,
      "step": 17960
    },
    {
      "epoch": 9.415618448637316,
      "grad_norm": 0.15277552604675293,
      "learning_rate": 3.1827599936714584e-05,
      "loss": 0.5412,
      "num_input_tokens_seen": 11741032,
      "step": 17965
    },
    {
      "epoch": 9.418238993710691,
      "grad_norm": 0.3540111482143402,
      "learning_rate": 3.1816599617928736e-05,
      "loss": 0.5045,
      "num_input_tokens_seen": 11744328,
      "step": 17970
    },
    {
      "epoch": 9.420859538784066,
      "grad_norm": 0.22672373056411743,
      "learning_rate": 3.1805597873188e-05,
      "loss": 0.3861,
      "num_input_tokens_seen": 11747976,
      "step": 17975
    },
    {
      "epoch": 9.423480083857442,
      "grad_norm": 0.1324695646762848,
      "learning_rate": 3.179459470479384e-05,
      "loss": 0.5896,
      "num_input_tokens_seen": 11751144,
      "step": 17980
    },
    {
      "epoch": 9.426100628930818,
      "grad_norm": 0.12731122970581055,
      "learning_rate": 3.1783590115047983e-05,
      "loss": 0.511,
      "num_input_tokens_seen": 11754312,
      "step": 17985
    },
    {
      "epoch": 9.428721174004194,
      "grad_norm": 0.13255809247493744,
      "learning_rate": 3.177258410625246e-05,
      "loss": 0.3741,
      "num_input_tokens_seen": 11757896,
      "step": 17990
    },
    {
      "epoch": 9.431341719077569,
      "grad_norm": 0.1183086633682251,
      "learning_rate": 3.176157668070962e-05,
      "loss": 0.4438,
      "num_input_tokens_seen": 11760840,
      "step": 17995
    },
    {
      "epoch": 9.433962264150944,
      "grad_norm": 0.1582033485174179,
      "learning_rate": 3.175056784072207e-05,
      "loss": 0.4235,
      "num_input_tokens_seen": 11764136,
      "step": 18000
    },
    {
      "epoch": 9.43658280922432,
      "grad_norm": 0.48192593455314636,
      "learning_rate": 3.173955758859274e-05,
      "loss": 0.6064,
      "num_input_tokens_seen": 11766696,
      "step": 18005
    },
    {
      "epoch": 9.439203354297694,
      "grad_norm": 0.28153151273727417,
      "learning_rate": 3.172854592662486e-05,
      "loss": 0.5452,
      "num_input_tokens_seen": 11769160,
      "step": 18010
    },
    {
      "epoch": 9.44182389937107,
      "grad_norm": 0.11222239583730698,
      "learning_rate": 3.1717532857121926e-05,
      "loss": 0.5272,
      "num_input_tokens_seen": 11772360,
      "step": 18015
    },
    {
      "epoch": 9.444444444444445,
      "grad_norm": 0.2621290683746338,
      "learning_rate": 3.170651838238776e-05,
      "loss": 0.5569,
      "num_input_tokens_seen": 11774984,
      "step": 18020
    },
    {
      "epoch": 9.44706498951782,
      "grad_norm": 0.12348195910453796,
      "learning_rate": 3.169550250472646e-05,
      "loss": 0.4635,
      "num_input_tokens_seen": 11777928,
      "step": 18025
    },
    {
      "epoch": 9.449685534591195,
      "grad_norm": 0.19952189922332764,
      "learning_rate": 3.168448522644243e-05,
      "loss": 0.5077,
      "num_input_tokens_seen": 11781096,
      "step": 18030
    },
    {
      "epoch": 9.45230607966457,
      "grad_norm": 0.10986637324094772,
      "learning_rate": 3.167346654984034e-05,
      "loss": 0.429,
      "num_input_tokens_seen": 11784104,
      "step": 18035
    },
    {
      "epoch": 9.454926624737945,
      "grad_norm": 0.11300304532051086,
      "learning_rate": 3.166244647722519e-05,
      "loss": 0.5217,
      "num_input_tokens_seen": 11787304,
      "step": 18040
    },
    {
      "epoch": 9.45754716981132,
      "grad_norm": 0.19439910352230072,
      "learning_rate": 3.1651425010902244e-05,
      "loss": 0.3768,
      "num_input_tokens_seen": 11790024,
      "step": 18045
    },
    {
      "epoch": 9.460167714884696,
      "grad_norm": 0.16758215427398682,
      "learning_rate": 3.164040215317706e-05,
      "loss": 0.4423,
      "num_input_tokens_seen": 11792840,
      "step": 18050
    },
    {
      "epoch": 9.46278825995807,
      "grad_norm": 0.2132693976163864,
      "learning_rate": 3.162937790635552e-05,
      "loss": 0.4658,
      "num_input_tokens_seen": 11795720,
      "step": 18055
    },
    {
      "epoch": 9.465408805031446,
      "grad_norm": 0.20296171307563782,
      "learning_rate": 3.1618352272743746e-05,
      "loss": 0.4734,
      "num_input_tokens_seen": 11798472,
      "step": 18060
    },
    {
      "epoch": 9.468029350104821,
      "grad_norm": 0.2555781304836273,
      "learning_rate": 3.160732525464818e-05,
      "loss": 0.476,
      "num_input_tokens_seen": 11801800,
      "step": 18065
    },
    {
      "epoch": 9.470649895178196,
      "grad_norm": 0.19155003130435944,
      "learning_rate": 3.1596296854375554e-05,
      "loss": 0.5502,
      "num_input_tokens_seen": 11805160,
      "step": 18070
    },
    {
      "epoch": 9.473270440251572,
      "grad_norm": 0.1712077111005783,
      "learning_rate": 3.1585267074232884e-05,
      "loss": 0.4518,
      "num_input_tokens_seen": 11811816,
      "step": 18075
    },
    {
      "epoch": 9.475890985324948,
      "grad_norm": 0.09524774551391602,
      "learning_rate": 3.157423591652746e-05,
      "loss": 0.5285,
      "num_input_tokens_seen": 11816040,
      "step": 18080
    },
    {
      "epoch": 9.478511530398324,
      "grad_norm": 0.14122621715068817,
      "learning_rate": 3.1563203383566894e-05,
      "loss": 0.5442,
      "num_input_tokens_seen": 11818920,
      "step": 18085
    },
    {
      "epoch": 9.481132075471699,
      "grad_norm": 0.20714227855205536,
      "learning_rate": 3.155216947765906e-05,
      "loss": 0.5612,
      "num_input_tokens_seen": 11821736,
      "step": 18090
    },
    {
      "epoch": 9.483752620545074,
      "grad_norm": 0.31506016850471497,
      "learning_rate": 3.154113420111211e-05,
      "loss": 0.5209,
      "num_input_tokens_seen": 11825256,
      "step": 18095
    },
    {
      "epoch": 9.48637316561845,
      "grad_norm": 0.11719614267349243,
      "learning_rate": 3.1530097556234526e-05,
      "loss": 0.5953,
      "num_input_tokens_seen": 11829416,
      "step": 18100
    },
    {
      "epoch": 9.488993710691824,
      "grad_norm": 0.14516332745552063,
      "learning_rate": 3.151905954533502e-05,
      "loss": 0.6157,
      "num_input_tokens_seen": 11832200,
      "step": 18105
    },
    {
      "epoch": 9.4916142557652,
      "grad_norm": 0.2680202126502991,
      "learning_rate": 3.150802017072263e-05,
      "loss": 0.5099,
      "num_input_tokens_seen": 11835592,
      "step": 18110
    },
    {
      "epoch": 9.494234800838575,
      "grad_norm": 0.18928822875022888,
      "learning_rate": 3.149697943470668e-05,
      "loss": 0.4481,
      "num_input_tokens_seen": 11839400,
      "step": 18115
    },
    {
      "epoch": 9.49685534591195,
      "grad_norm": 0.22961711883544922,
      "learning_rate": 3.1485937339596736e-05,
      "loss": 0.5974,
      "num_input_tokens_seen": 11842216,
      "step": 18120
    },
    {
      "epoch": 9.499475890985325,
      "grad_norm": 0.16986443102359772,
      "learning_rate": 3.147489388770271e-05,
      "loss": 0.4072,
      "num_input_tokens_seen": 11844936,
      "step": 18125
    },
    {
      "epoch": 9.5020964360587,
      "grad_norm": 0.21800008416175842,
      "learning_rate": 3.146384908133475e-05,
      "loss": 0.5015,
      "num_input_tokens_seen": 11847912,
      "step": 18130
    },
    {
      "epoch": 9.504716981132075,
      "grad_norm": 0.08950740844011307,
      "learning_rate": 3.14528029228033e-05,
      "loss": 0.3794,
      "num_input_tokens_seen": 11853192,
      "step": 18135
    },
    {
      "epoch": 9.50733752620545,
      "grad_norm": 0.1958509236574173,
      "learning_rate": 3.144175541441911e-05,
      "loss": 0.5643,
      "num_input_tokens_seen": 11856808,
      "step": 18140
    },
    {
      "epoch": 9.509958071278826,
      "grad_norm": 0.25250929594039917,
      "learning_rate": 3.1430706558493166e-05,
      "loss": 0.7472,
      "num_input_tokens_seen": 11860360,
      "step": 18145
    },
    {
      "epoch": 9.5125786163522,
      "grad_norm": 0.13288958370685577,
      "learning_rate": 3.141965635733677e-05,
      "loss": 0.3811,
      "num_input_tokens_seen": 11863048,
      "step": 18150
    },
    {
      "epoch": 9.515199161425576,
      "grad_norm": 0.13491995632648468,
      "learning_rate": 3.140860481326152e-05,
      "loss": 0.4739,
      "num_input_tokens_seen": 11867016,
      "step": 18155
    },
    {
      "epoch": 9.517819706498951,
      "grad_norm": 0.2640838027000427,
      "learning_rate": 3.139755192857925e-05,
      "loss": 0.4749,
      "num_input_tokens_seen": 11870728,
      "step": 18160
    },
    {
      "epoch": 9.520440251572326,
      "grad_norm": 0.11935146898031235,
      "learning_rate": 3.1386497705602094e-05,
      "loss": 0.4941,
      "num_input_tokens_seen": 11873928,
      "step": 18165
    },
    {
      "epoch": 9.523060796645701,
      "grad_norm": 0.20491957664489746,
      "learning_rate": 3.137544214664249e-05,
      "loss": 0.4397,
      "num_input_tokens_seen": 11876968,
      "step": 18170
    },
    {
      "epoch": 9.525681341719078,
      "grad_norm": 0.11041510850191116,
      "learning_rate": 3.136438525401312e-05,
      "loss": 0.4732,
      "num_input_tokens_seen": 11880648,
      "step": 18175
    },
    {
      "epoch": 9.528301886792454,
      "grad_norm": 0.18012474477291107,
      "learning_rate": 3.135332703002696e-05,
      "loss": 0.4933,
      "num_input_tokens_seen": 11883560,
      "step": 18180
    },
    {
      "epoch": 9.530922431865829,
      "grad_norm": 0.18524828553199768,
      "learning_rate": 3.1342267476997276e-05,
      "loss": 0.5429,
      "num_input_tokens_seen": 11888296,
      "step": 18185
    },
    {
      "epoch": 9.533542976939204,
      "grad_norm": 0.16662269830703735,
      "learning_rate": 3.1331206597237584e-05,
      "loss": 0.4599,
      "num_input_tokens_seen": 11892872,
      "step": 18190
    },
    {
      "epoch": 9.536163522012579,
      "grad_norm": 0.24300791323184967,
      "learning_rate": 3.1320144393061715e-05,
      "loss": 0.439,
      "num_input_tokens_seen": 11895944,
      "step": 18195
    },
    {
      "epoch": 9.538784067085954,
      "grad_norm": 0.30374908447265625,
      "learning_rate": 3.130908086678373e-05,
      "loss": 0.5251,
      "num_input_tokens_seen": 11898696,
      "step": 18200
    },
    {
      "epoch": 9.54140461215933,
      "grad_norm": 0.11676192283630371,
      "learning_rate": 3.1298016020718014e-05,
      "loss": 0.3828,
      "num_input_tokens_seen": 11904744,
      "step": 18205
    },
    {
      "epoch": 9.544025157232705,
      "grad_norm": 0.1840476393699646,
      "learning_rate": 3.128694985717919e-05,
      "loss": 0.4394,
      "num_input_tokens_seen": 11907656,
      "step": 18210
    },
    {
      "epoch": 9.54664570230608,
      "grad_norm": 0.17600831389427185,
      "learning_rate": 3.1275882378482185e-05,
      "loss": 0.5547,
      "num_input_tokens_seen": 11910504,
      "step": 18215
    },
    {
      "epoch": 9.549266247379455,
      "grad_norm": 0.2937624156475067,
      "learning_rate": 3.1264813586942185e-05,
      "loss": 0.578,
      "num_input_tokens_seen": 11913192,
      "step": 18220
    },
    {
      "epoch": 9.55188679245283,
      "grad_norm": 0.19018079340457916,
      "learning_rate": 3.125374348487466e-05,
      "loss": 0.6184,
      "num_input_tokens_seen": 11916680,
      "step": 18225
    },
    {
      "epoch": 9.554507337526205,
      "grad_norm": 0.41176334023475647,
      "learning_rate": 3.1242672074595334e-05,
      "loss": 0.4352,
      "num_input_tokens_seen": 11919656,
      "step": 18230
    },
    {
      "epoch": 9.55712788259958,
      "grad_norm": 0.1285465657711029,
      "learning_rate": 3.123159935842024e-05,
      "loss": 0.4512,
      "num_input_tokens_seen": 11922632,
      "step": 18235
    },
    {
      "epoch": 9.559748427672956,
      "grad_norm": 0.12114673107862473,
      "learning_rate": 3.122052533866565e-05,
      "loss": 0.5129,
      "num_input_tokens_seen": 11926280,
      "step": 18240
    },
    {
      "epoch": 9.56236897274633,
      "grad_norm": 0.11922896653413773,
      "learning_rate": 3.120945001764813e-05,
      "loss": 0.3417,
      "num_input_tokens_seen": 11929768,
      "step": 18245
    },
    {
      "epoch": 9.564989517819706,
      "grad_norm": 0.12022707611322403,
      "learning_rate": 3.11983733976845e-05,
      "loss": 0.4679,
      "num_input_tokens_seen": 11933480,
      "step": 18250
    },
    {
      "epoch": 9.567610062893081,
      "grad_norm": 0.19502714276313782,
      "learning_rate": 3.1187295481091876e-05,
      "loss": 0.3935,
      "num_input_tokens_seen": 11936200,
      "step": 18255
    },
    {
      "epoch": 9.570230607966456,
      "grad_norm": 0.2235952913761139,
      "learning_rate": 3.1176216270187625e-05,
      "loss": 0.5961,
      "num_input_tokens_seen": 11938472,
      "step": 18260
    },
    {
      "epoch": 9.572851153039831,
      "grad_norm": 0.1109403520822525,
      "learning_rate": 3.116513576728939e-05,
      "loss": 0.5474,
      "num_input_tokens_seen": 11942280,
      "step": 18265
    },
    {
      "epoch": 9.575471698113208,
      "grad_norm": 0.12512503564357758,
      "learning_rate": 3.1154053974715085e-05,
      "loss": 0.4295,
      "num_input_tokens_seen": 11945800,
      "step": 18270
    },
    {
      "epoch": 9.578092243186584,
      "grad_norm": 0.12743821740150452,
      "learning_rate": 3.1142970894782906e-05,
      "loss": 0.4557,
      "num_input_tokens_seen": 11948200,
      "step": 18275
    },
    {
      "epoch": 9.580712788259959,
      "grad_norm": 0.1749606728553772,
      "learning_rate": 3.1131886529811296e-05,
      "loss": 0.517,
      "num_input_tokens_seen": 11951656,
      "step": 18280
    },
    {
      "epoch": 9.583333333333334,
      "grad_norm": 0.15478569269180298,
      "learning_rate": 3.1120800882118985e-05,
      "loss": 0.4873,
      "num_input_tokens_seen": 11955208,
      "step": 18285
    },
    {
      "epoch": 9.585953878406709,
      "grad_norm": 0.18949362635612488,
      "learning_rate": 3.1109713954024945e-05,
      "loss": 0.3923,
      "num_input_tokens_seen": 11957768,
      "step": 18290
    },
    {
      "epoch": 9.588574423480084,
      "grad_norm": 0.1680799275636673,
      "learning_rate": 3.109862574784846e-05,
      "loss": 0.4693,
      "num_input_tokens_seen": 11960552,
      "step": 18295
    },
    {
      "epoch": 9.59119496855346,
      "grad_norm": 0.2009238749742508,
      "learning_rate": 3.108753626590905e-05,
      "loss": 0.5423,
      "num_input_tokens_seen": 11964552,
      "step": 18300
    },
    {
      "epoch": 9.593815513626835,
      "grad_norm": 0.37165844440460205,
      "learning_rate": 3.1076445510526486e-05,
      "loss": 0.47,
      "num_input_tokens_seen": 11966824,
      "step": 18305
    },
    {
      "epoch": 9.59643605870021,
      "grad_norm": 0.18043093383312225,
      "learning_rate": 3.106535348402086e-05,
      "loss": 0.4711,
      "num_input_tokens_seen": 11969832,
      "step": 18310
    },
    {
      "epoch": 9.599056603773585,
      "grad_norm": 0.16025732457637787,
      "learning_rate": 3.105426018871247e-05,
      "loss": 0.4216,
      "num_input_tokens_seen": 11972424,
      "step": 18315
    },
    {
      "epoch": 9.60167714884696,
      "grad_norm": 0.21205765008926392,
      "learning_rate": 3.104316562692193e-05,
      "loss": 0.4963,
      "num_input_tokens_seen": 11976552,
      "step": 18320
    },
    {
      "epoch": 9.604297693920335,
      "grad_norm": 0.19511628150939941,
      "learning_rate": 3.103206980097009e-05,
      "loss": 0.4656,
      "num_input_tokens_seen": 11979720,
      "step": 18325
    },
    {
      "epoch": 9.60691823899371,
      "grad_norm": 0.1651429533958435,
      "learning_rate": 3.102097271317805e-05,
      "loss": 0.5566,
      "num_input_tokens_seen": 11982536,
      "step": 18330
    },
    {
      "epoch": 9.609538784067086,
      "grad_norm": 0.11885642260313034,
      "learning_rate": 3.100987436586721e-05,
      "loss": 0.3691,
      "num_input_tokens_seen": 11985384,
      "step": 18335
    },
    {
      "epoch": 9.61215932914046,
      "grad_norm": 0.15550315380096436,
      "learning_rate": 3.099877476135923e-05,
      "loss": 0.4363,
      "num_input_tokens_seen": 11988712,
      "step": 18340
    },
    {
      "epoch": 9.614779874213836,
      "grad_norm": 0.15466240048408508,
      "learning_rate": 3.0987673901976e-05,
      "loss": 0.4581,
      "num_input_tokens_seen": 11993352,
      "step": 18345
    },
    {
      "epoch": 9.617400419287211,
      "grad_norm": 0.13441120088100433,
      "learning_rate": 3.0976571790039697e-05,
      "loss": 0.4936,
      "num_input_tokens_seen": 11995848,
      "step": 18350
    },
    {
      "epoch": 9.620020964360586,
      "grad_norm": 0.17471347749233246,
      "learning_rate": 3.096546842787277e-05,
      "loss": 0.5505,
      "num_input_tokens_seen": 11999336,
      "step": 18355
    },
    {
      "epoch": 9.622641509433961,
      "grad_norm": 0.15748532116413116,
      "learning_rate": 3.095436381779789e-05,
      "loss": 0.4949,
      "num_input_tokens_seen": 12003080,
      "step": 18360
    },
    {
      "epoch": 9.625262054507338,
      "grad_norm": 0.10783877968788147,
      "learning_rate": 3.0943257962138045e-05,
      "loss": 0.3103,
      "num_input_tokens_seen": 12007240,
      "step": 18365
    },
    {
      "epoch": 9.627882599580714,
      "grad_norm": 0.16698221862316132,
      "learning_rate": 3.093215086321644e-05,
      "loss": 0.5192,
      "num_input_tokens_seen": 12009928,
      "step": 18370
    },
    {
      "epoch": 9.630503144654089,
      "grad_norm": 0.13167397677898407,
      "learning_rate": 3.092104252335654e-05,
      "loss": 0.4519,
      "num_input_tokens_seen": 12013384,
      "step": 18375
    },
    {
      "epoch": 9.633123689727464,
      "grad_norm": 0.2024066299200058,
      "learning_rate": 3.09099329448821e-05,
      "loss": 0.5263,
      "num_input_tokens_seen": 12016520,
      "step": 18380
    },
    {
      "epoch": 9.635744234800839,
      "grad_norm": 0.20448380708694458,
      "learning_rate": 3.089882213011711e-05,
      "loss": 0.4523,
      "num_input_tokens_seen": 12020040,
      "step": 18385
    },
    {
      "epoch": 9.638364779874214,
      "grad_norm": 0.1310124695301056,
      "learning_rate": 3.088771008138583e-05,
      "loss": 0.4687,
      "num_input_tokens_seen": 12023464,
      "step": 18390
    },
    {
      "epoch": 9.64098532494759,
      "grad_norm": 0.2127106785774231,
      "learning_rate": 3.087659680101278e-05,
      "loss": 0.3481,
      "num_input_tokens_seen": 12026664,
      "step": 18395
    },
    {
      "epoch": 9.643605870020965,
      "grad_norm": 0.10459689795970917,
      "learning_rate": 3.086548229132271e-05,
      "loss": 0.4157,
      "num_input_tokens_seen": 12030440,
      "step": 18400
    },
    {
      "epoch": 9.64622641509434,
      "grad_norm": 0.17229372262954712,
      "learning_rate": 3.085436655464067e-05,
      "loss": 0.5253,
      "num_input_tokens_seen": 12032872,
      "step": 18405
    },
    {
      "epoch": 9.648846960167715,
      "grad_norm": 0.12836934626102448,
      "learning_rate": 3.084324959329193e-05,
      "loss": 0.3934,
      "num_input_tokens_seen": 12037864,
      "step": 18410
    },
    {
      "epoch": 9.65146750524109,
      "grad_norm": 0.20883244276046753,
      "learning_rate": 3.083213140960204e-05,
      "loss": 0.3418,
      "num_input_tokens_seen": 12040904,
      "step": 18415
    },
    {
      "epoch": 9.654088050314465,
      "grad_norm": 0.1446009874343872,
      "learning_rate": 3.08210120058968e-05,
      "loss": 0.4783,
      "num_input_tokens_seen": 12044296,
      "step": 18420
    },
    {
      "epoch": 9.65670859538784,
      "grad_norm": 0.13710594177246094,
      "learning_rate": 3.080989138450225e-05,
      "loss": 0.3692,
      "num_input_tokens_seen": 12047208,
      "step": 18425
    },
    {
      "epoch": 9.659329140461216,
      "grad_norm": 0.1865144819021225,
      "learning_rate": 3.0798769547744713e-05,
      "loss": 0.5493,
      "num_input_tokens_seen": 12050696,
      "step": 18430
    },
    {
      "epoch": 9.66194968553459,
      "grad_norm": 0.18144531548023224,
      "learning_rate": 3.0787646497950736e-05,
      "loss": 0.5308,
      "num_input_tokens_seen": 12054536,
      "step": 18435
    },
    {
      "epoch": 9.664570230607966,
      "grad_norm": 0.20674777030944824,
      "learning_rate": 3.0776522237447146e-05,
      "loss": 0.4889,
      "num_input_tokens_seen": 12056776,
      "step": 18440
    },
    {
      "epoch": 9.667190775681341,
      "grad_norm": 0.13645292818546295,
      "learning_rate": 3.076539676856101e-05,
      "loss": 0.5511,
      "num_input_tokens_seen": 12059720,
      "step": 18445
    },
    {
      "epoch": 9.669811320754716,
      "grad_norm": 0.18770788609981537,
      "learning_rate": 3.0754270093619634e-05,
      "loss": 0.4722,
      "num_input_tokens_seen": 12063304,
      "step": 18450
    },
    {
      "epoch": 9.672431865828091,
      "grad_norm": 0.1356370598077774,
      "learning_rate": 3.074314221495061e-05,
      "loss": 0.5369,
      "num_input_tokens_seen": 12067592,
      "step": 18455
    },
    {
      "epoch": 9.675052410901468,
      "grad_norm": 0.15113386511802673,
      "learning_rate": 3.0732013134881756e-05,
      "loss": 0.4913,
      "num_input_tokens_seen": 12070856,
      "step": 18460
    },
    {
      "epoch": 9.677672955974844,
      "grad_norm": 0.16270790994167328,
      "learning_rate": 3.072088285574114e-05,
      "loss": 0.4976,
      "num_input_tokens_seen": 12074280,
      "step": 18465
    },
    {
      "epoch": 9.680293501048219,
      "grad_norm": 0.217583566904068,
      "learning_rate": 3.070975137985711e-05,
      "loss": 0.4333,
      "num_input_tokens_seen": 12077288,
      "step": 18470
    },
    {
      "epoch": 9.682914046121594,
      "grad_norm": 0.18136842548847198,
      "learning_rate": 3.0698618709558226e-05,
      "loss": 0.4126,
      "num_input_tokens_seen": 12082056,
      "step": 18475
    },
    {
      "epoch": 9.685534591194969,
      "grad_norm": 0.15326933562755585,
      "learning_rate": 3.068748484717332e-05,
      "loss": 0.5617,
      "num_input_tokens_seen": 12085128,
      "step": 18480
    },
    {
      "epoch": 9.688155136268344,
      "grad_norm": 0.14102594554424286,
      "learning_rate": 3.067634979503147e-05,
      "loss": 0.605,
      "num_input_tokens_seen": 12089032,
      "step": 18485
    },
    {
      "epoch": 9.69077568134172,
      "grad_norm": 0.16476215422153473,
      "learning_rate": 3.0665213555462e-05,
      "loss": 0.5744,
      "num_input_tokens_seen": 12091496,
      "step": 18490
    },
    {
      "epoch": 9.693396226415095,
      "grad_norm": 0.17421087622642517,
      "learning_rate": 3.0654076130794494e-05,
      "loss": 0.5959,
      "num_input_tokens_seen": 12095208,
      "step": 18495
    },
    {
      "epoch": 9.69601677148847,
      "grad_norm": 0.15922339260578156,
      "learning_rate": 3.064293752335876e-05,
      "loss": 0.5351,
      "num_input_tokens_seen": 12099176,
      "step": 18500
    },
    {
      "epoch": 9.698637316561845,
      "grad_norm": 0.1835254430770874,
      "learning_rate": 3.063179773548487e-05,
      "loss": 0.5828,
      "num_input_tokens_seen": 12102408,
      "step": 18505
    },
    {
      "epoch": 9.70125786163522,
      "grad_norm": 0.14096078276634216,
      "learning_rate": 3.0620656769503164e-05,
      "loss": 0.579,
      "num_input_tokens_seen": 12105704,
      "step": 18510
    },
    {
      "epoch": 9.703878406708595,
      "grad_norm": 0.12370225042104721,
      "learning_rate": 3.060951462774417e-05,
      "loss": 0.57,
      "num_input_tokens_seen": 12108520,
      "step": 18515
    },
    {
      "epoch": 9.70649895178197,
      "grad_norm": 0.15922676026821136,
      "learning_rate": 3.0598371312538714e-05,
      "loss": 0.4599,
      "num_input_tokens_seen": 12111912,
      "step": 18520
    },
    {
      "epoch": 9.709119496855346,
      "grad_norm": 0.22588443756103516,
      "learning_rate": 3.058722682621785e-05,
      "loss": 0.4562,
      "num_input_tokens_seen": 12116040,
      "step": 18525
    },
    {
      "epoch": 9.71174004192872,
      "grad_norm": 0.12493490427732468,
      "learning_rate": 3.0576081171112875e-05,
      "loss": 0.4397,
      "num_input_tokens_seen": 12118632,
      "step": 18530
    },
    {
      "epoch": 9.714360587002096,
      "grad_norm": 0.16927245259284973,
      "learning_rate": 3.056493434955534e-05,
      "loss": 0.3466,
      "num_input_tokens_seen": 12121192,
      "step": 18535
    },
    {
      "epoch": 9.716981132075471,
      "grad_norm": 0.17830204963684082,
      "learning_rate": 3.0553786363877027e-05,
      "loss": 0.4547,
      "num_input_tokens_seen": 12123976,
      "step": 18540
    },
    {
      "epoch": 9.719601677148846,
      "grad_norm": 0.14569583535194397,
      "learning_rate": 3.054263721640996e-05,
      "loss": 0.6184,
      "num_input_tokens_seen": 12127816,
      "step": 18545
    },
    {
      "epoch": 9.722222222222221,
      "grad_norm": 0.11516307294368744,
      "learning_rate": 3.053148690948642e-05,
      "loss": 0.5651,
      "num_input_tokens_seen": 12131848,
      "step": 18550
    },
    {
      "epoch": 9.724842767295598,
      "grad_norm": 0.18481850624084473,
      "learning_rate": 3.0520335445438935e-05,
      "loss": 0.4727,
      "num_input_tokens_seen": 12135432,
      "step": 18555
    },
    {
      "epoch": 9.727463312368974,
      "grad_norm": 0.15310664474964142,
      "learning_rate": 3.0509182826600246e-05,
      "loss": 0.529,
      "num_input_tokens_seen": 12138376,
      "step": 18560
    },
    {
      "epoch": 9.730083857442349,
      "grad_norm": 0.1179194226861,
      "learning_rate": 3.049802905530337e-05,
      "loss": 0.5907,
      "num_input_tokens_seen": 12141480,
      "step": 18565
    },
    {
      "epoch": 9.732704402515724,
      "grad_norm": 0.1693059653043747,
      "learning_rate": 3.0486874133881533e-05,
      "loss": 0.4314,
      "num_input_tokens_seen": 12144168,
      "step": 18570
    },
    {
      "epoch": 9.735324947589099,
      "grad_norm": 0.16073809564113617,
      "learning_rate": 3.047571806466823e-05,
      "loss": 0.6012,
      "num_input_tokens_seen": 12148168,
      "step": 18575
    },
    {
      "epoch": 9.737945492662474,
      "grad_norm": 0.11536400765180588,
      "learning_rate": 3.0464560849997177e-05,
      "loss": 0.3419,
      "num_input_tokens_seen": 12150696,
      "step": 18580
    },
    {
      "epoch": 9.74056603773585,
      "grad_norm": 0.17802724242210388,
      "learning_rate": 3.0453402492202325e-05,
      "loss": 0.3838,
      "num_input_tokens_seen": 12153928,
      "step": 18585
    },
    {
      "epoch": 9.743186582809225,
      "grad_norm": 0.1350228637456894,
      "learning_rate": 3.0442242993617904e-05,
      "loss": 0.4496,
      "num_input_tokens_seen": 12157384,
      "step": 18590
    },
    {
      "epoch": 9.7458071278826,
      "grad_norm": 0.11933938413858414,
      "learning_rate": 3.0431082356578334e-05,
      "loss": 0.4792,
      "num_input_tokens_seen": 12160200,
      "step": 18595
    },
    {
      "epoch": 9.748427672955975,
      "grad_norm": 0.1897611916065216,
      "learning_rate": 3.0419920583418294e-05,
      "loss": 0.6618,
      "num_input_tokens_seen": 12162504,
      "step": 18600
    },
    {
      "epoch": 9.75104821802935,
      "grad_norm": 0.17222008109092712,
      "learning_rate": 3.0408757676472694e-05,
      "loss": 0.488,
      "num_input_tokens_seen": 12165288,
      "step": 18605
    },
    {
      "epoch": 9.753668763102725,
      "grad_norm": 0.21336397528648376,
      "learning_rate": 3.0397593638076705e-05,
      "loss": 0.4597,
      "num_input_tokens_seen": 12167912,
      "step": 18610
    },
    {
      "epoch": 9.7562893081761,
      "grad_norm": 0.5044729113578796,
      "learning_rate": 3.0386428470565702e-05,
      "loss": 0.5618,
      "num_input_tokens_seen": 12171624,
      "step": 18615
    },
    {
      "epoch": 9.758909853249476,
      "grad_norm": 0.280274361371994,
      "learning_rate": 3.0375262176275308e-05,
      "loss": 0.515,
      "num_input_tokens_seen": 12174632,
      "step": 18620
    },
    {
      "epoch": 9.76153039832285,
      "grad_norm": 0.26175570487976074,
      "learning_rate": 3.03640947575414e-05,
      "loss": 0.5985,
      "num_input_tokens_seen": 12177896,
      "step": 18625
    },
    {
      "epoch": 9.764150943396226,
      "grad_norm": 0.1817547082901001,
      "learning_rate": 3.0352926216700067e-05,
      "loss": 0.5146,
      "num_input_tokens_seen": 12181256,
      "step": 18630
    },
    {
      "epoch": 9.766771488469601,
      "grad_norm": 0.16235651075839996,
      "learning_rate": 3.034175655608763e-05,
      "loss": 0.5488,
      "num_input_tokens_seen": 12183848,
      "step": 18635
    },
    {
      "epoch": 9.769392033542976,
      "grad_norm": 0.14278504252433777,
      "learning_rate": 3.0330585778040677e-05,
      "loss": 0.4615,
      "num_input_tokens_seen": 12187208,
      "step": 18640
    },
    {
      "epoch": 9.772012578616351,
      "grad_norm": 0.1180848777294159,
      "learning_rate": 3.031941388489598e-05,
      "loss": 0.5009,
      "num_input_tokens_seen": 12190568,
      "step": 18645
    },
    {
      "epoch": 9.774633123689728,
      "grad_norm": 0.19960008561611176,
      "learning_rate": 3.0308240878990584e-05,
      "loss": 0.3658,
      "num_input_tokens_seen": 12193224,
      "step": 18650
    },
    {
      "epoch": 9.777253668763104,
      "grad_norm": 0.17981456220149994,
      "learning_rate": 3.0297066762661773e-05,
      "loss": 0.4655,
      "num_input_tokens_seen": 12195816,
      "step": 18655
    },
    {
      "epoch": 9.779874213836479,
      "grad_norm": 0.2994875907897949,
      "learning_rate": 3.0285891538247015e-05,
      "loss": 0.4803,
      "num_input_tokens_seen": 12198856,
      "step": 18660
    },
    {
      "epoch": 9.782494758909854,
      "grad_norm": 0.19680438935756683,
      "learning_rate": 3.0274715208084053e-05,
      "loss": 0.564,
      "num_input_tokens_seen": 12201992,
      "step": 18665
    },
    {
      "epoch": 9.785115303983229,
      "grad_norm": 0.18281762301921844,
      "learning_rate": 3.0263537774510847e-05,
      "loss": 0.516,
      "num_input_tokens_seen": 12205608,
      "step": 18670
    },
    {
      "epoch": 9.787735849056604,
      "grad_norm": 0.12456802278757095,
      "learning_rate": 3.0252359239865585e-05,
      "loss": 0.577,
      "num_input_tokens_seen": 12209352,
      "step": 18675
    },
    {
      "epoch": 9.79035639412998,
      "grad_norm": 0.18427973985671997,
      "learning_rate": 3.0241179606486704e-05,
      "loss": 0.4896,
      "num_input_tokens_seen": 12213768,
      "step": 18680
    },
    {
      "epoch": 9.792976939203355,
      "grad_norm": 0.18702265620231628,
      "learning_rate": 3.0229998876712827e-05,
      "loss": 0.4927,
      "num_input_tokens_seen": 12216776,
      "step": 18685
    },
    {
      "epoch": 9.79559748427673,
      "grad_norm": 0.1324544996023178,
      "learning_rate": 3.0218817052882858e-05,
      "loss": 0.4063,
      "num_input_tokens_seen": 12221000,
      "step": 18690
    },
    {
      "epoch": 9.798218029350105,
      "grad_norm": 0.2680393159389496,
      "learning_rate": 3.020763413733591e-05,
      "loss": 0.4759,
      "num_input_tokens_seen": 12224200,
      "step": 18695
    },
    {
      "epoch": 9.80083857442348,
      "grad_norm": 0.1631188988685608,
      "learning_rate": 3.0196450132411302e-05,
      "loss": 0.493,
      "num_input_tokens_seen": 12226920,
      "step": 18700
    },
    {
      "epoch": 9.803459119496855,
      "grad_norm": 0.12958025932312012,
      "learning_rate": 3.0185265040448616e-05,
      "loss": 0.5305,
      "num_input_tokens_seen": 12229960,
      "step": 18705
    },
    {
      "epoch": 9.80607966457023,
      "grad_norm": 0.2424205243587494,
      "learning_rate": 3.017407886378764e-05,
      "loss": 0.4428,
      "num_input_tokens_seen": 12232040,
      "step": 18710
    },
    {
      "epoch": 9.808700209643606,
      "grad_norm": 0.14446045458316803,
      "learning_rate": 3.016289160476839e-05,
      "loss": 0.4178,
      "num_input_tokens_seen": 12234472,
      "step": 18715
    },
    {
      "epoch": 9.81132075471698,
      "grad_norm": 0.26921847462654114,
      "learning_rate": 3.0151703265731125e-05,
      "loss": 0.3678,
      "num_input_tokens_seen": 12237448,
      "step": 18720
    },
    {
      "epoch": 9.813941299790356,
      "grad_norm": 0.12479125708341599,
      "learning_rate": 3.0140513849016305e-05,
      "loss": 0.479,
      "num_input_tokens_seen": 12240264,
      "step": 18725
    },
    {
      "epoch": 9.816561844863731,
      "grad_norm": 0.1319788545370102,
      "learning_rate": 3.0129323356964627e-05,
      "loss": 0.4241,
      "num_input_tokens_seen": 12243176,
      "step": 18730
    },
    {
      "epoch": 9.819182389937106,
      "grad_norm": 0.19833491742610931,
      "learning_rate": 3.0118131791917032e-05,
      "loss": 0.5,
      "num_input_tokens_seen": 12245864,
      "step": 18735
    },
    {
      "epoch": 9.821802935010481,
      "grad_norm": 0.1452035754919052,
      "learning_rate": 3.010693915621466e-05,
      "loss": 0.4585,
      "num_input_tokens_seen": 12248712,
      "step": 18740
    },
    {
      "epoch": 9.824423480083858,
      "grad_norm": 0.32711413502693176,
      "learning_rate": 3.009574545219887e-05,
      "loss": 0.5473,
      "num_input_tokens_seen": 12251144,
      "step": 18745
    },
    {
      "epoch": 9.827044025157234,
      "grad_norm": 0.1587843894958496,
      "learning_rate": 3.0084550682211272e-05,
      "loss": 0.4805,
      "num_input_tokens_seen": 12255336,
      "step": 18750
    },
    {
      "epoch": 9.829664570230609,
      "grad_norm": 0.33415383100509644,
      "learning_rate": 3.0073354848593677e-05,
      "loss": 0.5425,
      "num_input_tokens_seen": 12257672,
      "step": 18755
    },
    {
      "epoch": 9.832285115303984,
      "grad_norm": 0.15819387137889862,
      "learning_rate": 3.0062157953688125e-05,
      "loss": 0.4744,
      "num_input_tokens_seen": 12261032,
      "step": 18760
    },
    {
      "epoch": 9.834905660377359,
      "grad_norm": 0.15135294198989868,
      "learning_rate": 3.0050959999836887e-05,
      "loss": 0.4615,
      "num_input_tokens_seen": 12264488,
      "step": 18765
    },
    {
      "epoch": 9.837526205450734,
      "grad_norm": 0.2178337424993515,
      "learning_rate": 3.003976098938244e-05,
      "loss": 0.4945,
      "num_input_tokens_seen": 12267432,
      "step": 18770
    },
    {
      "epoch": 9.84014675052411,
      "grad_norm": 0.23321782052516937,
      "learning_rate": 3.0028560924667487e-05,
      "loss": 0.3794,
      "num_input_tokens_seen": 12270888,
      "step": 18775
    },
    {
      "epoch": 9.842767295597485,
      "grad_norm": 0.12761719524860382,
      "learning_rate": 3.001735980803496e-05,
      "loss": 0.4682,
      "num_input_tokens_seen": 12273896,
      "step": 18780
    },
    {
      "epoch": 9.84538784067086,
      "grad_norm": 0.17131273448467255,
      "learning_rate": 3.0006157641828004e-05,
      "loss": 0.543,
      "num_input_tokens_seen": 12276328,
      "step": 18785
    },
    {
      "epoch": 9.848008385744235,
      "grad_norm": 0.1767570674419403,
      "learning_rate": 2.9994954428389977e-05,
      "loss": 0.4275,
      "num_input_tokens_seen": 12279016,
      "step": 18790
    },
    {
      "epoch": 9.85062893081761,
      "grad_norm": 0.25052016973495483,
      "learning_rate": 2.9983750170064472e-05,
      "loss": 0.3827,
      "num_input_tokens_seen": 12281512,
      "step": 18795
    },
    {
      "epoch": 9.853249475890985,
      "grad_norm": 0.1748497635126114,
      "learning_rate": 2.9972544869195292e-05,
      "loss": 0.5588,
      "num_input_tokens_seen": 12285160,
      "step": 18800
    },
    {
      "epoch": 9.85587002096436,
      "grad_norm": 0.12875191867351532,
      "learning_rate": 2.9961338528126447e-05,
      "loss": 0.4903,
      "num_input_tokens_seen": 12288616,
      "step": 18805
    },
    {
      "epoch": 9.858490566037736,
      "grad_norm": 0.15217837691307068,
      "learning_rate": 2.9950131149202192e-05,
      "loss": 0.5652,
      "num_input_tokens_seen": 12292680,
      "step": 18810
    },
    {
      "epoch": 9.86111111111111,
      "grad_norm": 0.10094999521970749,
      "learning_rate": 2.9938922734766973e-05,
      "loss": 0.4292,
      "num_input_tokens_seen": 12297160,
      "step": 18815
    },
    {
      "epoch": 9.863731656184486,
      "grad_norm": 0.1393190324306488,
      "learning_rate": 2.992771328716546e-05,
      "loss": 0.6378,
      "num_input_tokens_seen": 12300520,
      "step": 18820
    },
    {
      "epoch": 9.866352201257861,
      "grad_norm": 0.1415702998638153,
      "learning_rate": 2.991650280874255e-05,
      "loss": 0.4987,
      "num_input_tokens_seen": 12303496,
      "step": 18825
    },
    {
      "epoch": 9.868972746331236,
      "grad_norm": 0.15239204466342926,
      "learning_rate": 2.990529130184334e-05,
      "loss": 0.5538,
      "num_input_tokens_seen": 12306568,
      "step": 18830
    },
    {
      "epoch": 9.871593291404611,
      "grad_norm": 0.16700586676597595,
      "learning_rate": 2.989407876881316e-05,
      "loss": 0.5771,
      "num_input_tokens_seen": 12309160,
      "step": 18835
    },
    {
      "epoch": 9.874213836477988,
      "grad_norm": 0.13757523894309998,
      "learning_rate": 2.988286521199753e-05,
      "loss": 0.4348,
      "num_input_tokens_seen": 12312776,
      "step": 18840
    },
    {
      "epoch": 9.876834381551364,
      "grad_norm": 0.18981553614139557,
      "learning_rate": 2.9871650633742198e-05,
      "loss": 0.421,
      "num_input_tokens_seen": 12316104,
      "step": 18845
    },
    {
      "epoch": 9.879454926624739,
      "grad_norm": 0.24476438760757446,
      "learning_rate": 2.9860435036393142e-05,
      "loss": 0.5791,
      "num_input_tokens_seen": 12318568,
      "step": 18850
    },
    {
      "epoch": 9.882075471698114,
      "grad_norm": 0.23508164286613464,
      "learning_rate": 2.9849218422296527e-05,
      "loss": 0.5382,
      "num_input_tokens_seen": 12321736,
      "step": 18855
    },
    {
      "epoch": 9.884696016771489,
      "grad_norm": 0.16726019978523254,
      "learning_rate": 2.983800079379873e-05,
      "loss": 0.509,
      "num_input_tokens_seen": 12325032,
      "step": 18860
    },
    {
      "epoch": 9.887316561844864,
      "grad_norm": 0.10407048463821411,
      "learning_rate": 2.982678215324638e-05,
      "loss": 0.5262,
      "num_input_tokens_seen": 12329160,
      "step": 18865
    },
    {
      "epoch": 9.88993710691824,
      "grad_norm": 0.12620820105075836,
      "learning_rate": 2.9815562502986256e-05,
      "loss": 0.7032,
      "num_input_tokens_seen": 12332392,
      "step": 18870
    },
    {
      "epoch": 9.892557651991615,
      "grad_norm": 0.10454914718866348,
      "learning_rate": 2.9804341845365398e-05,
      "loss": 0.4462,
      "num_input_tokens_seen": 12336456,
      "step": 18875
    },
    {
      "epoch": 9.89517819706499,
      "grad_norm": 0.14271146059036255,
      "learning_rate": 2.9793120182731037e-05,
      "loss": 0.4389,
      "num_input_tokens_seen": 12339080,
      "step": 18880
    },
    {
      "epoch": 9.897798742138365,
      "grad_norm": 0.19107124209403992,
      "learning_rate": 2.9781897517430617e-05,
      "loss": 0.419,
      "num_input_tokens_seen": 12341768,
      "step": 18885
    },
    {
      "epoch": 9.90041928721174,
      "grad_norm": 0.17078085243701935,
      "learning_rate": 2.977067385181179e-05,
      "loss": 0.4256,
      "num_input_tokens_seen": 12344936,
      "step": 18890
    },
    {
      "epoch": 9.903039832285115,
      "grad_norm": 0.2549859583377838,
      "learning_rate": 2.9759449188222426e-05,
      "loss": 0.4828,
      "num_input_tokens_seen": 12349416,
      "step": 18895
    },
    {
      "epoch": 9.90566037735849,
      "grad_norm": 0.37829867005348206,
      "learning_rate": 2.9748223529010582e-05,
      "loss": 0.5454,
      "num_input_tokens_seen": 12352360,
      "step": 18900
    },
    {
      "epoch": 9.908280922431866,
      "grad_norm": 0.17925307154655457,
      "learning_rate": 2.9736996876524546e-05,
      "loss": 0.549,
      "num_input_tokens_seen": 12355688,
      "step": 18905
    },
    {
      "epoch": 9.91090146750524,
      "grad_norm": 0.12819857895374298,
      "learning_rate": 2.9725769233112816e-05,
      "loss": 0.4845,
      "num_input_tokens_seen": 12359560,
      "step": 18910
    },
    {
      "epoch": 9.913522012578616,
      "grad_norm": 0.14171087741851807,
      "learning_rate": 2.9714540601124074e-05,
      "loss": 0.5391,
      "num_input_tokens_seen": 12363656,
      "step": 18915
    },
    {
      "epoch": 9.916142557651991,
      "grad_norm": 0.12264204770326614,
      "learning_rate": 2.9703310982907218e-05,
      "loss": 0.4134,
      "num_input_tokens_seen": 12367688,
      "step": 18920
    },
    {
      "epoch": 9.918763102725366,
      "grad_norm": 0.20448938012123108,
      "learning_rate": 2.9692080380811365e-05,
      "loss": 0.4824,
      "num_input_tokens_seen": 12371400,
      "step": 18925
    },
    {
      "epoch": 9.921383647798741,
      "grad_norm": 0.11438040435314178,
      "learning_rate": 2.9680848797185828e-05,
      "loss": 0.486,
      "num_input_tokens_seen": 12374152,
      "step": 18930
    },
    {
      "epoch": 9.924004192872118,
      "grad_norm": 0.29747694730758667,
      "learning_rate": 2.966961623438012e-05,
      "loss": 0.3828,
      "num_input_tokens_seen": 12378472,
      "step": 18935
    },
    {
      "epoch": 9.926624737945493,
      "grad_norm": 0.23134174942970276,
      "learning_rate": 2.9658382694743973e-05,
      "loss": 0.536,
      "num_input_tokens_seen": 12381480,
      "step": 18940
    },
    {
      "epoch": 9.929245283018869,
      "grad_norm": 0.12059253454208374,
      "learning_rate": 2.9647148180627317e-05,
      "loss": 0.4426,
      "num_input_tokens_seen": 12384648,
      "step": 18945
    },
    {
      "epoch": 9.931865828092244,
      "grad_norm": 0.19426752626895905,
      "learning_rate": 2.9635912694380275e-05,
      "loss": 0.3895,
      "num_input_tokens_seen": 12387432,
      "step": 18950
    },
    {
      "epoch": 9.934486373165619,
      "grad_norm": 0.1296989470720291,
      "learning_rate": 2.962467623835319e-05,
      "loss": 0.4564,
      "num_input_tokens_seen": 12390984,
      "step": 18955
    },
    {
      "epoch": 9.937106918238994,
      "grad_norm": 0.16728608310222626,
      "learning_rate": 2.961343881489659e-05,
      "loss": 0.4163,
      "num_input_tokens_seen": 12394760,
      "step": 18960
    },
    {
      "epoch": 9.93972746331237,
      "grad_norm": 0.21309345960617065,
      "learning_rate": 2.9602200426361226e-05,
      "loss": 0.5455,
      "num_input_tokens_seen": 12397768,
      "step": 18965
    },
    {
      "epoch": 9.942348008385745,
      "grad_norm": 0.18371060490608215,
      "learning_rate": 2.9590961075098038e-05,
      "loss": 0.4205,
      "num_input_tokens_seen": 12400968,
      "step": 18970
    },
    {
      "epoch": 9.94496855345912,
      "grad_norm": 0.14579181373119354,
      "learning_rate": 2.957972076345817e-05,
      "loss": 0.5478,
      "num_input_tokens_seen": 12406664,
      "step": 18975
    },
    {
      "epoch": 9.947589098532495,
      "grad_norm": 0.16790884733200073,
      "learning_rate": 2.9568479493792967e-05,
      "loss": 0.4725,
      "num_input_tokens_seen": 12409448,
      "step": 18980
    },
    {
      "epoch": 9.95020964360587,
      "grad_norm": 0.12232164293527603,
      "learning_rate": 2.955723726845397e-05,
      "loss": 0.3739,
      "num_input_tokens_seen": 12412744,
      "step": 18985
    },
    {
      "epoch": 9.952830188679245,
      "grad_norm": 0.13164786994457245,
      "learning_rate": 2.9545994089792928e-05,
      "loss": 0.6551,
      "num_input_tokens_seen": 12415656,
      "step": 18990
    },
    {
      "epoch": 9.95545073375262,
      "grad_norm": 0.1521766185760498,
      "learning_rate": 2.9534749960161795e-05,
      "loss": 0.5865,
      "num_input_tokens_seen": 12418472,
      "step": 18995
    },
    {
      "epoch": 9.958071278825996,
      "grad_norm": 0.18445096909999847,
      "learning_rate": 2.9523504881912694e-05,
      "loss": 0.4268,
      "num_input_tokens_seen": 12421768,
      "step": 19000
    },
    {
      "epoch": 9.96069182389937,
      "grad_norm": 0.13911646604537964,
      "learning_rate": 2.9512258857397978e-05,
      "loss": 0.4596,
      "num_input_tokens_seen": 12425800,
      "step": 19005
    },
    {
      "epoch": 9.963312368972746,
      "grad_norm": 0.18107731640338898,
      "learning_rate": 2.9501011888970203e-05,
      "loss": 0.5819,
      "num_input_tokens_seen": 12429288,
      "step": 19010
    },
    {
      "epoch": 9.965932914046121,
      "grad_norm": 0.13703414797782898,
      "learning_rate": 2.9489763978982075e-05,
      "loss": 0.4336,
      "num_input_tokens_seen": 12433160,
      "step": 19015
    },
    {
      "epoch": 9.968553459119496,
      "grad_norm": 0.32334578037261963,
      "learning_rate": 2.9478515129786556e-05,
      "loss": 0.4783,
      "num_input_tokens_seen": 12435944,
      "step": 19020
    },
    {
      "epoch": 9.971174004192871,
      "grad_norm": 0.11176835745573044,
      "learning_rate": 2.946726534373676e-05,
      "loss": 0.4231,
      "num_input_tokens_seen": 12438216,
      "step": 19025
    },
    {
      "epoch": 9.973794549266248,
      "grad_norm": 0.26483583450317383,
      "learning_rate": 2.945601462318601e-05,
      "loss": 0.4386,
      "num_input_tokens_seen": 12440904,
      "step": 19030
    },
    {
      "epoch": 9.976415094339622,
      "grad_norm": 0.2278711199760437,
      "learning_rate": 2.944476297048786e-05,
      "loss": 0.5449,
      "num_input_tokens_seen": 12443752,
      "step": 19035
    },
    {
      "epoch": 9.979035639412999,
      "grad_norm": 0.2993406355381012,
      "learning_rate": 2.9433510387995987e-05,
      "loss": 0.7075,
      "num_input_tokens_seen": 12446152,
      "step": 19040
    },
    {
      "epoch": 9.981656184486374,
      "grad_norm": 0.15308336913585663,
      "learning_rate": 2.9422256878064325e-05,
      "loss": 0.4867,
      "num_input_tokens_seen": 12449160,
      "step": 19045
    },
    {
      "epoch": 9.984276729559749,
      "grad_norm": 0.1429644674062729,
      "learning_rate": 2.9411002443046982e-05,
      "loss": 0.5173,
      "num_input_tokens_seen": 12453064,
      "step": 19050
    },
    {
      "epoch": 9.986897274633124,
      "grad_norm": 0.24389348924160004,
      "learning_rate": 2.9399747085298245e-05,
      "loss": 0.4813,
      "num_input_tokens_seen": 12455944,
      "step": 19055
    },
    {
      "epoch": 9.9895178197065,
      "grad_norm": 0.22168733179569244,
      "learning_rate": 2.9388490807172614e-05,
      "loss": 0.5291,
      "num_input_tokens_seen": 12459528,
      "step": 19060
    },
    {
      "epoch": 9.992138364779874,
      "grad_norm": 0.11071538180112839,
      "learning_rate": 2.9377233611024775e-05,
      "loss": 0.5056,
      "num_input_tokens_seen": 12463048,
      "step": 19065
    },
    {
      "epoch": 9.99475890985325,
      "grad_norm": 0.13581354916095734,
      "learning_rate": 2.9365975499209596e-05,
      "loss": 0.4685,
      "num_input_tokens_seen": 12467240,
      "step": 19070
    },
    {
      "epoch": 9.997379454926625,
      "grad_norm": 0.16562338173389435,
      "learning_rate": 2.9354716474082154e-05,
      "loss": 0.6806,
      "num_input_tokens_seen": 12470216,
      "step": 19075
    },
    {
      "epoch": 10.0,
      "grad_norm": 0.34829288721084595,
      "learning_rate": 2.9343456537997714e-05,
      "loss": 0.6831,
      "num_input_tokens_seen": 12472912,
      "step": 19080
    },
    {
      "epoch": 10.0,
      "eval_loss": 0.4971450865268707,
      "eval_runtime": 10.1386,
      "eval_samples_per_second": 83.64,
      "eval_steps_per_second": 20.91,
      "num_input_tokens_seen": 12472912,
      "step": 19080
    },
    {
      "epoch": 10.002620545073375,
      "grad_norm": 0.17619389295578003,
      "learning_rate": 2.9332195693311714e-05,
      "loss": 0.4159,
      "num_input_tokens_seen": 12475568,
      "step": 19085
    },
    {
      "epoch": 10.00524109014675,
      "grad_norm": 0.5941154360771179,
      "learning_rate": 2.9320933942379803e-05,
      "loss": 0.4889,
      "num_input_tokens_seen": 12478128,
      "step": 19090
    },
    {
      "epoch": 10.007861635220126,
      "grad_norm": 0.204456627368927,
      "learning_rate": 2.9309671287557805e-05,
      "loss": 0.4453,
      "num_input_tokens_seen": 12481520,
      "step": 19095
    },
    {
      "epoch": 10.0104821802935,
      "grad_norm": 0.10477451980113983,
      "learning_rate": 2.9298407731201737e-05,
      "loss": 0.5611,
      "num_input_tokens_seen": 12485264,
      "step": 19100
    },
    {
      "epoch": 10.013102725366876,
      "grad_norm": 0.18845832347869873,
      "learning_rate": 2.928714327566781e-05,
      "loss": 0.4041,
      "num_input_tokens_seen": 12488464,
      "step": 19105
    },
    {
      "epoch": 10.015723270440251,
      "grad_norm": 0.3404669165611267,
      "learning_rate": 2.9275877923312427e-05,
      "loss": 0.4513,
      "num_input_tokens_seen": 12491504,
      "step": 19110
    },
    {
      "epoch": 10.018343815513626,
      "grad_norm": 0.20609238743782043,
      "learning_rate": 2.9264611676492165e-05,
      "loss": 0.7896,
      "num_input_tokens_seen": 12494576,
      "step": 19115
    },
    {
      "epoch": 10.020964360587001,
      "grad_norm": 0.16646111011505127,
      "learning_rate": 2.9253344537563782e-05,
      "loss": 0.5075,
      "num_input_tokens_seen": 12497648,
      "step": 19120
    },
    {
      "epoch": 10.023584905660377,
      "grad_norm": 0.1865685135126114,
      "learning_rate": 2.924207650888426e-05,
      "loss": 0.474,
      "num_input_tokens_seen": 12500208,
      "step": 19125
    },
    {
      "epoch": 10.026205450733753,
      "grad_norm": 0.18192531168460846,
      "learning_rate": 2.9230807592810723e-05,
      "loss": 0.5793,
      "num_input_tokens_seen": 12503248,
      "step": 19130
    },
    {
      "epoch": 10.028825995807129,
      "grad_norm": 0.17478542029857635,
      "learning_rate": 2.92195377917005e-05,
      "loss": 0.4787,
      "num_input_tokens_seen": 12508080,
      "step": 19135
    },
    {
      "epoch": 10.031446540880504,
      "grad_norm": 0.10712284594774246,
      "learning_rate": 2.920826710791112e-05,
      "loss": 0.4476,
      "num_input_tokens_seen": 12511856,
      "step": 19140
    },
    {
      "epoch": 10.034067085953879,
      "grad_norm": 0.10625909268856049,
      "learning_rate": 2.919699554380026e-05,
      "loss": 0.4554,
      "num_input_tokens_seen": 12515920,
      "step": 19145
    },
    {
      "epoch": 10.036687631027254,
      "grad_norm": 0.22155456244945526,
      "learning_rate": 2.9185723101725825e-05,
      "loss": 0.5605,
      "num_input_tokens_seen": 12519024,
      "step": 19150
    },
    {
      "epoch": 10.03930817610063,
      "grad_norm": 0.25212299823760986,
      "learning_rate": 2.9174449784045867e-05,
      "loss": 0.5233,
      "num_input_tokens_seen": 12521840,
      "step": 19155
    },
    {
      "epoch": 10.041928721174004,
      "grad_norm": 0.1660975068807602,
      "learning_rate": 2.916317559311863e-05,
      "loss": 0.5358,
      "num_input_tokens_seen": 12525008,
      "step": 19160
    },
    {
      "epoch": 10.04454926624738,
      "grad_norm": 0.14444385468959808,
      "learning_rate": 2.9151900531302566e-05,
      "loss": 0.4121,
      "num_input_tokens_seen": 12528048,
      "step": 19165
    },
    {
      "epoch": 10.047169811320755,
      "grad_norm": 0.16487401723861694,
      "learning_rate": 2.9140624600956272e-05,
      "loss": 0.4253,
      "num_input_tokens_seen": 12531536,
      "step": 19170
    },
    {
      "epoch": 10.04979035639413,
      "grad_norm": 0.2071526199579239,
      "learning_rate": 2.912934780443855e-05,
      "loss": 0.5115,
      "num_input_tokens_seen": 12535248,
      "step": 19175
    },
    {
      "epoch": 10.052410901467505,
      "grad_norm": 0.13966104388237,
      "learning_rate": 2.9118070144108388e-05,
      "loss": 0.4064,
      "num_input_tokens_seen": 12538128,
      "step": 19180
    },
    {
      "epoch": 10.05503144654088,
      "grad_norm": 0.1339869499206543,
      "learning_rate": 2.9106791622324924e-05,
      "loss": 0.4286,
      "num_input_tokens_seen": 12541136,
      "step": 19185
    },
    {
      "epoch": 10.057651991614255,
      "grad_norm": 0.13749916851520538,
      "learning_rate": 2.909551224144751e-05,
      "loss": 0.4458,
      "num_input_tokens_seen": 12544272,
      "step": 19190
    },
    {
      "epoch": 10.06027253668763,
      "grad_norm": 0.18712349236011505,
      "learning_rate": 2.9084232003835653e-05,
      "loss": 0.4415,
      "num_input_tokens_seen": 12547184,
      "step": 19195
    },
    {
      "epoch": 10.062893081761006,
      "grad_norm": 0.14230571687221527,
      "learning_rate": 2.9072950911849058e-05,
      "loss": 0.4639,
      "num_input_tokens_seen": 12550416,
      "step": 19200
    },
    {
      "epoch": 10.065513626834381,
      "grad_norm": 0.18681497871875763,
      "learning_rate": 2.90616689678476e-05,
      "loss": 0.4377,
      "num_input_tokens_seen": 12552688,
      "step": 19205
    },
    {
      "epoch": 10.068134171907756,
      "grad_norm": 0.179862841963768,
      "learning_rate": 2.9050386174191337e-05,
      "loss": 0.5585,
      "num_input_tokens_seen": 12555536,
      "step": 19210
    },
    {
      "epoch": 10.070754716981131,
      "grad_norm": 0.1459597498178482,
      "learning_rate": 2.903910253324048e-05,
      "loss": 0.5228,
      "num_input_tokens_seen": 12558672,
      "step": 19215
    },
    {
      "epoch": 10.073375262054507,
      "grad_norm": 0.24992196261882782,
      "learning_rate": 2.9027818047355477e-05,
      "loss": 0.4768,
      "num_input_tokens_seen": 12562192,
      "step": 19220
    },
    {
      "epoch": 10.075995807127883,
      "grad_norm": 0.1382083147764206,
      "learning_rate": 2.9016532718896873e-05,
      "loss": 0.4675,
      "num_input_tokens_seen": 12565584,
      "step": 19225
    },
    {
      "epoch": 10.078616352201259,
      "grad_norm": 0.21090276539325714,
      "learning_rate": 2.900524655022545e-05,
      "loss": 0.4718,
      "num_input_tokens_seen": 12568720,
      "step": 19230
    },
    {
      "epoch": 10.081236897274634,
      "grad_norm": 0.16073617339134216,
      "learning_rate": 2.899395954370215e-05,
      "loss": 0.3896,
      "num_input_tokens_seen": 12572432,
      "step": 19235
    },
    {
      "epoch": 10.083857442348009,
      "grad_norm": 0.19281256198883057,
      "learning_rate": 2.898267170168807e-05,
      "loss": 0.5156,
      "num_input_tokens_seen": 12575632,
      "step": 19240
    },
    {
      "epoch": 10.086477987421384,
      "grad_norm": 0.09782528877258301,
      "learning_rate": 2.8971383026544507e-05,
      "loss": 0.4613,
      "num_input_tokens_seen": 12579888,
      "step": 19245
    },
    {
      "epoch": 10.08909853249476,
      "grad_norm": 0.2694251239299774,
      "learning_rate": 2.896009352063293e-05,
      "loss": 0.4083,
      "num_input_tokens_seen": 12582416,
      "step": 19250
    },
    {
      "epoch": 10.091719077568134,
      "grad_norm": 0.12287485599517822,
      "learning_rate": 2.894880318631496e-05,
      "loss": 0.4172,
      "num_input_tokens_seen": 12585360,
      "step": 19255
    },
    {
      "epoch": 10.09433962264151,
      "grad_norm": 0.10754785686731339,
      "learning_rate": 2.8937512025952417e-05,
      "loss": 0.4615,
      "num_input_tokens_seen": 12589456,
      "step": 19260
    },
    {
      "epoch": 10.096960167714885,
      "grad_norm": 0.18066613376140594,
      "learning_rate": 2.8926220041907283e-05,
      "loss": 0.4823,
      "num_input_tokens_seen": 12592752,
      "step": 19265
    },
    {
      "epoch": 10.09958071278826,
      "grad_norm": 0.13881002366542816,
      "learning_rate": 2.8914927236541706e-05,
      "loss": 0.585,
      "num_input_tokens_seen": 12596304,
      "step": 19270
    },
    {
      "epoch": 10.102201257861635,
      "grad_norm": 0.17131710052490234,
      "learning_rate": 2.8903633612218012e-05,
      "loss": 0.5236,
      "num_input_tokens_seen": 12600048,
      "step": 19275
    },
    {
      "epoch": 10.10482180293501,
      "grad_norm": 0.12704814970493317,
      "learning_rate": 2.8892339171298703e-05,
      "loss": 0.4086,
      "num_input_tokens_seen": 12602992,
      "step": 19280
    },
    {
      "epoch": 10.107442348008385,
      "grad_norm": 0.1663946956396103,
      "learning_rate": 2.888104391614645e-05,
      "loss": 0.4832,
      "num_input_tokens_seen": 12606960,
      "step": 19285
    },
    {
      "epoch": 10.11006289308176,
      "grad_norm": 0.14102241396903992,
      "learning_rate": 2.886974784912408e-05,
      "loss": 0.3717,
      "num_input_tokens_seen": 12609584,
      "step": 19290
    },
    {
      "epoch": 10.112683438155136,
      "grad_norm": 0.12241286039352417,
      "learning_rate": 2.8858450972594612e-05,
      "loss": 0.476,
      "num_input_tokens_seen": 12612368,
      "step": 19295
    },
    {
      "epoch": 10.115303983228511,
      "grad_norm": 0.16280245780944824,
      "learning_rate": 2.8847153288921226e-05,
      "loss": 0.4395,
      "num_input_tokens_seen": 12615472,
      "step": 19300
    },
    {
      "epoch": 10.117924528301886,
      "grad_norm": 0.2177734673023224,
      "learning_rate": 2.883585480046725e-05,
      "loss": 0.7102,
      "num_input_tokens_seen": 12621456,
      "step": 19305
    },
    {
      "epoch": 10.120545073375261,
      "grad_norm": 0.36904191970825195,
      "learning_rate": 2.8824555509596223e-05,
      "loss": 0.505,
      "num_input_tokens_seen": 12624624,
      "step": 19310
    },
    {
      "epoch": 10.123165618448636,
      "grad_norm": 0.13708451390266418,
      "learning_rate": 2.8813255418671815e-05,
      "loss": 0.5841,
      "num_input_tokens_seen": 12628112,
      "step": 19315
    },
    {
      "epoch": 10.125786163522013,
      "grad_norm": 0.1478472650051117,
      "learning_rate": 2.8801954530057874e-05,
      "loss": 0.5403,
      "num_input_tokens_seen": 12631536,
      "step": 19320
    },
    {
      "epoch": 10.128406708595389,
      "grad_norm": 0.30225613713264465,
      "learning_rate": 2.879065284611842e-05,
      "loss": 0.5,
      "num_input_tokens_seen": 12634704,
      "step": 19325
    },
    {
      "epoch": 10.131027253668764,
      "grad_norm": 0.14213676750659943,
      "learning_rate": 2.8779350369217634e-05,
      "loss": 0.456,
      "num_input_tokens_seen": 12638640,
      "step": 19330
    },
    {
      "epoch": 10.133647798742139,
      "grad_norm": 0.1278422474861145,
      "learning_rate": 2.876804710171987e-05,
      "loss": 0.4368,
      "num_input_tokens_seen": 12641392,
      "step": 19335
    },
    {
      "epoch": 10.136268343815514,
      "grad_norm": 0.10861237347126007,
      "learning_rate": 2.8756743045989643e-05,
      "loss": 0.3719,
      "num_input_tokens_seen": 12648656,
      "step": 19340
    },
    {
      "epoch": 10.13888888888889,
      "grad_norm": 0.21663898229599,
      "learning_rate": 2.8745438204391623e-05,
      "loss": 0.4194,
      "num_input_tokens_seen": 12651760,
      "step": 19345
    },
    {
      "epoch": 10.141509433962264,
      "grad_norm": 0.2097902148962021,
      "learning_rate": 2.8734132579290675e-05,
      "loss": 0.4975,
      "num_input_tokens_seen": 12655120,
      "step": 19350
    },
    {
      "epoch": 10.14412997903564,
      "grad_norm": 0.33432865142822266,
      "learning_rate": 2.8722826173051782e-05,
      "loss": 0.442,
      "num_input_tokens_seen": 12657936,
      "step": 19355
    },
    {
      "epoch": 10.146750524109015,
      "grad_norm": 0.25727665424346924,
      "learning_rate": 2.871151898804012e-05,
      "loss": 0.4551,
      "num_input_tokens_seen": 12661648,
      "step": 19360
    },
    {
      "epoch": 10.14937106918239,
      "grad_norm": 0.16212595999240875,
      "learning_rate": 2.8700211026621043e-05,
      "loss": 0.6409,
      "num_input_tokens_seen": 12665328,
      "step": 19365
    },
    {
      "epoch": 10.151991614255765,
      "grad_norm": 0.17951726913452148,
      "learning_rate": 2.8688902291160025e-05,
      "loss": 0.4664,
      "num_input_tokens_seen": 12668464,
      "step": 19370
    },
    {
      "epoch": 10.15461215932914,
      "grad_norm": 0.16044525802135468,
      "learning_rate": 2.867759278402273e-05,
      "loss": 0.4236,
      "num_input_tokens_seen": 12672016,
      "step": 19375
    },
    {
      "epoch": 10.157232704402515,
      "grad_norm": 0.2698766589164734,
      "learning_rate": 2.8666282507574994e-05,
      "loss": 0.549,
      "num_input_tokens_seen": 12675312,
      "step": 19380
    },
    {
      "epoch": 10.15985324947589,
      "grad_norm": 0.2336615025997162,
      "learning_rate": 2.8654971464182773e-05,
      "loss": 0.6497,
      "num_input_tokens_seen": 12677872,
      "step": 19385
    },
    {
      "epoch": 10.162473794549266,
      "grad_norm": 0.15606600046157837,
      "learning_rate": 2.8643659656212222e-05,
      "loss": 0.4552,
      "num_input_tokens_seen": 12684112,
      "step": 19390
    },
    {
      "epoch": 10.165094339622641,
      "grad_norm": 0.18213054537773132,
      "learning_rate": 2.8632347086029642e-05,
      "loss": 0.3408,
      "num_input_tokens_seen": 12686736,
      "step": 19395
    },
    {
      "epoch": 10.167714884696016,
      "grad_norm": 0.08522078394889832,
      "learning_rate": 2.8621033756001486e-05,
      "loss": 0.6669,
      "num_input_tokens_seen": 12690352,
      "step": 19400
    },
    {
      "epoch": 10.170335429769391,
      "grad_norm": 0.11065545678138733,
      "learning_rate": 2.8609719668494394e-05,
      "loss": 0.4754,
      "num_input_tokens_seen": 12693264,
      "step": 19405
    },
    {
      "epoch": 10.172955974842766,
      "grad_norm": 0.13316510617733002,
      "learning_rate": 2.8598404825875113e-05,
      "loss": 0.4307,
      "num_input_tokens_seen": 12695888,
      "step": 19410
    },
    {
      "epoch": 10.175576519916143,
      "grad_norm": 0.19055113196372986,
      "learning_rate": 2.8587089230510604e-05,
      "loss": 0.5353,
      "num_input_tokens_seen": 12699216,
      "step": 19415
    },
    {
      "epoch": 10.178197064989519,
      "grad_norm": 0.1322232186794281,
      "learning_rate": 2.857577288476795e-05,
      "loss": 0.4201,
      "num_input_tokens_seen": 12703344,
      "step": 19420
    },
    {
      "epoch": 10.180817610062894,
      "grad_norm": 0.15125538408756256,
      "learning_rate": 2.85644557910144e-05,
      "loss": 0.4193,
      "num_input_tokens_seen": 12706480,
      "step": 19425
    },
    {
      "epoch": 10.183438155136269,
      "grad_norm": 0.15560047328472137,
      "learning_rate": 2.8553137951617366e-05,
      "loss": 0.4939,
      "num_input_tokens_seen": 12709744,
      "step": 19430
    },
    {
      "epoch": 10.186058700209644,
      "grad_norm": 0.12741880118846893,
      "learning_rate": 2.854181936894441e-05,
      "loss": 0.5495,
      "num_input_tokens_seen": 12712816,
      "step": 19435
    },
    {
      "epoch": 10.18867924528302,
      "grad_norm": 0.18923035264015198,
      "learning_rate": 2.8530500045363247e-05,
      "loss": 0.4896,
      "num_input_tokens_seen": 12716304,
      "step": 19440
    },
    {
      "epoch": 10.191299790356394,
      "grad_norm": 0.15036897361278534,
      "learning_rate": 2.851917998324175e-05,
      "loss": 0.4554,
      "num_input_tokens_seen": 12719216,
      "step": 19445
    },
    {
      "epoch": 10.19392033542977,
      "grad_norm": 0.22731639444828033,
      "learning_rate": 2.8507859184947957e-05,
      "loss": 0.5834,
      "num_input_tokens_seen": 12723312,
      "step": 19450
    },
    {
      "epoch": 10.196540880503145,
      "grad_norm": 0.09202077239751816,
      "learning_rate": 2.8496537652850043e-05,
      "loss": 0.4338,
      "num_input_tokens_seen": 12726320,
      "step": 19455
    },
    {
      "epoch": 10.19916142557652,
      "grad_norm": 0.1846662014722824,
      "learning_rate": 2.8485215389316333e-05,
      "loss": 0.5903,
      "num_input_tokens_seen": 12730128,
      "step": 19460
    },
    {
      "epoch": 10.201781970649895,
      "grad_norm": 0.10878392308950424,
      "learning_rate": 2.8473892396715328e-05,
      "loss": 0.6951,
      "num_input_tokens_seen": 12735024,
      "step": 19465
    },
    {
      "epoch": 10.20440251572327,
      "grad_norm": 0.1824294924736023,
      "learning_rate": 2.846256867741567e-05,
      "loss": 0.3504,
      "num_input_tokens_seen": 12737872,
      "step": 19470
    },
    {
      "epoch": 10.207023060796645,
      "grad_norm": 0.15362805128097534,
      "learning_rate": 2.845124423378614e-05,
      "loss": 0.4436,
      "num_input_tokens_seen": 12740400,
      "step": 19475
    },
    {
      "epoch": 10.20964360587002,
      "grad_norm": 0.13209238648414612,
      "learning_rate": 2.8439919068195697e-05,
      "loss": 0.3423,
      "num_input_tokens_seen": 12743664,
      "step": 19480
    },
    {
      "epoch": 10.212264150943396,
      "grad_norm": 0.15094639360904694,
      "learning_rate": 2.842859318301343e-05,
      "loss": 0.6059,
      "num_input_tokens_seen": 12746512,
      "step": 19485
    },
    {
      "epoch": 10.214884696016771,
      "grad_norm": 0.10086678713560104,
      "learning_rate": 2.841726658060858e-05,
      "loss": 0.4601,
      "num_input_tokens_seen": 12750160,
      "step": 19490
    },
    {
      "epoch": 10.217505241090146,
      "grad_norm": 0.09218183159828186,
      "learning_rate": 2.8405939263350555e-05,
      "loss": 0.4739,
      "num_input_tokens_seen": 12753520,
      "step": 19495
    },
    {
      "epoch": 10.220125786163521,
      "grad_norm": 0.0984727218747139,
      "learning_rate": 2.8394611233608886e-05,
      "loss": 0.4925,
      "num_input_tokens_seen": 12756624,
      "step": 19500
    },
    {
      "epoch": 10.222746331236896,
      "grad_norm": 0.17836397886276245,
      "learning_rate": 2.8383282493753283e-05,
      "loss": 0.4446,
      "num_input_tokens_seen": 12759248,
      "step": 19505
    },
    {
      "epoch": 10.225366876310272,
      "grad_norm": 0.18964248895645142,
      "learning_rate": 2.8371953046153575e-05,
      "loss": 0.3993,
      "num_input_tokens_seen": 12762384,
      "step": 19510
    },
    {
      "epoch": 10.227987421383649,
      "grad_norm": 0.18691997230052948,
      "learning_rate": 2.8360622893179757e-05,
      "loss": 0.525,
      "num_input_tokens_seen": 12766224,
      "step": 19515
    },
    {
      "epoch": 10.230607966457024,
      "grad_norm": 0.16888873279094696,
      "learning_rate": 2.834929203720197e-05,
      "loss": 0.4746,
      "num_input_tokens_seen": 12768912,
      "step": 19520
    },
    {
      "epoch": 10.233228511530399,
      "grad_norm": 0.13453620672225952,
      "learning_rate": 2.83379604805905e-05,
      "loss": 0.4362,
      "num_input_tokens_seen": 12772336,
      "step": 19525
    },
    {
      "epoch": 10.235849056603774,
      "grad_norm": 0.2503962516784668,
      "learning_rate": 2.832662822571578e-05,
      "loss": 0.4215,
      "num_input_tokens_seen": 12775216,
      "step": 19530
    },
    {
      "epoch": 10.23846960167715,
      "grad_norm": 0.1439778059720993,
      "learning_rate": 2.8315295274948395e-05,
      "loss": 0.4501,
      "num_input_tokens_seen": 12778512,
      "step": 19535
    },
    {
      "epoch": 10.241090146750524,
      "grad_norm": 0.1706630438566208,
      "learning_rate": 2.8303961630659047e-05,
      "loss": 0.3647,
      "num_input_tokens_seen": 12782576,
      "step": 19540
    },
    {
      "epoch": 10.2437106918239,
      "grad_norm": 0.1547662764787674,
      "learning_rate": 2.8292627295218627e-05,
      "loss": 0.4471,
      "num_input_tokens_seen": 12785584,
      "step": 19545
    },
    {
      "epoch": 10.246331236897275,
      "grad_norm": 0.1291886866092682,
      "learning_rate": 2.828129227099815e-05,
      "loss": 0.4649,
      "num_input_tokens_seen": 12788432,
      "step": 19550
    },
    {
      "epoch": 10.24895178197065,
      "grad_norm": 0.22107499837875366,
      "learning_rate": 2.826995656036875e-05,
      "loss": 0.7189,
      "num_input_tokens_seen": 12791312,
      "step": 19555
    },
    {
      "epoch": 10.251572327044025,
      "grad_norm": 0.13948306441307068,
      "learning_rate": 2.8258620165701747e-05,
      "loss": 0.4135,
      "num_input_tokens_seen": 12795120,
      "step": 19560
    },
    {
      "epoch": 10.2541928721174,
      "grad_norm": 0.16613081097602844,
      "learning_rate": 2.8247283089368586e-05,
      "loss": 0.4142,
      "num_input_tokens_seen": 12798032,
      "step": 19565
    },
    {
      "epoch": 10.256813417190775,
      "grad_norm": 0.20621073246002197,
      "learning_rate": 2.8235945333740842e-05,
      "loss": 0.4575,
      "num_input_tokens_seen": 12800688,
      "step": 19570
    },
    {
      "epoch": 10.25943396226415,
      "grad_norm": 0.10993679612874985,
      "learning_rate": 2.822460690119026e-05,
      "loss": 0.5477,
      "num_input_tokens_seen": 12804240,
      "step": 19575
    },
    {
      "epoch": 10.262054507337526,
      "grad_norm": 0.14705383777618408,
      "learning_rate": 2.82132677940887e-05,
      "loss": 0.4812,
      "num_input_tokens_seen": 12807216,
      "step": 19580
    },
    {
      "epoch": 10.264675052410901,
      "grad_norm": 0.14055654406547546,
      "learning_rate": 2.820192801480817e-05,
      "loss": 0.4279,
      "num_input_tokens_seen": 12810352,
      "step": 19585
    },
    {
      "epoch": 10.267295597484276,
      "grad_norm": 0.19405193626880646,
      "learning_rate": 2.8190587565720842e-05,
      "loss": 0.4923,
      "num_input_tokens_seen": 12813328,
      "step": 19590
    },
    {
      "epoch": 10.269916142557651,
      "grad_norm": 0.15681879222393036,
      "learning_rate": 2.8179246449198986e-05,
      "loss": 0.4775,
      "num_input_tokens_seen": 12816336,
      "step": 19595
    },
    {
      "epoch": 10.272536687631026,
      "grad_norm": 0.24582593142986298,
      "learning_rate": 2.8167904667615048e-05,
      "loss": 0.4119,
      "num_input_tokens_seen": 12820624,
      "step": 19600
    },
    {
      "epoch": 10.275157232704402,
      "grad_norm": 0.15083831548690796,
      "learning_rate": 2.8156562223341592e-05,
      "loss": 0.5527,
      "num_input_tokens_seen": 12824176,
      "step": 19605
    },
    {
      "epoch": 10.277777777777779,
      "grad_norm": 0.20041437447071075,
      "learning_rate": 2.814521911875133e-05,
      "loss": 0.5568,
      "num_input_tokens_seen": 12827376,
      "step": 19610
    },
    {
      "epoch": 10.280398322851154,
      "grad_norm": 0.20994549989700317,
      "learning_rate": 2.8133875356217114e-05,
      "loss": 0.3314,
      "num_input_tokens_seen": 12830704,
      "step": 19615
    },
    {
      "epoch": 10.283018867924529,
      "grad_norm": 0.21265143156051636,
      "learning_rate": 2.8122530938111923e-05,
      "loss": 0.5862,
      "num_input_tokens_seen": 12833616,
      "step": 19620
    },
    {
      "epoch": 10.285639412997904,
      "grad_norm": 0.11998819559812546,
      "learning_rate": 2.8111185866808888e-05,
      "loss": 0.5002,
      "num_input_tokens_seen": 12836688,
      "step": 19625
    },
    {
      "epoch": 10.28825995807128,
      "grad_norm": 0.1533779352903366,
      "learning_rate": 2.809984014468126e-05,
      "loss": 0.5221,
      "num_input_tokens_seen": 12840656,
      "step": 19630
    },
    {
      "epoch": 10.290880503144654,
      "grad_norm": 0.1270076334476471,
      "learning_rate": 2.8088493774102443e-05,
      "loss": 0.4805,
      "num_input_tokens_seen": 12843408,
      "step": 19635
    },
    {
      "epoch": 10.29350104821803,
      "grad_norm": 0.1441001296043396,
      "learning_rate": 2.807714675744596e-05,
      "loss": 0.5759,
      "num_input_tokens_seen": 12846224,
      "step": 19640
    },
    {
      "epoch": 10.296121593291405,
      "grad_norm": 0.24471817910671234,
      "learning_rate": 2.8065799097085475e-05,
      "loss": 0.5862,
      "num_input_tokens_seen": 12850288,
      "step": 19645
    },
    {
      "epoch": 10.29874213836478,
      "grad_norm": 0.2000216245651245,
      "learning_rate": 2.8054450795394804e-05,
      "loss": 0.5055,
      "num_input_tokens_seen": 12853424,
      "step": 19650
    },
    {
      "epoch": 10.301362683438155,
      "grad_norm": 0.33258989453315735,
      "learning_rate": 2.8043101854747878e-05,
      "loss": 0.5458,
      "num_input_tokens_seen": 12856656,
      "step": 19655
    },
    {
      "epoch": 10.30398322851153,
      "grad_norm": 0.1553894430398941,
      "learning_rate": 2.8031752277518748e-05,
      "loss": 0.5595,
      "num_input_tokens_seen": 12859248,
      "step": 19660
    },
    {
      "epoch": 10.306603773584905,
      "grad_norm": 0.13782547414302826,
      "learning_rate": 2.802040206608164e-05,
      "loss": 0.3913,
      "num_input_tokens_seen": 12862160,
      "step": 19665
    },
    {
      "epoch": 10.30922431865828,
      "grad_norm": 0.3571415841579437,
      "learning_rate": 2.8009051222810877e-05,
      "loss": 0.5668,
      "num_input_tokens_seen": 12864304,
      "step": 19670
    },
    {
      "epoch": 10.311844863731656,
      "grad_norm": 0.15897728502750397,
      "learning_rate": 2.7997699750080918e-05,
      "loss": 0.5403,
      "num_input_tokens_seen": 12868752,
      "step": 19675
    },
    {
      "epoch": 10.314465408805031,
      "grad_norm": 0.15519912540912628,
      "learning_rate": 2.7986347650266386e-05,
      "loss": 0.4597,
      "num_input_tokens_seen": 12872112,
      "step": 19680
    },
    {
      "epoch": 10.317085953878406,
      "grad_norm": 0.26737332344055176,
      "learning_rate": 2.797499492574199e-05,
      "loss": 0.4854,
      "num_input_tokens_seen": 12874576,
      "step": 19685
    },
    {
      "epoch": 10.319706498951781,
      "grad_norm": 0.17957699298858643,
      "learning_rate": 2.7963641578882594e-05,
      "loss": 0.4941,
      "num_input_tokens_seen": 12877232,
      "step": 19690
    },
    {
      "epoch": 10.322327044025156,
      "grad_norm": 0.29669028520584106,
      "learning_rate": 2.7952287612063193e-05,
      "loss": 0.4433,
      "num_input_tokens_seen": 12880656,
      "step": 19695
    },
    {
      "epoch": 10.324947589098532,
      "grad_norm": 0.19460353255271912,
      "learning_rate": 2.79409330276589e-05,
      "loss": 0.4216,
      "num_input_tokens_seen": 12883472,
      "step": 19700
    },
    {
      "epoch": 10.327568134171909,
      "grad_norm": 0.22645461559295654,
      "learning_rate": 2.7929577828044974e-05,
      "loss": 0.4539,
      "num_input_tokens_seen": 12885776,
      "step": 19705
    },
    {
      "epoch": 10.330188679245284,
      "grad_norm": 0.2482338696718216,
      "learning_rate": 2.7918222015596784e-05,
      "loss": 0.5241,
      "num_input_tokens_seen": 12888688,
      "step": 19710
    },
    {
      "epoch": 10.332809224318659,
      "grad_norm": 0.16269174218177795,
      "learning_rate": 2.7906865592689847e-05,
      "loss": 0.5241,
      "num_input_tokens_seen": 12891664,
      "step": 19715
    },
    {
      "epoch": 10.335429769392034,
      "grad_norm": 0.28545090556144714,
      "learning_rate": 2.789550856169979e-05,
      "loss": 0.4949,
      "num_input_tokens_seen": 12894832,
      "step": 19720
    },
    {
      "epoch": 10.33805031446541,
      "grad_norm": 0.11826161295175552,
      "learning_rate": 2.7884150925002374e-05,
      "loss": 0.3531,
      "num_input_tokens_seen": 12897744,
      "step": 19725
    },
    {
      "epoch": 10.340670859538784,
      "grad_norm": 0.19439180195331573,
      "learning_rate": 2.787279268497349e-05,
      "loss": 0.4411,
      "num_input_tokens_seen": 12901008,
      "step": 19730
    },
    {
      "epoch": 10.34329140461216,
      "grad_norm": 0.1738458275794983,
      "learning_rate": 2.786143384398915e-05,
      "loss": 0.4513,
      "num_input_tokens_seen": 12903792,
      "step": 19735
    },
    {
      "epoch": 10.345911949685535,
      "grad_norm": 0.18355853855609894,
      "learning_rate": 2.785007440442549e-05,
      "loss": 0.4188,
      "num_input_tokens_seen": 12907120,
      "step": 19740
    },
    {
      "epoch": 10.34853249475891,
      "grad_norm": 0.21943214535713196,
      "learning_rate": 2.7838714368658785e-05,
      "loss": 0.5717,
      "num_input_tokens_seen": 12910320,
      "step": 19745
    },
    {
      "epoch": 10.351153039832285,
      "grad_norm": 0.20481079816818237,
      "learning_rate": 2.7827353739065427e-05,
      "loss": 0.5326,
      "num_input_tokens_seen": 12913008,
      "step": 19750
    },
    {
      "epoch": 10.35377358490566,
      "grad_norm": 0.15442804992198944,
      "learning_rate": 2.781599251802191e-05,
      "loss": 0.576,
      "num_input_tokens_seen": 12916432,
      "step": 19755
    },
    {
      "epoch": 10.356394129979035,
      "grad_norm": 0.1624644547700882,
      "learning_rate": 2.780463070790489e-05,
      "loss": 0.5011,
      "num_input_tokens_seen": 12919536,
      "step": 19760
    },
    {
      "epoch": 10.35901467505241,
      "grad_norm": 0.13420209288597107,
      "learning_rate": 2.7793268311091125e-05,
      "loss": 0.4386,
      "num_input_tokens_seen": 12922768,
      "step": 19765
    },
    {
      "epoch": 10.361635220125786,
      "grad_norm": 0.14668361842632294,
      "learning_rate": 2.7781905329957496e-05,
      "loss": 0.4672,
      "num_input_tokens_seen": 12925840,
      "step": 19770
    },
    {
      "epoch": 10.364255765199161,
      "grad_norm": 0.21486034989356995,
      "learning_rate": 2.7770541766881008e-05,
      "loss": 0.406,
      "num_input_tokens_seen": 12928656,
      "step": 19775
    },
    {
      "epoch": 10.366876310272536,
      "grad_norm": 0.14177089929580688,
      "learning_rate": 2.7759177624238786e-05,
      "loss": 0.4651,
      "num_input_tokens_seen": 12931504,
      "step": 19780
    },
    {
      "epoch": 10.369496855345911,
      "grad_norm": 0.13913238048553467,
      "learning_rate": 2.7747812904408094e-05,
      "loss": 0.472,
      "num_input_tokens_seen": 12935056,
      "step": 19785
    },
    {
      "epoch": 10.372117400419286,
      "grad_norm": 0.3603043854236603,
      "learning_rate": 2.7736447609766286e-05,
      "loss": 0.4379,
      "num_input_tokens_seen": 12938064,
      "step": 19790
    },
    {
      "epoch": 10.374737945492662,
      "grad_norm": 0.15918636322021484,
      "learning_rate": 2.772508174269085e-05,
      "loss": 0.4978,
      "num_input_tokens_seen": 12941456,
      "step": 19795
    },
    {
      "epoch": 10.377358490566039,
      "grad_norm": 0.15396685898303986,
      "learning_rate": 2.7713715305559414e-05,
      "loss": 0.4281,
      "num_input_tokens_seen": 12945232,
      "step": 19800
    },
    {
      "epoch": 10.379979035639414,
      "grad_norm": 0.1659289002418518,
      "learning_rate": 2.7702348300749692e-05,
      "loss": 0.379,
      "num_input_tokens_seen": 12947760,
      "step": 19805
    },
    {
      "epoch": 10.382599580712789,
      "grad_norm": 0.14559932053089142,
      "learning_rate": 2.7690980730639533e-05,
      "loss": 0.4457,
      "num_input_tokens_seen": 12951216,
      "step": 19810
    },
    {
      "epoch": 10.385220125786164,
      "grad_norm": 0.17667697370052338,
      "learning_rate": 2.7679612597606902e-05,
      "loss": 0.3929,
      "num_input_tokens_seen": 12954544,
      "step": 19815
    },
    {
      "epoch": 10.38784067085954,
      "grad_norm": 0.30228832364082336,
      "learning_rate": 2.7668243904029894e-05,
      "loss": 0.621,
      "num_input_tokens_seen": 12959568,
      "step": 19820
    },
    {
      "epoch": 10.390461215932914,
      "grad_norm": 0.13482782244682312,
      "learning_rate": 2.76568746522867e-05,
      "loss": 0.4891,
      "num_input_tokens_seen": 12962992,
      "step": 19825
    },
    {
      "epoch": 10.39308176100629,
      "grad_norm": 0.2587512731552124,
      "learning_rate": 2.7645504844755637e-05,
      "loss": 0.463,
      "num_input_tokens_seen": 12966608,
      "step": 19830
    },
    {
      "epoch": 10.395702306079665,
      "grad_norm": 0.10975059121847153,
      "learning_rate": 2.7634134483815143e-05,
      "loss": 0.4758,
      "num_input_tokens_seen": 12969744,
      "step": 19835
    },
    {
      "epoch": 10.39832285115304,
      "grad_norm": 0.26723048090934753,
      "learning_rate": 2.762276357184377e-05,
      "loss": 0.5816,
      "num_input_tokens_seen": 12972784,
      "step": 19840
    },
    {
      "epoch": 10.400943396226415,
      "grad_norm": 0.14812612533569336,
      "learning_rate": 2.7611392111220176e-05,
      "loss": 0.4813,
      "num_input_tokens_seen": 12976944,
      "step": 19845
    },
    {
      "epoch": 10.40356394129979,
      "grad_norm": 0.15417072176933289,
      "learning_rate": 2.7600020104323154e-05,
      "loss": 0.3646,
      "num_input_tokens_seen": 12979984,
      "step": 19850
    },
    {
      "epoch": 10.406184486373165,
      "grad_norm": 0.329166978597641,
      "learning_rate": 2.7588647553531576e-05,
      "loss": 0.5247,
      "num_input_tokens_seen": 12982704,
      "step": 19855
    },
    {
      "epoch": 10.40880503144654,
      "grad_norm": 0.1716093271970749,
      "learning_rate": 2.7577274461224463e-05,
      "loss": 0.4841,
      "num_input_tokens_seen": 12985680,
      "step": 19860
    },
    {
      "epoch": 10.411425576519916,
      "grad_norm": 0.1961328238248825,
      "learning_rate": 2.756590082978095e-05,
      "loss": 0.4423,
      "num_input_tokens_seen": 12989200,
      "step": 19865
    },
    {
      "epoch": 10.414046121593291,
      "grad_norm": 0.10592420399188995,
      "learning_rate": 2.7554526661580243e-05,
      "loss": 0.4382,
      "num_input_tokens_seen": 12992560,
      "step": 19870
    },
    {
      "epoch": 10.416666666666666,
      "grad_norm": 0.24271756410598755,
      "learning_rate": 2.7543151959001708e-05,
      "loss": 0.4183,
      "num_input_tokens_seen": 12995472,
      "step": 19875
    },
    {
      "epoch": 10.419287211740041,
      "grad_norm": 0.18628337979316711,
      "learning_rate": 2.7531776724424797e-05,
      "loss": 0.5133,
      "num_input_tokens_seen": 12999024,
      "step": 19880
    },
    {
      "epoch": 10.421907756813416,
      "grad_norm": 0.12593437731266022,
      "learning_rate": 2.752040096022907e-05,
      "loss": 0.4313,
      "num_input_tokens_seen": 13001744,
      "step": 19885
    },
    {
      "epoch": 10.424528301886792,
      "grad_norm": 0.2053132802248001,
      "learning_rate": 2.750902466879423e-05,
      "loss": 0.5417,
      "num_input_tokens_seen": 13004272,
      "step": 19890
    },
    {
      "epoch": 10.427148846960169,
      "grad_norm": 0.3195745646953583,
      "learning_rate": 2.7497647852500048e-05,
      "loss": 0.5273,
      "num_input_tokens_seen": 13007472,
      "step": 19895
    },
    {
      "epoch": 10.429769392033544,
      "grad_norm": 0.1969156265258789,
      "learning_rate": 2.7486270513726424e-05,
      "loss": 0.4529,
      "num_input_tokens_seen": 13011120,
      "step": 19900
    },
    {
      "epoch": 10.432389937106919,
      "grad_norm": 0.19509455561637878,
      "learning_rate": 2.7474892654853384e-05,
      "loss": 0.4805,
      "num_input_tokens_seen": 13013872,
      "step": 19905
    },
    {
      "epoch": 10.435010482180294,
      "grad_norm": 0.13432827591896057,
      "learning_rate": 2.7463514278261034e-05,
      "loss": 0.4995,
      "num_input_tokens_seen": 13016592,
      "step": 19910
    },
    {
      "epoch": 10.43763102725367,
      "grad_norm": 0.23408302664756775,
      "learning_rate": 2.7452135386329596e-05,
      "loss": 0.3792,
      "num_input_tokens_seen": 13019888,
      "step": 19915
    },
    {
      "epoch": 10.440251572327044,
      "grad_norm": 0.1405448466539383,
      "learning_rate": 2.7440755981439414e-05,
      "loss": 0.6093,
      "num_input_tokens_seen": 13023664,
      "step": 19920
    },
    {
      "epoch": 10.44287211740042,
      "grad_norm": 0.14633838832378387,
      "learning_rate": 2.7429376065970924e-05,
      "loss": 0.4288,
      "num_input_tokens_seen": 13027152,
      "step": 19925
    },
    {
      "epoch": 10.445492662473795,
      "grad_norm": 0.13302844762802124,
      "learning_rate": 2.7417995642304678e-05,
      "loss": 0.4969,
      "num_input_tokens_seen": 13030160,
      "step": 19930
    },
    {
      "epoch": 10.44811320754717,
      "grad_norm": 0.14689864218235016,
      "learning_rate": 2.7406614712821333e-05,
      "loss": 0.4409,
      "num_input_tokens_seen": 13034000,
      "step": 19935
    },
    {
      "epoch": 10.450733752620545,
      "grad_norm": 0.15243510901927948,
      "learning_rate": 2.739523327990164e-05,
      "loss": 0.4254,
      "num_input_tokens_seen": 13037392,
      "step": 19940
    },
    {
      "epoch": 10.45335429769392,
      "grad_norm": 0.2038809061050415,
      "learning_rate": 2.7383851345926477e-05,
      "loss": 0.4308,
      "num_input_tokens_seen": 13040496,
      "step": 19945
    },
    {
      "epoch": 10.455974842767295,
      "grad_norm": 0.662039041519165,
      "learning_rate": 2.7372468913276804e-05,
      "loss": 0.4758,
      "num_input_tokens_seen": 13043248,
      "step": 19950
    },
    {
      "epoch": 10.45859538784067,
      "grad_norm": 0.2507651746273041,
      "learning_rate": 2.7361085984333705e-05,
      "loss": 0.4603,
      "num_input_tokens_seen": 13046512,
      "step": 19955
    },
    {
      "epoch": 10.461215932914046,
      "grad_norm": 0.170226588845253,
      "learning_rate": 2.7349702561478356e-05,
      "loss": 0.4698,
      "num_input_tokens_seen": 13049616,
      "step": 19960
    },
    {
      "epoch": 10.463836477987421,
      "grad_norm": 0.17163187265396118,
      "learning_rate": 2.733831864709203e-05,
      "loss": 0.4047,
      "num_input_tokens_seen": 13052048,
      "step": 19965
    },
    {
      "epoch": 10.466457023060796,
      "grad_norm": 0.2617413103580475,
      "learning_rate": 2.732693424355613e-05,
      "loss": 0.5656,
      "num_input_tokens_seen": 13054096,
      "step": 19970
    },
    {
      "epoch": 10.469077568134171,
      "grad_norm": 0.11913618445396423,
      "learning_rate": 2.731554935325213e-05,
      "loss": 0.3772,
      "num_input_tokens_seen": 13057488,
      "step": 19975
    },
    {
      "epoch": 10.471698113207546,
      "grad_norm": 0.1523989737033844,
      "learning_rate": 2.7304163978561613e-05,
      "loss": 0.4153,
      "num_input_tokens_seen": 13060336,
      "step": 19980
    },
    {
      "epoch": 10.474318658280922,
      "grad_norm": 0.13809600472450256,
      "learning_rate": 2.729277812186629e-05,
      "loss": 0.468,
      "num_input_tokens_seen": 13063728,
      "step": 19985
    },
    {
      "epoch": 10.476939203354299,
      "grad_norm": 0.1754143387079239,
      "learning_rate": 2.7281391785547945e-05,
      "loss": 0.4973,
      "num_input_tokens_seen": 13066256,
      "step": 19990
    },
    {
      "epoch": 10.479559748427674,
      "grad_norm": 0.2557976245880127,
      "learning_rate": 2.7270004971988456e-05,
      "loss": 0.5513,
      "num_input_tokens_seen": 13069424,
      "step": 19995
    },
    {
      "epoch": 10.482180293501049,
      "grad_norm": 0.14643293619155884,
      "learning_rate": 2.7258617683569825e-05,
      "loss": 0.4374,
      "num_input_tokens_seen": 13072560,
      "step": 20000
    },
    {
      "epoch": 10.484800838574424,
      "grad_norm": 0.1461789906024933,
      "learning_rate": 2.7247229922674143e-05,
      "loss": 0.6033,
      "num_input_tokens_seen": 13076176,
      "step": 20005
    },
    {
      "epoch": 10.4874213836478,
      "grad_norm": 0.1430204063653946,
      "learning_rate": 2.7235841691683605e-05,
      "loss": 0.4683,
      "num_input_tokens_seen": 13079984,
      "step": 20010
    },
    {
      "epoch": 10.490041928721174,
      "grad_norm": 0.1494777351617813,
      "learning_rate": 2.7224452992980487e-05,
      "loss": 0.4274,
      "num_input_tokens_seen": 13082928,
      "step": 20015
    },
    {
      "epoch": 10.49266247379455,
      "grad_norm": 0.12793654203414917,
      "learning_rate": 2.7213063828947182e-05,
      "loss": 0.5705,
      "num_input_tokens_seen": 13085776,
      "step": 20020
    },
    {
      "epoch": 10.495283018867925,
      "grad_norm": 0.12117847800254822,
      "learning_rate": 2.7201674201966176e-05,
      "loss": 0.4086,
      "num_input_tokens_seen": 13088656,
      "step": 20025
    },
    {
      "epoch": 10.4979035639413,
      "grad_norm": 0.1304643154144287,
      "learning_rate": 2.7190284114420035e-05,
      "loss": 0.5014,
      "num_input_tokens_seen": 13091728,
      "step": 20030
    },
    {
      "epoch": 10.500524109014675,
      "grad_norm": 0.12860649824142456,
      "learning_rate": 2.717889356869146e-05,
      "loss": 0.5977,
      "num_input_tokens_seen": 13096048,
      "step": 20035
    },
    {
      "epoch": 10.50314465408805,
      "grad_norm": 0.17968139052391052,
      "learning_rate": 2.7167502567163194e-05,
      "loss": 0.5894,
      "num_input_tokens_seen": 13098608,
      "step": 20040
    },
    {
      "epoch": 10.505765199161425,
      "grad_norm": 0.17091137170791626,
      "learning_rate": 2.715611111221813e-05,
      "loss": 0.497,
      "num_input_tokens_seen": 13101872,
      "step": 20045
    },
    {
      "epoch": 10.5083857442348,
      "grad_norm": 0.17848967015743256,
      "learning_rate": 2.7144719206239217e-05,
      "loss": 0.4888,
      "num_input_tokens_seen": 13109072,
      "step": 20050
    },
    {
      "epoch": 10.511006289308176,
      "grad_norm": 0.29173263907432556,
      "learning_rate": 2.7133326851609508e-05,
      "loss": 0.4763,
      "num_input_tokens_seen": 13112080,
      "step": 20055
    },
    {
      "epoch": 10.51362683438155,
      "grad_norm": 0.13407306373119354,
      "learning_rate": 2.7121934050712166e-05,
      "loss": 0.3839,
      "num_input_tokens_seen": 13114704,
      "step": 20060
    },
    {
      "epoch": 10.516247379454926,
      "grad_norm": 0.16720566153526306,
      "learning_rate": 2.711054080593043e-05,
      "loss": 0.5988,
      "num_input_tokens_seen": 13117360,
      "step": 20065
    },
    {
      "epoch": 10.518867924528301,
      "grad_norm": 0.13324300944805145,
      "learning_rate": 2.7099147119647627e-05,
      "loss": 0.4392,
      "num_input_tokens_seen": 13121104,
      "step": 20070
    },
    {
      "epoch": 10.521488469601676,
      "grad_norm": 0.1548394411802292,
      "learning_rate": 2.7087752994247206e-05,
      "loss": 0.4589,
      "num_input_tokens_seen": 13124464,
      "step": 20075
    },
    {
      "epoch": 10.524109014675052,
      "grad_norm": 0.20153720676898956,
      "learning_rate": 2.707635843211267e-05,
      "loss": 0.3882,
      "num_input_tokens_seen": 13127024,
      "step": 20080
    },
    {
      "epoch": 10.526729559748428,
      "grad_norm": 0.174974724650383,
      "learning_rate": 2.7064963435627637e-05,
      "loss": 0.4756,
      "num_input_tokens_seen": 13129936,
      "step": 20085
    },
    {
      "epoch": 10.529350104821804,
      "grad_norm": 0.14790694415569305,
      "learning_rate": 2.7053568007175823e-05,
      "loss": 0.4799,
      "num_input_tokens_seen": 13132656,
      "step": 20090
    },
    {
      "epoch": 10.531970649895179,
      "grad_norm": 0.16085699200630188,
      "learning_rate": 2.7042172149140994e-05,
      "loss": 0.5134,
      "num_input_tokens_seen": 13136528,
      "step": 20095
    },
    {
      "epoch": 10.534591194968554,
      "grad_norm": 0.15915867686271667,
      "learning_rate": 2.703077586390706e-05,
      "loss": 0.4219,
      "num_input_tokens_seen": 13139568,
      "step": 20100
    },
    {
      "epoch": 10.53721174004193,
      "grad_norm": 0.14146330952644348,
      "learning_rate": 2.7019379153857982e-05,
      "loss": 0.4201,
      "num_input_tokens_seen": 13142704,
      "step": 20105
    },
    {
      "epoch": 10.539832285115304,
      "grad_norm": 0.1504357010126114,
      "learning_rate": 2.700798202137782e-05,
      "loss": 0.5328,
      "num_input_tokens_seen": 13145936,
      "step": 20110
    },
    {
      "epoch": 10.54245283018868,
      "grad_norm": 0.1462569683790207,
      "learning_rate": 2.6996584468850732e-05,
      "loss": 0.5756,
      "num_input_tokens_seen": 13149360,
      "step": 20115
    },
    {
      "epoch": 10.545073375262055,
      "grad_norm": 0.21340790390968323,
      "learning_rate": 2.6985186498660946e-05,
      "loss": 0.5692,
      "num_input_tokens_seen": 13151920,
      "step": 20120
    },
    {
      "epoch": 10.54769392033543,
      "grad_norm": 0.15492786467075348,
      "learning_rate": 2.69737881131928e-05,
      "loss": 0.5335,
      "num_input_tokens_seen": 13155024,
      "step": 20125
    },
    {
      "epoch": 10.550314465408805,
      "grad_norm": 0.19259151816368103,
      "learning_rate": 2.6962389314830692e-05,
      "loss": 0.4995,
      "num_input_tokens_seen": 13158320,
      "step": 20130
    },
    {
      "epoch": 10.55293501048218,
      "grad_norm": 0.16882336139678955,
      "learning_rate": 2.6950990105959135e-05,
      "loss": 0.4682,
      "num_input_tokens_seen": 13161040,
      "step": 20135
    },
    {
      "epoch": 10.555555555555555,
      "grad_norm": 0.29852160811424255,
      "learning_rate": 2.6939590488962703e-05,
      "loss": 0.4065,
      "num_input_tokens_seen": 13163984,
      "step": 20140
    },
    {
      "epoch": 10.55817610062893,
      "grad_norm": 0.19128836691379547,
      "learning_rate": 2.6928190466226067e-05,
      "loss": 0.5888,
      "num_input_tokens_seen": 13167888,
      "step": 20145
    },
    {
      "epoch": 10.560796645702306,
      "grad_norm": 0.10712496191263199,
      "learning_rate": 2.691679004013398e-05,
      "loss": 0.4317,
      "num_input_tokens_seen": 13171312,
      "step": 20150
    },
    {
      "epoch": 10.56341719077568,
      "grad_norm": 0.20573151111602783,
      "learning_rate": 2.6905389213071285e-05,
      "loss": 0.514,
      "num_input_tokens_seen": 13174192,
      "step": 20155
    },
    {
      "epoch": 10.566037735849056,
      "grad_norm": 0.12080226838588715,
      "learning_rate": 2.6893987987422907e-05,
      "loss": 0.5738,
      "num_input_tokens_seen": 13178736,
      "step": 20160
    },
    {
      "epoch": 10.568658280922431,
      "grad_norm": 0.11883092671632767,
      "learning_rate": 2.6882586365573854e-05,
      "loss": 0.4381,
      "num_input_tokens_seen": 13181392,
      "step": 20165
    },
    {
      "epoch": 10.571278825995806,
      "grad_norm": 0.12699566781520844,
      "learning_rate": 2.6871184349909206e-05,
      "loss": 0.4773,
      "num_input_tokens_seen": 13184496,
      "step": 20170
    },
    {
      "epoch": 10.573899371069182,
      "grad_norm": 0.18763354420661926,
      "learning_rate": 2.685978194281414e-05,
      "loss": 0.5803,
      "num_input_tokens_seen": 13187760,
      "step": 20175
    },
    {
      "epoch": 10.576519916142558,
      "grad_norm": 0.2600409984588623,
      "learning_rate": 2.6848379146673914e-05,
      "loss": 0.401,
      "num_input_tokens_seen": 13190640,
      "step": 20180
    },
    {
      "epoch": 10.579140461215934,
      "grad_norm": 0.21701908111572266,
      "learning_rate": 2.6836975963873843e-05,
      "loss": 0.4883,
      "num_input_tokens_seen": 13193840,
      "step": 20185
    },
    {
      "epoch": 10.581761006289309,
      "grad_norm": 0.1562628149986267,
      "learning_rate": 2.6825572396799377e-05,
      "loss": 0.5507,
      "num_input_tokens_seen": 13197648,
      "step": 20190
    },
    {
      "epoch": 10.584381551362684,
      "grad_norm": 0.16639086604118347,
      "learning_rate": 2.6814168447835986e-05,
      "loss": 0.2978,
      "num_input_tokens_seen": 13201808,
      "step": 20195
    },
    {
      "epoch": 10.58700209643606,
      "grad_norm": 0.19121335446834564,
      "learning_rate": 2.6802764119369245e-05,
      "loss": 0.5877,
      "num_input_tokens_seen": 13204976,
      "step": 20200
    },
    {
      "epoch": 10.589622641509434,
      "grad_norm": 0.20536208152770996,
      "learning_rate": 2.6791359413784828e-05,
      "loss": 0.5854,
      "num_input_tokens_seen": 13207856,
      "step": 20205
    },
    {
      "epoch": 10.59224318658281,
      "grad_norm": 0.21317221224308014,
      "learning_rate": 2.677995433346845e-05,
      "loss": 0.373,
      "num_input_tokens_seen": 13210768,
      "step": 20210
    },
    {
      "epoch": 10.594863731656185,
      "grad_norm": 0.202205091714859,
      "learning_rate": 2.6768548880805934e-05,
      "loss": 0.5436,
      "num_input_tokens_seen": 13213776,
      "step": 20215
    },
    {
      "epoch": 10.59748427672956,
      "grad_norm": 0.12497900426387787,
      "learning_rate": 2.6757143058183177e-05,
      "loss": 0.6163,
      "num_input_tokens_seen": 13217104,
      "step": 20220
    },
    {
      "epoch": 10.600104821802935,
      "grad_norm": 0.10826189815998077,
      "learning_rate": 2.6745736867986132e-05,
      "loss": 0.427,
      "num_input_tokens_seen": 13220048,
      "step": 20225
    },
    {
      "epoch": 10.60272536687631,
      "grad_norm": 0.18681395053863525,
      "learning_rate": 2.673433031260085e-05,
      "loss": 0.4628,
      "num_input_tokens_seen": 13223376,
      "step": 20230
    },
    {
      "epoch": 10.605345911949685,
      "grad_norm": 0.16077353060245514,
      "learning_rate": 2.6722923394413456e-05,
      "loss": 0.5653,
      "num_input_tokens_seen": 13226512,
      "step": 20235
    },
    {
      "epoch": 10.60796645702306,
      "grad_norm": 0.09887852519750595,
      "learning_rate": 2.6711516115810136e-05,
      "loss": 0.3923,
      "num_input_tokens_seen": 13229744,
      "step": 20240
    },
    {
      "epoch": 10.610587002096436,
      "grad_norm": 0.13596001267433167,
      "learning_rate": 2.6700108479177188e-05,
      "loss": 0.4655,
      "num_input_tokens_seen": 13233552,
      "step": 20245
    },
    {
      "epoch": 10.61320754716981,
      "grad_norm": 0.20795822143554688,
      "learning_rate": 2.6688700486900926e-05,
      "loss": 0.4282,
      "num_input_tokens_seen": 13237008,
      "step": 20250
    },
    {
      "epoch": 10.615828092243186,
      "grad_norm": 0.19032399356365204,
      "learning_rate": 2.6677292141367787e-05,
      "loss": 0.3399,
      "num_input_tokens_seen": 13240080,
      "step": 20255
    },
    {
      "epoch": 10.618448637316561,
      "grad_norm": 0.20794418454170227,
      "learning_rate": 2.6665883444964278e-05,
      "loss": 0.4978,
      "num_input_tokens_seen": 13243408,
      "step": 20260
    },
    {
      "epoch": 10.621069182389936,
      "grad_norm": 0.14287076890468597,
      "learning_rate": 2.665447440007695e-05,
      "loss": 0.5694,
      "num_input_tokens_seen": 13247152,
      "step": 20265
    },
    {
      "epoch": 10.623689727463312,
      "grad_norm": 0.10387358069419861,
      "learning_rate": 2.664306500909246e-05,
      "loss": 0.3365,
      "num_input_tokens_seen": 13251376,
      "step": 20270
    },
    {
      "epoch": 10.626310272536688,
      "grad_norm": 0.14554502069950104,
      "learning_rate": 2.663165527439751e-05,
      "loss": 0.4229,
      "num_input_tokens_seen": 13254992,
      "step": 20275
    },
    {
      "epoch": 10.628930817610064,
      "grad_norm": 0.1368219554424286,
      "learning_rate": 2.662024519837889e-05,
      "loss": 0.4651,
      "num_input_tokens_seen": 13257712,
      "step": 20280
    },
    {
      "epoch": 10.631551362683439,
      "grad_norm": 0.1395539939403534,
      "learning_rate": 2.6608834783423465e-05,
      "loss": 0.3973,
      "num_input_tokens_seen": 13262192,
      "step": 20285
    },
    {
      "epoch": 10.634171907756814,
      "grad_norm": 0.1371302306652069,
      "learning_rate": 2.6597424031918156e-05,
      "loss": 0.5232,
      "num_input_tokens_seen": 13266832,
      "step": 20290
    },
    {
      "epoch": 10.63679245283019,
      "grad_norm": 0.2585635483264923,
      "learning_rate": 2.6586012946249954e-05,
      "loss": 0.5558,
      "num_input_tokens_seen": 13269424,
      "step": 20295
    },
    {
      "epoch": 10.639412997903564,
      "grad_norm": 0.17920686304569244,
      "learning_rate": 2.6574601528805953e-05,
      "loss": 0.6431,
      "num_input_tokens_seen": 13272144,
      "step": 20300
    },
    {
      "epoch": 10.64203354297694,
      "grad_norm": 0.13252219557762146,
      "learning_rate": 2.6563189781973268e-05,
      "loss": 0.4322,
      "num_input_tokens_seen": 13275376,
      "step": 20305
    },
    {
      "epoch": 10.644654088050315,
      "grad_norm": 0.1289781779050827,
      "learning_rate": 2.655177770813912e-05,
      "loss": 0.4301,
      "num_input_tokens_seen": 13278320,
      "step": 20310
    },
    {
      "epoch": 10.64727463312369,
      "grad_norm": 0.24712684750556946,
      "learning_rate": 2.654036530969077e-05,
      "loss": 0.5228,
      "num_input_tokens_seen": 13280816,
      "step": 20315
    },
    {
      "epoch": 10.649895178197065,
      "grad_norm": 0.192186638712883,
      "learning_rate": 2.6528952589015576e-05,
      "loss": 0.6868,
      "num_input_tokens_seen": 13284784,
      "step": 20320
    },
    {
      "epoch": 10.65251572327044,
      "grad_norm": 0.13974469900131226,
      "learning_rate": 2.6517539548500947e-05,
      "loss": 0.6431,
      "num_input_tokens_seen": 13287376,
      "step": 20325
    },
    {
      "epoch": 10.655136268343815,
      "grad_norm": 0.14908920228481293,
      "learning_rate": 2.6506126190534354e-05,
      "loss": 0.3185,
      "num_input_tokens_seen": 13290928,
      "step": 20330
    },
    {
      "epoch": 10.65775681341719,
      "grad_norm": 0.2758485972881317,
      "learning_rate": 2.649471251750334e-05,
      "loss": 0.5812,
      "num_input_tokens_seen": 13294192,
      "step": 20335
    },
    {
      "epoch": 10.660377358490566,
      "grad_norm": 0.22740522027015686,
      "learning_rate": 2.6483298531795527e-05,
      "loss": 0.5008,
      "num_input_tokens_seen": 13297456,
      "step": 20340
    },
    {
      "epoch": 10.66299790356394,
      "grad_norm": 0.1389443278312683,
      "learning_rate": 2.6471884235798583e-05,
      "loss": 0.6638,
      "num_input_tokens_seen": 13301008,
      "step": 20345
    },
    {
      "epoch": 10.665618448637316,
      "grad_norm": 0.278085857629776,
      "learning_rate": 2.6460469631900247e-05,
      "loss": 0.565,
      "num_input_tokens_seen": 13304368,
      "step": 20350
    },
    {
      "epoch": 10.668238993710691,
      "grad_norm": 0.19158746302127838,
      "learning_rate": 2.644905472248832e-05,
      "loss": 0.611,
      "num_input_tokens_seen": 13306864,
      "step": 20355
    },
    {
      "epoch": 10.670859538784066,
      "grad_norm": 0.25173577666282654,
      "learning_rate": 2.6437639509950686e-05,
      "loss": 0.4703,
      "num_input_tokens_seen": 13310352,
      "step": 20360
    },
    {
      "epoch": 10.673480083857442,
      "grad_norm": 0.24885645508766174,
      "learning_rate": 2.6426223996675264e-05,
      "loss": 0.533,
      "num_input_tokens_seen": 13313488,
      "step": 20365
    },
    {
      "epoch": 10.676100628930818,
      "grad_norm": 0.23764264583587646,
      "learning_rate": 2.641480818505005e-05,
      "loss": 0.4586,
      "num_input_tokens_seen": 13316112,
      "step": 20370
    },
    {
      "epoch": 10.678721174004194,
      "grad_norm": 0.1534949243068695,
      "learning_rate": 2.6403392077463108e-05,
      "loss": 0.4777,
      "num_input_tokens_seen": 13320528,
      "step": 20375
    },
    {
      "epoch": 10.681341719077569,
      "grad_norm": 0.12002287060022354,
      "learning_rate": 2.639197567630256e-05,
      "loss": 0.5023,
      "num_input_tokens_seen": 13323888,
      "step": 20380
    },
    {
      "epoch": 10.683962264150944,
      "grad_norm": 0.159285768866539,
      "learning_rate": 2.638055898395657e-05,
      "loss": 0.5701,
      "num_input_tokens_seen": 13327568,
      "step": 20385
    },
    {
      "epoch": 10.68658280922432,
      "grad_norm": 0.23345963656902313,
      "learning_rate": 2.6369142002813403e-05,
      "loss": 0.6027,
      "num_input_tokens_seen": 13330096,
      "step": 20390
    },
    {
      "epoch": 10.689203354297694,
      "grad_norm": 0.2113945186138153,
      "learning_rate": 2.6357724735261335e-05,
      "loss": 0.4416,
      "num_input_tokens_seen": 13332560,
      "step": 20395
    },
    {
      "epoch": 10.69182389937107,
      "grad_norm": 0.14019392430782318,
      "learning_rate": 2.6346307183688752e-05,
      "loss": 0.5983,
      "num_input_tokens_seen": 13336080,
      "step": 20400
    },
    {
      "epoch": 10.694444444444445,
      "grad_norm": 0.49764156341552734,
      "learning_rate": 2.6334889350484064e-05,
      "loss": 0.4564,
      "num_input_tokens_seen": 13339696,
      "step": 20405
    },
    {
      "epoch": 10.69706498951782,
      "grad_norm": 0.10855230689048767,
      "learning_rate": 2.632347123803575e-05,
      "loss": 0.4441,
      "num_input_tokens_seen": 13343760,
      "step": 20410
    },
    {
      "epoch": 10.699685534591195,
      "grad_norm": 0.2084486335515976,
      "learning_rate": 2.6312052848732353e-05,
      "loss": 0.5469,
      "num_input_tokens_seen": 13346928,
      "step": 20415
    },
    {
      "epoch": 10.70230607966457,
      "grad_norm": 0.16730178892612457,
      "learning_rate": 2.6300634184962475e-05,
      "loss": 0.5803,
      "num_input_tokens_seen": 13349968,
      "step": 20420
    },
    {
      "epoch": 10.704926624737945,
      "grad_norm": 0.3250099718570709,
      "learning_rate": 2.628921524911475e-05,
      "loss": 0.5297,
      "num_input_tokens_seen": 13353232,
      "step": 20425
    },
    {
      "epoch": 10.70754716981132,
      "grad_norm": 0.2020486742258072,
      "learning_rate": 2.6277796043577924e-05,
      "loss": 0.4625,
      "num_input_tokens_seen": 13356496,
      "step": 20430
    },
    {
      "epoch": 10.710167714884696,
      "grad_norm": 0.2069292515516281,
      "learning_rate": 2.6266376570740725e-05,
      "loss": 0.5022,
      "num_input_tokens_seen": 13360656,
      "step": 20435
    },
    {
      "epoch": 10.71278825995807,
      "grad_norm": 0.2563443183898926,
      "learning_rate": 2.6254956832992005e-05,
      "loss": 0.3742,
      "num_input_tokens_seen": 13363984,
      "step": 20440
    },
    {
      "epoch": 10.715408805031446,
      "grad_norm": 0.28599289059638977,
      "learning_rate": 2.6243536832720627e-05,
      "loss": 0.5189,
      "num_input_tokens_seen": 13367664,
      "step": 20445
    },
    {
      "epoch": 10.718029350104821,
      "grad_norm": 0.2972331941127777,
      "learning_rate": 2.623211657231553e-05,
      "loss": 0.384,
      "num_input_tokens_seen": 13369744,
      "step": 20450
    },
    {
      "epoch": 10.720649895178196,
      "grad_norm": 0.1737794131040573,
      "learning_rate": 2.6220696054165704e-05,
      "loss": 0.4803,
      "num_input_tokens_seen": 13373584,
      "step": 20455
    },
    {
      "epoch": 10.723270440251572,
      "grad_norm": 0.12063518911600113,
      "learning_rate": 2.6209275280660194e-05,
      "loss": 0.537,
      "num_input_tokens_seen": 13377232,
      "step": 20460
    },
    {
      "epoch": 10.725890985324948,
      "grad_norm": 0.18710580468177795,
      "learning_rate": 2.6197854254188085e-05,
      "loss": 0.6145,
      "num_input_tokens_seen": 13380944,
      "step": 20465
    },
    {
      "epoch": 10.728511530398324,
      "grad_norm": 0.2161768674850464,
      "learning_rate": 2.618643297713854e-05,
      "loss": 0.4817,
      "num_input_tokens_seen": 13385072,
      "step": 20470
    },
    {
      "epoch": 10.731132075471699,
      "grad_norm": 0.18040154874324799,
      "learning_rate": 2.617501145190075e-05,
      "loss": 0.4692,
      "num_input_tokens_seen": 13388016,
      "step": 20475
    },
    {
      "epoch": 10.733752620545074,
      "grad_norm": 0.16616059839725494,
      "learning_rate": 2.616358968086397e-05,
      "loss": 0.499,
      "num_input_tokens_seen": 13391760,
      "step": 20480
    },
    {
      "epoch": 10.73637316561845,
      "grad_norm": 0.13148348033428192,
      "learning_rate": 2.6152167666417493e-05,
      "loss": 0.5656,
      "num_input_tokens_seen": 13396144,
      "step": 20485
    },
    {
      "epoch": 10.738993710691824,
      "grad_norm": 0.2018650323152542,
      "learning_rate": 2.6140745410950694e-05,
      "loss": 0.3922,
      "num_input_tokens_seen": 13399824,
      "step": 20490
    },
    {
      "epoch": 10.7416142557652,
      "grad_norm": 0.16665661334991455,
      "learning_rate": 2.6129322916852973e-05,
      "loss": 0.454,
      "num_input_tokens_seen": 13403248,
      "step": 20495
    },
    {
      "epoch": 10.744234800838575,
      "grad_norm": 0.16241638362407684,
      "learning_rate": 2.611790018651377e-05,
      "loss": 0.4406,
      "num_input_tokens_seen": 13405904,
      "step": 20500
    },
    {
      "epoch": 10.74685534591195,
      "grad_norm": 0.19606104493141174,
      "learning_rate": 2.6106477222322613e-05,
      "loss": 0.3901,
      "num_input_tokens_seen": 13409360,
      "step": 20505
    },
    {
      "epoch": 10.749475890985325,
      "grad_norm": 0.12451237440109253,
      "learning_rate": 2.6095054026669037e-05,
      "loss": 0.4555,
      "num_input_tokens_seen": 13412304,
      "step": 20510
    },
    {
      "epoch": 10.7520964360587,
      "grad_norm": 0.2052038311958313,
      "learning_rate": 2.608363060194265e-05,
      "loss": 0.5139,
      "num_input_tokens_seen": 13415568,
      "step": 20515
    },
    {
      "epoch": 10.754716981132075,
      "grad_norm": 0.29246827960014343,
      "learning_rate": 2.6072206950533117e-05,
      "loss": 0.5238,
      "num_input_tokens_seen": 13418160,
      "step": 20520
    },
    {
      "epoch": 10.75733752620545,
      "grad_norm": 0.17645728588104248,
      "learning_rate": 2.606078307483011e-05,
      "loss": 0.3472,
      "num_input_tokens_seen": 13421168,
      "step": 20525
    },
    {
      "epoch": 10.759958071278826,
      "grad_norm": 0.11048577725887299,
      "learning_rate": 2.6049358977223392e-05,
      "loss": 0.4519,
      "num_input_tokens_seen": 13423824,
      "step": 20530
    },
    {
      "epoch": 10.7625786163522,
      "grad_norm": 0.1501067578792572,
      "learning_rate": 2.6037934660102747e-05,
      "loss": 0.4101,
      "num_input_tokens_seen": 13426928,
      "step": 20535
    },
    {
      "epoch": 10.765199161425576,
      "grad_norm": 0.14457200467586517,
      "learning_rate": 2.6026510125858006e-05,
      "loss": 0.5223,
      "num_input_tokens_seen": 13429712,
      "step": 20540
    },
    {
      "epoch": 10.767819706498951,
      "grad_norm": 0.1112373098731041,
      "learning_rate": 2.6015085376879063e-05,
      "loss": 0.3996,
      "num_input_tokens_seen": 13433648,
      "step": 20545
    },
    {
      "epoch": 10.770440251572326,
      "grad_norm": 0.2123623490333557,
      "learning_rate": 2.600366041555584e-05,
      "loss": 0.4455,
      "num_input_tokens_seen": 13436048,
      "step": 20550
    },
    {
      "epoch": 10.773060796645701,
      "grad_norm": 0.15376488864421844,
      "learning_rate": 2.5992235244278302e-05,
      "loss": 0.5111,
      "num_input_tokens_seen": 13438672,
      "step": 20555
    },
    {
      "epoch": 10.775681341719078,
      "grad_norm": 0.12574006617069244,
      "learning_rate": 2.5980809865436485e-05,
      "loss": 0.3954,
      "num_input_tokens_seen": 13440976,
      "step": 20560
    },
    {
      "epoch": 10.778301886792454,
      "grad_norm": 0.1425301879644394,
      "learning_rate": 2.5969384281420424e-05,
      "loss": 0.3733,
      "num_input_tokens_seen": 13444528,
      "step": 20565
    },
    {
      "epoch": 10.780922431865829,
      "grad_norm": 0.26806458830833435,
      "learning_rate": 2.5957958494620227e-05,
      "loss": 0.5415,
      "num_input_tokens_seen": 13447664,
      "step": 20570
    },
    {
      "epoch": 10.783542976939204,
      "grad_norm": 0.11334314942359924,
      "learning_rate": 2.594653250742605e-05,
      "loss": 0.5274,
      "num_input_tokens_seen": 13451312,
      "step": 20575
    },
    {
      "epoch": 10.786163522012579,
      "grad_norm": 0.17624294757843018,
      "learning_rate": 2.5935106322228065e-05,
      "loss": 0.4915,
      "num_input_tokens_seen": 13453776,
      "step": 20580
    },
    {
      "epoch": 10.788784067085954,
      "grad_norm": 0.28603437542915344,
      "learning_rate": 2.5923679941416514e-05,
      "loss": 0.4617,
      "num_input_tokens_seen": 13456656,
      "step": 20585
    },
    {
      "epoch": 10.79140461215933,
      "grad_norm": 0.21021121740341187,
      "learning_rate": 2.591225336738165e-05,
      "loss": 0.4732,
      "num_input_tokens_seen": 13459440,
      "step": 20590
    },
    {
      "epoch": 10.794025157232705,
      "grad_norm": 0.16630612313747406,
      "learning_rate": 2.5900826602513782e-05,
      "loss": 0.5893,
      "num_input_tokens_seen": 13462608,
      "step": 20595
    },
    {
      "epoch": 10.79664570230608,
      "grad_norm": 0.15348459780216217,
      "learning_rate": 2.5889399649203277e-05,
      "loss": 0.4601,
      "num_input_tokens_seen": 13465424,
      "step": 20600
    },
    {
      "epoch": 10.799266247379455,
      "grad_norm": 0.24120095372200012,
      "learning_rate": 2.5877972509840514e-05,
      "loss": 0.4661,
      "num_input_tokens_seen": 13468688,
      "step": 20605
    },
    {
      "epoch": 10.80188679245283,
      "grad_norm": 0.18783250451087952,
      "learning_rate": 2.5866545186815912e-05,
      "loss": 0.4397,
      "num_input_tokens_seen": 13471312,
      "step": 20610
    },
    {
      "epoch": 10.804507337526205,
      "grad_norm": 0.28583386540412903,
      "learning_rate": 2.5855117682519954e-05,
      "loss": 0.3358,
      "num_input_tokens_seen": 13474352,
      "step": 20615
    },
    {
      "epoch": 10.80712788259958,
      "grad_norm": 0.13595852255821228,
      "learning_rate": 2.584368999934312e-05,
      "loss": 0.5403,
      "num_input_tokens_seen": 13478704,
      "step": 20620
    },
    {
      "epoch": 10.809748427672956,
      "grad_norm": 0.2146490067243576,
      "learning_rate": 2.5832262139675966e-05,
      "loss": 0.4604,
      "num_input_tokens_seen": 13481904,
      "step": 20625
    },
    {
      "epoch": 10.81236897274633,
      "grad_norm": 0.1612163484096527,
      "learning_rate": 2.582083410590907e-05,
      "loss": 0.6284,
      "num_input_tokens_seen": 13484624,
      "step": 20630
    },
    {
      "epoch": 10.814989517819706,
      "grad_norm": 0.140507772564888,
      "learning_rate": 2.5809405900433036e-05,
      "loss": 0.4208,
      "num_input_tokens_seen": 13488368,
      "step": 20635
    },
    {
      "epoch": 10.817610062893081,
      "grad_norm": 0.15342576801776886,
      "learning_rate": 2.579797752563853e-05,
      "loss": 0.6248,
      "num_input_tokens_seen": 13490672,
      "step": 20640
    },
    {
      "epoch": 10.820230607966456,
      "grad_norm": 0.20389677584171295,
      "learning_rate": 2.5786548983916232e-05,
      "loss": 0.444,
      "num_input_tokens_seen": 13495984,
      "step": 20645
    },
    {
      "epoch": 10.822851153039831,
      "grad_norm": 0.23295168578624725,
      "learning_rate": 2.577512027765685e-05,
      "loss": 0.4923,
      "num_input_tokens_seen": 13498416,
      "step": 20650
    },
    {
      "epoch": 10.825471698113208,
      "grad_norm": 0.12745100259780884,
      "learning_rate": 2.5763691409251155e-05,
      "loss": 0.5264,
      "num_input_tokens_seen": 13502512,
      "step": 20655
    },
    {
      "epoch": 10.828092243186584,
      "grad_norm": 0.11466500908136368,
      "learning_rate": 2.5752262381089932e-05,
      "loss": 0.5323,
      "num_input_tokens_seen": 13506416,
      "step": 20660
    },
    {
      "epoch": 10.830712788259959,
      "grad_norm": 0.22217972576618195,
      "learning_rate": 2.5740833195563996e-05,
      "loss": 0.6167,
      "num_input_tokens_seen": 13509392,
      "step": 20665
    },
    {
      "epoch": 10.833333333333334,
      "grad_norm": 0.3523816466331482,
      "learning_rate": 2.572940385506421e-05,
      "loss": 0.5724,
      "num_input_tokens_seen": 13514032,
      "step": 20670
    },
    {
      "epoch": 10.835953878406709,
      "grad_norm": 0.20005638897418976,
      "learning_rate": 2.5717974361981457e-05,
      "loss": 0.5376,
      "num_input_tokens_seen": 13517968,
      "step": 20675
    },
    {
      "epoch": 10.838574423480084,
      "grad_norm": 0.09593161940574646,
      "learning_rate": 2.5706544718706665e-05,
      "loss": 0.5625,
      "num_input_tokens_seen": 13521616,
      "step": 20680
    },
    {
      "epoch": 10.84119496855346,
      "grad_norm": 0.15726575255393982,
      "learning_rate": 2.569511492763077e-05,
      "loss": 0.355,
      "num_input_tokens_seen": 13524592,
      "step": 20685
    },
    {
      "epoch": 10.843815513626835,
      "grad_norm": 0.1575109362602234,
      "learning_rate": 2.5683684991144764e-05,
      "loss": 0.5316,
      "num_input_tokens_seen": 13528048,
      "step": 20690
    },
    {
      "epoch": 10.84643605870021,
      "grad_norm": 0.1416410654783249,
      "learning_rate": 2.5672254911639666e-05,
      "loss": 0.4927,
      "num_input_tokens_seen": 13531120,
      "step": 20695
    },
    {
      "epoch": 10.849056603773585,
      "grad_norm": 0.18790209293365479,
      "learning_rate": 2.5660824691506496e-05,
      "loss": 0.5247,
      "num_input_tokens_seen": 13534288,
      "step": 20700
    },
    {
      "epoch": 10.85167714884696,
      "grad_norm": 0.12188680469989777,
      "learning_rate": 2.5649394333136356e-05,
      "loss": 0.6226,
      "num_input_tokens_seen": 13538032,
      "step": 20705
    },
    {
      "epoch": 10.854297693920335,
      "grad_norm": 0.239188551902771,
      "learning_rate": 2.5637963838920316e-05,
      "loss": 0.5087,
      "num_input_tokens_seen": 13541008,
      "step": 20710
    },
    {
      "epoch": 10.85691823899371,
      "grad_norm": 0.22185324132442474,
      "learning_rate": 2.5626533211249525e-05,
      "loss": 0.5241,
      "num_input_tokens_seen": 13544304,
      "step": 20715
    },
    {
      "epoch": 10.859538784067086,
      "grad_norm": 0.16512343287467957,
      "learning_rate": 2.5615102452515138e-05,
      "loss": 0.5231,
      "num_input_tokens_seen": 13548048,
      "step": 20720
    },
    {
      "epoch": 10.86215932914046,
      "grad_norm": 0.200151726603508,
      "learning_rate": 2.5603671565108328e-05,
      "loss": 0.3601,
      "num_input_tokens_seen": 13554288,
      "step": 20725
    },
    {
      "epoch": 10.864779874213836,
      "grad_norm": 0.2482290416955948,
      "learning_rate": 2.5592240551420317e-05,
      "loss": 0.5515,
      "num_input_tokens_seen": 13558320,
      "step": 20730
    },
    {
      "epoch": 10.867400419287211,
      "grad_norm": 0.18132247030735016,
      "learning_rate": 2.5580809413842343e-05,
      "loss": 0.4015,
      "num_input_tokens_seen": 13561328,
      "step": 20735
    },
    {
      "epoch": 10.870020964360586,
      "grad_norm": 0.19128620624542236,
      "learning_rate": 2.5569378154765655e-05,
      "loss": 0.6196,
      "num_input_tokens_seen": 13566224,
      "step": 20740
    },
    {
      "epoch": 10.872641509433961,
      "grad_norm": 0.2789011299610138,
      "learning_rate": 2.5557946776581572e-05,
      "loss": 0.6284,
      "num_input_tokens_seen": 13568752,
      "step": 20745
    },
    {
      "epoch": 10.875262054507338,
      "grad_norm": 0.154091477394104,
      "learning_rate": 2.554651528168137e-05,
      "loss": 0.4176,
      "num_input_tokens_seen": 13571664,
      "step": 20750
    },
    {
      "epoch": 10.877882599580714,
      "grad_norm": 0.14478228986263275,
      "learning_rate": 2.5535083672456407e-05,
      "loss": 0.4728,
      "num_input_tokens_seen": 13574704,
      "step": 20755
    },
    {
      "epoch": 10.880503144654089,
      "grad_norm": 0.12291364371776581,
      "learning_rate": 2.5523651951298055e-05,
      "loss": 0.5269,
      "num_input_tokens_seen": 13577264,
      "step": 20760
    },
    {
      "epoch": 10.883123689727464,
      "grad_norm": 0.23310257494449615,
      "learning_rate": 2.5512220120597686e-05,
      "loss": 0.4061,
      "num_input_tokens_seen": 13580240,
      "step": 20765
    },
    {
      "epoch": 10.885744234800839,
      "grad_norm": 0.11247768998146057,
      "learning_rate": 2.550078818274671e-05,
      "loss": 0.416,
      "num_input_tokens_seen": 13583472,
      "step": 20770
    },
    {
      "epoch": 10.888364779874214,
      "grad_norm": 0.12174230068922043,
      "learning_rate": 2.548935614013656e-05,
      "loss": 0.4796,
      "num_input_tokens_seen": 13586640,
      "step": 20775
    },
    {
      "epoch": 10.89098532494759,
      "grad_norm": 0.12950177490711212,
      "learning_rate": 2.5477923995158683e-05,
      "loss": 0.4876,
      "num_input_tokens_seen": 13589680,
      "step": 20780
    },
    {
      "epoch": 10.893605870020965,
      "grad_norm": 0.2909754812717438,
      "learning_rate": 2.5466491750204562e-05,
      "loss": 0.6837,
      "num_input_tokens_seen": 13593072,
      "step": 20785
    },
    {
      "epoch": 10.89622641509434,
      "grad_norm": 0.13834179937839508,
      "learning_rate": 2.5455059407665693e-05,
      "loss": 0.425,
      "num_input_tokens_seen": 13596624,
      "step": 20790
    },
    {
      "epoch": 10.898846960167715,
      "grad_norm": 0.12465330958366394,
      "learning_rate": 2.544362696993358e-05,
      "loss": 0.4125,
      "num_input_tokens_seen": 13599088,
      "step": 20795
    },
    {
      "epoch": 10.90146750524109,
      "grad_norm": 0.15733283758163452,
      "learning_rate": 2.5432194439399775e-05,
      "loss": 0.4444,
      "num_input_tokens_seen": 13601552,
      "step": 20800
    },
    {
      "epoch": 10.904088050314465,
      "grad_norm": 0.1675613820552826,
      "learning_rate": 2.542076181845582e-05,
      "loss": 0.3995,
      "num_input_tokens_seen": 13604144,
      "step": 20805
    },
    {
      "epoch": 10.90670859538784,
      "grad_norm": 0.2614853084087372,
      "learning_rate": 2.540932910949329e-05,
      "loss": 0.5007,
      "num_input_tokens_seen": 13606928,
      "step": 20810
    },
    {
      "epoch": 10.909329140461216,
      "grad_norm": 0.14431080222129822,
      "learning_rate": 2.5397896314903785e-05,
      "loss": 0.456,
      "num_input_tokens_seen": 13610320,
      "step": 20815
    },
    {
      "epoch": 10.91194968553459,
      "grad_norm": 0.08204982429742813,
      "learning_rate": 2.5386463437078905e-05,
      "loss": 0.7238,
      "num_input_tokens_seen": 13613680,
      "step": 20820
    },
    {
      "epoch": 10.914570230607966,
      "grad_norm": 0.20563144981861115,
      "learning_rate": 2.5375030478410282e-05,
      "loss": 0.3907,
      "num_input_tokens_seen": 13617264,
      "step": 20825
    },
    {
      "epoch": 10.917190775681341,
      "grad_norm": 0.20345088839530945,
      "learning_rate": 2.536359744128957e-05,
      "loss": 0.4615,
      "num_input_tokens_seen": 13620208,
      "step": 20830
    },
    {
      "epoch": 10.919811320754716,
      "grad_norm": 0.1871671974658966,
      "learning_rate": 2.5352164328108413e-05,
      "loss": 0.4288,
      "num_input_tokens_seen": 13623248,
      "step": 20835
    },
    {
      "epoch": 10.922431865828091,
      "grad_norm": 0.18190716207027435,
      "learning_rate": 2.53407311412585e-05,
      "loss": 0.4664,
      "num_input_tokens_seen": 13626672,
      "step": 20840
    },
    {
      "epoch": 10.925052410901468,
      "grad_norm": 0.12643055617809296,
      "learning_rate": 2.5329297883131526e-05,
      "loss": 0.5225,
      "num_input_tokens_seen": 13629776,
      "step": 20845
    },
    {
      "epoch": 10.927672955974844,
      "grad_norm": 0.15670083463191986,
      "learning_rate": 2.5317864556119182e-05,
      "loss": 0.5382,
      "num_input_tokens_seen": 13632336,
      "step": 20850
    },
    {
      "epoch": 10.930293501048219,
      "grad_norm": 0.23147745430469513,
      "learning_rate": 2.53064311626132e-05,
      "loss": 0.5345,
      "num_input_tokens_seen": 13635600,
      "step": 20855
    },
    {
      "epoch": 10.932914046121594,
      "grad_norm": 0.1664314866065979,
      "learning_rate": 2.5294997705005318e-05,
      "loss": 0.4393,
      "num_input_tokens_seen": 13639248,
      "step": 20860
    },
    {
      "epoch": 10.935534591194969,
      "grad_norm": 0.23335446417331696,
      "learning_rate": 2.5283564185687282e-05,
      "loss": 0.4337,
      "num_input_tokens_seen": 13641872,
      "step": 20865
    },
    {
      "epoch": 10.938155136268344,
      "grad_norm": 0.108344167470932,
      "learning_rate": 2.5272130607050848e-05,
      "loss": 0.4401,
      "num_input_tokens_seen": 13645424,
      "step": 20870
    },
    {
      "epoch": 10.94077568134172,
      "grad_norm": 0.11517685651779175,
      "learning_rate": 2.5260696971487802e-05,
      "loss": 0.4318,
      "num_input_tokens_seen": 13648368,
      "step": 20875
    },
    {
      "epoch": 10.943396226415095,
      "grad_norm": 0.15290090441703796,
      "learning_rate": 2.5249263281389923e-05,
      "loss": 0.5333,
      "num_input_tokens_seen": 13650704,
      "step": 20880
    },
    {
      "epoch": 10.94601677148847,
      "grad_norm": 0.17337509989738464,
      "learning_rate": 2.5237829539149004e-05,
      "loss": 0.6614,
      "num_input_tokens_seen": 13653552,
      "step": 20885
    },
    {
      "epoch": 10.948637316561845,
      "grad_norm": 0.13502858579158783,
      "learning_rate": 2.5226395747156866e-05,
      "loss": 0.5145,
      "num_input_tokens_seen": 13657488,
      "step": 20890
    },
    {
      "epoch": 10.95125786163522,
      "grad_norm": 0.18395495414733887,
      "learning_rate": 2.5214961907805317e-05,
      "loss": 0.4681,
      "num_input_tokens_seen": 13660976,
      "step": 20895
    },
    {
      "epoch": 10.953878406708595,
      "grad_norm": 0.136240154504776,
      "learning_rate": 2.520352802348619e-05,
      "loss": 0.4152,
      "num_input_tokens_seen": 13664080,
      "step": 20900
    },
    {
      "epoch": 10.95649895178197,
      "grad_norm": 0.160923570394516,
      "learning_rate": 2.5192094096591323e-05,
      "loss": 0.6492,
      "num_input_tokens_seen": 13666960,
      "step": 20905
    },
    {
      "epoch": 10.959119496855346,
      "grad_norm": 0.12130378931760788,
      "learning_rate": 2.5180660129512556e-05,
      "loss": 0.4597,
      "num_input_tokens_seen": 13670832,
      "step": 20910
    },
    {
      "epoch": 10.96174004192872,
      "grad_norm": 0.20256735384464264,
      "learning_rate": 2.516922612464176e-05,
      "loss": 0.469,
      "num_input_tokens_seen": 13673456,
      "step": 20915
    },
    {
      "epoch": 10.964360587002096,
      "grad_norm": 0.183502197265625,
      "learning_rate": 2.5157792084370783e-05,
      "loss": 0.4113,
      "num_input_tokens_seen": 13676208,
      "step": 20920
    },
    {
      "epoch": 10.966981132075471,
      "grad_norm": 0.1475711166858673,
      "learning_rate": 2.5146358011091502e-05,
      "loss": 0.5637,
      "num_input_tokens_seen": 13679184,
      "step": 20925
    },
    {
      "epoch": 10.969601677148846,
      "grad_norm": 0.25516006350517273,
      "learning_rate": 2.5134923907195805e-05,
      "loss": 0.5402,
      "num_input_tokens_seen": 13682896,
      "step": 20930
    },
    {
      "epoch": 10.972222222222221,
      "grad_norm": 0.2055857628583908,
      "learning_rate": 2.5123489775075558e-05,
      "loss": 0.6012,
      "num_input_tokens_seen": 13686064,
      "step": 20935
    },
    {
      "epoch": 10.974842767295598,
      "grad_norm": 0.11208902299404144,
      "learning_rate": 2.511205561712266e-05,
      "loss": 0.584,
      "num_input_tokens_seen": 13689424,
      "step": 20940
    },
    {
      "epoch": 10.977463312368974,
      "grad_norm": 0.11751967668533325,
      "learning_rate": 2.5100621435729004e-05,
      "loss": 0.6066,
      "num_input_tokens_seen": 13692496,
      "step": 20945
    },
    {
      "epoch": 10.980083857442349,
      "grad_norm": 0.5020133256912231,
      "learning_rate": 2.5089187233286492e-05,
      "loss": 0.5306,
      "num_input_tokens_seen": 13695728,
      "step": 20950
    },
    {
      "epoch": 10.982704402515724,
      "grad_norm": 0.11550825834274292,
      "learning_rate": 2.5077753012187026e-05,
      "loss": 0.5413,
      "num_input_tokens_seen": 13698640,
      "step": 20955
    },
    {
      "epoch": 10.985324947589099,
      "grad_norm": 0.34351712465286255,
      "learning_rate": 2.506631877482252e-05,
      "loss": 0.524,
      "num_input_tokens_seen": 13702608,
      "step": 20960
    },
    {
      "epoch": 10.987945492662474,
      "grad_norm": 0.17760956287384033,
      "learning_rate": 2.505488452358488e-05,
      "loss": 0.5701,
      "num_input_tokens_seen": 13706352,
      "step": 20965
    },
    {
      "epoch": 10.99056603773585,
      "grad_norm": 0.16950920224189758,
      "learning_rate": 2.504345026086603e-05,
      "loss": 0.4537,
      "num_input_tokens_seen": 13711056,
      "step": 20970
    },
    {
      "epoch": 10.993186582809225,
      "grad_norm": 0.22235874831676483,
      "learning_rate": 2.5032015989057877e-05,
      "loss": 0.419,
      "num_input_tokens_seen": 13714480,
      "step": 20975
    },
    {
      "epoch": 10.9958071278826,
      "grad_norm": 0.10042800009250641,
      "learning_rate": 2.5020581710552348e-05,
      "loss": 0.447,
      "num_input_tokens_seen": 13717136,
      "step": 20980
    },
    {
      "epoch": 10.998427672955975,
      "grad_norm": 0.14763371646404266,
      "learning_rate": 2.5009147427741358e-05,
      "loss": 0.5967,
      "num_input_tokens_seen": 13720464,
      "step": 20985
    },
    {
      "epoch": 11.0,
      "eval_loss": 0.49557408690452576,
      "eval_runtime": 10.2109,
      "eval_samples_per_second": 83.048,
      "eval_steps_per_second": 20.762,
      "num_input_tokens_seen": 13721824,
      "step": 20988
    },
    {
      "epoch": 11.00104821802935,
      "grad_norm": 0.1380603313446045,
      "learning_rate": 2.4997713143016822e-05,
      "loss": 0.4261,
      "num_input_tokens_seen": 13723072,
      "step": 20990
    },
    {
      "epoch": 11.003668763102725,
      "grad_norm": 0.1614612638950348,
      "learning_rate": 2.498627885877068e-05,
      "loss": 0.4447,
      "num_input_tokens_seen": 13726400,
      "step": 20995
    },
    {
      "epoch": 11.0062893081761,
      "grad_norm": 0.24533474445343018,
      "learning_rate": 2.4974844577394846e-05,
      "loss": 0.4137,
      "num_input_tokens_seen": 13729280,
      "step": 21000
    },
    {
      "epoch": 11.008909853249476,
      "grad_norm": 0.20511910319328308,
      "learning_rate": 2.496341030128123e-05,
      "loss": 0.4179,
      "num_input_tokens_seen": 13732416,
      "step": 21005
    },
    {
      "epoch": 11.01153039832285,
      "grad_norm": 0.23500210046768188,
      "learning_rate": 2.495197603282177e-05,
      "loss": 0.6236,
      "num_input_tokens_seen": 13735232,
      "step": 21010
    },
    {
      "epoch": 11.014150943396226,
      "grad_norm": 0.1945894956588745,
      "learning_rate": 2.4940541774408368e-05,
      "loss": 0.4678,
      "num_input_tokens_seen": 13738656,
      "step": 21015
    },
    {
      "epoch": 11.016771488469601,
      "grad_norm": 0.08934716880321503,
      "learning_rate": 2.4929107528432953e-05,
      "loss": 0.4131,
      "num_input_tokens_seen": 13741568,
      "step": 21020
    },
    {
      "epoch": 11.019392033542976,
      "grad_norm": 0.14088189601898193,
      "learning_rate": 2.4917673297287446e-05,
      "loss": 0.3849,
      "num_input_tokens_seen": 13744192,
      "step": 21025
    },
    {
      "epoch": 11.022012578616351,
      "grad_norm": 0.7425488829612732,
      "learning_rate": 2.4906239083363728e-05,
      "loss": 0.5707,
      "num_input_tokens_seen": 13746656,
      "step": 21030
    },
    {
      "epoch": 11.024633123689727,
      "grad_norm": 0.27368995547294617,
      "learning_rate": 2.4894804889053732e-05,
      "loss": 0.521,
      "num_input_tokens_seen": 13749920,
      "step": 21035
    },
    {
      "epoch": 11.027253668763104,
      "grad_norm": 0.1336825042963028,
      "learning_rate": 2.4883370716749347e-05,
      "loss": 0.6615,
      "num_input_tokens_seen": 13752864,
      "step": 21040
    },
    {
      "epoch": 11.029874213836479,
      "grad_norm": 0.2131584882736206,
      "learning_rate": 2.4871936568842486e-05,
      "loss": 0.4645,
      "num_input_tokens_seen": 13756192,
      "step": 21045
    },
    {
      "epoch": 11.032494758909854,
      "grad_norm": 0.17084567248821259,
      "learning_rate": 2.4860502447725033e-05,
      "loss": 0.4555,
      "num_input_tokens_seen": 13759968,
      "step": 21050
    },
    {
      "epoch": 11.035115303983229,
      "grad_norm": 0.10771068930625916,
      "learning_rate": 2.4849068355788873e-05,
      "loss": 0.4549,
      "num_input_tokens_seen": 13763072,
      "step": 21055
    },
    {
      "epoch": 11.037735849056604,
      "grad_norm": 0.19785679876804352,
      "learning_rate": 2.4837634295425892e-05,
      "loss": 0.4079,
      "num_input_tokens_seen": 13765728,
      "step": 21060
    },
    {
      "epoch": 11.04035639412998,
      "grad_norm": 0.19424286484718323,
      "learning_rate": 2.4826200269027957e-05,
      "loss": 0.4754,
      "num_input_tokens_seen": 13769504,
      "step": 21065
    },
    {
      "epoch": 11.042976939203355,
      "grad_norm": 0.16108860075473785,
      "learning_rate": 2.481476627898695e-05,
      "loss": 0.5382,
      "num_input_tokens_seen": 13773760,
      "step": 21070
    },
    {
      "epoch": 11.04559748427673,
      "grad_norm": 0.1513192057609558,
      "learning_rate": 2.4803332327694724e-05,
      "loss": 0.4959,
      "num_input_tokens_seen": 13777088,
      "step": 21075
    },
    {
      "epoch": 11.048218029350105,
      "grad_norm": 0.22195346653461456,
      "learning_rate": 2.4791898417543136e-05,
      "loss": 0.4635,
      "num_input_tokens_seen": 13781120,
      "step": 21080
    },
    {
      "epoch": 11.05083857442348,
      "grad_norm": 0.16972734034061432,
      "learning_rate": 2.4780464550924014e-05,
      "loss": 0.48,
      "num_input_tokens_seen": 13784064,
      "step": 21085
    },
    {
      "epoch": 11.053459119496855,
      "grad_norm": 0.4249837398529053,
      "learning_rate": 2.4769030730229213e-05,
      "loss": 0.5483,
      "num_input_tokens_seen": 13787104,
      "step": 21090
    },
    {
      "epoch": 11.05607966457023,
      "grad_norm": 0.20257896184921265,
      "learning_rate": 2.475759695785054e-05,
      "loss": 0.4977,
      "num_input_tokens_seen": 13789824,
      "step": 21095
    },
    {
      "epoch": 11.058700209643606,
      "grad_norm": 0.16523677110671997,
      "learning_rate": 2.4746163236179822e-05,
      "loss": 0.5507,
      "num_input_tokens_seen": 13793152,
      "step": 21100
    },
    {
      "epoch": 11.06132075471698,
      "grad_norm": 0.1324256807565689,
      "learning_rate": 2.4734729567608858e-05,
      "loss": 0.6212,
      "num_input_tokens_seen": 13796288,
      "step": 21105
    },
    {
      "epoch": 11.063941299790356,
      "grad_norm": 0.24139411747455597,
      "learning_rate": 2.472329595452945e-05,
      "loss": 0.5678,
      "num_input_tokens_seen": 13799072,
      "step": 21110
    },
    {
      "epoch": 11.066561844863731,
      "grad_norm": 0.12446754425764084,
      "learning_rate": 2.4711862399333367e-05,
      "loss": 0.4449,
      "num_input_tokens_seen": 13802560,
      "step": 21115
    },
    {
      "epoch": 11.069182389937106,
      "grad_norm": 0.1470658779144287,
      "learning_rate": 2.4700428904412377e-05,
      "loss": 0.4787,
      "num_input_tokens_seen": 13806592,
      "step": 21120
    },
    {
      "epoch": 11.071802935010481,
      "grad_norm": 0.17421486973762512,
      "learning_rate": 2.4688995472158245e-05,
      "loss": 0.414,
      "num_input_tokens_seen": 13809568,
      "step": 21125
    },
    {
      "epoch": 11.074423480083857,
      "grad_norm": 0.24540013074874878,
      "learning_rate": 2.4677562104962712e-05,
      "loss": 0.3878,
      "num_input_tokens_seen": 13812960,
      "step": 21130
    },
    {
      "epoch": 11.077044025157234,
      "grad_norm": 0.13900677859783173,
      "learning_rate": 2.4666128805217513e-05,
      "loss": 0.7086,
      "num_input_tokens_seen": 13815712,
      "step": 21135
    },
    {
      "epoch": 11.079664570230609,
      "grad_norm": 0.3107653260231018,
      "learning_rate": 2.465469557531436e-05,
      "loss": 0.517,
      "num_input_tokens_seen": 13818560,
      "step": 21140
    },
    {
      "epoch": 11.082285115303984,
      "grad_norm": 0.14099867641925812,
      "learning_rate": 2.4643262417644944e-05,
      "loss": 0.4318,
      "num_input_tokens_seen": 13823264,
      "step": 21145
    },
    {
      "epoch": 11.084905660377359,
      "grad_norm": 0.12864509224891663,
      "learning_rate": 2.463182933460096e-05,
      "loss": 0.5111,
      "num_input_tokens_seen": 13826464,
      "step": 21150
    },
    {
      "epoch": 11.087526205450734,
      "grad_norm": 0.18782588839530945,
      "learning_rate": 2.4620396328574087e-05,
      "loss": 0.4181,
      "num_input_tokens_seen": 13829280,
      "step": 21155
    },
    {
      "epoch": 11.09014675052411,
      "grad_norm": 0.2293669730424881,
      "learning_rate": 2.4608963401955965e-05,
      "loss": 0.4257,
      "num_input_tokens_seen": 13833344,
      "step": 21160
    },
    {
      "epoch": 11.092767295597485,
      "grad_norm": 0.09346948564052582,
      "learning_rate": 2.4597530557138247e-05,
      "loss": 0.5031,
      "num_input_tokens_seen": 13837088,
      "step": 21165
    },
    {
      "epoch": 11.09538784067086,
      "grad_norm": 0.20324739813804626,
      "learning_rate": 2.4586097796512538e-05,
      "loss": 0.5357,
      "num_input_tokens_seen": 13839488,
      "step": 21170
    },
    {
      "epoch": 11.098008385744235,
      "grad_norm": 0.16652870178222656,
      "learning_rate": 2.457466512247045e-05,
      "loss": 0.3871,
      "num_input_tokens_seen": 13842528,
      "step": 21175
    },
    {
      "epoch": 11.10062893081761,
      "grad_norm": 0.15310059487819672,
      "learning_rate": 2.4563232537403563e-05,
      "loss": 0.3693,
      "num_input_tokens_seen": 13845312,
      "step": 21180
    },
    {
      "epoch": 11.103249475890985,
      "grad_norm": 0.1758144348859787,
      "learning_rate": 2.4551800043703442e-05,
      "loss": 0.4255,
      "num_input_tokens_seen": 13848256,
      "step": 21185
    },
    {
      "epoch": 11.10587002096436,
      "grad_norm": 0.12939687073230743,
      "learning_rate": 2.454036764376165e-05,
      "loss": 0.4532,
      "num_input_tokens_seen": 13852064,
      "step": 21190
    },
    {
      "epoch": 11.108490566037736,
      "grad_norm": 0.20995992422103882,
      "learning_rate": 2.4528935339969705e-05,
      "loss": 0.5578,
      "num_input_tokens_seen": 13855040,
      "step": 21195
    },
    {
      "epoch": 11.11111111111111,
      "grad_norm": 0.2168520987033844,
      "learning_rate": 2.45175031347191e-05,
      "loss": 0.5312,
      "num_input_tokens_seen": 13858528,
      "step": 21200
    },
    {
      "epoch": 11.113731656184486,
      "grad_norm": 0.13868017494678497,
      "learning_rate": 2.4506071030401344e-05,
      "loss": 0.4906,
      "num_input_tokens_seen": 13862048,
      "step": 21205
    },
    {
      "epoch": 11.116352201257861,
      "grad_norm": 0.14995940029621124,
      "learning_rate": 2.4494639029407884e-05,
      "loss": 0.4561,
      "num_input_tokens_seen": 13864864,
      "step": 21210
    },
    {
      "epoch": 11.118972746331236,
      "grad_norm": 0.1066553145647049,
      "learning_rate": 2.448320713413018e-05,
      "loss": 0.4254,
      "num_input_tokens_seen": 13868256,
      "step": 21215
    },
    {
      "epoch": 11.121593291404611,
      "grad_norm": 0.3384319543838501,
      "learning_rate": 2.4471775346959654e-05,
      "loss": 0.5195,
      "num_input_tokens_seen": 13870944,
      "step": 21220
    },
    {
      "epoch": 11.124213836477987,
      "grad_norm": 0.14371028542518616,
      "learning_rate": 2.446034367028769e-05,
      "loss": 0.4501,
      "num_input_tokens_seen": 13874016,
      "step": 21225
    },
    {
      "epoch": 11.126834381551364,
      "grad_norm": 0.17819127440452576,
      "learning_rate": 2.444891210650567e-05,
      "loss": 0.5845,
      "num_input_tokens_seen": 13881184,
      "step": 21230
    },
    {
      "epoch": 11.129454926624739,
      "grad_norm": 0.13000139594078064,
      "learning_rate": 2.4437480658004958e-05,
      "loss": 0.5414,
      "num_input_tokens_seen": 13884224,
      "step": 21235
    },
    {
      "epoch": 11.132075471698114,
      "grad_norm": 0.1704341471195221,
      "learning_rate": 2.4426049327176868e-05,
      "loss": 0.3791,
      "num_input_tokens_seen": 13888064,
      "step": 21240
    },
    {
      "epoch": 11.134696016771489,
      "grad_norm": 0.13512958586215973,
      "learning_rate": 2.4414618116412716e-05,
      "loss": 0.3593,
      "num_input_tokens_seen": 13890912,
      "step": 21245
    },
    {
      "epoch": 11.137316561844864,
      "grad_norm": 0.17980235815048218,
      "learning_rate": 2.440318702810378e-05,
      "loss": 0.5675,
      "num_input_tokens_seen": 13894432,
      "step": 21250
    },
    {
      "epoch": 11.13993710691824,
      "grad_norm": 0.2082352638244629,
      "learning_rate": 2.43917560646413e-05,
      "loss": 0.5683,
      "num_input_tokens_seen": 13898048,
      "step": 21255
    },
    {
      "epoch": 11.142557651991615,
      "grad_norm": 0.4236050546169281,
      "learning_rate": 2.4380325228416518e-05,
      "loss": 0.7028,
      "num_input_tokens_seen": 13901184,
      "step": 21260
    },
    {
      "epoch": 11.14517819706499,
      "grad_norm": 0.18845106661319733,
      "learning_rate": 2.4368894521820624e-05,
      "loss": 0.3424,
      "num_input_tokens_seen": 13904864,
      "step": 21265
    },
    {
      "epoch": 11.147798742138365,
      "grad_norm": 0.1423502266407013,
      "learning_rate": 2.43574639472448e-05,
      "loss": 0.3872,
      "num_input_tokens_seen": 13908096,
      "step": 21270
    },
    {
      "epoch": 11.15041928721174,
      "grad_norm": 0.09266326576471329,
      "learning_rate": 2.4346033507080178e-05,
      "loss": 0.3512,
      "num_input_tokens_seen": 13911392,
      "step": 21275
    },
    {
      "epoch": 11.153039832285115,
      "grad_norm": 0.13499440252780914,
      "learning_rate": 2.4334603203717903e-05,
      "loss": 0.5534,
      "num_input_tokens_seen": 13915104,
      "step": 21280
    },
    {
      "epoch": 11.15566037735849,
      "grad_norm": 0.16119010746479034,
      "learning_rate": 2.432317303954904e-05,
      "loss": 0.4887,
      "num_input_tokens_seen": 13918176,
      "step": 21285
    },
    {
      "epoch": 11.158280922431866,
      "grad_norm": 0.19382722675800323,
      "learning_rate": 2.4311743016964648e-05,
      "loss": 0.4046,
      "num_input_tokens_seen": 13921120,
      "step": 21290
    },
    {
      "epoch": 11.16090146750524,
      "grad_norm": 0.13208797574043274,
      "learning_rate": 2.4300313138355772e-05,
      "loss": 0.3966,
      "num_input_tokens_seen": 13924224,
      "step": 21295
    },
    {
      "epoch": 11.163522012578616,
      "grad_norm": 0.13543839752674103,
      "learning_rate": 2.4288883406113396e-05,
      "loss": 0.5248,
      "num_input_tokens_seen": 13927488,
      "step": 21300
    },
    {
      "epoch": 11.166142557651991,
      "grad_norm": 0.2724040746688843,
      "learning_rate": 2.4277453822628512e-05,
      "loss": 0.3928,
      "num_input_tokens_seen": 13930880,
      "step": 21305
    },
    {
      "epoch": 11.168763102725366,
      "grad_norm": 0.20071637630462646,
      "learning_rate": 2.4266024390292037e-05,
      "loss": 0.494,
      "num_input_tokens_seen": 13933728,
      "step": 21310
    },
    {
      "epoch": 11.171383647798741,
      "grad_norm": 0.19907252490520477,
      "learning_rate": 2.425459511149489e-05,
      "loss": 0.5387,
      "num_input_tokens_seen": 13937664,
      "step": 21315
    },
    {
      "epoch": 11.174004192872117,
      "grad_norm": 0.11471583694219589,
      "learning_rate": 2.424316598862793e-05,
      "loss": 0.3906,
      "num_input_tokens_seen": 13940544,
      "step": 21320
    },
    {
      "epoch": 11.176624737945493,
      "grad_norm": 0.10464145988225937,
      "learning_rate": 2.4231737024082023e-05,
      "loss": 0.4669,
      "num_input_tokens_seen": 13943712,
      "step": 21325
    },
    {
      "epoch": 11.179245283018869,
      "grad_norm": 0.11363422870635986,
      "learning_rate": 2.4220308220247954e-05,
      "loss": 0.3891,
      "num_input_tokens_seen": 13947200,
      "step": 21330
    },
    {
      "epoch": 11.181865828092244,
      "grad_norm": 0.2077786773443222,
      "learning_rate": 2.4208879579516528e-05,
      "loss": 0.8462,
      "num_input_tokens_seen": 13950240,
      "step": 21335
    },
    {
      "epoch": 11.184486373165619,
      "grad_norm": 0.1091032475233078,
      "learning_rate": 2.419745110427846e-05,
      "loss": 0.3752,
      "num_input_tokens_seen": 13953440,
      "step": 21340
    },
    {
      "epoch": 11.187106918238994,
      "grad_norm": 0.1509476751089096,
      "learning_rate": 2.418602279692446e-05,
      "loss": 0.4475,
      "num_input_tokens_seen": 13956416,
      "step": 21345
    },
    {
      "epoch": 11.18972746331237,
      "grad_norm": 0.12116574496030807,
      "learning_rate": 2.417459465984521e-05,
      "loss": 0.4454,
      "num_input_tokens_seen": 13959840,
      "step": 21350
    },
    {
      "epoch": 11.192348008385745,
      "grad_norm": 0.2705865502357483,
      "learning_rate": 2.4163166695431336e-05,
      "loss": 0.631,
      "num_input_tokens_seen": 13963168,
      "step": 21355
    },
    {
      "epoch": 11.19496855345912,
      "grad_norm": 0.1757218837738037,
      "learning_rate": 2.4151738906073446e-05,
      "loss": 0.3685,
      "num_input_tokens_seen": 13965632,
      "step": 21360
    },
    {
      "epoch": 11.197589098532495,
      "grad_norm": 0.18483972549438477,
      "learning_rate": 2.4140311294162105e-05,
      "loss": 0.4876,
      "num_input_tokens_seen": 13969184,
      "step": 21365
    },
    {
      "epoch": 11.20020964360587,
      "grad_norm": 0.1766013652086258,
      "learning_rate": 2.412888386208782e-05,
      "loss": 0.5148,
      "num_input_tokens_seen": 13972800,
      "step": 21370
    },
    {
      "epoch": 11.202830188679245,
      "grad_norm": 0.24110528826713562,
      "learning_rate": 2.4117456612241103e-05,
      "loss": 0.664,
      "num_input_tokens_seen": 13976128,
      "step": 21375
    },
    {
      "epoch": 11.20545073375262,
      "grad_norm": 0.10392973572015762,
      "learning_rate": 2.410602954701239e-05,
      "loss": 0.4809,
      "num_input_tokens_seen": 13979424,
      "step": 21380
    },
    {
      "epoch": 11.208071278825996,
      "grad_norm": 0.13567550480365753,
      "learning_rate": 2.409460266879209e-05,
      "loss": 0.3775,
      "num_input_tokens_seen": 13981856,
      "step": 21385
    },
    {
      "epoch": 11.21069182389937,
      "grad_norm": 0.1257496327161789,
      "learning_rate": 2.4083175979970592e-05,
      "loss": 0.3658,
      "num_input_tokens_seen": 13985248,
      "step": 21390
    },
    {
      "epoch": 11.213312368972746,
      "grad_norm": 0.12641268968582153,
      "learning_rate": 2.4071749482938226e-05,
      "loss": 0.5163,
      "num_input_tokens_seen": 13988576,
      "step": 21395
    },
    {
      "epoch": 11.215932914046121,
      "grad_norm": 0.2568434774875641,
      "learning_rate": 2.4060323180085266e-05,
      "loss": 0.534,
      "num_input_tokens_seen": 13991232,
      "step": 21400
    },
    {
      "epoch": 11.218553459119496,
      "grad_norm": 0.18739286065101624,
      "learning_rate": 2.404889707380198e-05,
      "loss": 0.4574,
      "num_input_tokens_seen": 13994592,
      "step": 21405
    },
    {
      "epoch": 11.221174004192871,
      "grad_norm": 0.19144019484519958,
      "learning_rate": 2.403747116647857e-05,
      "loss": 0.4603,
      "num_input_tokens_seen": 13997280,
      "step": 21410
    },
    {
      "epoch": 11.223794549266247,
      "grad_norm": 0.1576446294784546,
      "learning_rate": 2.4026045460505216e-05,
      "loss": 0.477,
      "num_input_tokens_seen": 14000384,
      "step": 21415
    },
    {
      "epoch": 11.226415094339623,
      "grad_norm": 0.1328977644443512,
      "learning_rate": 2.4014619958272047e-05,
      "loss": 0.471,
      "num_input_tokens_seen": 14003872,
      "step": 21420
    },
    {
      "epoch": 11.229035639412999,
      "grad_norm": 0.11489875614643097,
      "learning_rate": 2.4003194662169132e-05,
      "loss": 0.4609,
      "num_input_tokens_seen": 14006976,
      "step": 21425
    },
    {
      "epoch": 11.231656184486374,
      "grad_norm": 0.15660269558429718,
      "learning_rate": 2.3991769574586522e-05,
      "loss": 0.7756,
      "num_input_tokens_seen": 14009472,
      "step": 21430
    },
    {
      "epoch": 11.234276729559749,
      "grad_norm": 0.3515683114528656,
      "learning_rate": 2.3980344697914215e-05,
      "loss": 0.4564,
      "num_input_tokens_seen": 14012480,
      "step": 21435
    },
    {
      "epoch": 11.236897274633124,
      "grad_norm": 0.13736432790756226,
      "learning_rate": 2.3968920034542168e-05,
      "loss": 0.3836,
      "num_input_tokens_seen": 14015072,
      "step": 21440
    },
    {
      "epoch": 11.2395178197065,
      "grad_norm": 0.14089898765087128,
      "learning_rate": 2.3957495586860278e-05,
      "loss": 0.5064,
      "num_input_tokens_seen": 14018496,
      "step": 21445
    },
    {
      "epoch": 11.242138364779874,
      "grad_norm": 0.18569685518741608,
      "learning_rate": 2.394607135725843e-05,
      "loss": 0.5934,
      "num_input_tokens_seen": 14021600,
      "step": 21450
    },
    {
      "epoch": 11.24475890985325,
      "grad_norm": 0.16085019707679749,
      "learning_rate": 2.3934647348126426e-05,
      "loss": 0.5137,
      "num_input_tokens_seen": 14024544,
      "step": 21455
    },
    {
      "epoch": 11.247379454926625,
      "grad_norm": 0.19393520057201385,
      "learning_rate": 2.3923223561854035e-05,
      "loss": 0.4793,
      "num_input_tokens_seen": 14027968,
      "step": 21460
    },
    {
      "epoch": 11.25,
      "grad_norm": 0.17670084536075592,
      "learning_rate": 2.391180000083099e-05,
      "loss": 0.5339,
      "num_input_tokens_seen": 14030688,
      "step": 21465
    },
    {
      "epoch": 11.252620545073375,
      "grad_norm": 0.14027784764766693,
      "learning_rate": 2.390037666744698e-05,
      "loss": 0.4352,
      "num_input_tokens_seen": 14033408,
      "step": 21470
    },
    {
      "epoch": 11.25524109014675,
      "grad_norm": 0.12942515313625336,
      "learning_rate": 2.388895356409162e-05,
      "loss": 0.4384,
      "num_input_tokens_seen": 14036416,
      "step": 21475
    },
    {
      "epoch": 11.257861635220126,
      "grad_norm": 0.14934515953063965,
      "learning_rate": 2.3877530693154506e-05,
      "loss": 0.4527,
      "num_input_tokens_seen": 14039872,
      "step": 21480
    },
    {
      "epoch": 11.2604821802935,
      "grad_norm": 0.13710898160934448,
      "learning_rate": 2.3866108057025162e-05,
      "loss": 0.5037,
      "num_input_tokens_seen": 14042624,
      "step": 21485
    },
    {
      "epoch": 11.263102725366876,
      "grad_norm": 0.17272472381591797,
      "learning_rate": 2.385468565809307e-05,
      "loss": 0.5271,
      "num_input_tokens_seen": 14045984,
      "step": 21490
    },
    {
      "epoch": 11.265723270440251,
      "grad_norm": 0.15008461475372314,
      "learning_rate": 2.3843263498747675e-05,
      "loss": 0.4881,
      "num_input_tokens_seen": 14049152,
      "step": 21495
    },
    {
      "epoch": 11.268343815513626,
      "grad_norm": 0.1449713259935379,
      "learning_rate": 2.383184158137836e-05,
      "loss": 0.4584,
      "num_input_tokens_seen": 14052320,
      "step": 21500
    },
    {
      "epoch": 11.270964360587001,
      "grad_norm": 0.1296524852514267,
      "learning_rate": 2.3820419908374465e-05,
      "loss": 0.4564,
      "num_input_tokens_seen": 14055584,
      "step": 21505
    },
    {
      "epoch": 11.273584905660377,
      "grad_norm": 0.18252681195735931,
      "learning_rate": 2.3808998482125265e-05,
      "loss": 0.6045,
      "num_input_tokens_seen": 14058816,
      "step": 21510
    },
    {
      "epoch": 11.276205450733753,
      "grad_norm": 0.13690422475337982,
      "learning_rate": 2.3797577305019987e-05,
      "loss": 0.6651,
      "num_input_tokens_seen": 14062528,
      "step": 21515
    },
    {
      "epoch": 11.278825995807129,
      "grad_norm": 0.2615548074245453,
      "learning_rate": 2.378615637944783e-05,
      "loss": 0.6379,
      "num_input_tokens_seen": 14065280,
      "step": 21520
    },
    {
      "epoch": 11.281446540880504,
      "grad_norm": 0.2607584297657013,
      "learning_rate": 2.3774735707797895e-05,
      "loss": 0.4359,
      "num_input_tokens_seen": 14068160,
      "step": 21525
    },
    {
      "epoch": 11.284067085953879,
      "grad_norm": 0.1666508913040161,
      "learning_rate": 2.376331529245928e-05,
      "loss": 0.4733,
      "num_input_tokens_seen": 14071712,
      "step": 21530
    },
    {
      "epoch": 11.286687631027254,
      "grad_norm": 0.09908217191696167,
      "learning_rate": 2.3751895135821006e-05,
      "loss": 0.4122,
      "num_input_tokens_seen": 14074912,
      "step": 21535
    },
    {
      "epoch": 11.28930817610063,
      "grad_norm": 0.19665400683879852,
      "learning_rate": 2.374047524027202e-05,
      "loss": 0.4812,
      "num_input_tokens_seen": 14078912,
      "step": 21540
    },
    {
      "epoch": 11.291928721174004,
      "grad_norm": 0.11504946649074554,
      "learning_rate": 2.3729055608201243e-05,
      "loss": 0.4424,
      "num_input_tokens_seen": 14082304,
      "step": 21545
    },
    {
      "epoch": 11.29454926624738,
      "grad_norm": 0.28548479080200195,
      "learning_rate": 2.3717636241997537e-05,
      "loss": 0.4988,
      "num_input_tokens_seen": 14085056,
      "step": 21550
    },
    {
      "epoch": 11.297169811320755,
      "grad_norm": 0.1349485069513321,
      "learning_rate": 2.3706217144049696e-05,
      "loss": 0.5472,
      "num_input_tokens_seen": 14088128,
      "step": 21555
    },
    {
      "epoch": 11.29979035639413,
      "grad_norm": 0.1545563042163849,
      "learning_rate": 2.369479831674647e-05,
      "loss": 0.4387,
      "num_input_tokens_seen": 14090912,
      "step": 21560
    },
    {
      "epoch": 11.302410901467505,
      "grad_norm": 0.09844419360160828,
      "learning_rate": 2.3683379762476555e-05,
      "loss": 0.6101,
      "num_input_tokens_seen": 14094464,
      "step": 21565
    },
    {
      "epoch": 11.30503144654088,
      "grad_norm": 0.1538505256175995,
      "learning_rate": 2.3671961483628562e-05,
      "loss": 0.4853,
      "num_input_tokens_seen": 14097280,
      "step": 21570
    },
    {
      "epoch": 11.307651991614255,
      "grad_norm": 0.17233000695705414,
      "learning_rate": 2.3660543482591078e-05,
      "loss": 0.5317,
      "num_input_tokens_seen": 14100224,
      "step": 21575
    },
    {
      "epoch": 11.31027253668763,
      "grad_norm": 0.11113125830888748,
      "learning_rate": 2.364912576175261e-05,
      "loss": 0.4636,
      "num_input_tokens_seen": 14103744,
      "step": 21580
    },
    {
      "epoch": 11.312893081761006,
      "grad_norm": 0.15137217938899994,
      "learning_rate": 2.363770832350163e-05,
      "loss": 0.4076,
      "num_input_tokens_seen": 14107008,
      "step": 21585
    },
    {
      "epoch": 11.315513626834381,
      "grad_norm": 0.14396452903747559,
      "learning_rate": 2.362629117022653e-05,
      "loss": 0.4607,
      "num_input_tokens_seen": 14110176,
      "step": 21590
    },
    {
      "epoch": 11.318134171907756,
      "grad_norm": 0.23294401168823242,
      "learning_rate": 2.3614874304315633e-05,
      "loss": 0.68,
      "num_input_tokens_seen": 14115968,
      "step": 21595
    },
    {
      "epoch": 11.320754716981131,
      "grad_norm": 0.10368028283119202,
      "learning_rate": 2.3603457728157237e-05,
      "loss": 0.3915,
      "num_input_tokens_seen": 14120864,
      "step": 21600
    },
    {
      "epoch": 11.323375262054507,
      "grad_norm": 0.11356614530086517,
      "learning_rate": 2.3592041444139547e-05,
      "loss": 0.5746,
      "num_input_tokens_seen": 14124576,
      "step": 21605
    },
    {
      "epoch": 11.325995807127883,
      "grad_norm": 0.1217467188835144,
      "learning_rate": 2.358062545465073e-05,
      "loss": 0.5231,
      "num_input_tokens_seen": 14127648,
      "step": 21610
    },
    {
      "epoch": 11.328616352201259,
      "grad_norm": 0.163148432970047,
      "learning_rate": 2.3569209762078866e-05,
      "loss": 0.4478,
      "num_input_tokens_seen": 14130400,
      "step": 21615
    },
    {
      "epoch": 11.331236897274634,
      "grad_norm": 0.10691268742084503,
      "learning_rate": 2.355779436881201e-05,
      "loss": 0.4594,
      "num_input_tokens_seen": 14134080,
      "step": 21620
    },
    {
      "epoch": 11.333857442348009,
      "grad_norm": 0.2097443789243698,
      "learning_rate": 2.3546379277238107e-05,
      "loss": 0.4469,
      "num_input_tokens_seen": 14138080,
      "step": 21625
    },
    {
      "epoch": 11.336477987421384,
      "grad_norm": 0.118393674492836,
      "learning_rate": 2.353496448974508e-05,
      "loss": 0.39,
      "num_input_tokens_seen": 14142432,
      "step": 21630
    },
    {
      "epoch": 11.33909853249476,
      "grad_norm": 0.16039955615997314,
      "learning_rate": 2.3523550008720763e-05,
      "loss": 0.481,
      "num_input_tokens_seen": 14145440,
      "step": 21635
    },
    {
      "epoch": 11.341719077568134,
      "grad_norm": 0.13501852750778198,
      "learning_rate": 2.3512135836552945e-05,
      "loss": 0.467,
      "num_input_tokens_seen": 14148672,
      "step": 21640
    },
    {
      "epoch": 11.34433962264151,
      "grad_norm": 0.1265019327402115,
      "learning_rate": 2.350072197562933e-05,
      "loss": 0.3997,
      "num_input_tokens_seen": 14152000,
      "step": 21645
    },
    {
      "epoch": 11.346960167714885,
      "grad_norm": 0.16148750483989716,
      "learning_rate": 2.3489308428337586e-05,
      "loss": 0.5658,
      "num_input_tokens_seen": 14154912,
      "step": 21650
    },
    {
      "epoch": 11.34958071278826,
      "grad_norm": 0.1265823245048523,
      "learning_rate": 2.3477895197065275e-05,
      "loss": 0.4751,
      "num_input_tokens_seen": 14158208,
      "step": 21655
    },
    {
      "epoch": 11.352201257861635,
      "grad_norm": 0.19180402159690857,
      "learning_rate": 2.346648228419992e-05,
      "loss": 0.5122,
      "num_input_tokens_seen": 14161920,
      "step": 21660
    },
    {
      "epoch": 11.35482180293501,
      "grad_norm": 0.19626615941524506,
      "learning_rate": 2.3455069692128982e-05,
      "loss": 0.4471,
      "num_input_tokens_seen": 14165120,
      "step": 21665
    },
    {
      "epoch": 11.357442348008385,
      "grad_norm": 0.159951314330101,
      "learning_rate": 2.3443657423239832e-05,
      "loss": 0.5254,
      "num_input_tokens_seen": 14168000,
      "step": 21670
    },
    {
      "epoch": 11.36006289308176,
      "grad_norm": 0.13085465133190155,
      "learning_rate": 2.34322454799198e-05,
      "loss": 0.3916,
      "num_input_tokens_seen": 14171584,
      "step": 21675
    },
    {
      "epoch": 11.362683438155136,
      "grad_norm": 0.32077842950820923,
      "learning_rate": 2.3420833864556133e-05,
      "loss": 0.5055,
      "num_input_tokens_seen": 14174336,
      "step": 21680
    },
    {
      "epoch": 11.365303983228511,
      "grad_norm": 0.1597016304731369,
      "learning_rate": 2.3409422579535995e-05,
      "loss": 0.4933,
      "num_input_tokens_seen": 14177344,
      "step": 21685
    },
    {
      "epoch": 11.367924528301886,
      "grad_norm": 0.16042181849479675,
      "learning_rate": 2.339801162724651e-05,
      "loss": 0.5142,
      "num_input_tokens_seen": 14180608,
      "step": 21690
    },
    {
      "epoch": 11.370545073375261,
      "grad_norm": 0.13249295949935913,
      "learning_rate": 2.338660101007471e-05,
      "loss": 0.5487,
      "num_input_tokens_seen": 14183872,
      "step": 21695
    },
    {
      "epoch": 11.373165618448636,
      "grad_norm": 0.24046285450458527,
      "learning_rate": 2.3375190730407578e-05,
      "loss": 0.5313,
      "num_input_tokens_seen": 14186720,
      "step": 21700
    },
    {
      "epoch": 11.375786163522013,
      "grad_norm": 0.14804120361804962,
      "learning_rate": 2.336378079063202e-05,
      "loss": 0.5842,
      "num_input_tokens_seen": 14190336,
      "step": 21705
    },
    {
      "epoch": 11.378406708595389,
      "grad_norm": 0.1369718462228775,
      "learning_rate": 2.335237119313484e-05,
      "loss": 0.4863,
      "num_input_tokens_seen": 14193760,
      "step": 21710
    },
    {
      "epoch": 11.381027253668764,
      "grad_norm": 0.1958620548248291,
      "learning_rate": 2.3340961940302807e-05,
      "loss": 0.4979,
      "num_input_tokens_seen": 14196096,
      "step": 21715
    },
    {
      "epoch": 11.383647798742139,
      "grad_norm": 0.19229057431221008,
      "learning_rate": 2.3329553034522616e-05,
      "loss": 0.5003,
      "num_input_tokens_seen": 14199296,
      "step": 21720
    },
    {
      "epoch": 11.386268343815514,
      "grad_norm": 0.10619350522756577,
      "learning_rate": 2.3318144478180866e-05,
      "loss": 0.477,
      "num_input_tokens_seen": 14202528,
      "step": 21725
    },
    {
      "epoch": 11.38888888888889,
      "grad_norm": 0.10960507392883301,
      "learning_rate": 2.330673627366411e-05,
      "loss": 0.5168,
      "num_input_tokens_seen": 14205568,
      "step": 21730
    },
    {
      "epoch": 11.391509433962264,
      "grad_norm": 0.12357842177152634,
      "learning_rate": 2.329532842335881e-05,
      "loss": 0.4784,
      "num_input_tokens_seen": 14209056,
      "step": 21735
    },
    {
      "epoch": 11.39412997903564,
      "grad_norm": 0.15565402805805206,
      "learning_rate": 2.328392092965135e-05,
      "loss": 0.2634,
      "num_input_tokens_seen": 14214720,
      "step": 21740
    },
    {
      "epoch": 11.396750524109015,
      "grad_norm": 0.12622810900211334,
      "learning_rate": 2.3272513794928054e-05,
      "loss": 0.4671,
      "num_input_tokens_seen": 14217664,
      "step": 21745
    },
    {
      "epoch": 11.39937106918239,
      "grad_norm": 0.19140109419822693,
      "learning_rate": 2.326110702157516e-05,
      "loss": 0.5149,
      "num_input_tokens_seen": 14221216,
      "step": 21750
    },
    {
      "epoch": 11.401991614255765,
      "grad_norm": 0.23499618470668793,
      "learning_rate": 2.3249700611978847e-05,
      "loss": 0.4994,
      "num_input_tokens_seen": 14224000,
      "step": 21755
    },
    {
      "epoch": 11.40461215932914,
      "grad_norm": 0.1386759728193283,
      "learning_rate": 2.323829456852519e-05,
      "loss": 0.4787,
      "num_input_tokens_seen": 14227872,
      "step": 21760
    },
    {
      "epoch": 11.407232704402515,
      "grad_norm": 0.15352123975753784,
      "learning_rate": 2.3226888893600223e-05,
      "loss": 0.5662,
      "num_input_tokens_seen": 14231936,
      "step": 21765
    },
    {
      "epoch": 11.40985324947589,
      "grad_norm": 0.16575725376605988,
      "learning_rate": 2.3215483589589865e-05,
      "loss": 0.3649,
      "num_input_tokens_seen": 14235328,
      "step": 21770
    },
    {
      "epoch": 11.412473794549266,
      "grad_norm": 0.25801241397857666,
      "learning_rate": 2.3204078658879974e-05,
      "loss": 0.4547,
      "num_input_tokens_seen": 14238848,
      "step": 21775
    },
    {
      "epoch": 11.415094339622641,
      "grad_norm": 0.1658267378807068,
      "learning_rate": 2.3192674103856337e-05,
      "loss": 0.5922,
      "num_input_tokens_seen": 14242336,
      "step": 21780
    },
    {
      "epoch": 11.417714884696016,
      "grad_norm": 0.26935818791389465,
      "learning_rate": 2.318126992690467e-05,
      "loss": 0.5212,
      "num_input_tokens_seen": 14245728,
      "step": 21785
    },
    {
      "epoch": 11.420335429769391,
      "grad_norm": 0.18367771804332733,
      "learning_rate": 2.3169866130410585e-05,
      "loss": 0.3525,
      "num_input_tokens_seen": 14248448,
      "step": 21790
    },
    {
      "epoch": 11.422955974842766,
      "grad_norm": 0.09673085063695908,
      "learning_rate": 2.315846271675962e-05,
      "loss": 0.4569,
      "num_input_tokens_seen": 14251648,
      "step": 21795
    },
    {
      "epoch": 11.425576519916142,
      "grad_norm": 0.14490745961666107,
      "learning_rate": 2.3147059688337244e-05,
      "loss": 0.3762,
      "num_input_tokens_seen": 14254624,
      "step": 21800
    },
    {
      "epoch": 11.428197064989519,
      "grad_norm": 0.21006865799427032,
      "learning_rate": 2.313565704752884e-05,
      "loss": 0.5024,
      "num_input_tokens_seen": 14257728,
      "step": 21805
    },
    {
      "epoch": 11.430817610062894,
      "grad_norm": 0.1132911890745163,
      "learning_rate": 2.3124254796719718e-05,
      "loss": 0.5495,
      "num_input_tokens_seen": 14262176,
      "step": 21810
    },
    {
      "epoch": 11.433438155136269,
      "grad_norm": 0.20898111164569855,
      "learning_rate": 2.3112852938295088e-05,
      "loss": 0.4584,
      "num_input_tokens_seen": 14265760,
      "step": 21815
    },
    {
      "epoch": 11.436058700209644,
      "grad_norm": 0.2917116582393646,
      "learning_rate": 2.3101451474640104e-05,
      "loss": 0.5409,
      "num_input_tokens_seen": 14268768,
      "step": 21820
    },
    {
      "epoch": 11.43867924528302,
      "grad_norm": 0.1291864812374115,
      "learning_rate": 2.3090050408139808e-05,
      "loss": 0.5896,
      "num_input_tokens_seen": 14272064,
      "step": 21825
    },
    {
      "epoch": 11.441299790356394,
      "grad_norm": 0.1279425472021103,
      "learning_rate": 2.307864974117917e-05,
      "loss": 0.5086,
      "num_input_tokens_seen": 14275456,
      "step": 21830
    },
    {
      "epoch": 11.44392033542977,
      "grad_norm": 0.15651635825634003,
      "learning_rate": 2.3067249476143087e-05,
      "loss": 0.5387,
      "num_input_tokens_seen": 14279712,
      "step": 21835
    },
    {
      "epoch": 11.446540880503145,
      "grad_norm": 0.1536741852760315,
      "learning_rate": 2.3055849615416363e-05,
      "loss": 0.4665,
      "num_input_tokens_seen": 14283616,
      "step": 21840
    },
    {
      "epoch": 11.44916142557652,
      "grad_norm": 0.18757760524749756,
      "learning_rate": 2.304445016138372e-05,
      "loss": 0.3419,
      "num_input_tokens_seen": 14286720,
      "step": 21845
    },
    {
      "epoch": 11.451781970649895,
      "grad_norm": 0.2506685256958008,
      "learning_rate": 2.303305111642981e-05,
      "loss": 0.4736,
      "num_input_tokens_seen": 14291232,
      "step": 21850
    },
    {
      "epoch": 11.45440251572327,
      "grad_norm": 0.2352968156337738,
      "learning_rate": 2.3021652482939144e-05,
      "loss": 0.508,
      "num_input_tokens_seen": 14294496,
      "step": 21855
    },
    {
      "epoch": 11.457023060796645,
      "grad_norm": 0.10156344622373581,
      "learning_rate": 2.3010254263296214e-05,
      "loss": 0.5025,
      "num_input_tokens_seen": 14297088,
      "step": 21860
    },
    {
      "epoch": 11.45964360587002,
      "grad_norm": 0.19527706503868103,
      "learning_rate": 2.2998856459885396e-05,
      "loss": 0.407,
      "num_input_tokens_seen": 14301056,
      "step": 21865
    },
    {
      "epoch": 11.462264150943396,
      "grad_norm": 0.1598333865404129,
      "learning_rate": 2.298745907509097e-05,
      "loss": 0.4081,
      "num_input_tokens_seen": 14303968,
      "step": 21870
    },
    {
      "epoch": 11.464884696016771,
      "grad_norm": 0.10005700588226318,
      "learning_rate": 2.2976062111297156e-05,
      "loss": 0.4912,
      "num_input_tokens_seen": 14308256,
      "step": 21875
    },
    {
      "epoch": 11.467505241090146,
      "grad_norm": 0.14017972350120544,
      "learning_rate": 2.296466557088805e-05,
      "loss": 0.609,
      "num_input_tokens_seen": 14311424,
      "step": 21880
    },
    {
      "epoch": 11.470125786163521,
      "grad_norm": 0.17760933935642242,
      "learning_rate": 2.2953269456247688e-05,
      "loss": 0.5068,
      "num_input_tokens_seen": 14314752,
      "step": 21885
    },
    {
      "epoch": 11.472746331236896,
      "grad_norm": 0.11269520223140717,
      "learning_rate": 2.2941873769760006e-05,
      "loss": 0.4552,
      "num_input_tokens_seen": 14317760,
      "step": 21890
    },
    {
      "epoch": 11.475366876310272,
      "grad_norm": 0.20692786574363708,
      "learning_rate": 2.2930478513808844e-05,
      "loss": 0.6207,
      "num_input_tokens_seen": 14321440,
      "step": 21895
    },
    {
      "epoch": 11.477987421383649,
      "grad_norm": 0.23165671527385712,
      "learning_rate": 2.2919083690777976e-05,
      "loss": 0.5835,
      "num_input_tokens_seen": 14324512,
      "step": 21900
    },
    {
      "epoch": 11.480607966457024,
      "grad_norm": 0.17016170918941498,
      "learning_rate": 2.2907689303051062e-05,
      "loss": 0.5759,
      "num_input_tokens_seen": 14327648,
      "step": 21905
    },
    {
      "epoch": 11.483228511530399,
      "grad_norm": 0.13630621135234833,
      "learning_rate": 2.289629535301167e-05,
      "loss": 0.6232,
      "num_input_tokens_seen": 14331040,
      "step": 21910
    },
    {
      "epoch": 11.485849056603774,
      "grad_norm": 0.15520672500133514,
      "learning_rate": 2.288490184304329e-05,
      "loss": 0.4526,
      "num_input_tokens_seen": 14333920,
      "step": 21915
    },
    {
      "epoch": 11.48846960167715,
      "grad_norm": 0.21172456443309784,
      "learning_rate": 2.2873508775529308e-05,
      "loss": 0.5114,
      "num_input_tokens_seen": 14337056,
      "step": 21920
    },
    {
      "epoch": 11.491090146750524,
      "grad_norm": 0.15883320569992065,
      "learning_rate": 2.2862116152853034e-05,
      "loss": 0.3697,
      "num_input_tokens_seen": 14339584,
      "step": 21925
    },
    {
      "epoch": 11.4937106918239,
      "grad_norm": 0.10326235741376877,
      "learning_rate": 2.285072397739767e-05,
      "loss": 0.3964,
      "num_input_tokens_seen": 14345024,
      "step": 21930
    },
    {
      "epoch": 11.496331236897275,
      "grad_norm": 0.13121376931667328,
      "learning_rate": 2.283933225154634e-05,
      "loss": 0.4594,
      "num_input_tokens_seen": 14348288,
      "step": 21935
    },
    {
      "epoch": 11.49895178197065,
      "grad_norm": 0.4461466372013092,
      "learning_rate": 2.2827940977682045e-05,
      "loss": 0.6282,
      "num_input_tokens_seen": 14352000,
      "step": 21940
    },
    {
      "epoch": 11.501572327044025,
      "grad_norm": 0.16442346572875977,
      "learning_rate": 2.2816550158187717e-05,
      "loss": 0.4079,
      "num_input_tokens_seen": 14354336,
      "step": 21945
    },
    {
      "epoch": 11.5041928721174,
      "grad_norm": 0.15990789234638214,
      "learning_rate": 2.2805159795446182e-05,
      "loss": 0.61,
      "num_input_tokens_seen": 14357728,
      "step": 21950
    },
    {
      "epoch": 11.506813417190775,
      "grad_norm": 0.1804056465625763,
      "learning_rate": 2.2793769891840184e-05,
      "loss": 0.5011,
      "num_input_tokens_seen": 14360384,
      "step": 21955
    },
    {
      "epoch": 11.50943396226415,
      "grad_norm": 0.12268757820129395,
      "learning_rate": 2.278238044975236e-05,
      "loss": 0.5114,
      "num_input_tokens_seen": 14364800,
      "step": 21960
    },
    {
      "epoch": 11.512054507337526,
      "grad_norm": 0.2330923229455948,
      "learning_rate": 2.2770991471565236e-05,
      "loss": 0.3431,
      "num_input_tokens_seen": 14367872,
      "step": 21965
    },
    {
      "epoch": 11.514675052410901,
      "grad_norm": 0.10054771602153778,
      "learning_rate": 2.2759602959661265e-05,
      "loss": 0.5322,
      "num_input_tokens_seen": 14371872,
      "step": 21970
    },
    {
      "epoch": 11.517295597484276,
      "grad_norm": 0.2719261050224304,
      "learning_rate": 2.2748214916422796e-05,
      "loss": 0.495,
      "num_input_tokens_seen": 14374752,
      "step": 21975
    },
    {
      "epoch": 11.519916142557651,
      "grad_norm": 0.6107059717178345,
      "learning_rate": 2.2736827344232082e-05,
      "loss": 0.5076,
      "num_input_tokens_seen": 14377792,
      "step": 21980
    },
    {
      "epoch": 11.522536687631026,
      "grad_norm": 0.09491610527038574,
      "learning_rate": 2.272544024547126e-05,
      "loss": 0.3602,
      "num_input_tokens_seen": 14382496,
      "step": 21985
    },
    {
      "epoch": 11.525157232704402,
      "grad_norm": 0.149775892496109,
      "learning_rate": 2.27140536225224e-05,
      "loss": 0.4593,
      "num_input_tokens_seen": 14389184,
      "step": 21990
    },
    {
      "epoch": 11.527777777777779,
      "grad_norm": 0.18003691732883453,
      "learning_rate": 2.270266747776744e-05,
      "loss": 0.4405,
      "num_input_tokens_seen": 14391904,
      "step": 21995
    },
    {
      "epoch": 11.530398322851154,
      "grad_norm": 0.12163138389587402,
      "learning_rate": 2.2691281813588226e-05,
      "loss": 0.435,
      "num_input_tokens_seen": 14394880,
      "step": 22000
    },
    {
      "epoch": 11.533018867924529,
      "grad_norm": 0.128933385014534,
      "learning_rate": 2.267989663236652e-05,
      "loss": 0.3934,
      "num_input_tokens_seen": 14399168,
      "step": 22005
    },
    {
      "epoch": 11.535639412997904,
      "grad_norm": 0.22455473244190216,
      "learning_rate": 2.2668511936483975e-05,
      "loss": 0.6597,
      "num_input_tokens_seen": 14402848,
      "step": 22010
    },
    {
      "epoch": 11.53825995807128,
      "grad_norm": 0.13822489976882935,
      "learning_rate": 2.2657127728322124e-05,
      "loss": 0.4478,
      "num_input_tokens_seen": 14406272,
      "step": 22015
    },
    {
      "epoch": 11.540880503144654,
      "grad_norm": 0.19622178375720978,
      "learning_rate": 2.264574401026244e-05,
      "loss": 0.445,
      "num_input_tokens_seen": 14409120,
      "step": 22020
    },
    {
      "epoch": 11.54350104821803,
      "grad_norm": 0.1423366516828537,
      "learning_rate": 2.2634360784686242e-05,
      "loss": 0.4146,
      "num_input_tokens_seen": 14412224,
      "step": 22025
    },
    {
      "epoch": 11.546121593291405,
      "grad_norm": 0.19467249512672424,
      "learning_rate": 2.2622978053974776e-05,
      "loss": 0.5531,
      "num_input_tokens_seen": 14415616,
      "step": 22030
    },
    {
      "epoch": 11.54874213836478,
      "grad_norm": 0.22653518617153168,
      "learning_rate": 2.2611595820509186e-05,
      "loss": 0.7365,
      "num_input_tokens_seen": 14419392,
      "step": 22035
    },
    {
      "epoch": 11.551362683438155,
      "grad_norm": 0.16694830358028412,
      "learning_rate": 2.2600214086670494e-05,
      "loss": 0.4357,
      "num_input_tokens_seen": 14423040,
      "step": 22040
    },
    {
      "epoch": 11.55398322851153,
      "grad_norm": 0.14956659078598022,
      "learning_rate": 2.2588832854839644e-05,
      "loss": 0.6025,
      "num_input_tokens_seen": 14426656,
      "step": 22045
    },
    {
      "epoch": 11.556603773584905,
      "grad_norm": 0.4134361147880554,
      "learning_rate": 2.2577452127397463e-05,
      "loss": 0.6459,
      "num_input_tokens_seen": 14429792,
      "step": 22050
    },
    {
      "epoch": 11.55922431865828,
      "grad_norm": 0.23053276538848877,
      "learning_rate": 2.256607190672464e-05,
      "loss": 0.4552,
      "num_input_tokens_seen": 14432992,
      "step": 22055
    },
    {
      "epoch": 11.561844863731656,
      "grad_norm": 0.1331683099269867,
      "learning_rate": 2.2554692195201817e-05,
      "loss": 0.504,
      "num_input_tokens_seen": 14436448,
      "step": 22060
    },
    {
      "epoch": 11.564465408805031,
      "grad_norm": 0.16059865057468414,
      "learning_rate": 2.254331299520948e-05,
      "loss": 0.5158,
      "num_input_tokens_seen": 14438880,
      "step": 22065
    },
    {
      "epoch": 11.567085953878406,
      "grad_norm": 0.21860885620117188,
      "learning_rate": 2.253193430912805e-05,
      "loss": 0.4827,
      "num_input_tokens_seen": 14441248,
      "step": 22070
    },
    {
      "epoch": 11.569706498951781,
      "grad_norm": 0.15501323342323303,
      "learning_rate": 2.2520556139337803e-05,
      "loss": 0.4345,
      "num_input_tokens_seen": 14444576,
      "step": 22075
    },
    {
      "epoch": 11.572327044025156,
      "grad_norm": 0.15238524973392487,
      "learning_rate": 2.250917848821892e-05,
      "loss": 0.455,
      "num_input_tokens_seen": 14447968,
      "step": 22080
    },
    {
      "epoch": 11.574947589098532,
      "grad_norm": 0.12270162254571915,
      "learning_rate": 2.2497801358151472e-05,
      "loss": 0.6581,
      "num_input_tokens_seen": 14450816,
      "step": 22085
    },
    {
      "epoch": 11.577568134171909,
      "grad_norm": 0.18273666501045227,
      "learning_rate": 2.248642475151545e-05,
      "loss": 0.3653,
      "num_input_tokens_seen": 14453728,
      "step": 22090
    },
    {
      "epoch": 11.580188679245284,
      "grad_norm": 0.1594831347465515,
      "learning_rate": 2.2475048670690678e-05,
      "loss": 0.6258,
      "num_input_tokens_seen": 14456512,
      "step": 22095
    },
    {
      "epoch": 11.582809224318659,
      "grad_norm": 0.15628185868263245,
      "learning_rate": 2.2463673118056927e-05,
      "loss": 0.6161,
      "num_input_tokens_seen": 14460160,
      "step": 22100
    },
    {
      "epoch": 11.585429769392034,
      "grad_norm": 0.13927863538265228,
      "learning_rate": 2.245229809599383e-05,
      "loss": 0.6457,
      "num_input_tokens_seen": 14462848,
      "step": 22105
    },
    {
      "epoch": 11.58805031446541,
      "grad_norm": 0.15575748682022095,
      "learning_rate": 2.2440923606880896e-05,
      "loss": 0.4378,
      "num_input_tokens_seen": 14466592,
      "step": 22110
    },
    {
      "epoch": 11.590670859538784,
      "grad_norm": 0.20957757532596588,
      "learning_rate": 2.242954965309755e-05,
      "loss": 0.4164,
      "num_input_tokens_seen": 14468960,
      "step": 22115
    },
    {
      "epoch": 11.59329140461216,
      "grad_norm": 0.16310149431228638,
      "learning_rate": 2.2418176237023088e-05,
      "loss": 0.418,
      "num_input_tokens_seen": 14472864,
      "step": 22120
    },
    {
      "epoch": 11.595911949685535,
      "grad_norm": 0.1325836479663849,
      "learning_rate": 2.2406803361036706e-05,
      "loss": 0.5192,
      "num_input_tokens_seen": 14476768,
      "step": 22125
    },
    {
      "epoch": 11.59853249475891,
      "grad_norm": 0.11760737746953964,
      "learning_rate": 2.2395431027517475e-05,
      "loss": 0.4319,
      "num_input_tokens_seen": 14479872,
      "step": 22130
    },
    {
      "epoch": 11.601153039832285,
      "grad_norm": 0.15132297575473785,
      "learning_rate": 2.2384059238844364e-05,
      "loss": 0.4953,
      "num_input_tokens_seen": 14482400,
      "step": 22135
    },
    {
      "epoch": 11.60377358490566,
      "grad_norm": 0.1814073622226715,
      "learning_rate": 2.2372687997396215e-05,
      "loss": 0.5241,
      "num_input_tokens_seen": 14485696,
      "step": 22140
    },
    {
      "epoch": 11.606394129979035,
      "grad_norm": 0.1340370625257492,
      "learning_rate": 2.2361317305551756e-05,
      "loss": 0.6161,
      "num_input_tokens_seen": 14488800,
      "step": 22145
    },
    {
      "epoch": 11.60901467505241,
      "grad_norm": 0.12955999374389648,
      "learning_rate": 2.2349947165689615e-05,
      "loss": 0.4463,
      "num_input_tokens_seen": 14491520,
      "step": 22150
    },
    {
      "epoch": 11.611635220125786,
      "grad_norm": 0.22342528402805328,
      "learning_rate": 2.2338577580188296e-05,
      "loss": 0.4106,
      "num_input_tokens_seen": 14494176,
      "step": 22155
    },
    {
      "epoch": 11.614255765199161,
      "grad_norm": 0.25212591886520386,
      "learning_rate": 2.2327208551426188e-05,
      "loss": 0.5329,
      "num_input_tokens_seen": 14496640,
      "step": 22160
    },
    {
      "epoch": 11.616876310272536,
      "grad_norm": 0.25895926356315613,
      "learning_rate": 2.2315840081781546e-05,
      "loss": 0.4636,
      "num_input_tokens_seen": 14499232,
      "step": 22165
    },
    {
      "epoch": 11.619496855345911,
      "grad_norm": 0.26660069823265076,
      "learning_rate": 2.230447217363255e-05,
      "loss": 0.5351,
      "num_input_tokens_seen": 14501600,
      "step": 22170
    },
    {
      "epoch": 11.622117400419286,
      "grad_norm": 0.1258116215467453,
      "learning_rate": 2.2293104829357212e-05,
      "loss": 0.4729,
      "num_input_tokens_seen": 14504544,
      "step": 22175
    },
    {
      "epoch": 11.624737945492662,
      "grad_norm": 0.3033285140991211,
      "learning_rate": 2.2281738051333467e-05,
      "loss": 0.4038,
      "num_input_tokens_seen": 14507904,
      "step": 22180
    },
    {
      "epoch": 11.627358490566039,
      "grad_norm": 0.10187256336212158,
      "learning_rate": 2.2270371841939105e-05,
      "loss": 0.451,
      "num_input_tokens_seen": 14511264,
      "step": 22185
    },
    {
      "epoch": 11.629979035639414,
      "grad_norm": 0.12135638296604156,
      "learning_rate": 2.2259006203551824e-05,
      "loss": 0.424,
      "num_input_tokens_seen": 14514752,
      "step": 22190
    },
    {
      "epoch": 11.632599580712789,
      "grad_norm": 0.12043897807598114,
      "learning_rate": 2.2247641138549168e-05,
      "loss": 0.4769,
      "num_input_tokens_seen": 14517312,
      "step": 22195
    },
    {
      "epoch": 11.635220125786164,
      "grad_norm": 0.1384189873933792,
      "learning_rate": 2.2236276649308583e-05,
      "loss": 0.4866,
      "num_input_tokens_seen": 14520800,
      "step": 22200
    },
    {
      "epoch": 11.63784067085954,
      "grad_norm": 0.14493156969547272,
      "learning_rate": 2.2224912738207395e-05,
      "loss": 0.4106,
      "num_input_tokens_seen": 14524000,
      "step": 22205
    },
    {
      "epoch": 11.640461215932914,
      "grad_norm": 0.08453737199306488,
      "learning_rate": 2.2213549407622796e-05,
      "loss": 0.5165,
      "num_input_tokens_seen": 14526944,
      "step": 22210
    },
    {
      "epoch": 11.64308176100629,
      "grad_norm": 0.14615784585475922,
      "learning_rate": 2.220218665993188e-05,
      "loss": 0.3595,
      "num_input_tokens_seen": 14530368,
      "step": 22215
    },
    {
      "epoch": 11.645702306079665,
      "grad_norm": 0.16467097401618958,
      "learning_rate": 2.21908244975116e-05,
      "loss": 0.4176,
      "num_input_tokens_seen": 14533824,
      "step": 22220
    },
    {
      "epoch": 11.64832285115304,
      "grad_norm": 0.29182881116867065,
      "learning_rate": 2.2179462922738776e-05,
      "loss": 0.4147,
      "num_input_tokens_seen": 14539232,
      "step": 22225
    },
    {
      "epoch": 11.650943396226415,
      "grad_norm": 0.1229160875082016,
      "learning_rate": 2.2168101937990135e-05,
      "loss": 0.4214,
      "num_input_tokens_seen": 14542784,
      "step": 22230
    },
    {
      "epoch": 11.65356394129979,
      "grad_norm": 0.18504326045513153,
      "learning_rate": 2.2156741545642253e-05,
      "loss": 0.4831,
      "num_input_tokens_seen": 14545696,
      "step": 22235
    },
    {
      "epoch": 11.656184486373165,
      "grad_norm": 0.15312336385250092,
      "learning_rate": 2.2145381748071606e-05,
      "loss": 0.3061,
      "num_input_tokens_seen": 14549344,
      "step": 22240
    },
    {
      "epoch": 11.65880503144654,
      "grad_norm": 0.15376685559749603,
      "learning_rate": 2.2134022547654538e-05,
      "loss": 0.4768,
      "num_input_tokens_seen": 14552096,
      "step": 22245
    },
    {
      "epoch": 11.661425576519916,
      "grad_norm": 0.15389664471149445,
      "learning_rate": 2.212266394676725e-05,
      "loss": 0.3668,
      "num_input_tokens_seen": 14556384,
      "step": 22250
    },
    {
      "epoch": 11.664046121593291,
      "grad_norm": 0.2193564623594284,
      "learning_rate": 2.2111305947785836e-05,
      "loss": 0.461,
      "num_input_tokens_seen": 14559680,
      "step": 22255
    },
    {
      "epoch": 11.666666666666666,
      "grad_norm": 0.13149169087409973,
      "learning_rate": 2.2099948553086263e-05,
      "loss": 0.4053,
      "num_input_tokens_seen": 14562592,
      "step": 22260
    },
    {
      "epoch": 11.669287211740041,
      "grad_norm": 0.1791049838066101,
      "learning_rate": 2.2088591765044365e-05,
      "loss": 0.5629,
      "num_input_tokens_seen": 14564960,
      "step": 22265
    },
    {
      "epoch": 11.671907756813416,
      "grad_norm": 0.16253262758255005,
      "learning_rate": 2.207723558603586e-05,
      "loss": 0.5231,
      "num_input_tokens_seen": 14568384,
      "step": 22270
    },
    {
      "epoch": 11.674528301886792,
      "grad_norm": 0.15644998848438263,
      "learning_rate": 2.206588001843633e-05,
      "loss": 0.4557,
      "num_input_tokens_seen": 14571680,
      "step": 22275
    },
    {
      "epoch": 11.677148846960169,
      "grad_norm": 0.318962037563324,
      "learning_rate": 2.2054525064621216e-05,
      "loss": 0.5063,
      "num_input_tokens_seen": 14575136,
      "step": 22280
    },
    {
      "epoch": 11.679769392033544,
      "grad_norm": 0.16652970016002655,
      "learning_rate": 2.2043170726965858e-05,
      "loss": 0.4379,
      "num_input_tokens_seen": 14577888,
      "step": 22285
    },
    {
      "epoch": 11.682389937106919,
      "grad_norm": 0.11927176266908646,
      "learning_rate": 2.2031817007845447e-05,
      "loss": 0.4473,
      "num_input_tokens_seen": 14581248,
      "step": 22290
    },
    {
      "epoch": 11.685010482180294,
      "grad_norm": 0.21284963190555573,
      "learning_rate": 2.202046390963506e-05,
      "loss": 0.5388,
      "num_input_tokens_seen": 14584928,
      "step": 22295
    },
    {
      "epoch": 11.68763102725367,
      "grad_norm": 0.37909138202667236,
      "learning_rate": 2.2009111434709627e-05,
      "loss": 0.6434,
      "num_input_tokens_seen": 14587552,
      "step": 22300
    },
    {
      "epoch": 11.690251572327044,
      "grad_norm": 0.24156992137432098,
      "learning_rate": 2.1997759585443968e-05,
      "loss": 0.4713,
      "num_input_tokens_seen": 14589760,
      "step": 22305
    },
    {
      "epoch": 11.69287211740042,
      "grad_norm": 0.21964874863624573,
      "learning_rate": 2.198640836421275e-05,
      "loss": 0.5508,
      "num_input_tokens_seen": 14592448,
      "step": 22310
    },
    {
      "epoch": 11.695492662473795,
      "grad_norm": 0.15352916717529297,
      "learning_rate": 2.1975057773390515e-05,
      "loss": 0.4779,
      "num_input_tokens_seen": 14595808,
      "step": 22315
    },
    {
      "epoch": 11.69811320754717,
      "grad_norm": 0.1335953176021576,
      "learning_rate": 2.1963707815351683e-05,
      "loss": 0.4409,
      "num_input_tokens_seen": 14598624,
      "step": 22320
    },
    {
      "epoch": 11.700733752620545,
      "grad_norm": 0.35864707827568054,
      "learning_rate": 2.1952358492470544e-05,
      "loss": 0.437,
      "num_input_tokens_seen": 14601536,
      "step": 22325
    },
    {
      "epoch": 11.70335429769392,
      "grad_norm": 0.12784892320632935,
      "learning_rate": 2.194100980712124e-05,
      "loss": 0.4707,
      "num_input_tokens_seen": 14604736,
      "step": 22330
    },
    {
      "epoch": 11.705974842767295,
      "grad_norm": 0.21845759451389313,
      "learning_rate": 2.1929661761677774e-05,
      "loss": 0.418,
      "num_input_tokens_seen": 14607840,
      "step": 22335
    },
    {
      "epoch": 11.70859538784067,
      "grad_norm": 0.15742729604244232,
      "learning_rate": 2.191831435851405e-05,
      "loss": 0.5494,
      "num_input_tokens_seen": 14610816,
      "step": 22340
    },
    {
      "epoch": 11.711215932914046,
      "grad_norm": 0.23561349511146545,
      "learning_rate": 2.1906967600003793e-05,
      "loss": 0.5466,
      "num_input_tokens_seen": 14614464,
      "step": 22345
    },
    {
      "epoch": 11.713836477987421,
      "grad_norm": 0.15136197209358215,
      "learning_rate": 2.1895621488520636e-05,
      "loss": 0.417,
      "num_input_tokens_seen": 14616704,
      "step": 22350
    },
    {
      "epoch": 11.716457023060796,
      "grad_norm": 0.10239632427692413,
      "learning_rate": 2.1884276026438044e-05,
      "loss": 0.502,
      "num_input_tokens_seen": 14620832,
      "step": 22355
    },
    {
      "epoch": 11.719077568134171,
      "grad_norm": 0.14237911999225616,
      "learning_rate": 2.187293121612937e-05,
      "loss": 0.5198,
      "num_input_tokens_seen": 14623328,
      "step": 22360
    },
    {
      "epoch": 11.721698113207546,
      "grad_norm": 0.13126392662525177,
      "learning_rate": 2.1861587059967804e-05,
      "loss": 0.4953,
      "num_input_tokens_seen": 14627520,
      "step": 22365
    },
    {
      "epoch": 11.724318658280922,
      "grad_norm": 0.23275960981845856,
      "learning_rate": 2.1850243560326414e-05,
      "loss": 0.4275,
      "num_input_tokens_seen": 14631712,
      "step": 22370
    },
    {
      "epoch": 11.726939203354299,
      "grad_norm": 0.13444215059280396,
      "learning_rate": 2.1838900719578143e-05,
      "loss": 0.4862,
      "num_input_tokens_seen": 14634912,
      "step": 22375
    },
    {
      "epoch": 11.729559748427674,
      "grad_norm": 0.18043385446071625,
      "learning_rate": 2.1827558540095772e-05,
      "loss": 0.5378,
      "num_input_tokens_seen": 14637440,
      "step": 22380
    },
    {
      "epoch": 11.732180293501049,
      "grad_norm": 0.07126526534557343,
      "learning_rate": 2.1816217024251966e-05,
      "loss": 0.495,
      "num_input_tokens_seen": 14642336,
      "step": 22385
    },
    {
      "epoch": 11.734800838574424,
      "grad_norm": 0.25570669770240784,
      "learning_rate": 2.1804876174419242e-05,
      "loss": 0.5155,
      "num_input_tokens_seen": 14646016,
      "step": 22390
    },
    {
      "epoch": 11.7374213836478,
      "grad_norm": 0.15631555020809174,
      "learning_rate": 2.179353599296996e-05,
      "loss": 0.4113,
      "num_input_tokens_seen": 14648384,
      "step": 22395
    },
    {
      "epoch": 11.740041928721174,
      "grad_norm": 0.1658967286348343,
      "learning_rate": 2.1782196482276368e-05,
      "loss": 0.5519,
      "num_input_tokens_seen": 14651552,
      "step": 22400
    },
    {
      "epoch": 11.74266247379455,
      "grad_norm": 0.12620548903942108,
      "learning_rate": 2.1770857644710567e-05,
      "loss": 0.5763,
      "num_input_tokens_seen": 14654304,
      "step": 22405
    },
    {
      "epoch": 11.745283018867925,
      "grad_norm": 0.23069600760936737,
      "learning_rate": 2.1759519482644495e-05,
      "loss": 0.4773,
      "num_input_tokens_seen": 14657472,
      "step": 22410
    },
    {
      "epoch": 11.7479035639413,
      "grad_norm": 0.14536269009113312,
      "learning_rate": 2.174818199844999e-05,
      "loss": 0.6283,
      "num_input_tokens_seen": 14660992,
      "step": 22415
    },
    {
      "epoch": 11.750524109014675,
      "grad_norm": 0.11084678024053574,
      "learning_rate": 2.173684519449872e-05,
      "loss": 0.4868,
      "num_input_tokens_seen": 14665344,
      "step": 22420
    },
    {
      "epoch": 11.75314465408805,
      "grad_norm": 0.20842356979846954,
      "learning_rate": 2.1725509073162188e-05,
      "loss": 0.5501,
      "num_input_tokens_seen": 14668384,
      "step": 22425
    },
    {
      "epoch": 11.755765199161425,
      "grad_norm": 0.171134352684021,
      "learning_rate": 2.171417363681181e-05,
      "loss": 0.5694,
      "num_input_tokens_seen": 14671488,
      "step": 22430
    },
    {
      "epoch": 11.7583857442348,
      "grad_norm": 0.2520579397678375,
      "learning_rate": 2.1702838887818816e-05,
      "loss": 0.5537,
      "num_input_tokens_seen": 14674592,
      "step": 22435
    },
    {
      "epoch": 11.761006289308176,
      "grad_norm": 0.1504664272069931,
      "learning_rate": 2.1691504828554315e-05,
      "loss": 0.5811,
      "num_input_tokens_seen": 14678176,
      "step": 22440
    },
    {
      "epoch": 11.76362683438155,
      "grad_norm": 0.2868511974811554,
      "learning_rate": 2.1680171461389265e-05,
      "loss": 0.4156,
      "num_input_tokens_seen": 14681184,
      "step": 22445
    },
    {
      "epoch": 11.766247379454926,
      "grad_norm": 0.09194225817918777,
      "learning_rate": 2.166883878869446e-05,
      "loss": 0.3774,
      "num_input_tokens_seen": 14684192,
      "step": 22450
    },
    {
      "epoch": 11.768867924528301,
      "grad_norm": 0.1520577371120453,
      "learning_rate": 2.1657506812840582e-05,
      "loss": 0.4623,
      "num_input_tokens_seen": 14687040,
      "step": 22455
    },
    {
      "epoch": 11.771488469601676,
      "grad_norm": 0.12114858627319336,
      "learning_rate": 2.164617553619814e-05,
      "loss": 0.6084,
      "num_input_tokens_seen": 14690592,
      "step": 22460
    },
    {
      "epoch": 11.774109014675052,
      "grad_norm": 0.17078810930252075,
      "learning_rate": 2.1634844961137517e-05,
      "loss": 0.4472,
      "num_input_tokens_seen": 14693792,
      "step": 22465
    },
    {
      "epoch": 11.776729559748428,
      "grad_norm": 0.2833832800388336,
      "learning_rate": 2.162351509002893e-05,
      "loss": 0.6751,
      "num_input_tokens_seen": 14696544,
      "step": 22470
    },
    {
      "epoch": 11.779350104821804,
      "grad_norm": 0.14143456518650055,
      "learning_rate": 2.161218592524248e-05,
      "loss": 0.4968,
      "num_input_tokens_seen": 14700416,
      "step": 22475
    },
    {
      "epoch": 11.781970649895179,
      "grad_norm": 0.1668720543384552,
      "learning_rate": 2.160085746914807e-05,
      "loss": 0.5837,
      "num_input_tokens_seen": 14703648,
      "step": 22480
    },
    {
      "epoch": 11.784591194968554,
      "grad_norm": 0.2227289378643036,
      "learning_rate": 2.1589529724115503e-05,
      "loss": 0.4441,
      "num_input_tokens_seen": 14706304,
      "step": 22485
    },
    {
      "epoch": 11.78721174004193,
      "grad_norm": 0.16521139442920685,
      "learning_rate": 2.1578202692514407e-05,
      "loss": 0.5126,
      "num_input_tokens_seen": 14708672,
      "step": 22490
    },
    {
      "epoch": 11.789832285115304,
      "grad_norm": 0.1923743635416031,
      "learning_rate": 2.1566876376714274e-05,
      "loss": 0.6306,
      "num_input_tokens_seen": 14711904,
      "step": 22495
    },
    {
      "epoch": 11.79245283018868,
      "grad_norm": 0.41844189167022705,
      "learning_rate": 2.155555077908443e-05,
      "loss": 0.5223,
      "num_input_tokens_seen": 14714080,
      "step": 22500
    },
    {
      "epoch": 11.795073375262055,
      "grad_norm": 0.24710392951965332,
      "learning_rate": 2.1544225901994077e-05,
      "loss": 0.3901,
      "num_input_tokens_seen": 14717152,
      "step": 22505
    },
    {
      "epoch": 11.79769392033543,
      "grad_norm": 0.16235949099063873,
      "learning_rate": 2.1532901747812234e-05,
      "loss": 0.4204,
      "num_input_tokens_seen": 14720064,
      "step": 22510
    },
    {
      "epoch": 11.800314465408805,
      "grad_norm": 0.1784198135137558,
      "learning_rate": 2.1521578318907787e-05,
      "loss": 0.5526,
      "num_input_tokens_seen": 14722848,
      "step": 22515
    },
    {
      "epoch": 11.80293501048218,
      "grad_norm": 0.18567579984664917,
      "learning_rate": 2.1510255617649476e-05,
      "loss": 0.398,
      "num_input_tokens_seen": 14725696,
      "step": 22520
    },
    {
      "epoch": 11.805555555555555,
      "grad_norm": 0.15196038782596588,
      "learning_rate": 2.149893364640587e-05,
      "loss": 0.4721,
      "num_input_tokens_seen": 14728800,
      "step": 22525
    },
    {
      "epoch": 11.80817610062893,
      "grad_norm": 0.1137351468205452,
      "learning_rate": 2.1487612407545417e-05,
      "loss": 0.368,
      "num_input_tokens_seen": 14732608,
      "step": 22530
    },
    {
      "epoch": 11.810796645702306,
      "grad_norm": 0.1267421692609787,
      "learning_rate": 2.147629190343637e-05,
      "loss": 0.4673,
      "num_input_tokens_seen": 14736640,
      "step": 22535
    },
    {
      "epoch": 11.81341719077568,
      "grad_norm": 0.17372453212738037,
      "learning_rate": 2.1464972136446852e-05,
      "loss": 0.3917,
      "num_input_tokens_seen": 14739296,
      "step": 22540
    },
    {
      "epoch": 11.816037735849056,
      "grad_norm": 0.1669887751340866,
      "learning_rate": 2.1453653108944838e-05,
      "loss": 0.3695,
      "num_input_tokens_seen": 14742048,
      "step": 22545
    },
    {
      "epoch": 11.818658280922431,
      "grad_norm": 0.2862309217453003,
      "learning_rate": 2.1442334823298128e-05,
      "loss": 0.4201,
      "num_input_tokens_seen": 14745056,
      "step": 22550
    },
    {
      "epoch": 11.821278825995806,
      "grad_norm": 0.15484197437763214,
      "learning_rate": 2.1431017281874386e-05,
      "loss": 0.6481,
      "num_input_tokens_seen": 14748064,
      "step": 22555
    },
    {
      "epoch": 11.823899371069182,
      "grad_norm": 0.21453934907913208,
      "learning_rate": 2.1419700487041123e-05,
      "loss": 0.5446,
      "num_input_tokens_seen": 14751264,
      "step": 22560
    },
    {
      "epoch": 11.826519916142558,
      "grad_norm": 0.1255691945552826,
      "learning_rate": 2.140838444116566e-05,
      "loss": 0.5826,
      "num_input_tokens_seen": 14755168,
      "step": 22565
    },
    {
      "epoch": 11.829140461215934,
      "grad_norm": 0.343636155128479,
      "learning_rate": 2.13970691466152e-05,
      "loss": 0.6588,
      "num_input_tokens_seen": 14757408,
      "step": 22570
    },
    {
      "epoch": 11.831761006289309,
      "grad_norm": 0.13244694471359253,
      "learning_rate": 2.1385754605756773e-05,
      "loss": 0.518,
      "num_input_tokens_seen": 14760288,
      "step": 22575
    },
    {
      "epoch": 11.834381551362684,
      "grad_norm": 0.1750529259443283,
      "learning_rate": 2.137444082095724e-05,
      "loss": 0.495,
      "num_input_tokens_seen": 14763616,
      "step": 22580
    },
    {
      "epoch": 11.83700209643606,
      "grad_norm": 0.20517563819885254,
      "learning_rate": 2.1363127794583335e-05,
      "loss": 0.4188,
      "num_input_tokens_seen": 14768544,
      "step": 22585
    },
    {
      "epoch": 11.839622641509434,
      "grad_norm": 0.14927485585212708,
      "learning_rate": 2.1351815529001605e-05,
      "loss": 0.5682,
      "num_input_tokens_seen": 14771456,
      "step": 22590
    },
    {
      "epoch": 11.84224318658281,
      "grad_norm": 0.12037352472543716,
      "learning_rate": 2.1340504026578438e-05,
      "loss": 0.583,
      "num_input_tokens_seen": 14774880,
      "step": 22595
    },
    {
      "epoch": 11.844863731656185,
      "grad_norm": 0.12237898260354996,
      "learning_rate": 2.1329193289680084e-05,
      "loss": 0.46,
      "num_input_tokens_seen": 14778432,
      "step": 22600
    },
    {
      "epoch": 11.84748427672956,
      "grad_norm": 0.17160014808177948,
      "learning_rate": 2.1317883320672608e-05,
      "loss": 0.4737,
      "num_input_tokens_seen": 14781984,
      "step": 22605
    },
    {
      "epoch": 11.850104821802935,
      "grad_norm": 0.2155626118183136,
      "learning_rate": 2.1306574121921942e-05,
      "loss": 0.584,
      "num_input_tokens_seen": 14785280,
      "step": 22610
    },
    {
      "epoch": 11.85272536687631,
      "grad_norm": 0.16833671927452087,
      "learning_rate": 2.1295265695793836e-05,
      "loss": 0.5531,
      "num_input_tokens_seen": 14789568,
      "step": 22615
    },
    {
      "epoch": 11.855345911949685,
      "grad_norm": 0.2232862263917923,
      "learning_rate": 2.128395804465387e-05,
      "loss": 0.3779,
      "num_input_tokens_seen": 14792608,
      "step": 22620
    },
    {
      "epoch": 11.85796645702306,
      "grad_norm": 0.035930462181568146,
      "learning_rate": 2.1272651170867487e-05,
      "loss": 0.3511,
      "num_input_tokens_seen": 14797760,
      "step": 22625
    },
    {
      "epoch": 11.860587002096436,
      "grad_norm": 0.19212792813777924,
      "learning_rate": 2.1261345076799955e-05,
      "loss": 0.5238,
      "num_input_tokens_seen": 14800768,
      "step": 22630
    },
    {
      "epoch": 11.86320754716981,
      "grad_norm": 0.11024747043848038,
      "learning_rate": 2.125003976481638e-05,
      "loss": 0.4429,
      "num_input_tokens_seen": 14804000,
      "step": 22635
    },
    {
      "epoch": 11.865828092243186,
      "grad_norm": 0.15612688660621643,
      "learning_rate": 2.123873523728171e-05,
      "loss": 0.347,
      "num_input_tokens_seen": 14807392,
      "step": 22640
    },
    {
      "epoch": 11.868448637316561,
      "grad_norm": 0.13038018345832825,
      "learning_rate": 2.122743149656072e-05,
      "loss": 0.4491,
      "num_input_tokens_seen": 14810816,
      "step": 22645
    },
    {
      "epoch": 11.871069182389936,
      "grad_norm": 0.25065165758132935,
      "learning_rate": 2.1216128545018012e-05,
      "loss": 0.4629,
      "num_input_tokens_seen": 14813056,
      "step": 22650
    },
    {
      "epoch": 11.873689727463312,
      "grad_norm": 0.10231825709342957,
      "learning_rate": 2.1204826385018052e-05,
      "loss": 0.5164,
      "num_input_tokens_seen": 14816736,
      "step": 22655
    },
    {
      "epoch": 11.876310272536688,
      "grad_norm": 0.12867975234985352,
      "learning_rate": 2.1193525018925106e-05,
      "loss": 0.559,
      "num_input_tokens_seen": 14820160,
      "step": 22660
    },
    {
      "epoch": 11.878930817610064,
      "grad_norm": 0.14494048058986664,
      "learning_rate": 2.1182224449103304e-05,
      "loss": 0.4954,
      "num_input_tokens_seen": 14822944,
      "step": 22665
    },
    {
      "epoch": 11.881551362683439,
      "grad_norm": 0.16312329471111298,
      "learning_rate": 2.117092467791659e-05,
      "loss": 0.4502,
      "num_input_tokens_seen": 14826464,
      "step": 22670
    },
    {
      "epoch": 11.884171907756814,
      "grad_norm": 0.3473546504974365,
      "learning_rate": 2.115962570772876e-05,
      "loss": 0.4538,
      "num_input_tokens_seen": 14830080,
      "step": 22675
    },
    {
      "epoch": 11.88679245283019,
      "grad_norm": 0.12514525651931763,
      "learning_rate": 2.1148327540903408e-05,
      "loss": 0.5227,
      "num_input_tokens_seen": 14833280,
      "step": 22680
    },
    {
      "epoch": 11.889412997903564,
      "grad_norm": 0.14657741785049438,
      "learning_rate": 2.113703017980399e-05,
      "loss": 0.5913,
      "num_input_tokens_seen": 14836768,
      "step": 22685
    },
    {
      "epoch": 11.89203354297694,
      "grad_norm": 0.19785329699516296,
      "learning_rate": 2.1125733626793793e-05,
      "loss": 0.5702,
      "num_input_tokens_seen": 14839456,
      "step": 22690
    },
    {
      "epoch": 11.894654088050315,
      "grad_norm": 0.14832130074501038,
      "learning_rate": 2.1114437884235915e-05,
      "loss": 0.505,
      "num_input_tokens_seen": 14842272,
      "step": 22695
    },
    {
      "epoch": 11.89727463312369,
      "grad_norm": 0.12831120193004608,
      "learning_rate": 2.1103142954493313e-05,
      "loss": 0.6327,
      "num_input_tokens_seen": 14846304,
      "step": 22700
    },
    {
      "epoch": 11.899895178197065,
      "grad_norm": 0.21179407835006714,
      "learning_rate": 2.1091848839928745e-05,
      "loss": 0.506,
      "num_input_tokens_seen": 14849184,
      "step": 22705
    },
    {
      "epoch": 11.90251572327044,
      "grad_norm": 0.2349219173192978,
      "learning_rate": 2.1080555542904805e-05,
      "loss": 0.4699,
      "num_input_tokens_seen": 14852704,
      "step": 22710
    },
    {
      "epoch": 11.905136268343815,
      "grad_norm": 0.1183244064450264,
      "learning_rate": 2.106926306578393e-05,
      "loss": 0.4225,
      "num_input_tokens_seen": 14856288,
      "step": 22715
    },
    {
      "epoch": 11.90775681341719,
      "grad_norm": 0.2504673898220062,
      "learning_rate": 2.1057971410928383e-05,
      "loss": 0.4799,
      "num_input_tokens_seen": 14859232,
      "step": 22720
    },
    {
      "epoch": 11.910377358490566,
      "grad_norm": 0.15851722657680511,
      "learning_rate": 2.104668058070024e-05,
      "loss": 0.4811,
      "num_input_tokens_seen": 14862624,
      "step": 22725
    },
    {
      "epoch": 11.91299790356394,
      "grad_norm": 0.24636399745941162,
      "learning_rate": 2.1035390577461427e-05,
      "loss": 0.4916,
      "num_input_tokens_seen": 14866368,
      "step": 22730
    },
    {
      "epoch": 11.915618448637316,
      "grad_norm": 0.14191019535064697,
      "learning_rate": 2.102410140357367e-05,
      "loss": 0.3887,
      "num_input_tokens_seen": 14869024,
      "step": 22735
    },
    {
      "epoch": 11.918238993710691,
      "grad_norm": 0.1466316431760788,
      "learning_rate": 2.1012813061398534e-05,
      "loss": 0.3953,
      "num_input_tokens_seen": 14871936,
      "step": 22740
    },
    {
      "epoch": 11.920859538784066,
      "grad_norm": 0.17704921960830688,
      "learning_rate": 2.1001525553297422e-05,
      "loss": 0.6049,
      "num_input_tokens_seen": 14874400,
      "step": 22745
    },
    {
      "epoch": 11.923480083857442,
      "grad_norm": 0.14862367510795593,
      "learning_rate": 2.099023888163154e-05,
      "loss": 0.4329,
      "num_input_tokens_seen": 14876864,
      "step": 22750
    },
    {
      "epoch": 11.926100628930818,
      "grad_norm": 0.31523367762565613,
      "learning_rate": 2.0978953048761947e-05,
      "loss": 0.4195,
      "num_input_tokens_seen": 14879872,
      "step": 22755
    },
    {
      "epoch": 11.928721174004194,
      "grad_norm": 0.144957035779953,
      "learning_rate": 2.0967668057049507e-05,
      "loss": 0.4757,
      "num_input_tokens_seen": 14883424,
      "step": 22760
    },
    {
      "epoch": 11.931341719077569,
      "grad_norm": 0.16079333424568176,
      "learning_rate": 2.0956383908854895e-05,
      "loss": 0.4435,
      "num_input_tokens_seen": 14887392,
      "step": 22765
    },
    {
      "epoch": 11.933962264150944,
      "grad_norm": 0.1828307807445526,
      "learning_rate": 2.0945100606538644e-05,
      "loss": 0.6296,
      "num_input_tokens_seen": 14890688,
      "step": 22770
    },
    {
      "epoch": 11.93658280922432,
      "grad_norm": 0.10304130613803864,
      "learning_rate": 2.0933818152461077e-05,
      "loss": 0.341,
      "num_input_tokens_seen": 14894464,
      "step": 22775
    },
    {
      "epoch": 11.939203354297694,
      "grad_norm": 0.1972535252571106,
      "learning_rate": 2.092253654898237e-05,
      "loss": 0.4787,
      "num_input_tokens_seen": 14897024,
      "step": 22780
    },
    {
      "epoch": 11.94182389937107,
      "grad_norm": 0.15495531260967255,
      "learning_rate": 2.0911255798462494e-05,
      "loss": 0.4677,
      "num_input_tokens_seen": 14900480,
      "step": 22785
    },
    {
      "epoch": 11.944444444444445,
      "grad_norm": 0.284637451171875,
      "learning_rate": 2.0899975903261266e-05,
      "loss": 0.5621,
      "num_input_tokens_seen": 14903136,
      "step": 22790
    },
    {
      "epoch": 11.94706498951782,
      "grad_norm": 0.1644841581583023,
      "learning_rate": 2.0888696865738295e-05,
      "loss": 0.4295,
      "num_input_tokens_seen": 14907040,
      "step": 22795
    },
    {
      "epoch": 11.949685534591195,
      "grad_norm": 0.19708369672298431,
      "learning_rate": 2.087741868825304e-05,
      "loss": 0.6479,
      "num_input_tokens_seen": 14910304,
      "step": 22800
    },
    {
      "epoch": 11.95230607966457,
      "grad_norm": 0.10305672138929367,
      "learning_rate": 2.0866141373164758e-05,
      "loss": 0.5332,
      "num_input_tokens_seen": 14913312,
      "step": 22805
    },
    {
      "epoch": 11.954926624737945,
      "grad_norm": 0.17559048533439636,
      "learning_rate": 2.0854864922832544e-05,
      "loss": 0.6005,
      "num_input_tokens_seen": 14916512,
      "step": 22810
    },
    {
      "epoch": 11.95754716981132,
      "grad_norm": 0.1219574362039566,
      "learning_rate": 2.0843589339615304e-05,
      "loss": 0.458,
      "num_input_tokens_seen": 14920416,
      "step": 22815
    },
    {
      "epoch": 11.960167714884696,
      "grad_norm": 0.17363740503787994,
      "learning_rate": 2.0832314625871742e-05,
      "loss": 0.5504,
      "num_input_tokens_seen": 14923840,
      "step": 22820
    },
    {
      "epoch": 11.96278825995807,
      "grad_norm": 0.1753455102443695,
      "learning_rate": 2.0821040783960423e-05,
      "loss": 0.5201,
      "num_input_tokens_seen": 14927520,
      "step": 22825
    },
    {
      "epoch": 11.965408805031446,
      "grad_norm": 0.23158268630504608,
      "learning_rate": 2.0809767816239686e-05,
      "loss": 0.4093,
      "num_input_tokens_seen": 14930240,
      "step": 22830
    },
    {
      "epoch": 11.968029350104821,
      "grad_norm": 0.38180941343307495,
      "learning_rate": 2.0798495725067725e-05,
      "loss": 0.4966,
      "num_input_tokens_seen": 14932928,
      "step": 22835
    },
    {
      "epoch": 11.970649895178196,
      "grad_norm": 0.2187260389328003,
      "learning_rate": 2.078722451280252e-05,
      "loss": 0.4893,
      "num_input_tokens_seen": 14935616,
      "step": 22840
    },
    {
      "epoch": 11.973270440251572,
      "grad_norm": 0.18795020878314972,
      "learning_rate": 2.0775954181801893e-05,
      "loss": 0.4296,
      "num_input_tokens_seen": 14938080,
      "step": 22845
    },
    {
      "epoch": 11.975890985324948,
      "grad_norm": 0.3834461569786072,
      "learning_rate": 2.0764684734423455e-05,
      "loss": 0.6236,
      "num_input_tokens_seen": 14940320,
      "step": 22850
    },
    {
      "epoch": 11.978511530398324,
      "grad_norm": 0.25164785981178284,
      "learning_rate": 2.075341617302465e-05,
      "loss": 0.5133,
      "num_input_tokens_seen": 14944000,
      "step": 22855
    },
    {
      "epoch": 11.981132075471699,
      "grad_norm": 0.120369091629982,
      "learning_rate": 2.074214849996274e-05,
      "loss": 0.3981,
      "num_input_tokens_seen": 14947168,
      "step": 22860
    },
    {
      "epoch": 11.983752620545074,
      "grad_norm": 0.13096444308757782,
      "learning_rate": 2.0730881717594776e-05,
      "loss": 0.482,
      "num_input_tokens_seen": 14950592,
      "step": 22865
    },
    {
      "epoch": 11.98637316561845,
      "grad_norm": 0.18885473906993866,
      "learning_rate": 2.0719615828277657e-05,
      "loss": 0.4506,
      "num_input_tokens_seen": 14954176,
      "step": 22870
    },
    {
      "epoch": 11.988993710691824,
      "grad_norm": 0.2172231674194336,
      "learning_rate": 2.0708350834368083e-05,
      "loss": 0.5247,
      "num_input_tokens_seen": 14957248,
      "step": 22875
    },
    {
      "epoch": 11.9916142557652,
      "grad_norm": 0.19951286911964417,
      "learning_rate": 2.069708673822255e-05,
      "loss": 0.561,
      "num_input_tokens_seen": 14960640,
      "step": 22880
    },
    {
      "epoch": 11.994234800838575,
      "grad_norm": 0.24274232983589172,
      "learning_rate": 2.068582354219737e-05,
      "loss": 0.4453,
      "num_input_tokens_seen": 14963392,
      "step": 22885
    },
    {
      "epoch": 11.99685534591195,
      "grad_norm": 0.13511233031749725,
      "learning_rate": 2.0674561248648697e-05,
      "loss": 0.4817,
      "num_input_tokens_seen": 14966528,
      "step": 22890
    },
    {
      "epoch": 11.999475890985325,
      "grad_norm": 0.24711519479751587,
      "learning_rate": 2.0663299859932456e-05,
      "loss": 0.4765,
      "num_input_tokens_seen": 14970464,
      "step": 22895
    },
    {
      "epoch": 12.0,
      "eval_loss": 0.49469390511512756,
      "eval_runtime": 10.155,
      "eval_samples_per_second": 83.506,
      "eval_steps_per_second": 20.876,
      "num_input_tokens_seen": 14970528,
      "step": 22896
    },
    {
      "epoch": 12.0020964360587,
      "grad_norm": 0.16485463082790375,
      "learning_rate": 2.065203937840442e-05,
      "loss": 0.4405,
      "num_input_tokens_seen": 14973600,
      "step": 22900
    },
    {
      "epoch": 12.004716981132075,
      "grad_norm": 0.1255054473876953,
      "learning_rate": 2.0640779806420136e-05,
      "loss": 0.5587,
      "num_input_tokens_seen": 14977536,
      "step": 22905
    },
    {
      "epoch": 12.00733752620545,
      "grad_norm": 0.10265139490365982,
      "learning_rate": 2.0629521146334978e-05,
      "loss": 0.3985,
      "num_input_tokens_seen": 14980640,
      "step": 22910
    },
    {
      "epoch": 12.009958071278826,
      "grad_norm": 0.2125757932662964,
      "learning_rate": 2.061826340050414e-05,
      "loss": 0.4454,
      "num_input_tokens_seen": 14983456,
      "step": 22915
    },
    {
      "epoch": 12.0125786163522,
      "grad_norm": 0.15259458124637604,
      "learning_rate": 2.0607006571282605e-05,
      "loss": 0.5264,
      "num_input_tokens_seen": 14987328,
      "step": 22920
    },
    {
      "epoch": 12.015199161425576,
      "grad_norm": 0.24613706767559052,
      "learning_rate": 2.0595750661025183e-05,
      "loss": 0.4481,
      "num_input_tokens_seen": 14990432,
      "step": 22925
    },
    {
      "epoch": 12.017819706498951,
      "grad_norm": 0.11667165160179138,
      "learning_rate": 2.0584495672086483e-05,
      "loss": 0.522,
      "num_input_tokens_seen": 14993824,
      "step": 22930
    },
    {
      "epoch": 12.020440251572326,
      "grad_norm": 0.12722916901111603,
      "learning_rate": 2.05732416068209e-05,
      "loss": 0.4391,
      "num_input_tokens_seen": 14996608,
      "step": 22935
    },
    {
      "epoch": 12.023060796645701,
      "grad_norm": 0.14943550527095795,
      "learning_rate": 2.0561988467582676e-05,
      "loss": 0.4354,
      "num_input_tokens_seen": 15000128,
      "step": 22940
    },
    {
      "epoch": 12.025681341719078,
      "grad_norm": 0.16338971257209778,
      "learning_rate": 2.0550736256725828e-05,
      "loss": 0.4888,
      "num_input_tokens_seen": 15003104,
      "step": 22945
    },
    {
      "epoch": 12.028301886792454,
      "grad_norm": 0.19659090042114258,
      "learning_rate": 2.0539484976604196e-05,
      "loss": 0.4348,
      "num_input_tokens_seen": 15005632,
      "step": 22950
    },
    {
      "epoch": 12.030922431865829,
      "grad_norm": 0.21170586347579956,
      "learning_rate": 2.052823462957142e-05,
      "loss": 0.4223,
      "num_input_tokens_seen": 15009024,
      "step": 22955
    },
    {
      "epoch": 12.033542976939204,
      "grad_norm": 0.14097441732883453,
      "learning_rate": 2.0516985217980953e-05,
      "loss": 0.4282,
      "num_input_tokens_seen": 15011680,
      "step": 22960
    },
    {
      "epoch": 12.036163522012579,
      "grad_norm": 0.1503215879201889,
      "learning_rate": 2.050573674418602e-05,
      "loss": 0.5228,
      "num_input_tokens_seen": 15014656,
      "step": 22965
    },
    {
      "epoch": 12.038784067085954,
      "grad_norm": 0.21999865770339966,
      "learning_rate": 2.0494489210539693e-05,
      "loss": 0.3847,
      "num_input_tokens_seen": 15017952,
      "step": 22970
    },
    {
      "epoch": 12.04140461215933,
      "grad_norm": 0.23692122101783752,
      "learning_rate": 2.0483242619394817e-05,
      "loss": 0.5688,
      "num_input_tokens_seen": 15020736,
      "step": 22975
    },
    {
      "epoch": 12.044025157232705,
      "grad_norm": 0.22788693010807037,
      "learning_rate": 2.0471996973104063e-05,
      "loss": 0.6209,
      "num_input_tokens_seen": 15024544,
      "step": 22980
    },
    {
      "epoch": 12.04664570230608,
      "grad_norm": 0.14895467460155487,
      "learning_rate": 2.0460752274019886e-05,
      "loss": 0.4983,
      "num_input_tokens_seen": 15027424,
      "step": 22985
    },
    {
      "epoch": 12.049266247379455,
      "grad_norm": 0.12377920001745224,
      "learning_rate": 2.0449508524494537e-05,
      "loss": 0.4432,
      "num_input_tokens_seen": 15030976,
      "step": 22990
    },
    {
      "epoch": 12.05188679245283,
      "grad_norm": 0.10358201712369919,
      "learning_rate": 2.0438265726880092e-05,
      "loss": 0.4795,
      "num_input_tokens_seen": 15034464,
      "step": 22995
    },
    {
      "epoch": 12.054507337526205,
      "grad_norm": 0.1122230663895607,
      "learning_rate": 2.042702388352841e-05,
      "loss": 0.552,
      "num_input_tokens_seen": 15039104,
      "step": 23000
    },
    {
      "epoch": 12.05712788259958,
      "grad_norm": 0.1635124236345291,
      "learning_rate": 2.041578299679117e-05,
      "loss": 0.4295,
      "num_input_tokens_seen": 15041920,
      "step": 23005
    },
    {
      "epoch": 12.059748427672956,
      "grad_norm": 0.15693572163581848,
      "learning_rate": 2.0404543069019818e-05,
      "loss": 0.552,
      "num_input_tokens_seen": 15045344,
      "step": 23010
    },
    {
      "epoch": 12.06236897274633,
      "grad_norm": 0.13563470542430878,
      "learning_rate": 2.0393304102565642e-05,
      "loss": 0.5367,
      "num_input_tokens_seen": 15048352,
      "step": 23015
    },
    {
      "epoch": 12.064989517819706,
      "grad_norm": 0.16850398480892181,
      "learning_rate": 2.0382066099779685e-05,
      "loss": 0.4097,
      "num_input_tokens_seen": 15051680,
      "step": 23020
    },
    {
      "epoch": 12.067610062893081,
      "grad_norm": 0.4979192614555359,
      "learning_rate": 2.0370829063012808e-05,
      "loss": 0.5317,
      "num_input_tokens_seen": 15054304,
      "step": 23025
    },
    {
      "epoch": 12.070230607966456,
      "grad_norm": 0.1401159167289734,
      "learning_rate": 2.035959299461568e-05,
      "loss": 0.4176,
      "num_input_tokens_seen": 15056640,
      "step": 23030
    },
    {
      "epoch": 12.072851153039831,
      "grad_norm": 0.1543651819229126,
      "learning_rate": 2.0348357896938762e-05,
      "loss": 0.5552,
      "num_input_tokens_seen": 15059456,
      "step": 23035
    },
    {
      "epoch": 12.075471698113208,
      "grad_norm": 0.13273338973522186,
      "learning_rate": 2.03371237723323e-05,
      "loss": 0.5836,
      "num_input_tokens_seen": 15062720,
      "step": 23040
    },
    {
      "epoch": 12.078092243186584,
      "grad_norm": 0.10088152438402176,
      "learning_rate": 2.032589062314636e-05,
      "loss": 0.4307,
      "num_input_tokens_seen": 15065568,
      "step": 23045
    },
    {
      "epoch": 12.080712788259959,
      "grad_norm": 0.186522975564003,
      "learning_rate": 2.0314658451730767e-05,
      "loss": 0.4114,
      "num_input_tokens_seen": 15068224,
      "step": 23050
    },
    {
      "epoch": 12.083333333333334,
      "grad_norm": 0.10852599143981934,
      "learning_rate": 2.030342726043517e-05,
      "loss": 0.573,
      "num_input_tokens_seen": 15072032,
      "step": 23055
    },
    {
      "epoch": 12.085953878406709,
      "grad_norm": 0.14901012182235718,
      "learning_rate": 2.0292197051609013e-05,
      "loss": 0.4196,
      "num_input_tokens_seen": 15075104,
      "step": 23060
    },
    {
      "epoch": 12.088574423480084,
      "grad_norm": 0.1408700942993164,
      "learning_rate": 2.0280967827601518e-05,
      "loss": 0.5715,
      "num_input_tokens_seen": 15078336,
      "step": 23065
    },
    {
      "epoch": 12.09119496855346,
      "grad_norm": 0.23310446739196777,
      "learning_rate": 2.0269739590761723e-05,
      "loss": 0.3342,
      "num_input_tokens_seen": 15080736,
      "step": 23070
    },
    {
      "epoch": 12.093815513626835,
      "grad_norm": 0.14859327673912048,
      "learning_rate": 2.0258512343438442e-05,
      "loss": 0.4882,
      "num_input_tokens_seen": 15083392,
      "step": 23075
    },
    {
      "epoch": 12.09643605870021,
      "grad_norm": 0.15317310392856598,
      "learning_rate": 2.0247286087980276e-05,
      "loss": 0.3527,
      "num_input_tokens_seen": 15086784,
      "step": 23080
    },
    {
      "epoch": 12.099056603773585,
      "grad_norm": 0.18643192946910858,
      "learning_rate": 2.0236060826735644e-05,
      "loss": 0.5187,
      "num_input_tokens_seen": 15089504,
      "step": 23085
    },
    {
      "epoch": 12.10167714884696,
      "grad_norm": 0.12812469899654388,
      "learning_rate": 2.0224836562052732e-05,
      "loss": 0.5222,
      "num_input_tokens_seen": 15092512,
      "step": 23090
    },
    {
      "epoch": 12.104297693920335,
      "grad_norm": 0.12177382409572601,
      "learning_rate": 2.0213613296279533e-05,
      "loss": 0.51,
      "num_input_tokens_seen": 15095808,
      "step": 23095
    },
    {
      "epoch": 12.10691823899371,
      "grad_norm": 0.16051675379276276,
      "learning_rate": 2.0202391031763844e-05,
      "loss": 0.6007,
      "num_input_tokens_seen": 15098720,
      "step": 23100
    },
    {
      "epoch": 12.109538784067086,
      "grad_norm": 0.18936432898044586,
      "learning_rate": 2.0191169770853195e-05,
      "loss": 0.6364,
      "num_input_tokens_seen": 15102176,
      "step": 23105
    },
    {
      "epoch": 12.11215932914046,
      "grad_norm": 0.282458633184433,
      "learning_rate": 2.0179949515894975e-05,
      "loss": 0.5145,
      "num_input_tokens_seen": 15104480,
      "step": 23110
    },
    {
      "epoch": 12.114779874213836,
      "grad_norm": 0.21099427342414856,
      "learning_rate": 2.016873026923633e-05,
      "loss": 0.4772,
      "num_input_tokens_seen": 15107264,
      "step": 23115
    },
    {
      "epoch": 12.117400419287211,
      "grad_norm": 0.16138415038585663,
      "learning_rate": 2.0157512033224187e-05,
      "loss": 0.311,
      "num_input_tokens_seen": 15111104,
      "step": 23120
    },
    {
      "epoch": 12.120020964360586,
      "grad_norm": 0.1864442080259323,
      "learning_rate": 2.0146294810205287e-05,
      "loss": 0.4002,
      "num_input_tokens_seen": 15114240,
      "step": 23125
    },
    {
      "epoch": 12.122641509433961,
      "grad_norm": 0.18048903346061707,
      "learning_rate": 2.0135078602526146e-05,
      "loss": 0.5854,
      "num_input_tokens_seen": 15117600,
      "step": 23130
    },
    {
      "epoch": 12.125262054507338,
      "grad_norm": 0.09535393863916397,
      "learning_rate": 2.0123863412533046e-05,
      "loss": 0.3764,
      "num_input_tokens_seen": 15120416,
      "step": 23135
    },
    {
      "epoch": 12.127882599580714,
      "grad_norm": 0.16244404017925262,
      "learning_rate": 2.0112649242572095e-05,
      "loss": 0.4167,
      "num_input_tokens_seen": 15124064,
      "step": 23140
    },
    {
      "epoch": 12.130503144654089,
      "grad_norm": 0.15329131484031677,
      "learning_rate": 2.0101436094989164e-05,
      "loss": 0.4145,
      "num_input_tokens_seen": 15127168,
      "step": 23145
    },
    {
      "epoch": 12.133123689727464,
      "grad_norm": 0.24198763072490692,
      "learning_rate": 2.0090223972129918e-05,
      "loss": 0.43,
      "num_input_tokens_seen": 15130208,
      "step": 23150
    },
    {
      "epoch": 12.135744234800839,
      "grad_norm": 0.16086100041866302,
      "learning_rate": 2.0079012876339797e-05,
      "loss": 0.6018,
      "num_input_tokens_seen": 15133760,
      "step": 23155
    },
    {
      "epoch": 12.138364779874214,
      "grad_norm": 0.15694232285022736,
      "learning_rate": 2.0067802809964055e-05,
      "loss": 0.6178,
      "num_input_tokens_seen": 15136992,
      "step": 23160
    },
    {
      "epoch": 12.14098532494759,
      "grad_norm": 0.27381619811058044,
      "learning_rate": 2.0056593775347686e-05,
      "loss": 0.4853,
      "num_input_tokens_seen": 15139296,
      "step": 23165
    },
    {
      "epoch": 12.143605870020965,
      "grad_norm": 0.13184630870819092,
      "learning_rate": 2.00453857748355e-05,
      "loss": 0.445,
      "num_input_tokens_seen": 15142656,
      "step": 23170
    },
    {
      "epoch": 12.14622641509434,
      "grad_norm": 0.16281408071517944,
      "learning_rate": 2.0034178810772082e-05,
      "loss": 0.5354,
      "num_input_tokens_seen": 15145824,
      "step": 23175
    },
    {
      "epoch": 12.148846960167715,
      "grad_norm": 0.15027087926864624,
      "learning_rate": 2.002297288550181e-05,
      "loss": 0.3896,
      "num_input_tokens_seen": 15148608,
      "step": 23180
    },
    {
      "epoch": 12.15146750524109,
      "grad_norm": 0.2304319143295288,
      "learning_rate": 2.0011768001368835e-05,
      "loss": 0.4559,
      "num_input_tokens_seen": 15151200,
      "step": 23185
    },
    {
      "epoch": 12.154088050314465,
      "grad_norm": 0.18688103556632996,
      "learning_rate": 2.0000564160717076e-05,
      "loss": 0.4937,
      "num_input_tokens_seen": 15154464,
      "step": 23190
    },
    {
      "epoch": 12.15670859538784,
      "grad_norm": 0.22376668453216553,
      "learning_rate": 1.9989361365890263e-05,
      "loss": 0.5878,
      "num_input_tokens_seen": 15158272,
      "step": 23195
    },
    {
      "epoch": 12.159329140461216,
      "grad_norm": 0.18507876992225647,
      "learning_rate": 1.997815961923188e-05,
      "loss": 0.3907,
      "num_input_tokens_seen": 15161152,
      "step": 23200
    },
    {
      "epoch": 12.16194968553459,
      "grad_norm": 0.15233919024467468,
      "learning_rate": 1.996695892308522e-05,
      "loss": 0.4479,
      "num_input_tokens_seen": 15163616,
      "step": 23205
    },
    {
      "epoch": 12.164570230607966,
      "grad_norm": 0.24733871221542358,
      "learning_rate": 1.9955759279793327e-05,
      "loss": 0.5197,
      "num_input_tokens_seen": 15167264,
      "step": 23210
    },
    {
      "epoch": 12.167190775681341,
      "grad_norm": 0.13354463875293732,
      "learning_rate": 1.9944560691699057e-05,
      "loss": 0.3989,
      "num_input_tokens_seen": 15170816,
      "step": 23215
    },
    {
      "epoch": 12.169811320754716,
      "grad_norm": 0.21013344824314117,
      "learning_rate": 1.9933363161145006e-05,
      "loss": 0.5329,
      "num_input_tokens_seen": 15173280,
      "step": 23220
    },
    {
      "epoch": 12.172431865828091,
      "grad_norm": 0.1616644710302353,
      "learning_rate": 1.9922166690473575e-05,
      "loss": 0.4721,
      "num_input_tokens_seen": 15176544,
      "step": 23225
    },
    {
      "epoch": 12.175052410901468,
      "grad_norm": 0.1636475771665573,
      "learning_rate": 1.991097128202695e-05,
      "loss": 0.3862,
      "num_input_tokens_seen": 15179488,
      "step": 23230
    },
    {
      "epoch": 12.177672955974844,
      "grad_norm": 1.376652479171753,
      "learning_rate": 1.9899776938147065e-05,
      "loss": 0.5077,
      "num_input_tokens_seen": 15182016,
      "step": 23235
    },
    {
      "epoch": 12.180293501048219,
      "grad_norm": 0.1248728409409523,
      "learning_rate": 1.9888583661175665e-05,
      "loss": 0.5238,
      "num_input_tokens_seen": 15185760,
      "step": 23240
    },
    {
      "epoch": 12.182914046121594,
      "grad_norm": 0.21914039552211761,
      "learning_rate": 1.9877391453454253e-05,
      "loss": 0.4326,
      "num_input_tokens_seen": 15188992,
      "step": 23245
    },
    {
      "epoch": 12.185534591194969,
      "grad_norm": 0.12402516603469849,
      "learning_rate": 1.98662003173241e-05,
      "loss": 0.5149,
      "num_input_tokens_seen": 15192768,
      "step": 23250
    },
    {
      "epoch": 12.188155136268344,
      "grad_norm": 0.18572252988815308,
      "learning_rate": 1.9855010255126266e-05,
      "loss": 0.5388,
      "num_input_tokens_seen": 15196256,
      "step": 23255
    },
    {
      "epoch": 12.19077568134172,
      "grad_norm": 0.1326279193162918,
      "learning_rate": 1.9843821269201598e-05,
      "loss": 0.4507,
      "num_input_tokens_seen": 15202848,
      "step": 23260
    },
    {
      "epoch": 12.193396226415095,
      "grad_norm": 0.3494330048561096,
      "learning_rate": 1.983263336189069e-05,
      "loss": 0.4661,
      "num_input_tokens_seen": 15206240,
      "step": 23265
    },
    {
      "epoch": 12.19601677148847,
      "grad_norm": 0.23365932703018188,
      "learning_rate": 1.9821446535533937e-05,
      "loss": 0.3542,
      "num_input_tokens_seen": 15208864,
      "step": 23270
    },
    {
      "epoch": 12.198637316561845,
      "grad_norm": 0.311392217874527,
      "learning_rate": 1.9810260792471492e-05,
      "loss": 0.4223,
      "num_input_tokens_seen": 15211968,
      "step": 23275
    },
    {
      "epoch": 12.20125786163522,
      "grad_norm": 0.288585901260376,
      "learning_rate": 1.979907613504327e-05,
      "loss": 0.39,
      "num_input_tokens_seen": 15214656,
      "step": 23280
    },
    {
      "epoch": 12.203878406708595,
      "grad_norm": 0.1633455604314804,
      "learning_rate": 1.978789256558899e-05,
      "loss": 0.4997,
      "num_input_tokens_seen": 15218048,
      "step": 23285
    },
    {
      "epoch": 12.20649895178197,
      "grad_norm": 0.2281385213136673,
      "learning_rate": 1.977671008644812e-05,
      "loss": 0.6219,
      "num_input_tokens_seen": 15222112,
      "step": 23290
    },
    {
      "epoch": 12.209119496855346,
      "grad_norm": 0.20135067403316498,
      "learning_rate": 1.9765528699959912e-05,
      "loss": 0.4704,
      "num_input_tokens_seen": 15225312,
      "step": 23295
    },
    {
      "epoch": 12.21174004192872,
      "grad_norm": 0.17249463498592377,
      "learning_rate": 1.975434840846339e-05,
      "loss": 0.5947,
      "num_input_tokens_seen": 15228288,
      "step": 23300
    },
    {
      "epoch": 12.214360587002096,
      "grad_norm": 0.14747966825962067,
      "learning_rate": 1.9743169214297326e-05,
      "loss": 0.3715,
      "num_input_tokens_seen": 15232352,
      "step": 23305
    },
    {
      "epoch": 12.216981132075471,
      "grad_norm": 0.11545268446207047,
      "learning_rate": 1.973199111980029e-05,
      "loss": 0.4251,
      "num_input_tokens_seen": 15237984,
      "step": 23310
    },
    {
      "epoch": 12.219601677148846,
      "grad_norm": 0.12083601951599121,
      "learning_rate": 1.9720814127310608e-05,
      "loss": 0.3575,
      "num_input_tokens_seen": 15241184,
      "step": 23315
    },
    {
      "epoch": 12.222222222222221,
      "grad_norm": 0.1891760677099228,
      "learning_rate": 1.9709638239166386e-05,
      "loss": 0.5103,
      "num_input_tokens_seen": 15243744,
      "step": 23320
    },
    {
      "epoch": 12.224842767295598,
      "grad_norm": 0.37359732389450073,
      "learning_rate": 1.9698463457705485e-05,
      "loss": 0.6918,
      "num_input_tokens_seen": 15246432,
      "step": 23325
    },
    {
      "epoch": 12.227463312368974,
      "grad_norm": 0.2191866785287857,
      "learning_rate": 1.968728978526556e-05,
      "loss": 0.4394,
      "num_input_tokens_seen": 15249440,
      "step": 23330
    },
    {
      "epoch": 12.230083857442349,
      "grad_norm": 0.14708897471427917,
      "learning_rate": 1.9676117224183986e-05,
      "loss": 0.4874,
      "num_input_tokens_seen": 15253120,
      "step": 23335
    },
    {
      "epoch": 12.232704402515724,
      "grad_norm": 0.24887891113758087,
      "learning_rate": 1.9664945776797956e-05,
      "loss": 0.4251,
      "num_input_tokens_seen": 15256224,
      "step": 23340
    },
    {
      "epoch": 12.235324947589099,
      "grad_norm": 0.20253591239452362,
      "learning_rate": 1.96537754454444e-05,
      "loss": 0.6365,
      "num_input_tokens_seen": 15258720,
      "step": 23345
    },
    {
      "epoch": 12.237945492662474,
      "grad_norm": 0.13014231622219086,
      "learning_rate": 1.964260623246003e-05,
      "loss": 0.5244,
      "num_input_tokens_seen": 15261760,
      "step": 23350
    },
    {
      "epoch": 12.24056603773585,
      "grad_norm": 0.14373339712619781,
      "learning_rate": 1.9631438140181324e-05,
      "loss": 0.5722,
      "num_input_tokens_seen": 15265472,
      "step": 23355
    },
    {
      "epoch": 12.243186582809225,
      "grad_norm": 0.09112958610057831,
      "learning_rate": 1.9620271170944503e-05,
      "loss": 0.4089,
      "num_input_tokens_seen": 15268896,
      "step": 23360
    },
    {
      "epoch": 12.2458071278826,
      "grad_norm": 0.5023011565208435,
      "learning_rate": 1.960910532708558e-05,
      "loss": 0.5727,
      "num_input_tokens_seen": 15272320,
      "step": 23365
    },
    {
      "epoch": 12.248427672955975,
      "grad_norm": 0.26789391040802,
      "learning_rate": 1.9597940610940315e-05,
      "loss": 0.5773,
      "num_input_tokens_seen": 15275776,
      "step": 23370
    },
    {
      "epoch": 12.25104821802935,
      "grad_norm": 0.16384342312812805,
      "learning_rate": 1.958677702484425e-05,
      "loss": 0.408,
      "num_input_tokens_seen": 15278912,
      "step": 23375
    },
    {
      "epoch": 12.253668763102725,
      "grad_norm": 0.15958192944526672,
      "learning_rate": 1.957561457113267e-05,
      "loss": 0.5382,
      "num_input_tokens_seen": 15281536,
      "step": 23380
    },
    {
      "epoch": 12.2562893081761,
      "grad_norm": 0.17999079823493958,
      "learning_rate": 1.9564453252140648e-05,
      "loss": 0.517,
      "num_input_tokens_seen": 15284544,
      "step": 23385
    },
    {
      "epoch": 12.258909853249476,
      "grad_norm": 0.10658400505781174,
      "learning_rate": 1.9553293070202984e-05,
      "loss": 0.4072,
      "num_input_tokens_seen": 15287968,
      "step": 23390
    },
    {
      "epoch": 12.26153039832285,
      "grad_norm": 0.26677021384239197,
      "learning_rate": 1.9542134027654267e-05,
      "loss": 0.4147,
      "num_input_tokens_seen": 15291136,
      "step": 23395
    },
    {
      "epoch": 12.264150943396226,
      "grad_norm": 0.17081314325332642,
      "learning_rate": 1.953097612682885e-05,
      "loss": 0.4879,
      "num_input_tokens_seen": 15293824,
      "step": 23400
    },
    {
      "epoch": 12.266771488469601,
      "grad_norm": 0.2322831004858017,
      "learning_rate": 1.9519819370060832e-05,
      "loss": 0.3743,
      "num_input_tokens_seen": 15296928,
      "step": 23405
    },
    {
      "epoch": 12.269392033542976,
      "grad_norm": 0.2608250677585602,
      "learning_rate": 1.9508663759684074e-05,
      "loss": 0.4168,
      "num_input_tokens_seen": 15299904,
      "step": 23410
    },
    {
      "epoch": 12.272012578616351,
      "grad_norm": 0.1423935741186142,
      "learning_rate": 1.9497509298032224e-05,
      "loss": 0.508,
      "num_input_tokens_seen": 15302624,
      "step": 23415
    },
    {
      "epoch": 12.274633123689728,
      "grad_norm": 0.16418540477752686,
      "learning_rate": 1.9486355987438647e-05,
      "loss": 0.4768,
      "num_input_tokens_seen": 15305408,
      "step": 23420
    },
    {
      "epoch": 12.277253668763104,
      "grad_norm": 0.23909448087215424,
      "learning_rate": 1.947520383023649e-05,
      "loss": 0.5737,
      "num_input_tokens_seen": 15308288,
      "step": 23425
    },
    {
      "epoch": 12.279874213836479,
      "grad_norm": 0.19956205785274506,
      "learning_rate": 1.9464052828758666e-05,
      "loss": 0.4353,
      "num_input_tokens_seen": 15311200,
      "step": 23430
    },
    {
      "epoch": 12.282494758909854,
      "grad_norm": 0.11246875673532486,
      "learning_rate": 1.945290298533783e-05,
      "loss": 0.611,
      "num_input_tokens_seen": 15314688,
      "step": 23435
    },
    {
      "epoch": 12.285115303983229,
      "grad_norm": 0.26368191838264465,
      "learning_rate": 1.9441754302306406e-05,
      "loss": 0.4661,
      "num_input_tokens_seen": 15317632,
      "step": 23440
    },
    {
      "epoch": 12.287735849056604,
      "grad_norm": 0.15272511541843414,
      "learning_rate": 1.9430606781996578e-05,
      "loss": 0.3887,
      "num_input_tokens_seen": 15320288,
      "step": 23445
    },
    {
      "epoch": 12.29035639412998,
      "grad_norm": 0.1370931714773178,
      "learning_rate": 1.941946042674026e-05,
      "loss": 0.4654,
      "num_input_tokens_seen": 15323520,
      "step": 23450
    },
    {
      "epoch": 12.292976939203355,
      "grad_norm": 0.21464984118938446,
      "learning_rate": 1.940831523886916e-05,
      "loss": 0.5714,
      "num_input_tokens_seen": 15327008,
      "step": 23455
    },
    {
      "epoch": 12.29559748427673,
      "grad_norm": 0.14200042188167572,
      "learning_rate": 1.9397171220714714e-05,
      "loss": 0.4613,
      "num_input_tokens_seen": 15329952,
      "step": 23460
    },
    {
      "epoch": 12.298218029350105,
      "grad_norm": 0.18121786415576935,
      "learning_rate": 1.9386028374608134e-05,
      "loss": 0.5175,
      "num_input_tokens_seen": 15333056,
      "step": 23465
    },
    {
      "epoch": 12.30083857442348,
      "grad_norm": 0.1865161806344986,
      "learning_rate": 1.9374886702880372e-05,
      "loss": 0.4553,
      "num_input_tokens_seen": 15336736,
      "step": 23470
    },
    {
      "epoch": 12.303459119496855,
      "grad_norm": 0.16549701988697052,
      "learning_rate": 1.936374620786213e-05,
      "loss": 0.4087,
      "num_input_tokens_seen": 15339712,
      "step": 23475
    },
    {
      "epoch": 12.30607966457023,
      "grad_norm": 0.1029786542057991,
      "learning_rate": 1.9352606891883885e-05,
      "loss": 0.5888,
      "num_input_tokens_seen": 15343232,
      "step": 23480
    },
    {
      "epoch": 12.308700209643606,
      "grad_norm": 0.23026053607463837,
      "learning_rate": 1.9341468757275844e-05,
      "loss": 0.4916,
      "num_input_tokens_seen": 15346816,
      "step": 23485
    },
    {
      "epoch": 12.31132075471698,
      "grad_norm": 0.14264778792858124,
      "learning_rate": 1.933033180636798e-05,
      "loss": 0.577,
      "num_input_tokens_seen": 15349760,
      "step": 23490
    },
    {
      "epoch": 12.313941299790356,
      "grad_norm": 0.13267122209072113,
      "learning_rate": 1.9319196041490027e-05,
      "loss": 0.4838,
      "num_input_tokens_seen": 15353184,
      "step": 23495
    },
    {
      "epoch": 12.316561844863731,
      "grad_norm": 0.13066691160202026,
      "learning_rate": 1.930806146497146e-05,
      "loss": 0.3942,
      "num_input_tokens_seen": 15356512,
      "step": 23500
    },
    {
      "epoch": 12.319182389937106,
      "grad_norm": 0.127927765250206,
      "learning_rate": 1.9296928079141485e-05,
      "loss": 0.4014,
      "num_input_tokens_seen": 15359104,
      "step": 23505
    },
    {
      "epoch": 12.321802935010481,
      "grad_norm": 0.16044734418392181,
      "learning_rate": 1.9285795886329093e-05,
      "loss": 0.5438,
      "num_input_tokens_seen": 15362528,
      "step": 23510
    },
    {
      "epoch": 12.324423480083858,
      "grad_norm": 0.1751527488231659,
      "learning_rate": 1.927466488886301e-05,
      "loss": 0.6394,
      "num_input_tokens_seen": 15366240,
      "step": 23515
    },
    {
      "epoch": 12.327044025157234,
      "grad_norm": 0.380386620759964,
      "learning_rate": 1.926353508907172e-05,
      "loss": 0.5551,
      "num_input_tokens_seen": 15368704,
      "step": 23520
    },
    {
      "epoch": 12.329664570230609,
      "grad_norm": 0.15233922004699707,
      "learning_rate": 1.9252406489283437e-05,
      "loss": 0.4234,
      "num_input_tokens_seen": 15373312,
      "step": 23525
    },
    {
      "epoch": 12.332285115303984,
      "grad_norm": 0.17390449345111847,
      "learning_rate": 1.9241279091826155e-05,
      "loss": 0.5193,
      "num_input_tokens_seen": 15377632,
      "step": 23530
    },
    {
      "epoch": 12.334905660377359,
      "grad_norm": 0.18058669567108154,
      "learning_rate": 1.923015289902758e-05,
      "loss": 0.5194,
      "num_input_tokens_seen": 15380864,
      "step": 23535
    },
    {
      "epoch": 12.337526205450734,
      "grad_norm": 0.13676367700099945,
      "learning_rate": 1.9219027913215193e-05,
      "loss": 0.4766,
      "num_input_tokens_seen": 15384480,
      "step": 23540
    },
    {
      "epoch": 12.34014675052411,
      "grad_norm": 0.17622770369052887,
      "learning_rate": 1.9207904136716214e-05,
      "loss": 0.4146,
      "num_input_tokens_seen": 15387968,
      "step": 23545
    },
    {
      "epoch": 12.342767295597485,
      "grad_norm": 0.23283354938030243,
      "learning_rate": 1.9196781571857608e-05,
      "loss": 0.4227,
      "num_input_tokens_seen": 15390528,
      "step": 23550
    },
    {
      "epoch": 12.34538784067086,
      "grad_norm": 0.2291691154241562,
      "learning_rate": 1.9185660220966097e-05,
      "loss": 0.4409,
      "num_input_tokens_seen": 15394368,
      "step": 23555
    },
    {
      "epoch": 12.348008385744235,
      "grad_norm": 0.1288011074066162,
      "learning_rate": 1.9174540086368135e-05,
      "loss": 0.4534,
      "num_input_tokens_seen": 15397600,
      "step": 23560
    },
    {
      "epoch": 12.35062893081761,
      "grad_norm": 0.14817462861537933,
      "learning_rate": 1.9163421170389918e-05,
      "loss": 0.4342,
      "num_input_tokens_seen": 15400288,
      "step": 23565
    },
    {
      "epoch": 12.353249475890985,
      "grad_norm": 0.13403336703777313,
      "learning_rate": 1.9152303475357402e-05,
      "loss": 0.5902,
      "num_input_tokens_seen": 15404032,
      "step": 23570
    },
    {
      "epoch": 12.35587002096436,
      "grad_norm": 0.22638943791389465,
      "learning_rate": 1.9141187003596295e-05,
      "loss": 0.6241,
      "num_input_tokens_seen": 15407392,
      "step": 23575
    },
    {
      "epoch": 12.358490566037736,
      "grad_norm": 0.3391338586807251,
      "learning_rate": 1.913007175743202e-05,
      "loss": 0.4535,
      "num_input_tokens_seen": 15410208,
      "step": 23580
    },
    {
      "epoch": 12.36111111111111,
      "grad_norm": 0.17889820039272308,
      "learning_rate": 1.9118957739189773e-05,
      "loss": 0.5689,
      "num_input_tokens_seen": 15412992,
      "step": 23585
    },
    {
      "epoch": 12.363731656184486,
      "grad_norm": 0.11418145149946213,
      "learning_rate": 1.910784495119447e-05,
      "loss": 0.5275,
      "num_input_tokens_seen": 15417056,
      "step": 23590
    },
    {
      "epoch": 12.366352201257861,
      "grad_norm": 0.3536938726902008,
      "learning_rate": 1.9096733395770778e-05,
      "loss": 0.4501,
      "num_input_tokens_seen": 15420032,
      "step": 23595
    },
    {
      "epoch": 12.368972746331236,
      "grad_norm": 0.1375582367181778,
      "learning_rate": 1.9085623075243114e-05,
      "loss": 0.5561,
      "num_input_tokens_seen": 15423776,
      "step": 23600
    },
    {
      "epoch": 12.371593291404611,
      "grad_norm": 0.17508040368556976,
      "learning_rate": 1.9074513991935623e-05,
      "loss": 0.5122,
      "num_input_tokens_seen": 15426880,
      "step": 23605
    },
    {
      "epoch": 12.374213836477987,
      "grad_norm": 0.14260978996753693,
      "learning_rate": 1.906340614817221e-05,
      "loss": 0.4632,
      "num_input_tokens_seen": 15430688,
      "step": 23610
    },
    {
      "epoch": 12.376834381551364,
      "grad_norm": 0.19055083394050598,
      "learning_rate": 1.9052299546276506e-05,
      "loss": 0.4061,
      "num_input_tokens_seen": 15433728,
      "step": 23615
    },
    {
      "epoch": 12.379454926624739,
      "grad_norm": 0.21777646243572235,
      "learning_rate": 1.904119418857187e-05,
      "loss": 0.5239,
      "num_input_tokens_seen": 15437568,
      "step": 23620
    },
    {
      "epoch": 12.382075471698114,
      "grad_norm": 0.13115574419498444,
      "learning_rate": 1.903009007738144e-05,
      "loss": 0.5197,
      "num_input_tokens_seen": 15440384,
      "step": 23625
    },
    {
      "epoch": 12.384696016771489,
      "grad_norm": 0.2019612342119217,
      "learning_rate": 1.9018987215028045e-05,
      "loss": 0.4524,
      "num_input_tokens_seen": 15444032,
      "step": 23630
    },
    {
      "epoch": 12.387316561844864,
      "grad_norm": 0.3140495717525482,
      "learning_rate": 1.9007885603834298e-05,
      "loss": 0.4281,
      "num_input_tokens_seen": 15446400,
      "step": 23635
    },
    {
      "epoch": 12.38993710691824,
      "grad_norm": 0.16342511773109436,
      "learning_rate": 1.8996785246122526e-05,
      "loss": 0.4679,
      "num_input_tokens_seen": 15450880,
      "step": 23640
    },
    {
      "epoch": 12.392557651991615,
      "grad_norm": 0.14607220888137817,
      "learning_rate": 1.898568614421478e-05,
      "loss": 0.5176,
      "num_input_tokens_seen": 15453856,
      "step": 23645
    },
    {
      "epoch": 12.39517819706499,
      "grad_norm": 0.25307497382164,
      "learning_rate": 1.897458830043288e-05,
      "loss": 0.4165,
      "num_input_tokens_seen": 15456320,
      "step": 23650
    },
    {
      "epoch": 12.397798742138365,
      "grad_norm": 0.15542401373386383,
      "learning_rate": 1.8963491717098372e-05,
      "loss": 0.5388,
      "num_input_tokens_seen": 15459616,
      "step": 23655
    },
    {
      "epoch": 12.40041928721174,
      "grad_norm": 0.15403471887111664,
      "learning_rate": 1.8952396396532522e-05,
      "loss": 0.5191,
      "num_input_tokens_seen": 15463520,
      "step": 23660
    },
    {
      "epoch": 12.403039832285115,
      "grad_norm": 0.18786287307739258,
      "learning_rate": 1.8941302341056356e-05,
      "loss": 0.4496,
      "num_input_tokens_seen": 15466976,
      "step": 23665
    },
    {
      "epoch": 12.40566037735849,
      "grad_norm": 0.23772606253623962,
      "learning_rate": 1.8930209552990625e-05,
      "loss": 0.6496,
      "num_input_tokens_seen": 15470784,
      "step": 23670
    },
    {
      "epoch": 12.408280922431866,
      "grad_norm": 0.15218542516231537,
      "learning_rate": 1.8919118034655796e-05,
      "loss": 0.3666,
      "num_input_tokens_seen": 15474816,
      "step": 23675
    },
    {
      "epoch": 12.41090146750524,
      "grad_norm": 0.31866174936294556,
      "learning_rate": 1.8908027788372107e-05,
      "loss": 0.5221,
      "num_input_tokens_seen": 15478016,
      "step": 23680
    },
    {
      "epoch": 12.413522012578616,
      "grad_norm": 0.11664348840713501,
      "learning_rate": 1.88969388164595e-05,
      "loss": 0.3807,
      "num_input_tokens_seen": 15481536,
      "step": 23685
    },
    {
      "epoch": 12.416142557651991,
      "grad_norm": 0.12486865371465683,
      "learning_rate": 1.8885851121237666e-05,
      "loss": 0.3817,
      "num_input_tokens_seen": 15485504,
      "step": 23690
    },
    {
      "epoch": 12.418763102725366,
      "grad_norm": 0.19297818839550018,
      "learning_rate": 1.8874764705026023e-05,
      "loss": 0.5666,
      "num_input_tokens_seen": 15489120,
      "step": 23695
    },
    {
      "epoch": 12.421383647798741,
      "grad_norm": 0.23753510415554047,
      "learning_rate": 1.8863679570143728e-05,
      "loss": 0.6167,
      "num_input_tokens_seen": 15491840,
      "step": 23700
    },
    {
      "epoch": 12.424004192872117,
      "grad_norm": 0.27517127990722656,
      "learning_rate": 1.885259571890966e-05,
      "loss": 0.3974,
      "num_input_tokens_seen": 15494880,
      "step": 23705
    },
    {
      "epoch": 12.426624737945493,
      "grad_norm": 0.2476547360420227,
      "learning_rate": 1.8841513153642427e-05,
      "loss": 0.4109,
      "num_input_tokens_seen": 15497312,
      "step": 23710
    },
    {
      "epoch": 12.429245283018869,
      "grad_norm": 0.14671654999256134,
      "learning_rate": 1.8830431876660388e-05,
      "loss": 0.3948,
      "num_input_tokens_seen": 15500416,
      "step": 23715
    },
    {
      "epoch": 12.431865828092244,
      "grad_norm": 0.29950210452079773,
      "learning_rate": 1.8819351890281607e-05,
      "loss": 0.5619,
      "num_input_tokens_seen": 15503520,
      "step": 23720
    },
    {
      "epoch": 12.434486373165619,
      "grad_norm": 0.19523456692695618,
      "learning_rate": 1.880827319682391e-05,
      "loss": 0.3555,
      "num_input_tokens_seen": 15506656,
      "step": 23725
    },
    {
      "epoch": 12.437106918238994,
      "grad_norm": 0.1703871488571167,
      "learning_rate": 1.879719579860481e-05,
      "loss": 0.5161,
      "num_input_tokens_seen": 15509792,
      "step": 23730
    },
    {
      "epoch": 12.43972746331237,
      "grad_norm": 0.20924533903598785,
      "learning_rate": 1.878611969794159e-05,
      "loss": 0.4401,
      "num_input_tokens_seen": 15512768,
      "step": 23735
    },
    {
      "epoch": 12.442348008385745,
      "grad_norm": 0.16926229000091553,
      "learning_rate": 1.8775044897151232e-05,
      "loss": 0.5625,
      "num_input_tokens_seen": 15515712,
      "step": 23740
    },
    {
      "epoch": 12.44496855345912,
      "grad_norm": 0.18644025921821594,
      "learning_rate": 1.876397139855047e-05,
      "loss": 0.5953,
      "num_input_tokens_seen": 15518528,
      "step": 23745
    },
    {
      "epoch": 12.447589098532495,
      "grad_norm": 0.1431746482849121,
      "learning_rate": 1.875289920445574e-05,
      "loss": 0.5631,
      "num_input_tokens_seen": 15521472,
      "step": 23750
    },
    {
      "epoch": 12.45020964360587,
      "grad_norm": 0.4650684595108032,
      "learning_rate": 1.8741828317183234e-05,
      "loss": 0.3381,
      "num_input_tokens_seen": 15524320,
      "step": 23755
    },
    {
      "epoch": 12.452830188679245,
      "grad_norm": 0.19281548261642456,
      "learning_rate": 1.873075873904885e-05,
      "loss": 0.5098,
      "num_input_tokens_seen": 15527168,
      "step": 23760
    },
    {
      "epoch": 12.45545073375262,
      "grad_norm": 0.20554353296756744,
      "learning_rate": 1.87196904723682e-05,
      "loss": 0.3837,
      "num_input_tokens_seen": 15532064,
      "step": 23765
    },
    {
      "epoch": 12.458071278825996,
      "grad_norm": 0.1481783390045166,
      "learning_rate": 1.8708623519456665e-05,
      "loss": 0.5142,
      "num_input_tokens_seen": 15534944,
      "step": 23770
    },
    {
      "epoch": 12.46069182389937,
      "grad_norm": 0.12330201268196106,
      "learning_rate": 1.8697557882629308e-05,
      "loss": 0.4553,
      "num_input_tokens_seen": 15538368,
      "step": 23775
    },
    {
      "epoch": 12.463312368972746,
      "grad_norm": 0.11407015472650528,
      "learning_rate": 1.868649356420094e-05,
      "loss": 0.4312,
      "num_input_tokens_seen": 15541280,
      "step": 23780
    },
    {
      "epoch": 12.465932914046121,
      "grad_norm": 0.19602467119693756,
      "learning_rate": 1.86754305664861e-05,
      "loss": 0.4293,
      "num_input_tokens_seen": 15544736,
      "step": 23785
    },
    {
      "epoch": 12.468553459119496,
      "grad_norm": 0.23412972688674927,
      "learning_rate": 1.866436889179902e-05,
      "loss": 0.5638,
      "num_input_tokens_seen": 15549568,
      "step": 23790
    },
    {
      "epoch": 12.471174004192871,
      "grad_norm": 0.19600412249565125,
      "learning_rate": 1.865330854245369e-05,
      "loss": 0.4359,
      "num_input_tokens_seen": 15553184,
      "step": 23795
    },
    {
      "epoch": 12.473794549266247,
      "grad_norm": 0.16944432258605957,
      "learning_rate": 1.8642249520763804e-05,
      "loss": 0.4398,
      "num_input_tokens_seen": 15557120,
      "step": 23800
    },
    {
      "epoch": 12.476415094339623,
      "grad_norm": 0.22479405999183655,
      "learning_rate": 1.863119182904278e-05,
      "loss": 0.4448,
      "num_input_tokens_seen": 15561632,
      "step": 23805
    },
    {
      "epoch": 12.479035639412999,
      "grad_norm": 0.18133915960788727,
      "learning_rate": 1.8620135469603777e-05,
      "loss": 0.3173,
      "num_input_tokens_seen": 15568736,
      "step": 23810
    },
    {
      "epoch": 12.481656184486374,
      "grad_norm": 0.13944637775421143,
      "learning_rate": 1.8609080444759653e-05,
      "loss": 0.5498,
      "num_input_tokens_seen": 15572352,
      "step": 23815
    },
    {
      "epoch": 12.484276729559749,
      "grad_norm": 0.2588962912559509,
      "learning_rate": 1.8598026756822978e-05,
      "loss": 0.4299,
      "num_input_tokens_seen": 15574688,
      "step": 23820
    },
    {
      "epoch": 12.486897274633124,
      "grad_norm": 0.15254192054271698,
      "learning_rate": 1.8586974408106074e-05,
      "loss": 0.543,
      "num_input_tokens_seen": 15577376,
      "step": 23825
    },
    {
      "epoch": 12.4895178197065,
      "grad_norm": 0.15835078060626984,
      "learning_rate": 1.8575923400920954e-05,
      "loss": 0.5091,
      "num_input_tokens_seen": 15581280,
      "step": 23830
    },
    {
      "epoch": 12.492138364779874,
      "grad_norm": 0.1531577855348587,
      "learning_rate": 1.856487373757937e-05,
      "loss": 0.5281,
      "num_input_tokens_seen": 15584544,
      "step": 23835
    },
    {
      "epoch": 12.49475890985325,
      "grad_norm": 0.15906760096549988,
      "learning_rate": 1.8553825420392792e-05,
      "loss": 0.4644,
      "num_input_tokens_seen": 15587520,
      "step": 23840
    },
    {
      "epoch": 12.497379454926625,
      "grad_norm": 0.3091713786125183,
      "learning_rate": 1.8542778451672385e-05,
      "loss": 0.7649,
      "num_input_tokens_seen": 15590624,
      "step": 23845
    },
    {
      "epoch": 12.5,
      "grad_norm": 0.1131662130355835,
      "learning_rate": 1.8531732833729063e-05,
      "loss": 0.5291,
      "num_input_tokens_seen": 15594560,
      "step": 23850
    },
    {
      "epoch": 12.502620545073375,
      "grad_norm": 0.12885959446430206,
      "learning_rate": 1.852068856887343e-05,
      "loss": 0.6255,
      "num_input_tokens_seen": 15598048,
      "step": 23855
    },
    {
      "epoch": 12.50524109014675,
      "grad_norm": 0.13586226105690002,
      "learning_rate": 1.8509645659415837e-05,
      "loss": 0.5241,
      "num_input_tokens_seen": 15602048,
      "step": 23860
    },
    {
      "epoch": 12.507861635220126,
      "grad_norm": 0.22054751217365265,
      "learning_rate": 1.849860410766632e-05,
      "loss": 0.5878,
      "num_input_tokens_seen": 15605152,
      "step": 23865
    },
    {
      "epoch": 12.5104821802935,
      "grad_norm": 0.2935551702976227,
      "learning_rate": 1.8487563915934664e-05,
      "loss": 0.5152,
      "num_input_tokens_seen": 15607808,
      "step": 23870
    },
    {
      "epoch": 12.513102725366876,
      "grad_norm": 0.044481512159109116,
      "learning_rate": 1.847652508653033e-05,
      "loss": 0.4866,
      "num_input_tokens_seen": 15613344,
      "step": 23875
    },
    {
      "epoch": 12.515723270440251,
      "grad_norm": 0.12421856820583344,
      "learning_rate": 1.846548762176252e-05,
      "loss": 0.5846,
      "num_input_tokens_seen": 15616128,
      "step": 23880
    },
    {
      "epoch": 12.518343815513626,
      "grad_norm": 0.13851316273212433,
      "learning_rate": 1.845445152394015e-05,
      "loss": 0.5312,
      "num_input_tokens_seen": 15619328,
      "step": 23885
    },
    {
      "epoch": 12.520964360587001,
      "grad_norm": 0.11609542369842529,
      "learning_rate": 1.8443416795371853e-05,
      "loss": 0.5177,
      "num_input_tokens_seen": 15622624,
      "step": 23890
    },
    {
      "epoch": 12.523584905660378,
      "grad_norm": 0.10576795786619186,
      "learning_rate": 1.8432383438365957e-05,
      "loss": 0.3959,
      "num_input_tokens_seen": 15626560,
      "step": 23895
    },
    {
      "epoch": 12.526205450733752,
      "grad_norm": 0.13392958045005798,
      "learning_rate": 1.8421351455230528e-05,
      "loss": 0.5704,
      "num_input_tokens_seen": 15629888,
      "step": 23900
    },
    {
      "epoch": 12.528825995807129,
      "grad_norm": 0.168319970369339,
      "learning_rate": 1.8410320848273315e-05,
      "loss": 0.7004,
      "num_input_tokens_seen": 15633312,
      "step": 23905
    },
    {
      "epoch": 12.531446540880504,
      "grad_norm": 0.17815935611724854,
      "learning_rate": 1.8399291619801794e-05,
      "loss": 0.5269,
      "num_input_tokens_seen": 15636096,
      "step": 23910
    },
    {
      "epoch": 12.534067085953879,
      "grad_norm": 0.4910888075828552,
      "learning_rate": 1.838826377212317e-05,
      "loss": 0.5009,
      "num_input_tokens_seen": 15638496,
      "step": 23915
    },
    {
      "epoch": 12.536687631027254,
      "grad_norm": 0.17933793365955353,
      "learning_rate": 1.8377237307544326e-05,
      "loss": 0.3912,
      "num_input_tokens_seen": 15642848,
      "step": 23920
    },
    {
      "epoch": 12.53930817610063,
      "grad_norm": 0.11068574339151382,
      "learning_rate": 1.8366212228371894e-05,
      "loss": 0.434,
      "num_input_tokens_seen": 15646080,
      "step": 23925
    },
    {
      "epoch": 12.541928721174004,
      "grad_norm": 0.14684323966503143,
      "learning_rate": 1.8355188536912173e-05,
      "loss": 0.4059,
      "num_input_tokens_seen": 15648928,
      "step": 23930
    },
    {
      "epoch": 12.54454926624738,
      "grad_norm": 0.15746676921844482,
      "learning_rate": 1.83441662354712e-05,
      "loss": 0.5258,
      "num_input_tokens_seen": 15652608,
      "step": 23935
    },
    {
      "epoch": 12.547169811320755,
      "grad_norm": 0.11032804101705551,
      "learning_rate": 1.8333145326354722e-05,
      "loss": 0.4067,
      "num_input_tokens_seen": 15656576,
      "step": 23940
    },
    {
      "epoch": 12.54979035639413,
      "grad_norm": 0.1600068211555481,
      "learning_rate": 1.8322125811868175e-05,
      "loss": 0.5382,
      "num_input_tokens_seen": 15659264,
      "step": 23945
    },
    {
      "epoch": 12.552410901467505,
      "grad_norm": 0.302293598651886,
      "learning_rate": 1.831110769431673e-05,
      "loss": 0.4422,
      "num_input_tokens_seen": 15662272,
      "step": 23950
    },
    {
      "epoch": 12.55503144654088,
      "grad_norm": 0.2021394968032837,
      "learning_rate": 1.8300090976005247e-05,
      "loss": 0.4074,
      "num_input_tokens_seen": 15665184,
      "step": 23955
    },
    {
      "epoch": 12.557651991614255,
      "grad_norm": 0.15233489871025085,
      "learning_rate": 1.8289075659238285e-05,
      "loss": 0.3988,
      "num_input_tokens_seen": 15668832,
      "step": 23960
    },
    {
      "epoch": 12.56027253668763,
      "grad_norm": 0.12459663301706314,
      "learning_rate": 1.8278061746320134e-05,
      "loss": 0.5761,
      "num_input_tokens_seen": 15672160,
      "step": 23965
    },
    {
      "epoch": 12.562893081761006,
      "grad_norm": 0.10484962165355682,
      "learning_rate": 1.8267049239554786e-05,
      "loss": 0.4196,
      "num_input_tokens_seen": 15675296,
      "step": 23970
    },
    {
      "epoch": 12.565513626834381,
      "grad_norm": 0.19400125741958618,
      "learning_rate": 1.8256038141245912e-05,
      "loss": 0.474,
      "num_input_tokens_seen": 15678784,
      "step": 23975
    },
    {
      "epoch": 12.568134171907756,
      "grad_norm": 0.1449608951807022,
      "learning_rate": 1.824502845369693e-05,
      "loss": 0.5133,
      "num_input_tokens_seen": 15680992,
      "step": 23980
    },
    {
      "epoch": 12.570754716981131,
      "grad_norm": 0.1799064725637436,
      "learning_rate": 1.8234020179210938e-05,
      "loss": 0.4715,
      "num_input_tokens_seen": 15683872,
      "step": 23985
    },
    {
      "epoch": 12.573375262054507,
      "grad_norm": 0.17955265939235687,
      "learning_rate": 1.8223013320090715e-05,
      "loss": 0.4947,
      "num_input_tokens_seen": 15686432,
      "step": 23990
    },
    {
      "epoch": 12.575995807127882,
      "grad_norm": 0.12420953810214996,
      "learning_rate": 1.82120078786388e-05,
      "loss": 0.3896,
      "num_input_tokens_seen": 15691296,
      "step": 23995
    },
    {
      "epoch": 12.578616352201259,
      "grad_norm": 0.12934023141860962,
      "learning_rate": 1.820100385715739e-05,
      "loss": 0.6119,
      "num_input_tokens_seen": 15695712,
      "step": 24000
    },
    {
      "epoch": 12.581236897274634,
      "grad_norm": 0.11596029996871948,
      "learning_rate": 1.8190001257948414e-05,
      "loss": 0.4024,
      "num_input_tokens_seen": 15699264,
      "step": 24005
    },
    {
      "epoch": 12.583857442348009,
      "grad_norm": 0.1409575343132019,
      "learning_rate": 1.8179000083313483e-05,
      "loss": 0.4953,
      "num_input_tokens_seen": 15701952,
      "step": 24010
    },
    {
      "epoch": 12.586477987421384,
      "grad_norm": 0.13601747155189514,
      "learning_rate": 1.8168000335553907e-05,
      "loss": 0.4006,
      "num_input_tokens_seen": 15704768,
      "step": 24015
    },
    {
      "epoch": 12.58909853249476,
      "grad_norm": 0.17316684126853943,
      "learning_rate": 1.8157002016970723e-05,
      "loss": 0.5796,
      "num_input_tokens_seen": 15708352,
      "step": 24020
    },
    {
      "epoch": 12.591719077568134,
      "grad_norm": 0.17918556928634644,
      "learning_rate": 1.8146005129864644e-05,
      "loss": 0.4983,
      "num_input_tokens_seen": 15711424,
      "step": 24025
    },
    {
      "epoch": 12.59433962264151,
      "grad_norm": 0.18345260620117188,
      "learning_rate": 1.81350096765361e-05,
      "loss": 0.5082,
      "num_input_tokens_seen": 15714848,
      "step": 24030
    },
    {
      "epoch": 12.596960167714885,
      "grad_norm": 0.3015385866165161,
      "learning_rate": 1.812401565928521e-05,
      "loss": 0.5218,
      "num_input_tokens_seen": 15717664,
      "step": 24035
    },
    {
      "epoch": 12.59958071278826,
      "grad_norm": 0.15876904129981995,
      "learning_rate": 1.8113023080411806e-05,
      "loss": 0.5127,
      "num_input_tokens_seen": 15720448,
      "step": 24040
    },
    {
      "epoch": 12.602201257861635,
      "grad_norm": 0.2598276436328888,
      "learning_rate": 1.810203194221539e-05,
      "loss": 0.4675,
      "num_input_tokens_seen": 15723552,
      "step": 24045
    },
    {
      "epoch": 12.60482180293501,
      "grad_norm": 0.2312670201063156,
      "learning_rate": 1.8091042246995206e-05,
      "loss": 0.4148,
      "num_input_tokens_seen": 15727616,
      "step": 24050
    },
    {
      "epoch": 12.607442348008385,
      "grad_norm": 0.2129259556531906,
      "learning_rate": 1.8080053997050154e-05,
      "loss": 0.5226,
      "num_input_tokens_seen": 15730240,
      "step": 24055
    },
    {
      "epoch": 12.61006289308176,
      "grad_norm": 0.13722297549247742,
      "learning_rate": 1.8069067194678867e-05,
      "loss": 0.5536,
      "num_input_tokens_seen": 15734048,
      "step": 24060
    },
    {
      "epoch": 12.612683438155136,
      "grad_norm": 0.22918455302715302,
      "learning_rate": 1.8058081842179642e-05,
      "loss": 0.4741,
      "num_input_tokens_seen": 15736896,
      "step": 24065
    },
    {
      "epoch": 12.615303983228511,
      "grad_norm": 0.19877424836158752,
      "learning_rate": 1.804709794185051e-05,
      "loss": 0.5355,
      "num_input_tokens_seen": 15739872,
      "step": 24070
    },
    {
      "epoch": 12.617924528301886,
      "grad_norm": 0.07123308628797531,
      "learning_rate": 1.8036115495989164e-05,
      "loss": 0.4836,
      "num_input_tokens_seen": 15743200,
      "step": 24075
    },
    {
      "epoch": 12.620545073375261,
      "grad_norm": 0.13736797869205475,
      "learning_rate": 1.8025134506892997e-05,
      "loss": 0.4681,
      "num_input_tokens_seen": 15748224,
      "step": 24080
    },
    {
      "epoch": 12.623165618448636,
      "grad_norm": 0.21815446019172668,
      "learning_rate": 1.8014154976859126e-05,
      "loss": 0.4548,
      "num_input_tokens_seen": 15751168,
      "step": 24085
    },
    {
      "epoch": 12.625786163522012,
      "grad_norm": 0.11214003711938858,
      "learning_rate": 1.800317690818433e-05,
      "loss": 0.4187,
      "num_input_tokens_seen": 15754592,
      "step": 24090
    },
    {
      "epoch": 12.628406708595389,
      "grad_norm": 0.17245334386825562,
      "learning_rate": 1.799220030316511e-05,
      "loss": 0.4669,
      "num_input_tokens_seen": 15758368,
      "step": 24095
    },
    {
      "epoch": 12.631027253668764,
      "grad_norm": 0.15951940417289734,
      "learning_rate": 1.7981225164097633e-05,
      "loss": 0.4691,
      "num_input_tokens_seen": 15761504,
      "step": 24100
    },
    {
      "epoch": 12.633647798742139,
      "grad_norm": 0.13695621490478516,
      "learning_rate": 1.7970251493277774e-05,
      "loss": 0.5224,
      "num_input_tokens_seen": 15764224,
      "step": 24105
    },
    {
      "epoch": 12.636268343815514,
      "grad_norm": 0.24995505809783936,
      "learning_rate": 1.7959279293001102e-05,
      "loss": 0.4905,
      "num_input_tokens_seen": 15767072,
      "step": 24110
    },
    {
      "epoch": 12.63888888888889,
      "grad_norm": 0.35949477553367615,
      "learning_rate": 1.7948308565562876e-05,
      "loss": 0.7136,
      "num_input_tokens_seen": 15769920,
      "step": 24115
    },
    {
      "epoch": 12.641509433962264,
      "grad_norm": 0.24847373366355896,
      "learning_rate": 1.793733931325805e-05,
      "loss": 0.5506,
      "num_input_tokens_seen": 15773472,
      "step": 24120
    },
    {
      "epoch": 12.64412997903564,
      "grad_norm": 0.1908087283372879,
      "learning_rate": 1.792637153838127e-05,
      "loss": 0.4676,
      "num_input_tokens_seen": 15776064,
      "step": 24125
    },
    {
      "epoch": 12.646750524109015,
      "grad_norm": 0.14210960268974304,
      "learning_rate": 1.7915405243226864e-05,
      "loss": 0.4689,
      "num_input_tokens_seen": 15779552,
      "step": 24130
    },
    {
      "epoch": 12.64937106918239,
      "grad_norm": 0.11192979663610458,
      "learning_rate": 1.7904440430088847e-05,
      "loss": 0.3099,
      "num_input_tokens_seen": 15782272,
      "step": 24135
    },
    {
      "epoch": 12.651991614255765,
      "grad_norm": 0.3743188679218292,
      "learning_rate": 1.7893477101260947e-05,
      "loss": 0.6109,
      "num_input_tokens_seen": 15785024,
      "step": 24140
    },
    {
      "epoch": 12.65461215932914,
      "grad_norm": 0.1614103764295578,
      "learning_rate": 1.7882515259036558e-05,
      "loss": 0.7059,
      "num_input_tokens_seen": 15788128,
      "step": 24145
    },
    {
      "epoch": 12.657232704402515,
      "grad_norm": 0.20971094071865082,
      "learning_rate": 1.787155490570878e-05,
      "loss": 0.4533,
      "num_input_tokens_seen": 15790944,
      "step": 24150
    },
    {
      "epoch": 12.65985324947589,
      "grad_norm": 0.17651012539863586,
      "learning_rate": 1.7860596043570398e-05,
      "loss": 0.5253,
      "num_input_tokens_seen": 15793920,
      "step": 24155
    },
    {
      "epoch": 12.662473794549266,
      "grad_norm": 0.08989953994750977,
      "learning_rate": 1.7849638674913863e-05,
      "loss": 0.3099,
      "num_input_tokens_seen": 15799552,
      "step": 24160
    },
    {
      "epoch": 12.665094339622641,
      "grad_norm": 0.14308124780654907,
      "learning_rate": 1.7838682802031344e-05,
      "loss": 0.4699,
      "num_input_tokens_seen": 15802656,
      "step": 24165
    },
    {
      "epoch": 12.667714884696016,
      "grad_norm": 0.22403712570667267,
      "learning_rate": 1.782772842721468e-05,
      "loss": 0.5161,
      "num_input_tokens_seen": 15806112,
      "step": 24170
    },
    {
      "epoch": 12.670335429769391,
      "grad_norm": 0.2028873860836029,
      "learning_rate": 1.7816775552755406e-05,
      "loss": 0.4821,
      "num_input_tokens_seen": 15808928,
      "step": 24175
    },
    {
      "epoch": 12.672955974842766,
      "grad_norm": 0.22223393619060516,
      "learning_rate": 1.780582418094473e-05,
      "loss": 0.5031,
      "num_input_tokens_seen": 15812320,
      "step": 24180
    },
    {
      "epoch": 12.675576519916142,
      "grad_norm": 0.10249220579862595,
      "learning_rate": 1.7794874314073573e-05,
      "loss": 0.4506,
      "num_input_tokens_seen": 15816608,
      "step": 24185
    },
    {
      "epoch": 12.678197064989519,
      "grad_norm": 0.15247978270053864,
      "learning_rate": 1.7783925954432503e-05,
      "loss": 0.4803,
      "num_input_tokens_seen": 15819680,
      "step": 24190
    },
    {
      "epoch": 12.680817610062894,
      "grad_norm": 0.18198595941066742,
      "learning_rate": 1.777297910431179e-05,
      "loss": 0.5206,
      "num_input_tokens_seen": 15822848,
      "step": 24195
    },
    {
      "epoch": 12.683438155136269,
      "grad_norm": 0.1075211837887764,
      "learning_rate": 1.7762033766001402e-05,
      "loss": 0.6208,
      "num_input_tokens_seen": 15825312,
      "step": 24200
    },
    {
      "epoch": 12.686058700209644,
      "grad_norm": 0.0891597643494606,
      "learning_rate": 1.7751089941790977e-05,
      "loss": 0.3577,
      "num_input_tokens_seen": 15828672,
      "step": 24205
    },
    {
      "epoch": 12.68867924528302,
      "grad_norm": 0.18323807418346405,
      "learning_rate": 1.7740147633969844e-05,
      "loss": 0.5215,
      "num_input_tokens_seen": 15831648,
      "step": 24210
    },
    {
      "epoch": 12.691299790356394,
      "grad_norm": 0.14940644800662994,
      "learning_rate": 1.772920684482699e-05,
      "loss": 0.5121,
      "num_input_tokens_seen": 15835456,
      "step": 24215
    },
    {
      "epoch": 12.69392033542977,
      "grad_norm": 0.17155924439430237,
      "learning_rate": 1.771826757665112e-05,
      "loss": 0.3845,
      "num_input_tokens_seen": 15839008,
      "step": 24220
    },
    {
      "epoch": 12.696540880503145,
      "grad_norm": 0.21967677772045135,
      "learning_rate": 1.770732983173059e-05,
      "loss": 0.5808,
      "num_input_tokens_seen": 15842048,
      "step": 24225
    },
    {
      "epoch": 12.69916142557652,
      "grad_norm": 0.21287265419960022,
      "learning_rate": 1.7696393612353467e-05,
      "loss": 0.503,
      "num_input_tokens_seen": 15844992,
      "step": 24230
    },
    {
      "epoch": 12.701781970649895,
      "grad_norm": 0.20351524651050568,
      "learning_rate": 1.7685458920807474e-05,
      "loss": 0.328,
      "num_input_tokens_seen": 15847008,
      "step": 24235
    },
    {
      "epoch": 12.70440251572327,
      "grad_norm": 0.1799158751964569,
      "learning_rate": 1.7674525759380033e-05,
      "loss": 0.6095,
      "num_input_tokens_seen": 15849760,
      "step": 24240
    },
    {
      "epoch": 12.707023060796645,
      "grad_norm": 0.12950684130191803,
      "learning_rate": 1.766359413035823e-05,
      "loss": 0.4192,
      "num_input_tokens_seen": 15852608,
      "step": 24245
    },
    {
      "epoch": 12.70964360587002,
      "grad_norm": 0.21552185714244843,
      "learning_rate": 1.7652664036028827e-05,
      "loss": 0.4514,
      "num_input_tokens_seen": 15855904,
      "step": 24250
    },
    {
      "epoch": 12.712264150943396,
      "grad_norm": 0.16250699758529663,
      "learning_rate": 1.7641735478678288e-05,
      "loss": 0.4357,
      "num_input_tokens_seen": 15860224,
      "step": 24255
    },
    {
      "epoch": 12.714884696016771,
      "grad_norm": 0.22509078681468964,
      "learning_rate": 1.763080846059274e-05,
      "loss": 0.441,
      "num_input_tokens_seen": 15863136,
      "step": 24260
    },
    {
      "epoch": 12.717505241090146,
      "grad_norm": 0.22117304801940918,
      "learning_rate": 1.7619882984057987e-05,
      "loss": 0.5355,
      "num_input_tokens_seen": 15866400,
      "step": 24265
    },
    {
      "epoch": 12.720125786163521,
      "grad_norm": 0.19462938606739044,
      "learning_rate": 1.760895905135953e-05,
      "loss": 0.5135,
      "num_input_tokens_seen": 15868896,
      "step": 24270
    },
    {
      "epoch": 12.722746331236896,
      "grad_norm": 0.21911202371120453,
      "learning_rate": 1.7598036664782508e-05,
      "loss": 0.5645,
      "num_input_tokens_seen": 15871232,
      "step": 24275
    },
    {
      "epoch": 12.725366876310272,
      "grad_norm": 0.146309494972229,
      "learning_rate": 1.7587115826611766e-05,
      "loss": 0.3662,
      "num_input_tokens_seen": 15876800,
      "step": 24280
    },
    {
      "epoch": 12.727987421383649,
      "grad_norm": 0.25216662883758545,
      "learning_rate": 1.7576196539131828e-05,
      "loss": 0.2921,
      "num_input_tokens_seen": 15879520,
      "step": 24285
    },
    {
      "epoch": 12.730607966457024,
      "grad_norm": 0.14638814330101013,
      "learning_rate": 1.7565278804626872e-05,
      "loss": 0.4743,
      "num_input_tokens_seen": 15882432,
      "step": 24290
    },
    {
      "epoch": 12.733228511530399,
      "grad_norm": 0.11037851870059967,
      "learning_rate": 1.755436262538078e-05,
      "loss": 0.5333,
      "num_input_tokens_seen": 15887648,
      "step": 24295
    },
    {
      "epoch": 12.735849056603774,
      "grad_norm": 0.30871349573135376,
      "learning_rate": 1.7543448003677082e-05,
      "loss": 0.4356,
      "num_input_tokens_seen": 15892704,
      "step": 24300
    },
    {
      "epoch": 12.73846960167715,
      "grad_norm": 0.11369581520557404,
      "learning_rate": 1.7532534941798985e-05,
      "loss": 0.463,
      "num_input_tokens_seen": 15895840,
      "step": 24305
    },
    {
      "epoch": 12.741090146750524,
      "grad_norm": 0.26288801431655884,
      "learning_rate": 1.752162344202939e-05,
      "loss": 0.6211,
      "num_input_tokens_seen": 15899488,
      "step": 24310
    },
    {
      "epoch": 12.7437106918239,
      "grad_norm": 0.1603904366493225,
      "learning_rate": 1.751071350665085e-05,
      "loss": 0.4273,
      "num_input_tokens_seen": 15902816,
      "step": 24315
    },
    {
      "epoch": 12.746331236897275,
      "grad_norm": 0.18460851907730103,
      "learning_rate": 1.7499805137945602e-05,
      "loss": 0.5215,
      "num_input_tokens_seen": 15905568,
      "step": 24320
    },
    {
      "epoch": 12.74895178197065,
      "grad_norm": 0.1277153491973877,
      "learning_rate": 1.7488898338195566e-05,
      "loss": 0.4532,
      "num_input_tokens_seen": 15909120,
      "step": 24325
    },
    {
      "epoch": 12.751572327044025,
      "grad_norm": 0.1298418790102005,
      "learning_rate": 1.7477993109682288e-05,
      "loss": 0.4566,
      "num_input_tokens_seen": 15911872,
      "step": 24330
    },
    {
      "epoch": 12.7541928721174,
      "grad_norm": 0.15938138961791992,
      "learning_rate": 1.7467089454687043e-05,
      "loss": 0.4779,
      "num_input_tokens_seen": 15914784,
      "step": 24335
    },
    {
      "epoch": 12.756813417190775,
      "grad_norm": 0.17241710424423218,
      "learning_rate": 1.745618737549074e-05,
      "loss": 0.4839,
      "num_input_tokens_seen": 15918944,
      "step": 24340
    },
    {
      "epoch": 12.75943396226415,
      "grad_norm": 0.20380975306034088,
      "learning_rate": 1.7445286874373974e-05,
      "loss": 0.5553,
      "num_input_tokens_seen": 15921312,
      "step": 24345
    },
    {
      "epoch": 12.762054507337526,
      "grad_norm": 0.18619824945926666,
      "learning_rate": 1.743438795361701e-05,
      "loss": 0.4886,
      "num_input_tokens_seen": 15924320,
      "step": 24350
    },
    {
      "epoch": 12.764675052410901,
      "grad_norm": 0.14127063751220703,
      "learning_rate": 1.742349061549978e-05,
      "loss": 0.4225,
      "num_input_tokens_seen": 15927936,
      "step": 24355
    },
    {
      "epoch": 12.767295597484276,
      "grad_norm": 0.19809025526046753,
      "learning_rate": 1.741259486230186e-05,
      "loss": 0.713,
      "num_input_tokens_seen": 15931104,
      "step": 24360
    },
    {
      "epoch": 12.769916142557651,
      "grad_norm": 0.35203248262405396,
      "learning_rate": 1.7401700696302546e-05,
      "loss": 0.5494,
      "num_input_tokens_seen": 15935008,
      "step": 24365
    },
    {
      "epoch": 12.772536687631026,
      "grad_norm": 0.2000313252210617,
      "learning_rate": 1.739080811978075e-05,
      "loss": 0.4841,
      "num_input_tokens_seen": 15937824,
      "step": 24370
    },
    {
      "epoch": 12.775157232704402,
      "grad_norm": 0.15843766927719116,
      "learning_rate": 1.7379917135015093e-05,
      "loss": 0.4284,
      "num_input_tokens_seen": 15941056,
      "step": 24375
    },
    {
      "epoch": 12.777777777777779,
      "grad_norm": 0.14634636044502258,
      "learning_rate": 1.736902774428384e-05,
      "loss": 0.5105,
      "num_input_tokens_seen": 15944640,
      "step": 24380
    },
    {
      "epoch": 12.780398322851154,
      "grad_norm": 0.1348332166671753,
      "learning_rate": 1.735813994986491e-05,
      "loss": 0.3883,
      "num_input_tokens_seen": 15948384,
      "step": 24385
    },
    {
      "epoch": 12.783018867924529,
      "grad_norm": 0.24074769020080566,
      "learning_rate": 1.7347253754035933e-05,
      "loss": 0.4343,
      "num_input_tokens_seen": 15951232,
      "step": 24390
    },
    {
      "epoch": 12.785639412997904,
      "grad_norm": 0.21343198418617249,
      "learning_rate": 1.7336369159074154e-05,
      "loss": 0.4529,
      "num_input_tokens_seen": 15954816,
      "step": 24395
    },
    {
      "epoch": 12.78825995807128,
      "grad_norm": 0.141118586063385,
      "learning_rate": 1.7325486167256522e-05,
      "loss": 0.6348,
      "num_input_tokens_seen": 15958336,
      "step": 24400
    },
    {
      "epoch": 12.790880503144654,
      "grad_norm": 0.09385410696268082,
      "learning_rate": 1.7314604780859623e-05,
      "loss": 0.5059,
      "num_input_tokens_seen": 15961984,
      "step": 24405
    },
    {
      "epoch": 12.79350104821803,
      "grad_norm": 0.15401533246040344,
      "learning_rate": 1.730372500215973e-05,
      "loss": 0.4376,
      "num_input_tokens_seen": 15965056,
      "step": 24410
    },
    {
      "epoch": 12.796121593291405,
      "grad_norm": 0.1444540172815323,
      "learning_rate": 1.7292846833432763e-05,
      "loss": 0.5022,
      "num_input_tokens_seen": 15967904,
      "step": 24415
    },
    {
      "epoch": 12.79874213836478,
      "grad_norm": 0.12687504291534424,
      "learning_rate": 1.7281970276954305e-05,
      "loss": 0.4241,
      "num_input_tokens_seen": 15971424,
      "step": 24420
    },
    {
      "epoch": 12.801362683438155,
      "grad_norm": 0.270757794380188,
      "learning_rate": 1.727109533499961e-05,
      "loss": 0.7426,
      "num_input_tokens_seen": 15974272,
      "step": 24425
    },
    {
      "epoch": 12.80398322851153,
      "grad_norm": 0.20427510142326355,
      "learning_rate": 1.72602220098436e-05,
      "loss": 0.4456,
      "num_input_tokens_seen": 15977600,
      "step": 24430
    },
    {
      "epoch": 12.806603773584905,
      "grad_norm": 0.14908593893051147,
      "learning_rate": 1.7249350303760846e-05,
      "loss": 0.5172,
      "num_input_tokens_seen": 15980640,
      "step": 24435
    },
    {
      "epoch": 12.80922431865828,
      "grad_norm": 0.13206112384796143,
      "learning_rate": 1.7238480219025593e-05,
      "loss": 0.4316,
      "num_input_tokens_seen": 15983520,
      "step": 24440
    },
    {
      "epoch": 12.811844863731656,
      "grad_norm": 0.2190696746110916,
      "learning_rate": 1.7227611757911723e-05,
      "loss": 0.534,
      "num_input_tokens_seen": 15986592,
      "step": 24445
    },
    {
      "epoch": 12.814465408805031,
      "grad_norm": 0.13753919303417206,
      "learning_rate": 1.7216744922692794e-05,
      "loss": 0.5288,
      "num_input_tokens_seen": 15989600,
      "step": 24450
    },
    {
      "epoch": 12.817085953878406,
      "grad_norm": 0.17121677100658417,
      "learning_rate": 1.7205879715642037e-05,
      "loss": 0.4555,
      "num_input_tokens_seen": 15992416,
      "step": 24455
    },
    {
      "epoch": 12.819706498951781,
      "grad_norm": 0.18765375018119812,
      "learning_rate": 1.719501613903232e-05,
      "loss": 0.3931,
      "num_input_tokens_seen": 15994816,
      "step": 24460
    },
    {
      "epoch": 12.822327044025156,
      "grad_norm": 0.2956971228122711,
      "learning_rate": 1.7184154195136192e-05,
      "loss": 0.537,
      "num_input_tokens_seen": 15997536,
      "step": 24465
    },
    {
      "epoch": 12.824947589098532,
      "grad_norm": 0.133445143699646,
      "learning_rate": 1.7173293886225837e-05,
      "loss": 0.452,
      "num_input_tokens_seen": 16001344,
      "step": 24470
    },
    {
      "epoch": 12.827568134171909,
      "grad_norm": 0.18120086193084717,
      "learning_rate": 1.71624352145731e-05,
      "loss": 0.3731,
      "num_input_tokens_seen": 16005440,
      "step": 24475
    },
    {
      "epoch": 12.830188679245284,
      "grad_norm": 0.11980416625738144,
      "learning_rate": 1.7151578182449508e-05,
      "loss": 0.4393,
      "num_input_tokens_seen": 16008992,
      "step": 24480
    },
    {
      "epoch": 12.832809224318659,
      "grad_norm": 0.14406818151474,
      "learning_rate": 1.7140722792126212e-05,
      "loss": 0.459,
      "num_input_tokens_seen": 16011712,
      "step": 24485
    },
    {
      "epoch": 12.835429769392034,
      "grad_norm": 0.11108652502298355,
      "learning_rate": 1.7129869045874053e-05,
      "loss": 0.4369,
      "num_input_tokens_seen": 16014528,
      "step": 24490
    },
    {
      "epoch": 12.83805031446541,
      "grad_norm": 0.15583717823028564,
      "learning_rate": 1.7119016945963506e-05,
      "loss": 0.385,
      "num_input_tokens_seen": 16016736,
      "step": 24495
    },
    {
      "epoch": 12.840670859538784,
      "grad_norm": 0.265992134809494,
      "learning_rate": 1.7108166494664694e-05,
      "loss": 0.3676,
      "num_input_tokens_seen": 16020128,
      "step": 24500
    },
    {
      "epoch": 12.84329140461216,
      "grad_norm": 0.15809403359889984,
      "learning_rate": 1.7097317694247417e-05,
      "loss": 0.4342,
      "num_input_tokens_seen": 16023008,
      "step": 24505
    },
    {
      "epoch": 12.845911949685535,
      "grad_norm": 0.1439720094203949,
      "learning_rate": 1.7086470546981122e-05,
      "loss": 0.4091,
      "num_input_tokens_seen": 16026656,
      "step": 24510
    },
    {
      "epoch": 12.84853249475891,
      "grad_norm": 0.2773262560367584,
      "learning_rate": 1.7075625055134904e-05,
      "loss": 0.4377,
      "num_input_tokens_seen": 16029472,
      "step": 24515
    },
    {
      "epoch": 12.851153039832285,
      "grad_norm": 0.13985596597194672,
      "learning_rate": 1.706478122097752e-05,
      "loss": 0.4731,
      "num_input_tokens_seen": 16033504,
      "step": 24520
    },
    {
      "epoch": 12.85377358490566,
      "grad_norm": 0.2834948003292084,
      "learning_rate": 1.7053939046777384e-05,
      "loss": 0.4322,
      "num_input_tokens_seen": 16036480,
      "step": 24525
    },
    {
      "epoch": 12.856394129979035,
      "grad_norm": 0.20352864265441895,
      "learning_rate": 1.7043098534802533e-05,
      "loss": 0.5813,
      "num_input_tokens_seen": 16039072,
      "step": 24530
    },
    {
      "epoch": 12.85901467505241,
      "grad_norm": 0.15329988300800323,
      "learning_rate": 1.7032259687320695e-05,
      "loss": 0.4278,
      "num_input_tokens_seen": 16042144,
      "step": 24535
    },
    {
      "epoch": 12.861635220125786,
      "grad_norm": 0.1628599762916565,
      "learning_rate": 1.7021422506599228e-05,
      "loss": 0.4923,
      "num_input_tokens_seen": 16044672,
      "step": 24540
    },
    {
      "epoch": 12.864255765199161,
      "grad_norm": 0.15416781604290009,
      "learning_rate": 1.7010586994905155e-05,
      "loss": 0.4671,
      "num_input_tokens_seen": 16046912,
      "step": 24545
    },
    {
      "epoch": 12.866876310272536,
      "grad_norm": 0.15677526593208313,
      "learning_rate": 1.6999753154505123e-05,
      "loss": 0.4404,
      "num_input_tokens_seen": 16049920,
      "step": 24550
    },
    {
      "epoch": 12.869496855345911,
      "grad_norm": 0.11935121566057205,
      "learning_rate": 1.6988920987665476e-05,
      "loss": 0.4032,
      "num_input_tokens_seen": 16053952,
      "step": 24555
    },
    {
      "epoch": 12.872117400419286,
      "grad_norm": 0.17613953351974487,
      "learning_rate": 1.6978090496652156e-05,
      "loss": 0.4799,
      "num_input_tokens_seen": 16056352,
      "step": 24560
    },
    {
      "epoch": 12.874737945492662,
      "grad_norm": 0.1407698094844818,
      "learning_rate": 1.6967261683730778e-05,
      "loss": 0.4641,
      "num_input_tokens_seen": 16060608,
      "step": 24565
    },
    {
      "epoch": 12.877358490566039,
      "grad_norm": 0.09177940338850021,
      "learning_rate": 1.695643455116662e-05,
      "loss": 0.4918,
      "num_input_tokens_seen": 16063904,
      "step": 24570
    },
    {
      "epoch": 12.879979035639414,
      "grad_norm": 0.16088996827602386,
      "learning_rate": 1.694560910122458e-05,
      "loss": 0.3812,
      "num_input_tokens_seen": 16067136,
      "step": 24575
    },
    {
      "epoch": 12.882599580712789,
      "grad_norm": 0.2062811255455017,
      "learning_rate": 1.693478533616924e-05,
      "loss": 0.4395,
      "num_input_tokens_seen": 16069696,
      "step": 24580
    },
    {
      "epoch": 12.885220125786164,
      "grad_norm": 0.14062835276126862,
      "learning_rate": 1.692396325826479e-05,
      "loss": 0.5085,
      "num_input_tokens_seen": 16072800,
      "step": 24585
    },
    {
      "epoch": 12.88784067085954,
      "grad_norm": 0.1850368082523346,
      "learning_rate": 1.6913142869775088e-05,
      "loss": 0.5152,
      "num_input_tokens_seen": 16075424,
      "step": 24590
    },
    {
      "epoch": 12.890461215932914,
      "grad_norm": 0.21935702860355377,
      "learning_rate": 1.6902324172963635e-05,
      "loss": 0.4448,
      "num_input_tokens_seen": 16078400,
      "step": 24595
    },
    {
      "epoch": 12.89308176100629,
      "grad_norm": 0.18783889710903168,
      "learning_rate": 1.6891507170093586e-05,
      "loss": 0.4552,
      "num_input_tokens_seen": 16086432,
      "step": 24600
    },
    {
      "epoch": 12.895702306079665,
      "grad_norm": 0.15723739564418793,
      "learning_rate": 1.6880691863427725e-05,
      "loss": 0.548,
      "num_input_tokens_seen": 16089472,
      "step": 24605
    },
    {
      "epoch": 12.89832285115304,
      "grad_norm": 0.1559837907552719,
      "learning_rate": 1.686987825522851e-05,
      "loss": 0.516,
      "num_input_tokens_seen": 16092800,
      "step": 24610
    },
    {
      "epoch": 12.900943396226415,
      "grad_norm": 0.10720384865999222,
      "learning_rate": 1.6859066347758002e-05,
      "loss": 0.3641,
      "num_input_tokens_seen": 16100992,
      "step": 24615
    },
    {
      "epoch": 12.90356394129979,
      "grad_norm": 0.2006036341190338,
      "learning_rate": 1.6848256143277933e-05,
      "loss": 0.4827,
      "num_input_tokens_seen": 16104320,
      "step": 24620
    },
    {
      "epoch": 12.906184486373165,
      "grad_norm": 0.1548212170600891,
      "learning_rate": 1.683744764404968e-05,
      "loss": 0.5223,
      "num_input_tokens_seen": 16108064,
      "step": 24625
    },
    {
      "epoch": 12.90880503144654,
      "grad_norm": 0.1753319650888443,
      "learning_rate": 1.6826640852334257e-05,
      "loss": 0.5027,
      "num_input_tokens_seen": 16111008,
      "step": 24630
    },
    {
      "epoch": 12.911425576519916,
      "grad_norm": 0.11866996437311172,
      "learning_rate": 1.6815835770392324e-05,
      "loss": 0.5277,
      "num_input_tokens_seen": 16114080,
      "step": 24635
    },
    {
      "epoch": 12.914046121593291,
      "grad_norm": 0.15878330171108246,
      "learning_rate": 1.6805032400484184e-05,
      "loss": 0.5448,
      "num_input_tokens_seen": 16116832,
      "step": 24640
    },
    {
      "epoch": 12.916666666666666,
      "grad_norm": 0.17535433173179626,
      "learning_rate": 1.6794230744869767e-05,
      "loss": 0.4153,
      "num_input_tokens_seen": 16119872,
      "step": 24645
    },
    {
      "epoch": 12.919287211740041,
      "grad_norm": 0.17636507749557495,
      "learning_rate": 1.6783430805808658e-05,
      "loss": 0.577,
      "num_input_tokens_seen": 16123200,
      "step": 24650
    },
    {
      "epoch": 12.921907756813416,
      "grad_norm": 0.2939537763595581,
      "learning_rate": 1.6772632585560084e-05,
      "loss": 0.4457,
      "num_input_tokens_seen": 16126048,
      "step": 24655
    },
    {
      "epoch": 12.924528301886792,
      "grad_norm": 0.16691316664218903,
      "learning_rate": 1.6761836086382908e-05,
      "loss": 0.566,
      "num_input_tokens_seen": 16129088,
      "step": 24660
    },
    {
      "epoch": 12.927148846960169,
      "grad_norm": 0.18821680545806885,
      "learning_rate": 1.6751041310535653e-05,
      "loss": 0.5917,
      "num_input_tokens_seen": 16131584,
      "step": 24665
    },
    {
      "epoch": 12.929769392033544,
      "grad_norm": 0.1672230064868927,
      "learning_rate": 1.6740248260276437e-05,
      "loss": 0.5317,
      "num_input_tokens_seen": 16133984,
      "step": 24670
    },
    {
      "epoch": 12.932389937106919,
      "grad_norm": 0.1546766757965088,
      "learning_rate": 1.6729456937863047e-05,
      "loss": 0.4853,
      "num_input_tokens_seen": 16136480,
      "step": 24675
    },
    {
      "epoch": 12.935010482180294,
      "grad_norm": 0.2430206835269928,
      "learning_rate": 1.6718667345552918e-05,
      "loss": 0.3986,
      "num_input_tokens_seen": 16139616,
      "step": 24680
    },
    {
      "epoch": 12.93763102725367,
      "grad_norm": 0.19508956372737885,
      "learning_rate": 1.6707879485603097e-05,
      "loss": 0.5594,
      "num_input_tokens_seen": 16143008,
      "step": 24685
    },
    {
      "epoch": 12.940251572327044,
      "grad_norm": 0.1786627322435379,
      "learning_rate": 1.669709336027029e-05,
      "loss": 0.5443,
      "num_input_tokens_seen": 16146304,
      "step": 24690
    },
    {
      "epoch": 12.94287211740042,
      "grad_norm": 0.24772624671459198,
      "learning_rate": 1.6686308971810833e-05,
      "loss": 0.4895,
      "num_input_tokens_seen": 16149184,
      "step": 24695
    },
    {
      "epoch": 12.945492662473795,
      "grad_norm": 0.13996157050132751,
      "learning_rate": 1.6675526322480677e-05,
      "loss": 0.5838,
      "num_input_tokens_seen": 16151968,
      "step": 24700
    },
    {
      "epoch": 12.94811320754717,
      "grad_norm": 0.21031321585178375,
      "learning_rate": 1.6664745414535454e-05,
      "loss": 0.5177,
      "num_input_tokens_seen": 16154848,
      "step": 24705
    },
    {
      "epoch": 12.950733752620545,
      "grad_norm": 0.1028059646487236,
      "learning_rate": 1.665396625023039e-05,
      "loss": 0.4538,
      "num_input_tokens_seen": 16158336,
      "step": 24710
    },
    {
      "epoch": 12.95335429769392,
      "grad_norm": 0.27083200216293335,
      "learning_rate": 1.6643188831820375e-05,
      "loss": 0.6049,
      "num_input_tokens_seen": 16161856,
      "step": 24715
    },
    {
      "epoch": 12.955974842767295,
      "grad_norm": 0.1330176591873169,
      "learning_rate": 1.6632413161559912e-05,
      "loss": 0.4541,
      "num_input_tokens_seen": 16165696,
      "step": 24720
    },
    {
      "epoch": 12.95859538784067,
      "grad_norm": 0.21006838977336884,
      "learning_rate": 1.6621639241703162e-05,
      "loss": 0.4394,
      "num_input_tokens_seen": 16169440,
      "step": 24725
    },
    {
      "epoch": 12.961215932914046,
      "grad_norm": 0.1239052340388298,
      "learning_rate": 1.6610867074503893e-05,
      "loss": 0.433,
      "num_input_tokens_seen": 16173088,
      "step": 24730
    },
    {
      "epoch": 12.963836477987421,
      "grad_norm": 0.1648169606924057,
      "learning_rate": 1.660009666221552e-05,
      "loss": 0.4964,
      "num_input_tokens_seen": 16176352,
      "step": 24735
    },
    {
      "epoch": 12.966457023060796,
      "grad_norm": 0.20599743723869324,
      "learning_rate": 1.6589328007091093e-05,
      "loss": 0.3455,
      "num_input_tokens_seen": 16180128,
      "step": 24740
    },
    {
      "epoch": 12.969077568134171,
      "grad_norm": 0.13762521743774414,
      "learning_rate": 1.65785611113833e-05,
      "loss": 0.5902,
      "num_input_tokens_seen": 16182880,
      "step": 24745
    },
    {
      "epoch": 12.971698113207546,
      "grad_norm": 0.1291622668504715,
      "learning_rate": 1.6567795977344447e-05,
      "loss": 0.4588,
      "num_input_tokens_seen": 16186880,
      "step": 24750
    },
    {
      "epoch": 12.974318658280922,
      "grad_norm": 0.18536315858364105,
      "learning_rate": 1.6557032607226466e-05,
      "loss": 0.5133,
      "num_input_tokens_seen": 16189600,
      "step": 24755
    },
    {
      "epoch": 12.976939203354299,
      "grad_norm": 0.222805455327034,
      "learning_rate": 1.654627100328095e-05,
      "loss": 0.6377,
      "num_input_tokens_seen": 16194976,
      "step": 24760
    },
    {
      "epoch": 12.979559748427674,
      "grad_norm": 0.1737290918827057,
      "learning_rate": 1.6535511167759085e-05,
      "loss": 0.5731,
      "num_input_tokens_seen": 16197536,
      "step": 24765
    },
    {
      "epoch": 12.982180293501049,
      "grad_norm": 0.1648322343826294,
      "learning_rate": 1.652475310291172e-05,
      "loss": 0.5221,
      "num_input_tokens_seen": 16201440,
      "step": 24770
    },
    {
      "epoch": 12.984800838574424,
      "grad_norm": 0.21399761736392975,
      "learning_rate": 1.6513996810989317e-05,
      "loss": 0.4818,
      "num_input_tokens_seen": 16204448,
      "step": 24775
    },
    {
      "epoch": 12.9874213836478,
      "grad_norm": 0.11920510232448578,
      "learning_rate": 1.6503242294241973e-05,
      "loss": 0.3846,
      "num_input_tokens_seen": 16207584,
      "step": 24780
    },
    {
      "epoch": 12.990041928721174,
      "grad_norm": 0.17029890418052673,
      "learning_rate": 1.6492489554919404e-05,
      "loss": 0.5716,
      "num_input_tokens_seen": 16209920,
      "step": 24785
    },
    {
      "epoch": 12.99266247379455,
      "grad_norm": 0.14324885606765747,
      "learning_rate": 1.6481738595270953e-05,
      "loss": 0.4333,
      "num_input_tokens_seen": 16212736,
      "step": 24790
    },
    {
      "epoch": 12.995283018867925,
      "grad_norm": 0.177387073636055,
      "learning_rate": 1.6470989417545608e-05,
      "loss": 0.5576,
      "num_input_tokens_seen": 16216480,
      "step": 24795
    },
    {
      "epoch": 12.9979035639413,
      "grad_norm": 0.28568512201309204,
      "learning_rate": 1.646024202399197e-05,
      "loss": 0.6152,
      "num_input_tokens_seen": 16218976,
      "step": 24800
    },
    {
      "epoch": 13.0,
      "eval_loss": 0.4937197268009186,
      "eval_runtime": 10.0983,
      "eval_samples_per_second": 83.974,
      "eval_steps_per_second": 20.994,
      "num_input_tokens_seen": 16220808,
      "step": 24804
    },
    {
      "epoch": 13.000524109014675,
      "grad_norm": 0.2124636322259903,
      "learning_rate": 1.6449496416858284e-05,
      "loss": 0.4565,
      "num_input_tokens_seen": 16221160,
      "step": 24805
    },
    {
      "epoch": 13.00314465408805,
      "grad_norm": 0.18930938839912415,
      "learning_rate": 1.64387525983924e-05,
      "loss": 0.606,
      "num_input_tokens_seen": 16223848,
      "step": 24810
    },
    {
      "epoch": 13.005765199161425,
      "grad_norm": 0.07417704164981842,
      "learning_rate": 1.6428010570841788e-05,
      "loss": 0.4103,
      "num_input_tokens_seen": 16226856,
      "step": 24815
    },
    {
      "epoch": 13.0083857442348,
      "grad_norm": 0.16965928673744202,
      "learning_rate": 1.6417270336453573e-05,
      "loss": 0.3936,
      "num_input_tokens_seen": 16229736,
      "step": 24820
    },
    {
      "epoch": 13.011006289308176,
      "grad_norm": 0.19488833844661713,
      "learning_rate": 1.6406531897474492e-05,
      "loss": 0.395,
      "num_input_tokens_seen": 16233064,
      "step": 24825
    },
    {
      "epoch": 13.01362683438155,
      "grad_norm": 0.11360309273004532,
      "learning_rate": 1.639579525615089e-05,
      "loss": 0.421,
      "num_input_tokens_seen": 16236456,
      "step": 24830
    },
    {
      "epoch": 13.016247379454926,
      "grad_norm": 0.2917569875717163,
      "learning_rate": 1.638506041472877e-05,
      "loss": 0.7184,
      "num_input_tokens_seen": 16239272,
      "step": 24835
    },
    {
      "epoch": 13.018867924528301,
      "grad_norm": 0.2196984887123108,
      "learning_rate": 1.6374327375453724e-05,
      "loss": 0.3452,
      "num_input_tokens_seen": 16242152,
      "step": 24840
    },
    {
      "epoch": 13.021488469601676,
      "grad_norm": 0.1653095930814743,
      "learning_rate": 1.6363596140570982e-05,
      "loss": 0.5,
      "num_input_tokens_seen": 16245832,
      "step": 24845
    },
    {
      "epoch": 13.024109014675052,
      "grad_norm": 0.13057337701320648,
      "learning_rate": 1.6352866712325395e-05,
      "loss": 0.3933,
      "num_input_tokens_seen": 16249064,
      "step": 24850
    },
    {
      "epoch": 13.026729559748428,
      "grad_norm": 0.2008482962846756,
      "learning_rate": 1.634213909296144e-05,
      "loss": 0.5218,
      "num_input_tokens_seen": 16251944,
      "step": 24855
    },
    {
      "epoch": 13.029350104821804,
      "grad_norm": 0.14821606874465942,
      "learning_rate": 1.6331413284723217e-05,
      "loss": 0.5395,
      "num_input_tokens_seen": 16255368,
      "step": 24860
    },
    {
      "epoch": 13.031970649895179,
      "grad_norm": 0.16614019870758057,
      "learning_rate": 1.6320689289854442e-05,
      "loss": 0.5162,
      "num_input_tokens_seen": 16258152,
      "step": 24865
    },
    {
      "epoch": 13.034591194968554,
      "grad_norm": 0.1015448346734047,
      "learning_rate": 1.6309967110598438e-05,
      "loss": 0.4836,
      "num_input_tokens_seen": 16261672,
      "step": 24870
    },
    {
      "epoch": 13.03721174004193,
      "grad_norm": 0.19775862991809845,
      "learning_rate": 1.6299246749198183e-05,
      "loss": 0.512,
      "num_input_tokens_seen": 16265416,
      "step": 24875
    },
    {
      "epoch": 13.039832285115304,
      "grad_norm": 0.17528535425662994,
      "learning_rate": 1.6288528207896236e-05,
      "loss": 0.4656,
      "num_input_tokens_seen": 16268008,
      "step": 24880
    },
    {
      "epoch": 13.04245283018868,
      "grad_norm": 0.14077718555927277,
      "learning_rate": 1.627781148893481e-05,
      "loss": 0.3317,
      "num_input_tokens_seen": 16271912,
      "step": 24885
    },
    {
      "epoch": 13.045073375262055,
      "grad_norm": 0.10508255660533905,
      "learning_rate": 1.6267096594555703e-05,
      "loss": 0.5048,
      "num_input_tokens_seen": 16275304,
      "step": 24890
    },
    {
      "epoch": 13.04769392033543,
      "grad_norm": 0.09364369511604309,
      "learning_rate": 1.6256383527000372e-05,
      "loss": 0.4391,
      "num_input_tokens_seen": 16279240,
      "step": 24895
    },
    {
      "epoch": 13.050314465408805,
      "grad_norm": 0.150995671749115,
      "learning_rate": 1.6245672288509843e-05,
      "loss": 0.7258,
      "num_input_tokens_seen": 16282376,
      "step": 24900
    },
    {
      "epoch": 13.05293501048218,
      "grad_norm": 0.16167108714580536,
      "learning_rate": 1.62349628813248e-05,
      "loss": 0.4364,
      "num_input_tokens_seen": 16285000,
      "step": 24905
    },
    {
      "epoch": 13.055555555555555,
      "grad_norm": 0.1610938161611557,
      "learning_rate": 1.6224255307685525e-05,
      "loss": 0.4905,
      "num_input_tokens_seen": 16287976,
      "step": 24910
    },
    {
      "epoch": 13.05817610062893,
      "grad_norm": 0.11762724816799164,
      "learning_rate": 1.6213549569831922e-05,
      "loss": 0.4088,
      "num_input_tokens_seen": 16291080,
      "step": 24915
    },
    {
      "epoch": 13.060796645702306,
      "grad_norm": 0.252291202545166,
      "learning_rate": 1.62028456700035e-05,
      "loss": 0.4513,
      "num_input_tokens_seen": 16294504,
      "step": 24920
    },
    {
      "epoch": 13.06341719077568,
      "grad_norm": 0.11247007548809052,
      "learning_rate": 1.619214361043942e-05,
      "loss": 0.4353,
      "num_input_tokens_seen": 16298696,
      "step": 24925
    },
    {
      "epoch": 13.066037735849056,
      "grad_norm": 0.15903779864311218,
      "learning_rate": 1.61814433933784e-05,
      "loss": 0.4695,
      "num_input_tokens_seen": 16302536,
      "step": 24930
    },
    {
      "epoch": 13.068658280922431,
      "grad_norm": 0.06548091769218445,
      "learning_rate": 1.617074502105881e-05,
      "loss": 0.4169,
      "num_input_tokens_seen": 16305992,
      "step": 24935
    },
    {
      "epoch": 13.071278825995806,
      "grad_norm": 0.15608832240104675,
      "learning_rate": 1.616004849571864e-05,
      "loss": 0.5605,
      "num_input_tokens_seen": 16309768,
      "step": 24940
    },
    {
      "epoch": 13.073899371069182,
      "grad_norm": 0.13787856698036194,
      "learning_rate": 1.6149353819595463e-05,
      "loss": 0.4272,
      "num_input_tokens_seen": 16313160,
      "step": 24945
    },
    {
      "epoch": 13.076519916142558,
      "grad_norm": 0.2420976758003235,
      "learning_rate": 1.6138660994926508e-05,
      "loss": 0.3847,
      "num_input_tokens_seen": 16315688,
      "step": 24950
    },
    {
      "epoch": 13.079140461215934,
      "grad_norm": 0.40904921293258667,
      "learning_rate": 1.612797002394857e-05,
      "loss": 0.5099,
      "num_input_tokens_seen": 16319208,
      "step": 24955
    },
    {
      "epoch": 13.081761006289309,
      "grad_norm": 0.12024249136447906,
      "learning_rate": 1.611728090889808e-05,
      "loss": 0.4436,
      "num_input_tokens_seen": 16322600,
      "step": 24960
    },
    {
      "epoch": 13.084381551362684,
      "grad_norm": 0.1695440411567688,
      "learning_rate": 1.610659365201109e-05,
      "loss": 0.4522,
      "num_input_tokens_seen": 16326568,
      "step": 24965
    },
    {
      "epoch": 13.08700209643606,
      "grad_norm": 0.1726449877023697,
      "learning_rate": 1.609590825552324e-05,
      "loss": 0.7038,
      "num_input_tokens_seen": 16331336,
      "step": 24970
    },
    {
      "epoch": 13.089622641509434,
      "grad_norm": 0.2557152211666107,
      "learning_rate": 1.6085224721669807e-05,
      "loss": 0.4851,
      "num_input_tokens_seen": 16334824,
      "step": 24975
    },
    {
      "epoch": 13.09224318658281,
      "grad_norm": 0.3436630964279175,
      "learning_rate": 1.607454305268566e-05,
      "loss": 0.4177,
      "num_input_tokens_seen": 16337992,
      "step": 24980
    },
    {
      "epoch": 13.094863731656185,
      "grad_norm": 0.21487754583358765,
      "learning_rate": 1.606386325080528e-05,
      "loss": 0.6109,
      "num_input_tokens_seen": 16340680,
      "step": 24985
    },
    {
      "epoch": 13.09748427672956,
      "grad_norm": 0.17098979651927948,
      "learning_rate": 1.6053185318262746e-05,
      "loss": 0.4272,
      "num_input_tokens_seen": 16343464,
      "step": 24990
    },
    {
      "epoch": 13.100104821802935,
      "grad_norm": 0.1732240915298462,
      "learning_rate": 1.6042509257291783e-05,
      "loss": 0.4926,
      "num_input_tokens_seen": 16346376,
      "step": 24995
    },
    {
      "epoch": 13.10272536687631,
      "grad_norm": 0.11863845586776733,
      "learning_rate": 1.6031835070125685e-05,
      "loss": 0.316,
      "num_input_tokens_seen": 16349032,
      "step": 25000
    },
    {
      "epoch": 13.105345911949685,
      "grad_norm": 0.1404566466808319,
      "learning_rate": 1.6021162758997382e-05,
      "loss": 0.5494,
      "num_input_tokens_seen": 16352104,
      "step": 25005
    },
    {
      "epoch": 13.10796645702306,
      "grad_norm": 0.14034831523895264,
      "learning_rate": 1.6010492326139398e-05,
      "loss": 0.5513,
      "num_input_tokens_seen": 16355496,
      "step": 25010
    },
    {
      "epoch": 13.110587002096436,
      "grad_norm": 0.1229669600725174,
      "learning_rate": 1.599982377378385e-05,
      "loss": 0.5471,
      "num_input_tokens_seen": 16359080,
      "step": 25015
    },
    {
      "epoch": 13.11320754716981,
      "grad_norm": 0.186301589012146,
      "learning_rate": 1.5989157104162496e-05,
      "loss": 0.5019,
      "num_input_tokens_seen": 16362408,
      "step": 25020
    },
    {
      "epoch": 13.115828092243186,
      "grad_norm": 0.2236403524875641,
      "learning_rate": 1.5978492319506673e-05,
      "loss": 0.4667,
      "num_input_tokens_seen": 16365384,
      "step": 25025
    },
    {
      "epoch": 13.118448637316561,
      "grad_norm": 0.15061861276626587,
      "learning_rate": 1.596782942204733e-05,
      "loss": 0.386,
      "num_input_tokens_seen": 16368840,
      "step": 25030
    },
    {
      "epoch": 13.121069182389936,
      "grad_norm": 0.17813299596309662,
      "learning_rate": 1.595716841401504e-05,
      "loss": 0.4493,
      "num_input_tokens_seen": 16372296,
      "step": 25035
    },
    {
      "epoch": 13.123689727463312,
      "grad_norm": 0.2313900738954544,
      "learning_rate": 1.594650929763994e-05,
      "loss": 0.3574,
      "num_input_tokens_seen": 16375400,
      "step": 25040
    },
    {
      "epoch": 13.126310272536688,
      "grad_norm": 0.14481650292873383,
      "learning_rate": 1.5935852075151812e-05,
      "loss": 0.5201,
      "num_input_tokens_seen": 16378728,
      "step": 25045
    },
    {
      "epoch": 13.128930817610064,
      "grad_norm": 0.16285690665245056,
      "learning_rate": 1.5925196748780016e-05,
      "loss": 0.4837,
      "num_input_tokens_seen": 16381672,
      "step": 25050
    },
    {
      "epoch": 13.131551362683439,
      "grad_norm": 0.22448429465293884,
      "learning_rate": 1.5914543320753528e-05,
      "loss": 0.6087,
      "num_input_tokens_seen": 16384744,
      "step": 25055
    },
    {
      "epoch": 13.134171907756814,
      "grad_norm": 0.15289779007434845,
      "learning_rate": 1.5903891793300932e-05,
      "loss": 0.6028,
      "num_input_tokens_seen": 16387816,
      "step": 25060
    },
    {
      "epoch": 13.13679245283019,
      "grad_norm": 0.14656178653240204,
      "learning_rate": 1.5893242168650403e-05,
      "loss": 0.4612,
      "num_input_tokens_seen": 16390600,
      "step": 25065
    },
    {
      "epoch": 13.139412997903564,
      "grad_norm": 0.13182803988456726,
      "learning_rate": 1.5882594449029705e-05,
      "loss": 0.4546,
      "num_input_tokens_seen": 16394280,
      "step": 25070
    },
    {
      "epoch": 13.14203354297694,
      "grad_norm": 0.14165106415748596,
      "learning_rate": 1.587194863666624e-05,
      "loss": 0.4866,
      "num_input_tokens_seen": 16396904,
      "step": 25075
    },
    {
      "epoch": 13.144654088050315,
      "grad_norm": 0.1920154094696045,
      "learning_rate": 1.5861304733786974e-05,
      "loss": 0.53,
      "num_input_tokens_seen": 16400136,
      "step": 25080
    },
    {
      "epoch": 13.14727463312369,
      "grad_norm": 0.15012717247009277,
      "learning_rate": 1.5850662742618505e-05,
      "loss": 0.4401,
      "num_input_tokens_seen": 16402568,
      "step": 25085
    },
    {
      "epoch": 13.149895178197065,
      "grad_norm": 0.13674138486385345,
      "learning_rate": 1.5840022665387004e-05,
      "loss": 0.5678,
      "num_input_tokens_seen": 16405832,
      "step": 25090
    },
    {
      "epoch": 13.15251572327044,
      "grad_norm": 0.2254413664340973,
      "learning_rate": 1.5829384504318268e-05,
      "loss": 0.7391,
      "num_input_tokens_seen": 16409288,
      "step": 25095
    },
    {
      "epoch": 13.155136268343815,
      "grad_norm": 0.1577533334493637,
      "learning_rate": 1.5818748261637662e-05,
      "loss": 0.4125,
      "num_input_tokens_seen": 16412584,
      "step": 25100
    },
    {
      "epoch": 13.15775681341719,
      "grad_norm": 0.11814791709184647,
      "learning_rate": 1.5808113939570173e-05,
      "loss": 0.4404,
      "num_input_tokens_seen": 16415080,
      "step": 25105
    },
    {
      "epoch": 13.160377358490566,
      "grad_norm": 0.14270223677158356,
      "learning_rate": 1.5797481540340386e-05,
      "loss": 0.4312,
      "num_input_tokens_seen": 16418184,
      "step": 25110
    },
    {
      "epoch": 13.16299790356394,
      "grad_norm": 0.1727658212184906,
      "learning_rate": 1.5786851066172466e-05,
      "loss": 0.6569,
      "num_input_tokens_seen": 16421672,
      "step": 25115
    },
    {
      "epoch": 13.165618448637316,
      "grad_norm": 0.12148435413837433,
      "learning_rate": 1.5776222519290208e-05,
      "loss": 0.44,
      "num_input_tokens_seen": 16425448,
      "step": 25120
    },
    {
      "epoch": 13.168238993710691,
      "grad_norm": 0.11930834501981735,
      "learning_rate": 1.576559590191696e-05,
      "loss": 0.5561,
      "num_input_tokens_seen": 16428872,
      "step": 25125
    },
    {
      "epoch": 13.170859538784066,
      "grad_norm": 0.2050793617963791,
      "learning_rate": 1.5754971216275692e-05,
      "loss": 0.6502,
      "num_input_tokens_seen": 16431560,
      "step": 25130
    },
    {
      "epoch": 13.173480083857442,
      "grad_norm": 0.18366852402687073,
      "learning_rate": 1.5744348464588975e-05,
      "loss": 0.487,
      "num_input_tokens_seen": 16435048,
      "step": 25135
    },
    {
      "epoch": 13.176100628930818,
      "grad_norm": 0.17158372700214386,
      "learning_rate": 1.5733727649078968e-05,
      "loss": 0.4327,
      "num_input_tokens_seen": 16438248,
      "step": 25140
    },
    {
      "epoch": 13.178721174004194,
      "grad_norm": 0.1951850950717926,
      "learning_rate": 1.5723108771967414e-05,
      "loss": 0.5131,
      "num_input_tokens_seen": 16441384,
      "step": 25145
    },
    {
      "epoch": 13.181341719077569,
      "grad_norm": 0.16678906977176666,
      "learning_rate": 1.5712491835475685e-05,
      "loss": 0.6165,
      "num_input_tokens_seen": 16445992,
      "step": 25150
    },
    {
      "epoch": 13.183962264150944,
      "grad_norm": 0.09141234308481216,
      "learning_rate": 1.5701876841824698e-05,
      "loss": 0.4346,
      "num_input_tokens_seen": 16449416,
      "step": 25155
    },
    {
      "epoch": 13.18658280922432,
      "grad_norm": 0.12524276971817017,
      "learning_rate": 1.5691263793234993e-05,
      "loss": 0.3842,
      "num_input_tokens_seen": 16454472,
      "step": 25160
    },
    {
      "epoch": 13.189203354297694,
      "grad_norm": 0.17030976712703705,
      "learning_rate": 1.5680652691926706e-05,
      "loss": 0.4926,
      "num_input_tokens_seen": 16456808,
      "step": 25165
    },
    {
      "epoch": 13.19182389937107,
      "grad_norm": 0.2940899133682251,
      "learning_rate": 1.5670043540119553e-05,
      "loss": 0.5504,
      "num_input_tokens_seen": 16459752,
      "step": 25170
    },
    {
      "epoch": 13.194444444444445,
      "grad_norm": 0.19486522674560547,
      "learning_rate": 1.5659436340032855e-05,
      "loss": 0.4934,
      "num_input_tokens_seen": 16462760,
      "step": 25175
    },
    {
      "epoch": 13.19706498951782,
      "grad_norm": 0.15394221246242523,
      "learning_rate": 1.5648831093885515e-05,
      "loss": 0.4935,
      "num_input_tokens_seen": 16466536,
      "step": 25180
    },
    {
      "epoch": 13.199685534591195,
      "grad_norm": 0.18879742920398712,
      "learning_rate": 1.5638227803896015e-05,
      "loss": 0.6912,
      "num_input_tokens_seen": 16469384,
      "step": 25185
    },
    {
      "epoch": 13.20230607966457,
      "grad_norm": 0.1710646003484726,
      "learning_rate": 1.562762647228246e-05,
      "loss": 0.5171,
      "num_input_tokens_seen": 16472488,
      "step": 25190
    },
    {
      "epoch": 13.204926624737945,
      "grad_norm": 0.16277767717838287,
      "learning_rate": 1.5617027101262515e-05,
      "loss": 0.5038,
      "num_input_tokens_seen": 16475528,
      "step": 25195
    },
    {
      "epoch": 13.20754716981132,
      "grad_norm": 0.14245623350143433,
      "learning_rate": 1.5606429693053463e-05,
      "loss": 0.3739,
      "num_input_tokens_seen": 16478216,
      "step": 25200
    },
    {
      "epoch": 13.210167714884696,
      "grad_norm": 0.13619780540466309,
      "learning_rate": 1.5595834249872145e-05,
      "loss": 0.5511,
      "num_input_tokens_seen": 16481480,
      "step": 25205
    },
    {
      "epoch": 13.21278825995807,
      "grad_norm": 0.14710722863674164,
      "learning_rate": 1.558524077393502e-05,
      "loss": 0.4404,
      "num_input_tokens_seen": 16485768,
      "step": 25210
    },
    {
      "epoch": 13.215408805031446,
      "grad_norm": 0.13613829016685486,
      "learning_rate": 1.557464926745811e-05,
      "loss": 0.4094,
      "num_input_tokens_seen": 16490152,
      "step": 25215
    },
    {
      "epoch": 13.218029350104821,
      "grad_norm": 0.38838517665863037,
      "learning_rate": 1.5564059732657045e-05,
      "loss": 0.4608,
      "num_input_tokens_seen": 16493096,
      "step": 25220
    },
    {
      "epoch": 13.220649895178196,
      "grad_norm": 0.2381121665239334,
      "learning_rate": 1.555347217174703e-05,
      "loss": 0.4351,
      "num_input_tokens_seen": 16495848,
      "step": 25225
    },
    {
      "epoch": 13.223270440251572,
      "grad_norm": 0.12964469194412231,
      "learning_rate": 1.554288658694287e-05,
      "loss": 0.5258,
      "num_input_tokens_seen": 16502888,
      "step": 25230
    },
    {
      "epoch": 13.225890985324948,
      "grad_norm": 0.19406414031982422,
      "learning_rate": 1.553230298045895e-05,
      "loss": 0.4916,
      "num_input_tokens_seen": 16505928,
      "step": 25235
    },
    {
      "epoch": 13.228511530398324,
      "grad_norm": 0.20121119916439056,
      "learning_rate": 1.5521721354509223e-05,
      "loss": 0.6139,
      "num_input_tokens_seen": 16508648,
      "step": 25240
    },
    {
      "epoch": 13.231132075471699,
      "grad_norm": 0.1829136312007904,
      "learning_rate": 1.5511141711307258e-05,
      "loss": 0.5065,
      "num_input_tokens_seen": 16512968,
      "step": 25245
    },
    {
      "epoch": 13.233752620545074,
      "grad_norm": 0.1090233102440834,
      "learning_rate": 1.5500564053066184e-05,
      "loss": 0.6047,
      "num_input_tokens_seen": 16516552,
      "step": 25250
    },
    {
      "epoch": 13.23637316561845,
      "grad_norm": 0.1367502063512802,
      "learning_rate": 1.5489988381998743e-05,
      "loss": 0.3801,
      "num_input_tokens_seen": 16519528,
      "step": 25255
    },
    {
      "epoch": 13.238993710691824,
      "grad_norm": 0.17257137596607208,
      "learning_rate": 1.5479414700317237e-05,
      "loss": 0.4448,
      "num_input_tokens_seen": 16522504,
      "step": 25260
    },
    {
      "epoch": 13.2416142557652,
      "grad_norm": 0.32701805233955383,
      "learning_rate": 1.546884301023356e-05,
      "loss": 0.6373,
      "num_input_tokens_seen": 16525384,
      "step": 25265
    },
    {
      "epoch": 13.244234800838575,
      "grad_norm": 0.18802864849567413,
      "learning_rate": 1.5458273313959186e-05,
      "loss": 0.5505,
      "num_input_tokens_seen": 16528840,
      "step": 25270
    },
    {
      "epoch": 13.24685534591195,
      "grad_norm": 0.3038291931152344,
      "learning_rate": 1.5447705613705177e-05,
      "loss": 0.504,
      "num_input_tokens_seen": 16532648,
      "step": 25275
    },
    {
      "epoch": 13.249475890985325,
      "grad_norm": 0.14452312886714935,
      "learning_rate": 1.5437139911682176e-05,
      "loss": 0.4369,
      "num_input_tokens_seen": 16535720,
      "step": 25280
    },
    {
      "epoch": 13.2520964360587,
      "grad_norm": 0.2573510706424713,
      "learning_rate": 1.54265762101004e-05,
      "loss": 0.4247,
      "num_input_tokens_seen": 16538312,
      "step": 25285
    },
    {
      "epoch": 13.254716981132075,
      "grad_norm": 0.21444487571716309,
      "learning_rate": 1.5416014511169663e-05,
      "loss": 0.6057,
      "num_input_tokens_seen": 16541640,
      "step": 25290
    },
    {
      "epoch": 13.25733752620545,
      "grad_norm": 0.09039225429296494,
      "learning_rate": 1.5405454817099363e-05,
      "loss": 0.4188,
      "num_input_tokens_seen": 16545640,
      "step": 25295
    },
    {
      "epoch": 13.259958071278826,
      "grad_norm": 0.17135988175868988,
      "learning_rate": 1.5394897130098446e-05,
      "loss": 0.4877,
      "num_input_tokens_seen": 16548488,
      "step": 25300
    },
    {
      "epoch": 13.2625786163522,
      "grad_norm": 0.15653620660305023,
      "learning_rate": 1.5384341452375465e-05,
      "loss": 0.4536,
      "num_input_tokens_seen": 16551592,
      "step": 25305
    },
    {
      "epoch": 13.265199161425576,
      "grad_norm": 0.15092764794826508,
      "learning_rate": 1.5373787786138556e-05,
      "loss": 0.528,
      "num_input_tokens_seen": 16554600,
      "step": 25310
    },
    {
      "epoch": 13.267819706498951,
      "grad_norm": 0.19681279361248016,
      "learning_rate": 1.5363236133595416e-05,
      "loss": 0.5288,
      "num_input_tokens_seen": 16557192,
      "step": 25315
    },
    {
      "epoch": 13.270440251572326,
      "grad_norm": 0.24991029500961304,
      "learning_rate": 1.5352686496953344e-05,
      "loss": 0.6462,
      "num_input_tokens_seen": 16560040,
      "step": 25320
    },
    {
      "epoch": 13.273060796645701,
      "grad_norm": 0.12876568734645844,
      "learning_rate": 1.534213887841919e-05,
      "loss": 0.4361,
      "num_input_tokens_seen": 16563304,
      "step": 25325
    },
    {
      "epoch": 13.275681341719078,
      "grad_norm": 0.2440088838338852,
      "learning_rate": 1.5331593280199395e-05,
      "loss": 0.4627,
      "num_input_tokens_seen": 16566312,
      "step": 25330
    },
    {
      "epoch": 13.278301886792454,
      "grad_norm": 0.27541980147361755,
      "learning_rate": 1.532104970449999e-05,
      "loss": 0.5204,
      "num_input_tokens_seen": 16569736,
      "step": 25335
    },
    {
      "epoch": 13.280922431865829,
      "grad_norm": 0.21673999726772308,
      "learning_rate": 1.5310508153526555e-05,
      "loss": 0.5047,
      "num_input_tokens_seen": 16572680,
      "step": 25340
    },
    {
      "epoch": 13.283542976939204,
      "grad_norm": 0.13222481310367584,
      "learning_rate": 1.5299968629484276e-05,
      "loss": 0.4612,
      "num_input_tokens_seen": 16576168,
      "step": 25345
    },
    {
      "epoch": 13.286163522012579,
      "grad_norm": 0.3185020983219147,
      "learning_rate": 1.5289431134577904e-05,
      "loss": 0.5041,
      "num_input_tokens_seen": 16578504,
      "step": 25350
    },
    {
      "epoch": 13.288784067085954,
      "grad_norm": 0.20522572100162506,
      "learning_rate": 1.527889567101174e-05,
      "loss": 0.5319,
      "num_input_tokens_seen": 16582472,
      "step": 25355
    },
    {
      "epoch": 13.29140461215933,
      "grad_norm": 0.15060727298259735,
      "learning_rate": 1.5268362240989707e-05,
      "loss": 0.5594,
      "num_input_tokens_seen": 16585128,
      "step": 25360
    },
    {
      "epoch": 13.294025157232705,
      "grad_norm": 0.1861688494682312,
      "learning_rate": 1.525783084671526e-05,
      "loss": 0.5376,
      "num_input_tokens_seen": 16587592,
      "step": 25365
    },
    {
      "epoch": 13.29664570230608,
      "grad_norm": 0.16003726422786713,
      "learning_rate": 1.5247301490391458e-05,
      "loss": 0.4983,
      "num_input_tokens_seen": 16594600,
      "step": 25370
    },
    {
      "epoch": 13.299266247379455,
      "grad_norm": 0.17561201751232147,
      "learning_rate": 1.5236774174220922e-05,
      "loss": 0.3565,
      "num_input_tokens_seen": 16598120,
      "step": 25375
    },
    {
      "epoch": 13.30188679245283,
      "grad_norm": 0.10432951897382736,
      "learning_rate": 1.5226248900405852e-05,
      "loss": 0.609,
      "num_input_tokens_seen": 16601768,
      "step": 25380
    },
    {
      "epoch": 13.304507337526205,
      "grad_norm": 0.16116176545619965,
      "learning_rate": 1.5215725671147996e-05,
      "loss": 0.3887,
      "num_input_tokens_seen": 16604776,
      "step": 25385
    },
    {
      "epoch": 13.30712788259958,
      "grad_norm": 0.16786770522594452,
      "learning_rate": 1.5205204488648709e-05,
      "loss": 0.4208,
      "num_input_tokens_seen": 16607688,
      "step": 25390
    },
    {
      "epoch": 13.309748427672956,
      "grad_norm": 0.10948912054300308,
      "learning_rate": 1.5194685355108893e-05,
      "loss": 0.4075,
      "num_input_tokens_seen": 16610920,
      "step": 25395
    },
    {
      "epoch": 13.31236897274633,
      "grad_norm": 0.22151313722133636,
      "learning_rate": 1.518416827272904e-05,
      "loss": 0.3861,
      "num_input_tokens_seen": 16614024,
      "step": 25400
    },
    {
      "epoch": 13.314989517819706,
      "grad_norm": 0.20676109194755554,
      "learning_rate": 1.5173653243709207e-05,
      "loss": 0.5026,
      "num_input_tokens_seen": 16616936,
      "step": 25405
    },
    {
      "epoch": 13.317610062893081,
      "grad_norm": 0.16543693840503693,
      "learning_rate": 1.5163140270249e-05,
      "loss": 0.4669,
      "num_input_tokens_seen": 16619432,
      "step": 25410
    },
    {
      "epoch": 13.320230607966456,
      "grad_norm": 0.21160204708576202,
      "learning_rate": 1.5152629354547626e-05,
      "loss": 0.4337,
      "num_input_tokens_seen": 16622152,
      "step": 25415
    },
    {
      "epoch": 13.322851153039831,
      "grad_norm": 0.22471337020397186,
      "learning_rate": 1.5142120498803842e-05,
      "loss": 0.5353,
      "num_input_tokens_seen": 16624808,
      "step": 25420
    },
    {
      "epoch": 13.325471698113208,
      "grad_norm": 0.10486361384391785,
      "learning_rate": 1.5131613705215991e-05,
      "loss": 0.6049,
      "num_input_tokens_seen": 16628744,
      "step": 25425
    },
    {
      "epoch": 13.328092243186584,
      "grad_norm": 0.15137790143489838,
      "learning_rate": 1.5121108975981965e-05,
      "loss": 0.5545,
      "num_input_tokens_seen": 16631368,
      "step": 25430
    },
    {
      "epoch": 13.330712788259959,
      "grad_norm": 0.1751723289489746,
      "learning_rate": 1.5110606313299248e-05,
      "loss": 0.4935,
      "num_input_tokens_seen": 16634536,
      "step": 25435
    },
    {
      "epoch": 13.333333333333334,
      "grad_norm": 0.14041456580162048,
      "learning_rate": 1.5100105719364855e-05,
      "loss": 0.521,
      "num_input_tokens_seen": 16637544,
      "step": 25440
    },
    {
      "epoch": 13.335953878406709,
      "grad_norm": 0.17809246480464935,
      "learning_rate": 1.50896071963754e-05,
      "loss": 0.4755,
      "num_input_tokens_seen": 16641096,
      "step": 25445
    },
    {
      "epoch": 13.338574423480084,
      "grad_norm": 0.14154602587223053,
      "learning_rate": 1.5079110746527055e-05,
      "loss": 0.5527,
      "num_input_tokens_seen": 16643688,
      "step": 25450
    },
    {
      "epoch": 13.34119496855346,
      "grad_norm": 0.2732677459716797,
      "learning_rate": 1.5068616372015565e-05,
      "loss": 0.5504,
      "num_input_tokens_seen": 16646888,
      "step": 25455
    },
    {
      "epoch": 13.343815513626835,
      "grad_norm": 0.1274368017911911,
      "learning_rate": 1.5058124075036223e-05,
      "loss": 0.4039,
      "num_input_tokens_seen": 16650312,
      "step": 25460
    },
    {
      "epoch": 13.34643605870021,
      "grad_norm": 0.16333244740962982,
      "learning_rate": 1.5047633857783908e-05,
      "loss": 0.444,
      "num_input_tokens_seen": 16653896,
      "step": 25465
    },
    {
      "epoch": 13.349056603773585,
      "grad_norm": 0.21845021843910217,
      "learning_rate": 1.5037145722453044e-05,
      "loss": 0.4105,
      "num_input_tokens_seen": 16657320,
      "step": 25470
    },
    {
      "epoch": 13.35167714884696,
      "grad_norm": 0.12768541276454926,
      "learning_rate": 1.5026659671237624e-05,
      "loss": 0.4518,
      "num_input_tokens_seen": 16660456,
      "step": 25475
    },
    {
      "epoch": 13.354297693920335,
      "grad_norm": 0.12724332511425018,
      "learning_rate": 1.501617570633123e-05,
      "loss": 0.4896,
      "num_input_tokens_seen": 16664424,
      "step": 25480
    },
    {
      "epoch": 13.35691823899371,
      "grad_norm": 0.19144344329833984,
      "learning_rate": 1.500569382992697e-05,
      "loss": 0.5612,
      "num_input_tokens_seen": 16667656,
      "step": 25485
    },
    {
      "epoch": 13.359538784067086,
      "grad_norm": 0.12822741270065308,
      "learning_rate": 1.4995214044217547e-05,
      "loss": 0.4422,
      "num_input_tokens_seen": 16671688,
      "step": 25490
    },
    {
      "epoch": 13.36215932914046,
      "grad_norm": 0.16905200481414795,
      "learning_rate": 1.49847363513952e-05,
      "loss": 0.4833,
      "num_input_tokens_seen": 16674856,
      "step": 25495
    },
    {
      "epoch": 13.364779874213836,
      "grad_norm": 0.15227772295475006,
      "learning_rate": 1.4974260753651748e-05,
      "loss": 0.4217,
      "num_input_tokens_seen": 16677864,
      "step": 25500
    },
    {
      "epoch": 13.367400419287211,
      "grad_norm": 0.20970334112644196,
      "learning_rate": 1.4963787253178568e-05,
      "loss": 0.4012,
      "num_input_tokens_seen": 16680808,
      "step": 25505
    },
    {
      "epoch": 13.370020964360586,
      "grad_norm": 0.5908733606338501,
      "learning_rate": 1.4953315852166594e-05,
      "loss": 0.4416,
      "num_input_tokens_seen": 16683208,
      "step": 25510
    },
    {
      "epoch": 13.372641509433961,
      "grad_norm": 0.14444242417812347,
      "learning_rate": 1.4942846552806322e-05,
      "loss": 0.4616,
      "num_input_tokens_seen": 16686952,
      "step": 25515
    },
    {
      "epoch": 13.375262054507338,
      "grad_norm": 0.23204566538333893,
      "learning_rate": 1.4932379357287829e-05,
      "loss": 0.604,
      "num_input_tokens_seen": 16689576,
      "step": 25520
    },
    {
      "epoch": 13.377882599580714,
      "grad_norm": 0.1338678002357483,
      "learning_rate": 1.49219142678007e-05,
      "loss": 0.469,
      "num_input_tokens_seen": 16693320,
      "step": 25525
    },
    {
      "epoch": 13.380503144654089,
      "grad_norm": 0.1489826887845993,
      "learning_rate": 1.4911451286534128e-05,
      "loss": 0.4551,
      "num_input_tokens_seen": 16698344,
      "step": 25530
    },
    {
      "epoch": 13.383123689727464,
      "grad_norm": 0.15033715963363647,
      "learning_rate": 1.490099041567686e-05,
      "loss": 0.4714,
      "num_input_tokens_seen": 16701576,
      "step": 25535
    },
    {
      "epoch": 13.385744234800839,
      "grad_norm": 0.3331167697906494,
      "learning_rate": 1.4890531657417176e-05,
      "loss": 0.4926,
      "num_input_tokens_seen": 16704904,
      "step": 25540
    },
    {
      "epoch": 13.388364779874214,
      "grad_norm": 0.12160903215408325,
      "learning_rate": 1.4880075013942938e-05,
      "loss": 0.5101,
      "num_input_tokens_seen": 16707304,
      "step": 25545
    },
    {
      "epoch": 13.39098532494759,
      "grad_norm": 0.1439315229654312,
      "learning_rate": 1.4869620487441558e-05,
      "loss": 0.5416,
      "num_input_tokens_seen": 16710120,
      "step": 25550
    },
    {
      "epoch": 13.393605870020965,
      "grad_norm": 0.13502663373947144,
      "learning_rate": 1.4859168080099988e-05,
      "loss": 0.4116,
      "num_input_tokens_seen": 16713736,
      "step": 25555
    },
    {
      "epoch": 13.39622641509434,
      "grad_norm": 0.1807701140642166,
      "learning_rate": 1.484871779410477e-05,
      "loss": 0.4454,
      "num_input_tokens_seen": 16716616,
      "step": 25560
    },
    {
      "epoch": 13.398846960167715,
      "grad_norm": 0.25671201944351196,
      "learning_rate": 1.4838269631641971e-05,
      "loss": 0.6825,
      "num_input_tokens_seen": 16719432,
      "step": 25565
    },
    {
      "epoch": 13.40146750524109,
      "grad_norm": 0.14529167115688324,
      "learning_rate": 1.482782359489724e-05,
      "loss": 0.4871,
      "num_input_tokens_seen": 16723176,
      "step": 25570
    },
    {
      "epoch": 13.404088050314465,
      "grad_norm": 0.29110580682754517,
      "learning_rate": 1.4817379686055754e-05,
      "loss": 0.5523,
      "num_input_tokens_seen": 16725608,
      "step": 25575
    },
    {
      "epoch": 13.40670859538784,
      "grad_norm": 0.21081334352493286,
      "learning_rate": 1.4806937907302282e-05,
      "loss": 0.4665,
      "num_input_tokens_seen": 16729320,
      "step": 25580
    },
    {
      "epoch": 13.409329140461216,
      "grad_norm": 0.20041252672672272,
      "learning_rate": 1.4796498260821103e-05,
      "loss": 0.5531,
      "num_input_tokens_seen": 16731976,
      "step": 25585
    },
    {
      "epoch": 13.41194968553459,
      "grad_norm": 0.12584836781024933,
      "learning_rate": 1.4786060748796076e-05,
      "loss": 0.5548,
      "num_input_tokens_seen": 16734792,
      "step": 25590
    },
    {
      "epoch": 13.414570230607966,
      "grad_norm": 0.12890790402889252,
      "learning_rate": 1.4775625373410617e-05,
      "loss": 0.6061,
      "num_input_tokens_seen": 16737608,
      "step": 25595
    },
    {
      "epoch": 13.417190775681341,
      "grad_norm": 0.20965121686458588,
      "learning_rate": 1.4765192136847685e-05,
      "loss": 0.4199,
      "num_input_tokens_seen": 16741000,
      "step": 25600
    },
    {
      "epoch": 13.419811320754716,
      "grad_norm": 0.21648600697517395,
      "learning_rate": 1.47547610412898e-05,
      "loss": 0.4423,
      "num_input_tokens_seen": 16743944,
      "step": 25605
    },
    {
      "epoch": 13.422431865828091,
      "grad_norm": 0.13723400235176086,
      "learning_rate": 1.4744332088919011e-05,
      "loss": 0.453,
      "num_input_tokens_seen": 16747176,
      "step": 25610
    },
    {
      "epoch": 13.425052410901468,
      "grad_norm": 0.14720913767814636,
      "learning_rate": 1.4733905281916954e-05,
      "loss": 0.5321,
      "num_input_tokens_seen": 16749928,
      "step": 25615
    },
    {
      "epoch": 13.427672955974844,
      "grad_norm": 0.142183318734169,
      "learning_rate": 1.472348062246478e-05,
      "loss": 0.4361,
      "num_input_tokens_seen": 16753128,
      "step": 25620
    },
    {
      "epoch": 13.430293501048219,
      "grad_norm": 0.1309351921081543,
      "learning_rate": 1.4713058112743228e-05,
      "loss": 0.3094,
      "num_input_tokens_seen": 16756104,
      "step": 25625
    },
    {
      "epoch": 13.432914046121594,
      "grad_norm": 0.3888019323348999,
      "learning_rate": 1.4702637754932553e-05,
      "loss": 0.5437,
      "num_input_tokens_seen": 16758824,
      "step": 25630
    },
    {
      "epoch": 13.435534591194969,
      "grad_norm": 0.27229559421539307,
      "learning_rate": 1.4692219551212587e-05,
      "loss": 0.4856,
      "num_input_tokens_seen": 16761224,
      "step": 25635
    },
    {
      "epoch": 13.438155136268344,
      "grad_norm": 0.13896878063678741,
      "learning_rate": 1.4681803503762695e-05,
      "loss": 0.5061,
      "num_input_tokens_seen": 16764072,
      "step": 25640
    },
    {
      "epoch": 13.44077568134172,
      "grad_norm": 0.19768011569976807,
      "learning_rate": 1.4671389614761788e-05,
      "loss": 0.3866,
      "num_input_tokens_seen": 16767560,
      "step": 25645
    },
    {
      "epoch": 13.443396226415095,
      "grad_norm": 0.152439147233963,
      "learning_rate": 1.4660977886388339e-05,
      "loss": 0.3991,
      "num_input_tokens_seen": 16770504,
      "step": 25650
    },
    {
      "epoch": 13.44601677148847,
      "grad_norm": 0.18268460035324097,
      "learning_rate": 1.4650568320820362e-05,
      "loss": 0.5094,
      "num_input_tokens_seen": 16773224,
      "step": 25655
    },
    {
      "epoch": 13.448637316561845,
      "grad_norm": 0.15452340245246887,
      "learning_rate": 1.4640160920235423e-05,
      "loss": 0.4803,
      "num_input_tokens_seen": 16775880,
      "step": 25660
    },
    {
      "epoch": 13.45125786163522,
      "grad_norm": 0.0785640999674797,
      "learning_rate": 1.4629755686810631e-05,
      "loss": 0.504,
      "num_input_tokens_seen": 16778920,
      "step": 25665
    },
    {
      "epoch": 13.453878406708595,
      "grad_norm": 0.14267735183238983,
      "learning_rate": 1.4619352622722632e-05,
      "loss": 0.4078,
      "num_input_tokens_seen": 16781544,
      "step": 25670
    },
    {
      "epoch": 13.45649895178197,
      "grad_norm": 0.30013149976730347,
      "learning_rate": 1.4608951730147635e-05,
      "loss": 0.5161,
      "num_input_tokens_seen": 16784264,
      "step": 25675
    },
    {
      "epoch": 13.459119496855346,
      "grad_norm": 0.1047462522983551,
      "learning_rate": 1.4598553011261395e-05,
      "loss": 0.4734,
      "num_input_tokens_seen": 16787176,
      "step": 25680
    },
    {
      "epoch": 13.46174004192872,
      "grad_norm": 0.1271359920501709,
      "learning_rate": 1.4588156468239192e-05,
      "loss": 0.4625,
      "num_input_tokens_seen": 16790376,
      "step": 25685
    },
    {
      "epoch": 13.464360587002096,
      "grad_norm": 0.2170378714799881,
      "learning_rate": 1.4577762103255882e-05,
      "loss": 0.4656,
      "num_input_tokens_seen": 16793608,
      "step": 25690
    },
    {
      "epoch": 13.466981132075471,
      "grad_norm": 0.2621290385723114,
      "learning_rate": 1.4567369918485828e-05,
      "loss": 0.6701,
      "num_input_tokens_seen": 16796552,
      "step": 25695
    },
    {
      "epoch": 13.469601677148846,
      "grad_norm": 0.1702498495578766,
      "learning_rate": 1.4556979916102976e-05,
      "loss": 0.5408,
      "num_input_tokens_seen": 16799432,
      "step": 25700
    },
    {
      "epoch": 13.472222222222221,
      "grad_norm": 0.24687273800373077,
      "learning_rate": 1.454659209828077e-05,
      "loss": 0.5138,
      "num_input_tokens_seen": 16802184,
      "step": 25705
    },
    {
      "epoch": 13.474842767295598,
      "grad_norm": 0.3162415623664856,
      "learning_rate": 1.4536206467192243e-05,
      "loss": 0.336,
      "num_input_tokens_seen": 16807208,
      "step": 25710
    },
    {
      "epoch": 13.477463312368974,
      "grad_norm": 0.2025993913412094,
      "learning_rate": 1.4525823025009944e-05,
      "loss": 0.4987,
      "num_input_tokens_seen": 16810856,
      "step": 25715
    },
    {
      "epoch": 13.480083857442349,
      "grad_norm": 0.17418478429317474,
      "learning_rate": 1.4515441773905983e-05,
      "loss": 0.4542,
      "num_input_tokens_seen": 16813736,
      "step": 25720
    },
    {
      "epoch": 13.482704402515724,
      "grad_norm": 0.16370970010757446,
      "learning_rate": 1.4505062716051988e-05,
      "loss": 0.4101,
      "num_input_tokens_seen": 16817928,
      "step": 25725
    },
    {
      "epoch": 13.485324947589099,
      "grad_norm": 0.15935227274894714,
      "learning_rate": 1.4494685853619133e-05,
      "loss": 0.377,
      "num_input_tokens_seen": 16820872,
      "step": 25730
    },
    {
      "epoch": 13.487945492662474,
      "grad_norm": 0.18160395324230194,
      "learning_rate": 1.4484311188778143e-05,
      "loss": 0.513,
      "num_input_tokens_seen": 16823464,
      "step": 25735
    },
    {
      "epoch": 13.49056603773585,
      "grad_norm": 0.11339402943849564,
      "learning_rate": 1.4473938723699285e-05,
      "loss": 0.462,
      "num_input_tokens_seen": 16826632,
      "step": 25740
    },
    {
      "epoch": 13.493186582809225,
      "grad_norm": 0.39800432324409485,
      "learning_rate": 1.4463568460552368e-05,
      "loss": 0.4977,
      "num_input_tokens_seen": 16829640,
      "step": 25745
    },
    {
      "epoch": 13.4958071278826,
      "grad_norm": 0.1911105066537857,
      "learning_rate": 1.4453200401506723e-05,
      "loss": 0.4574,
      "num_input_tokens_seen": 16832456,
      "step": 25750
    },
    {
      "epoch": 13.498427672955975,
      "grad_norm": 0.12960012257099152,
      "learning_rate": 1.4442834548731218e-05,
      "loss": 0.4812,
      "num_input_tokens_seen": 16835208,
      "step": 25755
    },
    {
      "epoch": 13.50104821802935,
      "grad_norm": 0.16308017075061798,
      "learning_rate": 1.4432470904394285e-05,
      "loss": 0.4822,
      "num_input_tokens_seen": 16838632,
      "step": 25760
    },
    {
      "epoch": 13.503668763102725,
      "grad_norm": 0.1983921080827713,
      "learning_rate": 1.4422109470663878e-05,
      "loss": 0.5441,
      "num_input_tokens_seen": 16842088,
      "step": 25765
    },
    {
      "epoch": 13.5062893081761,
      "grad_norm": 0.2308402955532074,
      "learning_rate": 1.441175024970749e-05,
      "loss": 0.4234,
      "num_input_tokens_seen": 16844488,
      "step": 25770
    },
    {
      "epoch": 13.508909853249476,
      "grad_norm": 0.46283408999443054,
      "learning_rate": 1.4401393243692163e-05,
      "loss": 0.5915,
      "num_input_tokens_seen": 16848712,
      "step": 25775
    },
    {
      "epoch": 13.51153039832285,
      "grad_norm": 0.19248995184898376,
      "learning_rate": 1.4391038454784455e-05,
      "loss": 0.412,
      "num_input_tokens_seen": 16852232,
      "step": 25780
    },
    {
      "epoch": 13.514150943396226,
      "grad_norm": 0.09499936550855637,
      "learning_rate": 1.4380685885150458e-05,
      "loss": 0.4673,
      "num_input_tokens_seen": 16855368,
      "step": 25785
    },
    {
      "epoch": 13.516771488469601,
      "grad_norm": 0.274586945772171,
      "learning_rate": 1.437033553695582e-05,
      "loss": 0.5043,
      "num_input_tokens_seen": 16858536,
      "step": 25790
    },
    {
      "epoch": 13.519392033542976,
      "grad_norm": 0.2026134431362152,
      "learning_rate": 1.4359987412365721e-05,
      "loss": 0.4388,
      "num_input_tokens_seen": 16862152,
      "step": 25795
    },
    {
      "epoch": 13.522012578616351,
      "grad_norm": 0.2038629651069641,
      "learning_rate": 1.4349641513544876e-05,
      "loss": 0.4765,
      "num_input_tokens_seen": 16864904,
      "step": 25800
    },
    {
      "epoch": 13.524633123689728,
      "grad_norm": 0.21836549043655396,
      "learning_rate": 1.4339297842657523e-05,
      "loss": 0.5449,
      "num_input_tokens_seen": 16868136,
      "step": 25805
    },
    {
      "epoch": 13.527253668763104,
      "grad_norm": 0.10703162848949432,
      "learning_rate": 1.4328956401867427e-05,
      "loss": 0.4874,
      "num_input_tokens_seen": 16871432,
      "step": 25810
    },
    {
      "epoch": 13.529874213836479,
      "grad_norm": 0.16276076436042786,
      "learning_rate": 1.4318617193337908e-05,
      "loss": 0.4976,
      "num_input_tokens_seen": 16874120,
      "step": 25815
    },
    {
      "epoch": 13.532494758909854,
      "grad_norm": 0.11113688349723816,
      "learning_rate": 1.4308280219231812e-05,
      "loss": 0.4942,
      "num_input_tokens_seen": 16878088,
      "step": 25820
    },
    {
      "epoch": 13.535115303983229,
      "grad_norm": 0.13517172634601593,
      "learning_rate": 1.4297945481711527e-05,
      "loss": 0.4515,
      "num_input_tokens_seen": 16880776,
      "step": 25825
    },
    {
      "epoch": 13.537735849056604,
      "grad_norm": 0.13457396626472473,
      "learning_rate": 1.4287612982938946e-05,
      "loss": 0.4268,
      "num_input_tokens_seen": 16884360,
      "step": 25830
    },
    {
      "epoch": 13.54035639412998,
      "grad_norm": 0.12086474895477295,
      "learning_rate": 1.4277282725075523e-05,
      "loss": 0.7243,
      "num_input_tokens_seen": 16887080,
      "step": 25835
    },
    {
      "epoch": 13.542976939203355,
      "grad_norm": 0.15143485367298126,
      "learning_rate": 1.426695471028221e-05,
      "loss": 0.4333,
      "num_input_tokens_seen": 16890536,
      "step": 25840
    },
    {
      "epoch": 13.54559748427673,
      "grad_norm": 0.12194278836250305,
      "learning_rate": 1.4256628940719524e-05,
      "loss": 0.3872,
      "num_input_tokens_seen": 16893992,
      "step": 25845
    },
    {
      "epoch": 13.548218029350105,
      "grad_norm": 0.17147287726402283,
      "learning_rate": 1.42463054185475e-05,
      "loss": 0.4848,
      "num_input_tokens_seen": 16896904,
      "step": 25850
    },
    {
      "epoch": 13.55083857442348,
      "grad_norm": 0.17326506972312927,
      "learning_rate": 1.4235984145925705e-05,
      "loss": 0.4989,
      "num_input_tokens_seen": 16899880,
      "step": 25855
    },
    {
      "epoch": 13.553459119496855,
      "grad_norm": 0.21705211699008942,
      "learning_rate": 1.4225665125013224e-05,
      "loss": 0.3707,
      "num_input_tokens_seen": 16903528,
      "step": 25860
    },
    {
      "epoch": 13.55607966457023,
      "grad_norm": 0.21952679753303528,
      "learning_rate": 1.4215348357968669e-05,
      "loss": 0.406,
      "num_input_tokens_seen": 16905736,
      "step": 25865
    },
    {
      "epoch": 13.558700209643606,
      "grad_norm": 0.14646801352500916,
      "learning_rate": 1.42050338469502e-05,
      "loss": 0.4181,
      "num_input_tokens_seen": 16909256,
      "step": 25870
    },
    {
      "epoch": 13.56132075471698,
      "grad_norm": 0.1697891652584076,
      "learning_rate": 1.4194721594115495e-05,
      "loss": 0.4457,
      "num_input_tokens_seen": 16911976,
      "step": 25875
    },
    {
      "epoch": 13.563941299790356,
      "grad_norm": 0.14228041470050812,
      "learning_rate": 1.4184411601621769e-05,
      "loss": 0.364,
      "num_input_tokens_seen": 16915592,
      "step": 25880
    },
    {
      "epoch": 13.566561844863731,
      "grad_norm": 0.10742329061031342,
      "learning_rate": 1.4174103871625738e-05,
      "loss": 0.4925,
      "num_input_tokens_seen": 16919080,
      "step": 25885
    },
    {
      "epoch": 13.569182389937106,
      "grad_norm": 0.16567575931549072,
      "learning_rate": 1.4163798406283674e-05,
      "loss": 0.598,
      "num_input_tokens_seen": 16922632,
      "step": 25890
    },
    {
      "epoch": 13.571802935010481,
      "grad_norm": 0.38576003909111023,
      "learning_rate": 1.415349520775135e-05,
      "loss": 0.5607,
      "num_input_tokens_seen": 16925736,
      "step": 25895
    },
    {
      "epoch": 13.574423480083858,
      "grad_norm": 0.18291518092155457,
      "learning_rate": 1.4143194278184086e-05,
      "loss": 0.3712,
      "num_input_tokens_seen": 16929448,
      "step": 25900
    },
    {
      "epoch": 13.577044025157234,
      "grad_norm": 0.1721651554107666,
      "learning_rate": 1.4132895619736724e-05,
      "loss": 0.5572,
      "num_input_tokens_seen": 16932584,
      "step": 25905
    },
    {
      "epoch": 13.579664570230609,
      "grad_norm": 0.21846391260623932,
      "learning_rate": 1.4122599234563613e-05,
      "loss": 0.7399,
      "num_input_tokens_seen": 16936072,
      "step": 25910
    },
    {
      "epoch": 13.582285115303984,
      "grad_norm": 0.19189633429050446,
      "learning_rate": 1.4112305124818647e-05,
      "loss": 0.3647,
      "num_input_tokens_seen": 16939464,
      "step": 25915
    },
    {
      "epoch": 13.584905660377359,
      "grad_norm": 0.14572003483772278,
      "learning_rate": 1.4102013292655242e-05,
      "loss": 0.4671,
      "num_input_tokens_seen": 16943912,
      "step": 25920
    },
    {
      "epoch": 13.587526205450734,
      "grad_norm": 0.12567396461963654,
      "learning_rate": 1.409172374022632e-05,
      "loss": 0.551,
      "num_input_tokens_seen": 16947464,
      "step": 25925
    },
    {
      "epoch": 13.59014675052411,
      "grad_norm": 0.16335241496562958,
      "learning_rate": 1.4081436469684339e-05,
      "loss": 0.436,
      "num_input_tokens_seen": 16950504,
      "step": 25930
    },
    {
      "epoch": 13.592767295597485,
      "grad_norm": 0.09740027040243149,
      "learning_rate": 1.4071151483181294e-05,
      "loss": 0.5229,
      "num_input_tokens_seen": 16953672,
      "step": 25935
    },
    {
      "epoch": 13.59538784067086,
      "grad_norm": 0.2054692655801773,
      "learning_rate": 1.4060868782868669e-05,
      "loss": 0.4493,
      "num_input_tokens_seen": 16956520,
      "step": 25940
    },
    {
      "epoch": 13.598008385744235,
      "grad_norm": 0.19011721014976501,
      "learning_rate": 1.405058837089749e-05,
      "loss": 0.4574,
      "num_input_tokens_seen": 16958888,
      "step": 25945
    },
    {
      "epoch": 13.60062893081761,
      "grad_norm": 0.20849427580833435,
      "learning_rate": 1.404031024941832e-05,
      "loss": 0.4562,
      "num_input_tokens_seen": 16962216,
      "step": 25950
    },
    {
      "epoch": 13.603249475890985,
      "grad_norm": 0.12140940874814987,
      "learning_rate": 1.40300344205812e-05,
      "loss": 0.594,
      "num_input_tokens_seen": 16966152,
      "step": 25955
    },
    {
      "epoch": 13.60587002096436,
      "grad_norm": 0.16644354164600372,
      "learning_rate": 1.4019760886535738e-05,
      "loss": 0.4854,
      "num_input_tokens_seen": 16969096,
      "step": 25960
    },
    {
      "epoch": 13.608490566037736,
      "grad_norm": 0.10214459896087646,
      "learning_rate": 1.4009489649431018e-05,
      "loss": 0.3541,
      "num_input_tokens_seen": 16971592,
      "step": 25965
    },
    {
      "epoch": 13.61111111111111,
      "grad_norm": 0.18145525455474854,
      "learning_rate": 1.3999220711415675e-05,
      "loss": 0.448,
      "num_input_tokens_seen": 16974056,
      "step": 25970
    },
    {
      "epoch": 13.613731656184486,
      "grad_norm": 0.1973467916250229,
      "learning_rate": 1.3988954074637867e-05,
      "loss": 0.4246,
      "num_input_tokens_seen": 16976008,
      "step": 25975
    },
    {
      "epoch": 13.616352201257861,
      "grad_norm": 0.14072130620479584,
      "learning_rate": 1.3978689741245238e-05,
      "loss": 0.511,
      "num_input_tokens_seen": 16979080,
      "step": 25980
    },
    {
      "epoch": 13.618972746331236,
      "grad_norm": 0.13767609000205994,
      "learning_rate": 1.3968427713384986e-05,
      "loss": 0.6022,
      "num_input_tokens_seen": 16982216,
      "step": 25985
    },
    {
      "epoch": 13.621593291404611,
      "grad_norm": 0.29479292035102844,
      "learning_rate": 1.3958167993203791e-05,
      "loss": 0.5484,
      "num_input_tokens_seen": 16985288,
      "step": 25990
    },
    {
      "epoch": 13.624213836477988,
      "grad_norm": 0.13969646394252777,
      "learning_rate": 1.3947910582847879e-05,
      "loss": 0.425,
      "num_input_tokens_seen": 16988264,
      "step": 25995
    },
    {
      "epoch": 13.626834381551364,
      "grad_norm": 0.2779316008090973,
      "learning_rate": 1.3937655484462985e-05,
      "loss": 0.4501,
      "num_input_tokens_seen": 16990920,
      "step": 26000
    },
    {
      "epoch": 13.629454926624739,
      "grad_norm": 0.12158552557229996,
      "learning_rate": 1.392740270019437e-05,
      "loss": 0.4664,
      "num_input_tokens_seen": 16993768,
      "step": 26005
    },
    {
      "epoch": 13.632075471698114,
      "grad_norm": 0.15655769407749176,
      "learning_rate": 1.3917152232186776e-05,
      "loss": 0.5149,
      "num_input_tokens_seen": 16996584,
      "step": 26010
    },
    {
      "epoch": 13.634696016771489,
      "grad_norm": 0.1420499086380005,
      "learning_rate": 1.3906904082584504e-05,
      "loss": 0.53,
      "num_input_tokens_seen": 17000040,
      "step": 26015
    },
    {
      "epoch": 13.637316561844864,
      "grad_norm": 0.15835542976856232,
      "learning_rate": 1.3896658253531334e-05,
      "loss": 0.4045,
      "num_input_tokens_seen": 17002696,
      "step": 26020
    },
    {
      "epoch": 13.63993710691824,
      "grad_norm": 0.09841752052307129,
      "learning_rate": 1.3886414747170581e-05,
      "loss": 0.5308,
      "num_input_tokens_seen": 17005992,
      "step": 26025
    },
    {
      "epoch": 13.642557651991615,
      "grad_norm": 0.14272689819335938,
      "learning_rate": 1.3876173565645072e-05,
      "loss": 0.3844,
      "num_input_tokens_seen": 17008904,
      "step": 26030
    },
    {
      "epoch": 13.64517819706499,
      "grad_norm": 0.34751978516578674,
      "learning_rate": 1.3865934711097156e-05,
      "loss": 0.409,
      "num_input_tokens_seen": 17011880,
      "step": 26035
    },
    {
      "epoch": 13.647798742138365,
      "grad_norm": 0.14064107835292816,
      "learning_rate": 1.3855698185668675e-05,
      "loss": 0.4413,
      "num_input_tokens_seen": 17015208,
      "step": 26040
    },
    {
      "epoch": 13.65041928721174,
      "grad_norm": 0.13356930017471313,
      "learning_rate": 1.3845463991500985e-05,
      "loss": 0.5778,
      "num_input_tokens_seen": 17018664,
      "step": 26045
    },
    {
      "epoch": 13.653039832285115,
      "grad_norm": 0.17491106688976288,
      "learning_rate": 1.3835232130734968e-05,
      "loss": 0.5539,
      "num_input_tokens_seen": 17022280,
      "step": 26050
    },
    {
      "epoch": 13.65566037735849,
      "grad_norm": 0.21272413432598114,
      "learning_rate": 1.3825002605511017e-05,
      "loss": 0.5229,
      "num_input_tokens_seen": 17026344,
      "step": 26055
    },
    {
      "epoch": 13.658280922431866,
      "grad_norm": 0.11223936080932617,
      "learning_rate": 1.3814775417969037e-05,
      "loss": 0.4135,
      "num_input_tokens_seen": 17030024,
      "step": 26060
    },
    {
      "epoch": 13.66090146750524,
      "grad_norm": 0.16148245334625244,
      "learning_rate": 1.3804550570248431e-05,
      "loss": 0.5856,
      "num_input_tokens_seen": 17033992,
      "step": 26065
    },
    {
      "epoch": 13.663522012578616,
      "grad_norm": 0.11372127383947372,
      "learning_rate": 1.3794328064488116e-05,
      "loss": 0.4301,
      "num_input_tokens_seen": 17040936,
      "step": 26070
    },
    {
      "epoch": 13.666142557651991,
      "grad_norm": 0.2188396453857422,
      "learning_rate": 1.3784107902826523e-05,
      "loss": 0.4052,
      "num_input_tokens_seen": 17043432,
      "step": 26075
    },
    {
      "epoch": 13.668763102725366,
      "grad_norm": 0.29518237709999084,
      "learning_rate": 1.3773890087401603e-05,
      "loss": 0.3928,
      "num_input_tokens_seen": 17046184,
      "step": 26080
    },
    {
      "epoch": 13.671383647798741,
      "grad_norm": 0.17458340525627136,
      "learning_rate": 1.3763674620350802e-05,
      "loss": 0.4149,
      "num_input_tokens_seen": 17049160,
      "step": 26085
    },
    {
      "epoch": 13.674004192872118,
      "grad_norm": 0.18972456455230713,
      "learning_rate": 1.3753461503811089e-05,
      "loss": 0.4225,
      "num_input_tokens_seen": 17052584,
      "step": 26090
    },
    {
      "epoch": 13.676624737945493,
      "grad_norm": 0.12646540999412537,
      "learning_rate": 1.3743250739918922e-05,
      "loss": 0.5136,
      "num_input_tokens_seen": 17055624,
      "step": 26095
    },
    {
      "epoch": 13.679245283018869,
      "grad_norm": 0.16441386938095093,
      "learning_rate": 1.3733042330810265e-05,
      "loss": 0.5763,
      "num_input_tokens_seen": 17058280,
      "step": 26100
    },
    {
      "epoch": 13.681865828092244,
      "grad_norm": 0.13005343079566956,
      "learning_rate": 1.3722836278620616e-05,
      "loss": 0.3618,
      "num_input_tokens_seen": 17061256,
      "step": 26105
    },
    {
      "epoch": 13.684486373165619,
      "grad_norm": 0.23769012093544006,
      "learning_rate": 1.3712632585484957e-05,
      "loss": 0.4218,
      "num_input_tokens_seen": 17065160,
      "step": 26110
    },
    {
      "epoch": 13.687106918238994,
      "grad_norm": 0.25267332792282104,
      "learning_rate": 1.37024312535378e-05,
      "loss": 0.5512,
      "num_input_tokens_seen": 17067816,
      "step": 26115
    },
    {
      "epoch": 13.68972746331237,
      "grad_norm": 0.10082416981458664,
      "learning_rate": 1.3692232284913133e-05,
      "loss": 0.4226,
      "num_input_tokens_seen": 17070952,
      "step": 26120
    },
    {
      "epoch": 13.692348008385745,
      "grad_norm": 0.17879711091518402,
      "learning_rate": 1.3682035681744459e-05,
      "loss": 0.4712,
      "num_input_tokens_seen": 17073928,
      "step": 26125
    },
    {
      "epoch": 13.69496855345912,
      "grad_norm": 0.13470551371574402,
      "learning_rate": 1.3671841446164795e-05,
      "loss": 0.4632,
      "num_input_tokens_seen": 17077192,
      "step": 26130
    },
    {
      "epoch": 13.697589098532495,
      "grad_norm": 0.21978072822093964,
      "learning_rate": 1.3661649580306669e-05,
      "loss": 0.5095,
      "num_input_tokens_seen": 17080360,
      "step": 26135
    },
    {
      "epoch": 13.70020964360587,
      "grad_norm": 0.1485520750284195,
      "learning_rate": 1.3651460086302087e-05,
      "loss": 0.533,
      "num_input_tokens_seen": 17083304,
      "step": 26140
    },
    {
      "epoch": 13.702830188679245,
      "grad_norm": 0.16133719682693481,
      "learning_rate": 1.3641272966282609e-05,
      "loss": 0.4961,
      "num_input_tokens_seen": 17086952,
      "step": 26145
    },
    {
      "epoch": 13.70545073375262,
      "grad_norm": 0.1151939406991005,
      "learning_rate": 1.3631088222379218e-05,
      "loss": 0.4504,
      "num_input_tokens_seen": 17091656,
      "step": 26150
    },
    {
      "epoch": 13.708071278825996,
      "grad_norm": 0.13079792261123657,
      "learning_rate": 1.3620905856722468e-05,
      "loss": 0.5353,
      "num_input_tokens_seen": 17094888,
      "step": 26155
    },
    {
      "epoch": 13.71069182389937,
      "grad_norm": 0.15580356121063232,
      "learning_rate": 1.3610725871442392e-05,
      "loss": 0.4786,
      "num_input_tokens_seen": 17098856,
      "step": 26160
    },
    {
      "epoch": 13.713312368972746,
      "grad_norm": 0.11858493089675903,
      "learning_rate": 1.3600548268668532e-05,
      "loss": 0.3738,
      "num_input_tokens_seen": 17101992,
      "step": 26165
    },
    {
      "epoch": 13.715932914046121,
      "grad_norm": 0.19124896824359894,
      "learning_rate": 1.3590373050529927e-05,
      "loss": 0.6069,
      "num_input_tokens_seen": 17104520,
      "step": 26170
    },
    {
      "epoch": 13.718553459119496,
      "grad_norm": 0.17733165621757507,
      "learning_rate": 1.3580200219155116e-05,
      "loss": 0.4521,
      "num_input_tokens_seen": 17107944,
      "step": 26175
    },
    {
      "epoch": 13.721174004192871,
      "grad_norm": 0.1729353815317154,
      "learning_rate": 1.3570029776672122e-05,
      "loss": 0.5217,
      "num_input_tokens_seen": 17111016,
      "step": 26180
    },
    {
      "epoch": 13.723794549266248,
      "grad_norm": 0.1634477972984314,
      "learning_rate": 1.35598617252085e-05,
      "loss": 0.5138,
      "num_input_tokens_seen": 17113544,
      "step": 26185
    },
    {
      "epoch": 13.726415094339622,
      "grad_norm": 0.17450366914272308,
      "learning_rate": 1.354969606689129e-05,
      "loss": 0.3932,
      "num_input_tokens_seen": 17116168,
      "step": 26190
    },
    {
      "epoch": 13.729035639412999,
      "grad_norm": 0.14460481703281403,
      "learning_rate": 1.353953280384704e-05,
      "loss": 0.4568,
      "num_input_tokens_seen": 17120264,
      "step": 26195
    },
    {
      "epoch": 13.731656184486374,
      "grad_norm": 0.3388247787952423,
      "learning_rate": 1.352937193820177e-05,
      "loss": 0.541,
      "num_input_tokens_seen": 17123080,
      "step": 26200
    },
    {
      "epoch": 13.734276729559749,
      "grad_norm": 0.15319602191448212,
      "learning_rate": 1.3519213472081038e-05,
      "loss": 0.4425,
      "num_input_tokens_seen": 17125960,
      "step": 26205
    },
    {
      "epoch": 13.736897274633124,
      "grad_norm": 0.18750156462192535,
      "learning_rate": 1.3509057407609854e-05,
      "loss": 0.5428,
      "num_input_tokens_seen": 17129544,
      "step": 26210
    },
    {
      "epoch": 13.7395178197065,
      "grad_norm": 0.2533303499221802,
      "learning_rate": 1.3498903746912769e-05,
      "loss": 0.5371,
      "num_input_tokens_seen": 17133384,
      "step": 26215
    },
    {
      "epoch": 13.742138364779874,
      "grad_norm": 0.2285735011100769,
      "learning_rate": 1.3488752492113807e-05,
      "loss": 0.6143,
      "num_input_tokens_seen": 17136008,
      "step": 26220
    },
    {
      "epoch": 13.74475890985325,
      "grad_norm": 0.22239716351032257,
      "learning_rate": 1.3478603645336508e-05,
      "loss": 0.5772,
      "num_input_tokens_seen": 17139144,
      "step": 26225
    },
    {
      "epoch": 13.747379454926625,
      "grad_norm": 0.1447904258966446,
      "learning_rate": 1.3468457208703877e-05,
      "loss": 0.358,
      "num_input_tokens_seen": 17142280,
      "step": 26230
    },
    {
      "epoch": 13.75,
      "grad_norm": 0.10844695568084717,
      "learning_rate": 1.3458313184338437e-05,
      "loss": 0.4711,
      "num_input_tokens_seen": 17146664,
      "step": 26235
    },
    {
      "epoch": 13.752620545073375,
      "grad_norm": 0.14567354321479797,
      "learning_rate": 1.3448171574362201e-05,
      "loss": 0.5546,
      "num_input_tokens_seen": 17149384,
      "step": 26240
    },
    {
      "epoch": 13.75524109014675,
      "grad_norm": 0.182490274310112,
      "learning_rate": 1.3438032380896681e-05,
      "loss": 0.5802,
      "num_input_tokens_seen": 17152328,
      "step": 26245
    },
    {
      "epoch": 13.757861635220126,
      "grad_norm": 0.11776015162467957,
      "learning_rate": 1.342789560606289e-05,
      "loss": 0.4642,
      "num_input_tokens_seen": 17155624,
      "step": 26250
    },
    {
      "epoch": 13.7604821802935,
      "grad_norm": 0.18825647234916687,
      "learning_rate": 1.341776125198131e-05,
      "loss": 0.4477,
      "num_input_tokens_seen": 17158376,
      "step": 26255
    },
    {
      "epoch": 13.763102725366876,
      "grad_norm": 0.12748773396015167,
      "learning_rate": 1.3407629320771948e-05,
      "loss": 0.4282,
      "num_input_tokens_seen": 17161320,
      "step": 26260
    },
    {
      "epoch": 13.765723270440251,
      "grad_norm": 0.16459283232688904,
      "learning_rate": 1.3397499814554266e-05,
      "loss": 0.545,
      "num_input_tokens_seen": 17165288,
      "step": 26265
    },
    {
      "epoch": 13.768343815513626,
      "grad_norm": 0.10836946219205856,
      "learning_rate": 1.3387372735447256e-05,
      "loss": 0.4441,
      "num_input_tokens_seen": 17168456,
      "step": 26270
    },
    {
      "epoch": 13.770964360587001,
      "grad_norm": 0.16720902919769287,
      "learning_rate": 1.3377248085569399e-05,
      "loss": 0.4788,
      "num_input_tokens_seen": 17171496,
      "step": 26275
    },
    {
      "epoch": 13.773584905660378,
      "grad_norm": 0.16360491514205933,
      "learning_rate": 1.336712586703863e-05,
      "loss": 0.4354,
      "num_input_tokens_seen": 17174824,
      "step": 26280
    },
    {
      "epoch": 13.776205450733752,
      "grad_norm": 0.22536137700080872,
      "learning_rate": 1.3357006081972414e-05,
      "loss": 0.6261,
      "num_input_tokens_seen": 17177064,
      "step": 26285
    },
    {
      "epoch": 13.778825995807129,
      "grad_norm": 0.24196012318134308,
      "learning_rate": 1.3346888732487706e-05,
      "loss": 0.4271,
      "num_input_tokens_seen": 17179752,
      "step": 26290
    },
    {
      "epoch": 13.781446540880504,
      "grad_norm": 0.108596570789814,
      "learning_rate": 1.3336773820700921e-05,
      "loss": 0.5563,
      "num_input_tokens_seen": 17184232,
      "step": 26295
    },
    {
      "epoch": 13.784067085953879,
      "grad_norm": 0.19632305204868317,
      "learning_rate": 1.3326661348727992e-05,
      "loss": 0.4035,
      "num_input_tokens_seen": 17187624,
      "step": 26300
    },
    {
      "epoch": 13.786687631027254,
      "grad_norm": 0.19089238345623016,
      "learning_rate": 1.3316551318684342e-05,
      "loss": 0.4108,
      "num_input_tokens_seen": 17190088,
      "step": 26305
    },
    {
      "epoch": 13.78930817610063,
      "grad_norm": 0.12397842109203339,
      "learning_rate": 1.3306443732684854e-05,
      "loss": 0.5009,
      "num_input_tokens_seen": 17194216,
      "step": 26310
    },
    {
      "epoch": 13.791928721174004,
      "grad_norm": 0.16383856534957886,
      "learning_rate": 1.3296338592843929e-05,
      "loss": 0.4909,
      "num_input_tokens_seen": 17197544,
      "step": 26315
    },
    {
      "epoch": 13.79454926624738,
      "grad_norm": 0.2520841658115387,
      "learning_rate": 1.3286235901275457e-05,
      "loss": 0.5651,
      "num_input_tokens_seen": 17200488,
      "step": 26320
    },
    {
      "epoch": 13.797169811320755,
      "grad_norm": 0.34367135167121887,
      "learning_rate": 1.327613566009279e-05,
      "loss": 0.5058,
      "num_input_tokens_seen": 17202824,
      "step": 26325
    },
    {
      "epoch": 13.79979035639413,
      "grad_norm": 0.2808428108692169,
      "learning_rate": 1.3266037871408798e-05,
      "loss": 0.3773,
      "num_input_tokens_seen": 17206632,
      "step": 26330
    },
    {
      "epoch": 13.802410901467505,
      "grad_norm": 0.03968525677919388,
      "learning_rate": 1.3255942537335805e-05,
      "loss": 0.3302,
      "num_input_tokens_seen": 17212648,
      "step": 26335
    },
    {
      "epoch": 13.80503144654088,
      "grad_norm": 0.10814213752746582,
      "learning_rate": 1.3245849659985652e-05,
      "loss": 0.4655,
      "num_input_tokens_seen": 17215176,
      "step": 26340
    },
    {
      "epoch": 13.807651991614255,
      "grad_norm": 0.3072148561477661,
      "learning_rate": 1.3235759241469658e-05,
      "loss": 0.4905,
      "num_input_tokens_seen": 17217672,
      "step": 26345
    },
    {
      "epoch": 13.81027253668763,
      "grad_norm": 0.09587962925434113,
      "learning_rate": 1.322567128389861e-05,
      "loss": 0.3907,
      "num_input_tokens_seen": 17221096,
      "step": 26350
    },
    {
      "epoch": 13.812893081761006,
      "grad_norm": 0.1648232489824295,
      "learning_rate": 1.3215585789382812e-05,
      "loss": 0.4166,
      "num_input_tokens_seen": 17223784,
      "step": 26355
    },
    {
      "epoch": 13.815513626834381,
      "grad_norm": 0.144674152135849,
      "learning_rate": 1.320550276003201e-05,
      "loss": 0.5352,
      "num_input_tokens_seen": 17227496,
      "step": 26360
    },
    {
      "epoch": 13.818134171907756,
      "grad_norm": 0.1950272023677826,
      "learning_rate": 1.3195422197955473e-05,
      "loss": 0.4845,
      "num_input_tokens_seen": 17230728,
      "step": 26365
    },
    {
      "epoch": 13.820754716981131,
      "grad_norm": 0.12486172467470169,
      "learning_rate": 1.3185344105261937e-05,
      "loss": 0.5205,
      "num_input_tokens_seen": 17233800,
      "step": 26370
    },
    {
      "epoch": 13.823375262054507,
      "grad_norm": 0.15475252270698547,
      "learning_rate": 1.3175268484059639e-05,
      "loss": 0.4629,
      "num_input_tokens_seen": 17236744,
      "step": 26375
    },
    {
      "epoch": 13.825995807127882,
      "grad_norm": 0.12998069822788239,
      "learning_rate": 1.316519533645626e-05,
      "loss": 0.4327,
      "num_input_tokens_seen": 17241224,
      "step": 26380
    },
    {
      "epoch": 13.828616352201259,
      "grad_norm": 0.24108104407787323,
      "learning_rate": 1.3155124664559004e-05,
      "loss": 0.5091,
      "num_input_tokens_seen": 17244808,
      "step": 26385
    },
    {
      "epoch": 13.831236897274634,
      "grad_norm": 0.16592907905578613,
      "learning_rate": 1.3145056470474525e-05,
      "loss": 0.4195,
      "num_input_tokens_seen": 17248168,
      "step": 26390
    },
    {
      "epoch": 13.833857442348009,
      "grad_norm": 0.11457110941410065,
      "learning_rate": 1.313499075630899e-05,
      "loss": 0.5264,
      "num_input_tokens_seen": 17252136,
      "step": 26395
    },
    {
      "epoch": 13.836477987421384,
      "grad_norm": 0.14176633954048157,
      "learning_rate": 1.312492752416802e-05,
      "loss": 0.5464,
      "num_input_tokens_seen": 17255880,
      "step": 26400
    },
    {
      "epoch": 13.83909853249476,
      "grad_norm": 0.5131126642227173,
      "learning_rate": 1.3114866776156748e-05,
      "loss": 0.6819,
      "num_input_tokens_seen": 17258728,
      "step": 26405
    },
    {
      "epoch": 13.841719077568134,
      "grad_norm": 0.12001121044158936,
      "learning_rate": 1.3104808514379748e-05,
      "loss": 0.5007,
      "num_input_tokens_seen": 17261992,
      "step": 26410
    },
    {
      "epoch": 13.84433962264151,
      "grad_norm": 0.14547491073608398,
      "learning_rate": 1.3094752740941094e-05,
      "loss": 0.4146,
      "num_input_tokens_seen": 17265896,
      "step": 26415
    },
    {
      "epoch": 13.846960167714885,
      "grad_norm": 0.11494239419698715,
      "learning_rate": 1.3084699457944344e-05,
      "loss": 0.5037,
      "num_input_tokens_seen": 17269032,
      "step": 26420
    },
    {
      "epoch": 13.84958071278826,
      "grad_norm": 0.17621523141860962,
      "learning_rate": 1.3074648667492528e-05,
      "loss": 0.5049,
      "num_input_tokens_seen": 17272456,
      "step": 26425
    },
    {
      "epoch": 13.852201257861635,
      "grad_norm": 0.1509450525045395,
      "learning_rate": 1.3064600371688171e-05,
      "loss": 0.5116,
      "num_input_tokens_seen": 17275496,
      "step": 26430
    },
    {
      "epoch": 13.85482180293501,
      "grad_norm": 0.22158442437648773,
      "learning_rate": 1.3054554572633249e-05,
      "loss": 0.3943,
      "num_input_tokens_seen": 17277672,
      "step": 26435
    },
    {
      "epoch": 13.857442348008385,
      "grad_norm": 0.08836471289396286,
      "learning_rate": 1.3044511272429222e-05,
      "loss": 0.509,
      "num_input_tokens_seen": 17281768,
      "step": 26440
    },
    {
      "epoch": 13.86006289308176,
      "grad_norm": 0.1586163341999054,
      "learning_rate": 1.3034470473177038e-05,
      "loss": 0.5052,
      "num_input_tokens_seen": 17284616,
      "step": 26445
    },
    {
      "epoch": 13.862683438155136,
      "grad_norm": 0.1627034693956375,
      "learning_rate": 1.3024432176977125e-05,
      "loss": 0.5812,
      "num_input_tokens_seen": 17287848,
      "step": 26450
    },
    {
      "epoch": 13.865303983228511,
      "grad_norm": 0.11738841235637665,
      "learning_rate": 1.3014396385929372e-05,
      "loss": 0.4075,
      "num_input_tokens_seen": 17294312,
      "step": 26455
    },
    {
      "epoch": 13.867924528301886,
      "grad_norm": 0.18907158076763153,
      "learning_rate": 1.3004363102133165e-05,
      "loss": 0.6223,
      "num_input_tokens_seen": 17297672,
      "step": 26460
    },
    {
      "epoch": 13.870545073375261,
      "grad_norm": 0.16474077105522156,
      "learning_rate": 1.2994332327687348e-05,
      "loss": 0.4834,
      "num_input_tokens_seen": 17301032,
      "step": 26465
    },
    {
      "epoch": 13.873165618448636,
      "grad_norm": 0.12679100036621094,
      "learning_rate": 1.2984304064690228e-05,
      "loss": 0.441,
      "num_input_tokens_seen": 17304200,
      "step": 26470
    },
    {
      "epoch": 13.875786163522012,
      "grad_norm": 0.22307521104812622,
      "learning_rate": 1.2974278315239619e-05,
      "loss": 0.5095,
      "num_input_tokens_seen": 17307208,
      "step": 26475
    },
    {
      "epoch": 13.878406708595389,
      "grad_norm": 0.13444942235946655,
      "learning_rate": 1.2964255081432786e-05,
      "loss": 0.4182,
      "num_input_tokens_seen": 17310056,
      "step": 26480
    },
    {
      "epoch": 13.881027253668764,
      "grad_norm": 0.09311802685260773,
      "learning_rate": 1.2954234365366488e-05,
      "loss": 0.4413,
      "num_input_tokens_seen": 17313960,
      "step": 26485
    },
    {
      "epoch": 13.883647798742139,
      "grad_norm": 0.11760953813791275,
      "learning_rate": 1.2944216169136936e-05,
      "loss": 0.4136,
      "num_input_tokens_seen": 17316712,
      "step": 26490
    },
    {
      "epoch": 13.886268343815514,
      "grad_norm": 0.2325531393289566,
      "learning_rate": 1.2934200494839815e-05,
      "loss": 0.4364,
      "num_input_tokens_seen": 17321096,
      "step": 26495
    },
    {
      "epoch": 13.88888888888889,
      "grad_norm": 0.21308113634586334,
      "learning_rate": 1.2924187344570293e-05,
      "loss": 0.6161,
      "num_input_tokens_seen": 17324168,
      "step": 26500
    },
    {
      "epoch": 13.891509433962264,
      "grad_norm": 0.1894020289182663,
      "learning_rate": 1.2914176720423011e-05,
      "loss": 0.4247,
      "num_input_tokens_seen": 17327016,
      "step": 26505
    },
    {
      "epoch": 13.89412997903564,
      "grad_norm": 0.18694144487380981,
      "learning_rate": 1.2904168624492088e-05,
      "loss": 0.5771,
      "num_input_tokens_seen": 17330088,
      "step": 26510
    },
    {
      "epoch": 13.896750524109015,
      "grad_norm": 0.20804964005947113,
      "learning_rate": 1.2894163058871092e-05,
      "loss": 0.4254,
      "num_input_tokens_seen": 17332328,
      "step": 26515
    },
    {
      "epoch": 13.89937106918239,
      "grad_norm": 0.31418576836586,
      "learning_rate": 1.2884160025653064e-05,
      "loss": 0.4691,
      "num_input_tokens_seen": 17336584,
      "step": 26520
    },
    {
      "epoch": 13.901991614255765,
      "grad_norm": 0.18242831528186798,
      "learning_rate": 1.2874159526930532e-05,
      "loss": 0.5134,
      "num_input_tokens_seen": 17339240,
      "step": 26525
    },
    {
      "epoch": 13.90461215932914,
      "grad_norm": 0.14162349700927734,
      "learning_rate": 1.286416156479549e-05,
      "loss": 0.4654,
      "num_input_tokens_seen": 17343688,
      "step": 26530
    },
    {
      "epoch": 13.907232704402515,
      "grad_norm": 0.1275283694267273,
      "learning_rate": 1.2854166141339394e-05,
      "loss": 0.4119,
      "num_input_tokens_seen": 17347336,
      "step": 26535
    },
    {
      "epoch": 13.90985324947589,
      "grad_norm": 0.13431967794895172,
      "learning_rate": 1.2844173258653183e-05,
      "loss": 0.4474,
      "num_input_tokens_seen": 17350952,
      "step": 26540
    },
    {
      "epoch": 13.912473794549266,
      "grad_norm": 0.1646914929151535,
      "learning_rate": 1.2834182918827247e-05,
      "loss": 0.5818,
      "num_input_tokens_seen": 17354376,
      "step": 26545
    },
    {
      "epoch": 13.915094339622641,
      "grad_norm": 0.37774449586868286,
      "learning_rate": 1.282419512395144e-05,
      "loss": 0.5664,
      "num_input_tokens_seen": 17358216,
      "step": 26550
    },
    {
      "epoch": 13.917714884696016,
      "grad_norm": 0.18970873951911926,
      "learning_rate": 1.2814209876115102e-05,
      "loss": 0.4671,
      "num_input_tokens_seen": 17361224,
      "step": 26555
    },
    {
      "epoch": 13.920335429769391,
      "grad_norm": 0.1592930108308792,
      "learning_rate": 1.2804227177407038e-05,
      "loss": 0.4776,
      "num_input_tokens_seen": 17364488,
      "step": 26560
    },
    {
      "epoch": 13.922955974842766,
      "grad_norm": 0.18297436833381653,
      "learning_rate": 1.279424702991552e-05,
      "loss": 0.49,
      "num_input_tokens_seen": 17368936,
      "step": 26565
    },
    {
      "epoch": 13.925576519916142,
      "grad_norm": 0.183937206864357,
      "learning_rate": 1.2784269435728264e-05,
      "loss": 0.5213,
      "num_input_tokens_seen": 17371912,
      "step": 26570
    },
    {
      "epoch": 13.928197064989519,
      "grad_norm": 0.18025580048561096,
      "learning_rate": 1.2774294396932488e-05,
      "loss": 0.4138,
      "num_input_tokens_seen": 17374856,
      "step": 26575
    },
    {
      "epoch": 13.930817610062894,
      "grad_norm": 0.2484734058380127,
      "learning_rate": 1.2764321915614835e-05,
      "loss": 0.3276,
      "num_input_tokens_seen": 17377832,
      "step": 26580
    },
    {
      "epoch": 13.933438155136269,
      "grad_norm": 0.2611376941204071,
      "learning_rate": 1.2754351993861444e-05,
      "loss": 0.4828,
      "num_input_tokens_seen": 17381064,
      "step": 26585
    },
    {
      "epoch": 13.936058700209644,
      "grad_norm": 0.14672304689884186,
      "learning_rate": 1.2744384633757922e-05,
      "loss": 0.4309,
      "num_input_tokens_seen": 17384968,
      "step": 26590
    },
    {
      "epoch": 13.93867924528302,
      "grad_norm": 0.1626577228307724,
      "learning_rate": 1.2734419837389306e-05,
      "loss": 0.5378,
      "num_input_tokens_seen": 17388104,
      "step": 26595
    },
    {
      "epoch": 13.941299790356394,
      "grad_norm": 0.2722126841545105,
      "learning_rate": 1.2724457606840123e-05,
      "loss": 0.5468,
      "num_input_tokens_seen": 17391880,
      "step": 26600
    },
    {
      "epoch": 13.94392033542977,
      "grad_norm": 0.13339243829250336,
      "learning_rate": 1.2714497944194376e-05,
      "loss": 0.4645,
      "num_input_tokens_seen": 17395240,
      "step": 26605
    },
    {
      "epoch": 13.946540880503145,
      "grad_norm": 0.20784206688404083,
      "learning_rate": 1.2704540851535484e-05,
      "loss": 0.3834,
      "num_input_tokens_seen": 17400680,
      "step": 26610
    },
    {
      "epoch": 13.94916142557652,
      "grad_norm": 0.2060338705778122,
      "learning_rate": 1.2694586330946376e-05,
      "loss": 0.8121,
      "num_input_tokens_seen": 17403080,
      "step": 26615
    },
    {
      "epoch": 13.951781970649895,
      "grad_norm": 0.11973952502012253,
      "learning_rate": 1.2684634384509427e-05,
      "loss": 0.5099,
      "num_input_tokens_seen": 17406600,
      "step": 26620
    },
    {
      "epoch": 13.95440251572327,
      "grad_norm": 0.1380576491355896,
      "learning_rate": 1.2674685014306456e-05,
      "loss": 0.5954,
      "num_input_tokens_seen": 17409320,
      "step": 26625
    },
    {
      "epoch": 13.957023060796645,
      "grad_norm": 0.22311724722385406,
      "learning_rate": 1.2664738222418776e-05,
      "loss": 0.4898,
      "num_input_tokens_seen": 17412616,
      "step": 26630
    },
    {
      "epoch": 13.95964360587002,
      "grad_norm": 0.16721267998218536,
      "learning_rate": 1.2654794010927118e-05,
      "loss": 0.3671,
      "num_input_tokens_seen": 17415912,
      "step": 26635
    },
    {
      "epoch": 13.962264150943396,
      "grad_norm": 0.22722838819026947,
      "learning_rate": 1.2644852381911715e-05,
      "loss": 0.5223,
      "num_input_tokens_seen": 17419656,
      "step": 26640
    },
    {
      "epoch": 13.964884696016771,
      "grad_norm": 0.1491553634405136,
      "learning_rate": 1.2634913337452248e-05,
      "loss": 0.4111,
      "num_input_tokens_seen": 17421768,
      "step": 26645
    },
    {
      "epoch": 13.967505241090146,
      "grad_norm": 0.1544128954410553,
      "learning_rate": 1.2624976879627832e-05,
      "loss": 0.3829,
      "num_input_tokens_seen": 17424968,
      "step": 26650
    },
    {
      "epoch": 13.970125786163521,
      "grad_norm": 0.12810683250427246,
      "learning_rate": 1.2615043010517069e-05,
      "loss": 0.4003,
      "num_input_tokens_seen": 17427912,
      "step": 26655
    },
    {
      "epoch": 13.972746331236896,
      "grad_norm": 0.16142016649246216,
      "learning_rate": 1.2605111732198027e-05,
      "loss": 0.4239,
      "num_input_tokens_seen": 17431272,
      "step": 26660
    },
    {
      "epoch": 13.975366876310272,
      "grad_norm": 0.17632512748241425,
      "learning_rate": 1.2595183046748188e-05,
      "loss": 0.4767,
      "num_input_tokens_seen": 17434728,
      "step": 26665
    },
    {
      "epoch": 13.977987421383649,
      "grad_norm": 0.24644240736961365,
      "learning_rate": 1.2585256956244545e-05,
      "loss": 0.399,
      "num_input_tokens_seen": 17438216,
      "step": 26670
    },
    {
      "epoch": 13.980607966457024,
      "grad_norm": 0.1400589793920517,
      "learning_rate": 1.2575333462763506e-05,
      "loss": 0.5196,
      "num_input_tokens_seen": 17441384,
      "step": 26675
    },
    {
      "epoch": 13.983228511530399,
      "grad_norm": 0.26153579354286194,
      "learning_rate": 1.2565412568380957e-05,
      "loss": 0.4209,
      "num_input_tokens_seen": 17444168,
      "step": 26680
    },
    {
      "epoch": 13.985849056603774,
      "grad_norm": 0.17635604739189148,
      "learning_rate": 1.2555494275172237e-05,
      "loss": 0.4858,
      "num_input_tokens_seen": 17448168,
      "step": 26685
    },
    {
      "epoch": 13.98846960167715,
      "grad_norm": 0.24153147637844086,
      "learning_rate": 1.2545578585212148e-05,
      "loss": 0.5055,
      "num_input_tokens_seen": 17450824,
      "step": 26690
    },
    {
      "epoch": 13.991090146750524,
      "grad_norm": 0.1084812730550766,
      "learning_rate": 1.2535665500574922e-05,
      "loss": 0.386,
      "num_input_tokens_seen": 17453832,
      "step": 26695
    },
    {
      "epoch": 13.9937106918239,
      "grad_norm": 0.08338215202093124,
      "learning_rate": 1.2525755023334285e-05,
      "loss": 0.4141,
      "num_input_tokens_seen": 17457704,
      "step": 26700
    },
    {
      "epoch": 13.996331236897275,
      "grad_norm": 0.14520776271820068,
      "learning_rate": 1.251584715556337e-05,
      "loss": 0.4835,
      "num_input_tokens_seen": 17460680,
      "step": 26705
    },
    {
      "epoch": 13.99895178197065,
      "grad_norm": 0.12975601851940155,
      "learning_rate": 1.2505941899334805e-05,
      "loss": 0.4111,
      "num_input_tokens_seen": 17464072,
      "step": 26710
    },
    {
      "epoch": 14.0,
      "eval_loss": 0.4928559362888336,
      "eval_runtime": 10.1754,
      "eval_samples_per_second": 83.338,
      "eval_steps_per_second": 20.834,
      "num_input_tokens_seen": 17464792,
      "step": 26712
    },
    {
      "epoch": 14.001572327044025,
      "grad_norm": 0.1473226696252823,
      "learning_rate": 1.2496039256720665e-05,
      "loss": 0.4881,
      "num_input_tokens_seen": 17466744,
      "step": 26715
    },
    {
      "epoch": 14.0041928721174,
      "grad_norm": 0.1626979559659958,
      "learning_rate": 1.2486139229792452e-05,
      "loss": 0.4279,
      "num_input_tokens_seen": 17470072,
      "step": 26720
    },
    {
      "epoch": 14.006813417190775,
      "grad_norm": 0.12052366137504578,
      "learning_rate": 1.2476241820621152e-05,
      "loss": 0.3586,
      "num_input_tokens_seen": 17473432,
      "step": 26725
    },
    {
      "epoch": 14.00943396226415,
      "grad_norm": 0.14019180834293365,
      "learning_rate": 1.2466347031277176e-05,
      "loss": 0.3971,
      "num_input_tokens_seen": 17476568,
      "step": 26730
    },
    {
      "epoch": 14.012054507337526,
      "grad_norm": 0.10338247567415237,
      "learning_rate": 1.2456454863830411e-05,
      "loss": 0.4463,
      "num_input_tokens_seen": 17479608,
      "step": 26735
    },
    {
      "epoch": 14.014675052410901,
      "grad_norm": 0.17512182891368866,
      "learning_rate": 1.2446565320350185e-05,
      "loss": 0.5096,
      "num_input_tokens_seen": 17482904,
      "step": 26740
    },
    {
      "epoch": 14.017295597484276,
      "grad_norm": 0.15652383863925934,
      "learning_rate": 1.2436678402905283e-05,
      "loss": 0.5049,
      "num_input_tokens_seen": 17485432,
      "step": 26745
    },
    {
      "epoch": 14.019916142557651,
      "grad_norm": 0.21535621583461761,
      "learning_rate": 1.2426794113563933e-05,
      "loss": 0.3128,
      "num_input_tokens_seen": 17487864,
      "step": 26750
    },
    {
      "epoch": 14.022536687631026,
      "grad_norm": 0.2006426304578781,
      "learning_rate": 1.24169124543938e-05,
      "loss": 0.4504,
      "num_input_tokens_seen": 17490424,
      "step": 26755
    },
    {
      "epoch": 14.025157232704403,
      "grad_norm": 0.13911962509155273,
      "learning_rate": 1.2407033427462026e-05,
      "loss": 0.4488,
      "num_input_tokens_seen": 17492888,
      "step": 26760
    },
    {
      "epoch": 14.027777777777779,
      "grad_norm": 0.18758495151996613,
      "learning_rate": 1.2397157034835192e-05,
      "loss": 0.6126,
      "num_input_tokens_seen": 17495448,
      "step": 26765
    },
    {
      "epoch": 14.030398322851154,
      "grad_norm": 0.1951223462820053,
      "learning_rate": 1.2387283278579325e-05,
      "loss": 0.3676,
      "num_input_tokens_seen": 17499160,
      "step": 26770
    },
    {
      "epoch": 14.033018867924529,
      "grad_norm": 0.1392652541399002,
      "learning_rate": 1.237741216075991e-05,
      "loss": 0.5309,
      "num_input_tokens_seen": 17501880,
      "step": 26775
    },
    {
      "epoch": 14.035639412997904,
      "grad_norm": 0.2402554750442505,
      "learning_rate": 1.2367543683441868e-05,
      "loss": 0.6032,
      "num_input_tokens_seen": 17504696,
      "step": 26780
    },
    {
      "epoch": 14.03825995807128,
      "grad_norm": 0.1756455898284912,
      "learning_rate": 1.2357677848689558e-05,
      "loss": 0.4808,
      "num_input_tokens_seen": 17507672,
      "step": 26785
    },
    {
      "epoch": 14.040880503144654,
      "grad_norm": 0.1157180592417717,
      "learning_rate": 1.2347814658566811e-05,
      "loss": 0.4142,
      "num_input_tokens_seen": 17510296,
      "step": 26790
    },
    {
      "epoch": 14.04350104821803,
      "grad_norm": 0.1768495887517929,
      "learning_rate": 1.2337954115136894e-05,
      "loss": 0.4766,
      "num_input_tokens_seen": 17512920,
      "step": 26795
    },
    {
      "epoch": 14.046121593291405,
      "grad_norm": 0.11672406643629074,
      "learning_rate": 1.2328096220462529e-05,
      "loss": 0.4948,
      "num_input_tokens_seen": 17516888,
      "step": 26800
    },
    {
      "epoch": 14.04874213836478,
      "grad_norm": 0.21441589295864105,
      "learning_rate": 1.2318240976605865e-05,
      "loss": 0.4288,
      "num_input_tokens_seen": 17520792,
      "step": 26805
    },
    {
      "epoch": 14.051362683438155,
      "grad_norm": 0.2038901150226593,
      "learning_rate": 1.23083883856285e-05,
      "loss": 0.4149,
      "num_input_tokens_seen": 17523896,
      "step": 26810
    },
    {
      "epoch": 14.05398322851153,
      "grad_norm": 0.16355149447917938,
      "learning_rate": 1.229853844959149e-05,
      "loss": 0.4938,
      "num_input_tokens_seen": 17527352,
      "step": 26815
    },
    {
      "epoch": 14.056603773584905,
      "grad_norm": 0.13758906722068787,
      "learning_rate": 1.228869117055533e-05,
      "loss": 0.4789,
      "num_input_tokens_seen": 17530872,
      "step": 26820
    },
    {
      "epoch": 14.05922431865828,
      "grad_norm": 0.1313660442829132,
      "learning_rate": 1.227884655057997e-05,
      "loss": 0.5171,
      "num_input_tokens_seen": 17533528,
      "step": 26825
    },
    {
      "epoch": 14.061844863731656,
      "grad_norm": 0.13661064207553864,
      "learning_rate": 1.2269004591724777e-05,
      "loss": 0.3464,
      "num_input_tokens_seen": 17537368,
      "step": 26830
    },
    {
      "epoch": 14.064465408805031,
      "grad_norm": 0.23276285827159882,
      "learning_rate": 1.2259165296048575e-05,
      "loss": 0.4788,
      "num_input_tokens_seen": 17541080,
      "step": 26835
    },
    {
      "epoch": 14.067085953878406,
      "grad_norm": 0.1482679694890976,
      "learning_rate": 1.2249328665609638e-05,
      "loss": 0.3958,
      "num_input_tokens_seen": 17545112,
      "step": 26840
    },
    {
      "epoch": 14.069706498951781,
      "grad_norm": 0.2032323032617569,
      "learning_rate": 1.2239494702465676e-05,
      "loss": 0.5424,
      "num_input_tokens_seen": 17547960,
      "step": 26845
    },
    {
      "epoch": 14.072327044025156,
      "grad_norm": 0.11360020935535431,
      "learning_rate": 1.2229663408673842e-05,
      "loss": 0.536,
      "num_input_tokens_seen": 17551416,
      "step": 26850
    },
    {
      "epoch": 14.074947589098532,
      "grad_norm": 0.14467599987983704,
      "learning_rate": 1.221983478629074e-05,
      "loss": 0.4558,
      "num_input_tokens_seen": 17555096,
      "step": 26855
    },
    {
      "epoch": 14.077568134171909,
      "grad_norm": 0.20058909058570862,
      "learning_rate": 1.22100088373724e-05,
      "loss": 0.5286,
      "num_input_tokens_seen": 17559448,
      "step": 26860
    },
    {
      "epoch": 14.080188679245284,
      "grad_norm": 0.15279491245746613,
      "learning_rate": 1.2200185563974283e-05,
      "loss": 0.4651,
      "num_input_tokens_seen": 17562584,
      "step": 26865
    },
    {
      "epoch": 14.082809224318659,
      "grad_norm": 0.14347581565380096,
      "learning_rate": 1.219036496815132e-05,
      "loss": 0.4162,
      "num_input_tokens_seen": 17566776,
      "step": 26870
    },
    {
      "epoch": 14.085429769392034,
      "grad_norm": 0.2718152701854706,
      "learning_rate": 1.2180547051957863e-05,
      "loss": 0.5612,
      "num_input_tokens_seen": 17569368,
      "step": 26875
    },
    {
      "epoch": 14.08805031446541,
      "grad_norm": 0.21703636646270752,
      "learning_rate": 1.2170731817447723e-05,
      "loss": 0.4968,
      "num_input_tokens_seen": 17571928,
      "step": 26880
    },
    {
      "epoch": 14.090670859538784,
      "grad_norm": 0.1600339263677597,
      "learning_rate": 1.216091926667412e-05,
      "loss": 0.4969,
      "num_input_tokens_seen": 17574392,
      "step": 26885
    },
    {
      "epoch": 14.09329140461216,
      "grad_norm": 0.663903534412384,
      "learning_rate": 1.2151109401689723e-05,
      "loss": 0.4125,
      "num_input_tokens_seen": 17578104,
      "step": 26890
    },
    {
      "epoch": 14.095911949685535,
      "grad_norm": 0.1381397396326065,
      "learning_rate": 1.214130222454665e-05,
      "loss": 0.4602,
      "num_input_tokens_seen": 17581368,
      "step": 26895
    },
    {
      "epoch": 14.09853249475891,
      "grad_norm": 0.13195718824863434,
      "learning_rate": 1.2131497737296452e-05,
      "loss": 0.3837,
      "num_input_tokens_seen": 17584024,
      "step": 26900
    },
    {
      "epoch": 14.101153039832285,
      "grad_norm": 0.2600973844528198,
      "learning_rate": 1.2121695941990124e-05,
      "loss": 0.4036,
      "num_input_tokens_seen": 17586168,
      "step": 26905
    },
    {
      "epoch": 14.10377358490566,
      "grad_norm": 0.18816561996936798,
      "learning_rate": 1.2111896840678075e-05,
      "loss": 0.4548,
      "num_input_tokens_seen": 17591096,
      "step": 26910
    },
    {
      "epoch": 14.106394129979035,
      "grad_norm": 0.29368558526039124,
      "learning_rate": 1.210210043541018e-05,
      "loss": 0.5948,
      "num_input_tokens_seen": 17594264,
      "step": 26915
    },
    {
      "epoch": 14.10901467505241,
      "grad_norm": 0.14310596883296967,
      "learning_rate": 1.2092306728235717e-05,
      "loss": 0.3788,
      "num_input_tokens_seen": 17597944,
      "step": 26920
    },
    {
      "epoch": 14.111635220125786,
      "grad_norm": 0.1558026820421219,
      "learning_rate": 1.2082515721203427e-05,
      "loss": 0.4653,
      "num_input_tokens_seen": 17600728,
      "step": 26925
    },
    {
      "epoch": 14.114255765199161,
      "grad_norm": 0.19615815579891205,
      "learning_rate": 1.2072727416361479e-05,
      "loss": 0.5491,
      "num_input_tokens_seen": 17603576,
      "step": 26930
    },
    {
      "epoch": 14.116876310272536,
      "grad_norm": 0.20617617666721344,
      "learning_rate": 1.2062941815757484e-05,
      "loss": 0.526,
      "num_input_tokens_seen": 17606392,
      "step": 26935
    },
    {
      "epoch": 14.119496855345911,
      "grad_norm": 0.11342829465866089,
      "learning_rate": 1.2053158921438454e-05,
      "loss": 0.4928,
      "num_input_tokens_seen": 17609944,
      "step": 26940
    },
    {
      "epoch": 14.122117400419286,
      "grad_norm": 0.1343696415424347,
      "learning_rate": 1.2043378735450885e-05,
      "loss": 0.5911,
      "num_input_tokens_seen": 17614424,
      "step": 26945
    },
    {
      "epoch": 14.124737945492662,
      "grad_norm": 0.15488606691360474,
      "learning_rate": 1.2033601259840657e-05,
      "loss": 0.5008,
      "num_input_tokens_seen": 17617368,
      "step": 26950
    },
    {
      "epoch": 14.127358490566039,
      "grad_norm": 0.19816520810127258,
      "learning_rate": 1.2023826496653118e-05,
      "loss": 0.4041,
      "num_input_tokens_seen": 17620792,
      "step": 26955
    },
    {
      "epoch": 14.129979035639414,
      "grad_norm": 0.1976056545972824,
      "learning_rate": 1.2014054447933044e-05,
      "loss": 0.5511,
      "num_input_tokens_seen": 17624152,
      "step": 26960
    },
    {
      "epoch": 14.132599580712789,
      "grad_norm": 0.1524377018213272,
      "learning_rate": 1.200428511572462e-05,
      "loss": 0.3999,
      "num_input_tokens_seen": 17627064,
      "step": 26965
    },
    {
      "epoch": 14.135220125786164,
      "grad_norm": 0.2826947271823883,
      "learning_rate": 1.1994518502071484e-05,
      "loss": 0.5052,
      "num_input_tokens_seen": 17629720,
      "step": 26970
    },
    {
      "epoch": 14.13784067085954,
      "grad_norm": 0.1044401228427887,
      "learning_rate": 1.1984754609016716e-05,
      "loss": 0.498,
      "num_input_tokens_seen": 17633048,
      "step": 26975
    },
    {
      "epoch": 14.140461215932914,
      "grad_norm": 0.13333900272846222,
      "learning_rate": 1.1974993438602785e-05,
      "loss": 0.5275,
      "num_input_tokens_seen": 17636472,
      "step": 26980
    },
    {
      "epoch": 14.14308176100629,
      "grad_norm": 0.39575743675231934,
      "learning_rate": 1.1965234992871641e-05,
      "loss": 0.5937,
      "num_input_tokens_seen": 17640120,
      "step": 26985
    },
    {
      "epoch": 14.145702306079665,
      "grad_norm": 0.14423339068889618,
      "learning_rate": 1.195547927386462e-05,
      "loss": 0.5983,
      "num_input_tokens_seen": 17643704,
      "step": 26990
    },
    {
      "epoch": 14.14832285115304,
      "grad_norm": 0.1526825726032257,
      "learning_rate": 1.1945726283622514e-05,
      "loss": 0.5822,
      "num_input_tokens_seen": 17646968,
      "step": 26995
    },
    {
      "epoch": 14.150943396226415,
      "grad_norm": 0.26169297099113464,
      "learning_rate": 1.1935976024185547e-05,
      "loss": 0.497,
      "num_input_tokens_seen": 17650552,
      "step": 27000
    },
    {
      "epoch": 14.15356394129979,
      "grad_norm": 0.2078411877155304,
      "learning_rate": 1.1926228497593341e-05,
      "loss": 0.4605,
      "num_input_tokens_seen": 17653528,
      "step": 27005
    },
    {
      "epoch": 14.156184486373165,
      "grad_norm": 0.11587031185626984,
      "learning_rate": 1.1916483705884985e-05,
      "loss": 0.442,
      "num_input_tokens_seen": 17656536,
      "step": 27010
    },
    {
      "epoch": 14.15880503144654,
      "grad_norm": 0.18738381564617157,
      "learning_rate": 1.1906741651098983e-05,
      "loss": 0.4192,
      "num_input_tokens_seen": 17662200,
      "step": 27015
    },
    {
      "epoch": 14.161425576519916,
      "grad_norm": 0.27956441044807434,
      "learning_rate": 1.1897002335273239e-05,
      "loss": 0.4253,
      "num_input_tokens_seen": 17665592,
      "step": 27020
    },
    {
      "epoch": 14.164046121593291,
      "grad_norm": 0.14670847356319427,
      "learning_rate": 1.1887265760445121e-05,
      "loss": 0.5644,
      "num_input_tokens_seen": 17668376,
      "step": 27025
    },
    {
      "epoch": 14.166666666666666,
      "grad_norm": 0.12389665096998215,
      "learning_rate": 1.1877531928651417e-05,
      "loss": 0.4152,
      "num_input_tokens_seen": 17672760,
      "step": 27030
    },
    {
      "epoch": 14.169287211740041,
      "grad_norm": 0.21020227670669556,
      "learning_rate": 1.1867800841928313e-05,
      "loss": 0.4764,
      "num_input_tokens_seen": 17675480,
      "step": 27035
    },
    {
      "epoch": 14.171907756813416,
      "grad_norm": 0.22610130906105042,
      "learning_rate": 1.1858072502311465e-05,
      "loss": 0.4769,
      "num_input_tokens_seen": 17678360,
      "step": 27040
    },
    {
      "epoch": 14.174528301886792,
      "grad_norm": 0.13335855305194855,
      "learning_rate": 1.1848346911835908e-05,
      "loss": 0.3824,
      "num_input_tokens_seen": 17682040,
      "step": 27045
    },
    {
      "epoch": 14.177148846960169,
      "grad_norm": 0.18199199438095093,
      "learning_rate": 1.1838624072536137e-05,
      "loss": 0.5808,
      "num_input_tokens_seen": 17685432,
      "step": 27050
    },
    {
      "epoch": 14.179769392033544,
      "grad_norm": 0.1242290735244751,
      "learning_rate": 1.1828903986446055e-05,
      "loss": 0.4858,
      "num_input_tokens_seen": 17688728,
      "step": 27055
    },
    {
      "epoch": 14.182389937106919,
      "grad_norm": 0.25191032886505127,
      "learning_rate": 1.1819186655599007e-05,
      "loss": 0.5275,
      "num_input_tokens_seen": 17692216,
      "step": 27060
    },
    {
      "epoch": 14.185010482180294,
      "grad_norm": 0.11383549124002457,
      "learning_rate": 1.1809472082027734e-05,
      "loss": 0.3759,
      "num_input_tokens_seen": 17696184,
      "step": 27065
    },
    {
      "epoch": 14.18763102725367,
      "grad_norm": 0.19407851994037628,
      "learning_rate": 1.1799760267764407e-05,
      "loss": 0.4156,
      "num_input_tokens_seen": 17698808,
      "step": 27070
    },
    {
      "epoch": 14.190251572327044,
      "grad_norm": 0.3007219731807709,
      "learning_rate": 1.1790051214840634e-05,
      "loss": 0.5767,
      "num_input_tokens_seen": 17702648,
      "step": 27075
    },
    {
      "epoch": 14.19287211740042,
      "grad_norm": 0.15433701872825623,
      "learning_rate": 1.1780344925287443e-05,
      "loss": 0.4025,
      "num_input_tokens_seen": 17705464,
      "step": 27080
    },
    {
      "epoch": 14.195492662473795,
      "grad_norm": 0.15195944905281067,
      "learning_rate": 1.1770641401135282e-05,
      "loss": 0.6122,
      "num_input_tokens_seen": 17707864,
      "step": 27085
    },
    {
      "epoch": 14.19811320754717,
      "grad_norm": 0.13550598919391632,
      "learning_rate": 1.1760940644414003e-05,
      "loss": 0.5383,
      "num_input_tokens_seen": 17711736,
      "step": 27090
    },
    {
      "epoch": 14.200733752620545,
      "grad_norm": 0.16363228857517242,
      "learning_rate": 1.1751242657152914e-05,
      "loss": 0.3897,
      "num_input_tokens_seen": 17715992,
      "step": 27095
    },
    {
      "epoch": 14.20335429769392,
      "grad_norm": 0.20674467086791992,
      "learning_rate": 1.1741547441380698e-05,
      "loss": 0.313,
      "num_input_tokens_seen": 17718616,
      "step": 27100
    },
    {
      "epoch": 14.205974842767295,
      "grad_norm": 0.16142487525939941,
      "learning_rate": 1.17318549991255e-05,
      "loss": 0.3911,
      "num_input_tokens_seen": 17721880,
      "step": 27105
    },
    {
      "epoch": 14.20859538784067,
      "grad_norm": 0.20299293100833893,
      "learning_rate": 1.1722165332414863e-05,
      "loss": 0.5891,
      "num_input_tokens_seen": 17724376,
      "step": 27110
    },
    {
      "epoch": 14.211215932914046,
      "grad_norm": 0.13678036630153656,
      "learning_rate": 1.1712478443275768e-05,
      "loss": 0.5093,
      "num_input_tokens_seen": 17727448,
      "step": 27115
    },
    {
      "epoch": 14.213836477987421,
      "grad_norm": 0.149604931473732,
      "learning_rate": 1.170279433373459e-05,
      "loss": 0.4617,
      "num_input_tokens_seen": 17730808,
      "step": 27120
    },
    {
      "epoch": 14.216457023060796,
      "grad_norm": 0.1541646420955658,
      "learning_rate": 1.1693113005817125e-05,
      "loss": 0.4862,
      "num_input_tokens_seen": 17733464,
      "step": 27125
    },
    {
      "epoch": 14.219077568134171,
      "grad_norm": 0.1304326206445694,
      "learning_rate": 1.1683434461548606e-05,
      "loss": 0.3581,
      "num_input_tokens_seen": 17736600,
      "step": 27130
    },
    {
      "epoch": 14.221698113207546,
      "grad_norm": 0.15688075125217438,
      "learning_rate": 1.1673758702953675e-05,
      "loss": 0.4587,
      "num_input_tokens_seen": 17739608,
      "step": 27135
    },
    {
      "epoch": 14.224318658280922,
      "grad_norm": 0.11751613020896912,
      "learning_rate": 1.1664085732056398e-05,
      "loss": 0.4723,
      "num_input_tokens_seen": 17742840,
      "step": 27140
    },
    {
      "epoch": 14.226939203354299,
      "grad_norm": 0.2910347878932953,
      "learning_rate": 1.1654415550880243e-05,
      "loss": 0.3967,
      "num_input_tokens_seen": 17744824,
      "step": 27145
    },
    {
      "epoch": 14.229559748427674,
      "grad_norm": 0.22477547824382782,
      "learning_rate": 1.1644748161448088e-05,
      "loss": 0.4905,
      "num_input_tokens_seen": 17747544,
      "step": 27150
    },
    {
      "epoch": 14.232180293501049,
      "grad_norm": 0.14827457070350647,
      "learning_rate": 1.1635083565782257e-05,
      "loss": 0.383,
      "num_input_tokens_seen": 17750840,
      "step": 27155
    },
    {
      "epoch": 14.234800838574424,
      "grad_norm": 0.14796453714370728,
      "learning_rate": 1.1625421765904465e-05,
      "loss": 0.6408,
      "num_input_tokens_seen": 17753432,
      "step": 27160
    },
    {
      "epoch": 14.2374213836478,
      "grad_norm": 0.21967925131320953,
      "learning_rate": 1.1615762763835858e-05,
      "loss": 0.4942,
      "num_input_tokens_seen": 17756536,
      "step": 27165
    },
    {
      "epoch": 14.240041928721174,
      "grad_norm": 0.08308501541614532,
      "learning_rate": 1.1606106561596993e-05,
      "loss": 0.3373,
      "num_input_tokens_seen": 17759896,
      "step": 27170
    },
    {
      "epoch": 14.24266247379455,
      "grad_norm": 0.17898239195346832,
      "learning_rate": 1.1596453161207829e-05,
      "loss": 0.5563,
      "num_input_tokens_seen": 17763352,
      "step": 27175
    },
    {
      "epoch": 14.245283018867925,
      "grad_norm": 0.16199485957622528,
      "learning_rate": 1.1586802564687741e-05,
      "loss": 0.4353,
      "num_input_tokens_seen": 17765816,
      "step": 27180
    },
    {
      "epoch": 14.2479035639413,
      "grad_norm": 0.18009786307811737,
      "learning_rate": 1.1577154774055528e-05,
      "loss": 0.4389,
      "num_input_tokens_seen": 17768056,
      "step": 27185
    },
    {
      "epoch": 14.250524109014675,
      "grad_norm": 0.2898913323879242,
      "learning_rate": 1.1567509791329401e-05,
      "loss": 0.4383,
      "num_input_tokens_seen": 17771896,
      "step": 27190
    },
    {
      "epoch": 14.25314465408805,
      "grad_norm": 0.13897597789764404,
      "learning_rate": 1.155786761852699e-05,
      "loss": 0.3907,
      "num_input_tokens_seen": 17775096,
      "step": 27195
    },
    {
      "epoch": 14.255765199161425,
      "grad_norm": 0.14378385245800018,
      "learning_rate": 1.1548228257665317e-05,
      "loss": 0.4583,
      "num_input_tokens_seen": 17778392,
      "step": 27200
    },
    {
      "epoch": 14.2583857442348,
      "grad_norm": 0.251681387424469,
      "learning_rate": 1.1538591710760815e-05,
      "loss": 0.4387,
      "num_input_tokens_seen": 17781176,
      "step": 27205
    },
    {
      "epoch": 14.261006289308176,
      "grad_norm": 0.1429704874753952,
      "learning_rate": 1.1528957979829355e-05,
      "loss": 0.6936,
      "num_input_tokens_seen": 17785272,
      "step": 27210
    },
    {
      "epoch": 14.26362683438155,
      "grad_norm": 0.17189976572990417,
      "learning_rate": 1.1519327066886198e-05,
      "loss": 0.5431,
      "num_input_tokens_seen": 17788280,
      "step": 27215
    },
    {
      "epoch": 14.266247379454926,
      "grad_norm": 0.14701399207115173,
      "learning_rate": 1.1509698973946032e-05,
      "loss": 0.4434,
      "num_input_tokens_seen": 17791064,
      "step": 27220
    },
    {
      "epoch": 14.268867924528301,
      "grad_norm": 0.11586274206638336,
      "learning_rate": 1.1500073703022926e-05,
      "loss": 0.3333,
      "num_input_tokens_seen": 17793656,
      "step": 27225
    },
    {
      "epoch": 14.271488469601676,
      "grad_norm": 0.17720665037631989,
      "learning_rate": 1.1490451256130394e-05,
      "loss": 0.4934,
      "num_input_tokens_seen": 17796536,
      "step": 27230
    },
    {
      "epoch": 14.274109014675052,
      "grad_norm": 0.16551536321640015,
      "learning_rate": 1.1480831635281328e-05,
      "loss": 0.5283,
      "num_input_tokens_seen": 17801304,
      "step": 27235
    },
    {
      "epoch": 14.276729559748428,
      "grad_norm": 0.1802019327878952,
      "learning_rate": 1.1471214842488046e-05,
      "loss": 0.4518,
      "num_input_tokens_seen": 17804472,
      "step": 27240
    },
    {
      "epoch": 14.279350104821804,
      "grad_norm": 0.21269077062606812,
      "learning_rate": 1.1461600879762277e-05,
      "loss": 0.4031,
      "num_input_tokens_seen": 17807928,
      "step": 27245
    },
    {
      "epoch": 14.281970649895179,
      "grad_norm": 0.3554603159427643,
      "learning_rate": 1.1451989749115157e-05,
      "loss": 0.6101,
      "num_input_tokens_seen": 17810680,
      "step": 27250
    },
    {
      "epoch": 14.284591194968554,
      "grad_norm": 0.12846490740776062,
      "learning_rate": 1.1442381452557219e-05,
      "loss": 0.405,
      "num_input_tokens_seen": 17814008,
      "step": 27255
    },
    {
      "epoch": 14.28721174004193,
      "grad_norm": 0.18090762197971344,
      "learning_rate": 1.1432775992098397e-05,
      "loss": 0.5246,
      "num_input_tokens_seen": 17816440,
      "step": 27260
    },
    {
      "epoch": 14.289832285115304,
      "grad_norm": 0.17599181830883026,
      "learning_rate": 1.1423173369748056e-05,
      "loss": 0.3935,
      "num_input_tokens_seen": 17819256,
      "step": 27265
    },
    {
      "epoch": 14.29245283018868,
      "grad_norm": 0.1403091847896576,
      "learning_rate": 1.1413573587514952e-05,
      "loss": 0.4058,
      "num_input_tokens_seen": 17822232,
      "step": 27270
    },
    {
      "epoch": 14.295073375262055,
      "grad_norm": 0.22970180213451385,
      "learning_rate": 1.140397664740726e-05,
      "loss": 0.4982,
      "num_input_tokens_seen": 17825432,
      "step": 27275
    },
    {
      "epoch": 14.29769392033543,
      "grad_norm": 0.31273457407951355,
      "learning_rate": 1.1394382551432534e-05,
      "loss": 0.5581,
      "num_input_tokens_seen": 17828952,
      "step": 27280
    },
    {
      "epoch": 14.300314465408805,
      "grad_norm": 0.19569942355155945,
      "learning_rate": 1.1384791301597767e-05,
      "loss": 0.5316,
      "num_input_tokens_seen": 17831672,
      "step": 27285
    },
    {
      "epoch": 14.30293501048218,
      "grad_norm": 0.1524016410112381,
      "learning_rate": 1.1375202899909317e-05,
      "loss": 0.4725,
      "num_input_tokens_seen": 17835384,
      "step": 27290
    },
    {
      "epoch": 14.305555555555555,
      "grad_norm": 0.1673910766839981,
      "learning_rate": 1.1365617348372984e-05,
      "loss": 0.5555,
      "num_input_tokens_seen": 17838552,
      "step": 27295
    },
    {
      "epoch": 14.30817610062893,
      "grad_norm": 0.24851790070533752,
      "learning_rate": 1.135603464899396e-05,
      "loss": 0.451,
      "num_input_tokens_seen": 17841080,
      "step": 27300
    },
    {
      "epoch": 14.310796645702306,
      "grad_norm": 0.2051054686307907,
      "learning_rate": 1.1346454803776821e-05,
      "loss": 0.4511,
      "num_input_tokens_seen": 17847224,
      "step": 27305
    },
    {
      "epoch": 14.31341719077568,
      "grad_norm": 0.16071321070194244,
      "learning_rate": 1.1336877814725566e-05,
      "loss": 0.4089,
      "num_input_tokens_seen": 17850840,
      "step": 27310
    },
    {
      "epoch": 14.316037735849056,
      "grad_norm": 0.18722547590732574,
      "learning_rate": 1.1327303683843607e-05,
      "loss": 0.5353,
      "num_input_tokens_seen": 17853176,
      "step": 27315
    },
    {
      "epoch": 14.318658280922431,
      "grad_norm": 0.13863863050937653,
      "learning_rate": 1.1317732413133723e-05,
      "loss": 0.438,
      "num_input_tokens_seen": 17856280,
      "step": 27320
    },
    {
      "epoch": 14.321278825995806,
      "grad_norm": 0.10934146493673325,
      "learning_rate": 1.1308164004598118e-05,
      "loss": 0.4369,
      "num_input_tokens_seen": 17859064,
      "step": 27325
    },
    {
      "epoch": 14.323899371069182,
      "grad_norm": 0.14799141883850098,
      "learning_rate": 1.129859846023841e-05,
      "loss": 0.5553,
      "num_input_tokens_seen": 17861880,
      "step": 27330
    },
    {
      "epoch": 14.326519916142558,
      "grad_norm": 0.22536221146583557,
      "learning_rate": 1.1289035782055582e-05,
      "loss": 0.611,
      "num_input_tokens_seen": 17868120,
      "step": 27335
    },
    {
      "epoch": 14.329140461215934,
      "grad_norm": 0.22705018520355225,
      "learning_rate": 1.1279475972050044e-05,
      "loss": 0.6201,
      "num_input_tokens_seen": 17871896,
      "step": 27340
    },
    {
      "epoch": 14.331761006289309,
      "grad_norm": 0.11778509616851807,
      "learning_rate": 1.1269919032221612e-05,
      "loss": 0.4291,
      "num_input_tokens_seen": 17875512,
      "step": 27345
    },
    {
      "epoch": 14.334381551362684,
      "grad_norm": 0.1633971482515335,
      "learning_rate": 1.1260364964569466e-05,
      "loss": 0.5202,
      "num_input_tokens_seen": 17879000,
      "step": 27350
    },
    {
      "epoch": 14.33700209643606,
      "grad_norm": 0.3723154067993164,
      "learning_rate": 1.125081377109223e-05,
      "loss": 0.5722,
      "num_input_tokens_seen": 17881368,
      "step": 27355
    },
    {
      "epoch": 14.339622641509434,
      "grad_norm": 0.16166509687900543,
      "learning_rate": 1.1241265453787882e-05,
      "loss": 0.6246,
      "num_input_tokens_seen": 17884408,
      "step": 27360
    },
    {
      "epoch": 14.34224318658281,
      "grad_norm": 0.2592201232910156,
      "learning_rate": 1.1231720014653837e-05,
      "loss": 0.6863,
      "num_input_tokens_seen": 17887224,
      "step": 27365
    },
    {
      "epoch": 14.344863731656185,
      "grad_norm": 0.1839495450258255,
      "learning_rate": 1.1222177455686896e-05,
      "loss": 0.4171,
      "num_input_tokens_seen": 17889720,
      "step": 27370
    },
    {
      "epoch": 14.34748427672956,
      "grad_norm": 0.09440121054649353,
      "learning_rate": 1.1212637778883237e-05,
      "loss": 0.3839,
      "num_input_tokens_seen": 17892888,
      "step": 27375
    },
    {
      "epoch": 14.350104821802935,
      "grad_norm": 0.2740993797779083,
      "learning_rate": 1.1203100986238468e-05,
      "loss": 0.446,
      "num_input_tokens_seen": 17896056,
      "step": 27380
    },
    {
      "epoch": 14.35272536687631,
      "grad_norm": 0.1660538762807846,
      "learning_rate": 1.1193567079747561e-05,
      "loss": 0.4703,
      "num_input_tokens_seen": 17898840,
      "step": 27385
    },
    {
      "epoch": 14.355345911949685,
      "grad_norm": 0.22171610593795776,
      "learning_rate": 1.1184036061404907e-05,
      "loss": 0.5111,
      "num_input_tokens_seen": 17901560,
      "step": 27390
    },
    {
      "epoch": 14.35796645702306,
      "grad_norm": 0.12834064662456512,
      "learning_rate": 1.117450793320429e-05,
      "loss": 0.4492,
      "num_input_tokens_seen": 17905432,
      "step": 27395
    },
    {
      "epoch": 14.360587002096436,
      "grad_norm": 0.17113064229488373,
      "learning_rate": 1.1164982697138893e-05,
      "loss": 0.4409,
      "num_input_tokens_seen": 17908088,
      "step": 27400
    },
    {
      "epoch": 14.36320754716981,
      "grad_norm": 0.20752014219760895,
      "learning_rate": 1.1155460355201267e-05,
      "loss": 0.5,
      "num_input_tokens_seen": 17910616,
      "step": 27405
    },
    {
      "epoch": 14.365828092243186,
      "grad_norm": 0.17972788214683533,
      "learning_rate": 1.1145940909383399e-05,
      "loss": 0.4139,
      "num_input_tokens_seen": 17913400,
      "step": 27410
    },
    {
      "epoch": 14.368448637316561,
      "grad_norm": 0.13246791064739227,
      "learning_rate": 1.1136424361676626e-05,
      "loss": 0.5577,
      "num_input_tokens_seen": 17916600,
      "step": 27415
    },
    {
      "epoch": 14.371069182389936,
      "grad_norm": 0.3164871037006378,
      "learning_rate": 1.1126910714071711e-05,
      "loss": 0.4754,
      "num_input_tokens_seen": 17919160,
      "step": 27420
    },
    {
      "epoch": 14.373689727463312,
      "grad_norm": 0.15395191311836243,
      "learning_rate": 1.1117399968558806e-05,
      "loss": 0.4014,
      "num_input_tokens_seen": 17921912,
      "step": 27425
    },
    {
      "epoch": 14.376310272536688,
      "grad_norm": 0.13754673302173615,
      "learning_rate": 1.1107892127127454e-05,
      "loss": 0.5297,
      "num_input_tokens_seen": 17926040,
      "step": 27430
    },
    {
      "epoch": 14.378930817610064,
      "grad_norm": 0.09833981841802597,
      "learning_rate": 1.109838719176658e-05,
      "loss": 0.4612,
      "num_input_tokens_seen": 17929656,
      "step": 27435
    },
    {
      "epoch": 14.381551362683439,
      "grad_norm": 0.13263724744319916,
      "learning_rate": 1.1088885164464497e-05,
      "loss": 0.399,
      "num_input_tokens_seen": 17935704,
      "step": 27440
    },
    {
      "epoch": 14.384171907756814,
      "grad_norm": 0.21520179510116577,
      "learning_rate": 1.1079386047208932e-05,
      "loss": 0.5574,
      "num_input_tokens_seen": 17938328,
      "step": 27445
    },
    {
      "epoch": 14.38679245283019,
      "grad_norm": 0.1792689710855484,
      "learning_rate": 1.1069889841986991e-05,
      "loss": 0.4287,
      "num_input_tokens_seen": 17941240,
      "step": 27450
    },
    {
      "epoch": 14.389412997903564,
      "grad_norm": 0.13223841786384583,
      "learning_rate": 1.1060396550785182e-05,
      "loss": 0.5732,
      "num_input_tokens_seen": 17944728,
      "step": 27455
    },
    {
      "epoch": 14.39203354297694,
      "grad_norm": 0.17644409835338593,
      "learning_rate": 1.1050906175589382e-05,
      "loss": 0.5248,
      "num_input_tokens_seen": 17948536,
      "step": 27460
    },
    {
      "epoch": 14.394654088050315,
      "grad_norm": 0.14366519451141357,
      "learning_rate": 1.1041418718384863e-05,
      "loss": 0.6001,
      "num_input_tokens_seen": 17951448,
      "step": 27465
    },
    {
      "epoch": 14.39727463312369,
      "grad_norm": 0.1573939472436905,
      "learning_rate": 1.1031934181156298e-05,
      "loss": 0.5039,
      "num_input_tokens_seen": 17954776,
      "step": 27470
    },
    {
      "epoch": 14.399895178197065,
      "grad_norm": 0.17027828097343445,
      "learning_rate": 1.1022452565887747e-05,
      "loss": 0.3721,
      "num_input_tokens_seen": 17958712,
      "step": 27475
    },
    {
      "epoch": 14.40251572327044,
      "grad_norm": 0.09228859841823578,
      "learning_rate": 1.1012973874562657e-05,
      "loss": 0.3752,
      "num_input_tokens_seen": 17961464,
      "step": 27480
    },
    {
      "epoch": 14.405136268343815,
      "grad_norm": 0.1806757152080536,
      "learning_rate": 1.1003498109163865e-05,
      "loss": 0.4605,
      "num_input_tokens_seen": 17965176,
      "step": 27485
    },
    {
      "epoch": 14.40775681341719,
      "grad_norm": 0.21589212119579315,
      "learning_rate": 1.0994025271673592e-05,
      "loss": 0.5286,
      "num_input_tokens_seen": 17968440,
      "step": 27490
    },
    {
      "epoch": 14.410377358490566,
      "grad_norm": 0.3616698384284973,
      "learning_rate": 1.0984555364073431e-05,
      "loss": 0.4825,
      "num_input_tokens_seen": 17971416,
      "step": 27495
    },
    {
      "epoch": 14.41299790356394,
      "grad_norm": 0.21798746287822723,
      "learning_rate": 1.0975088388344393e-05,
      "loss": 0.6145,
      "num_input_tokens_seen": 17974360,
      "step": 27500
    },
    {
      "epoch": 14.415618448637316,
      "grad_norm": 0.10613401234149933,
      "learning_rate": 1.096562434646686e-05,
      "loss": 0.4611,
      "num_input_tokens_seen": 17977912,
      "step": 27505
    },
    {
      "epoch": 14.418238993710691,
      "grad_norm": 0.12031621485948563,
      "learning_rate": 1.0956163240420611e-05,
      "loss": 0.362,
      "num_input_tokens_seen": 17980792,
      "step": 27510
    },
    {
      "epoch": 14.420859538784066,
      "grad_norm": 0.26918771862983704,
      "learning_rate": 1.094670507218479e-05,
      "loss": 0.4407,
      "num_input_tokens_seen": 17983416,
      "step": 27515
    },
    {
      "epoch": 14.423480083857442,
      "grad_norm": 0.14904610812664032,
      "learning_rate": 1.093724984373793e-05,
      "loss": 0.4558,
      "num_input_tokens_seen": 17986328,
      "step": 27520
    },
    {
      "epoch": 14.426100628930818,
      "grad_norm": 0.2208884209394455,
      "learning_rate": 1.0927797557057968e-05,
      "loss": 0.451,
      "num_input_tokens_seen": 17990584,
      "step": 27525
    },
    {
      "epoch": 14.428721174004194,
      "grad_norm": 0.11564580351114273,
      "learning_rate": 1.0918348214122209e-05,
      "loss": 0.4717,
      "num_input_tokens_seen": 17993848,
      "step": 27530
    },
    {
      "epoch": 14.431341719077569,
      "grad_norm": 0.17035138607025146,
      "learning_rate": 1.0908901816907366e-05,
      "loss": 0.4941,
      "num_input_tokens_seen": 17997208,
      "step": 27535
    },
    {
      "epoch": 14.433962264150944,
      "grad_norm": 0.3350023925304413,
      "learning_rate": 1.08994583673895e-05,
      "loss": 0.5495,
      "num_input_tokens_seen": 18000056,
      "step": 27540
    },
    {
      "epoch": 14.43658280922432,
      "grad_norm": 0.17966578900814056,
      "learning_rate": 1.0890017867544068e-05,
      "loss": 0.635,
      "num_input_tokens_seen": 18002968,
      "step": 27545
    },
    {
      "epoch": 14.439203354297694,
      "grad_norm": 0.10858426988124847,
      "learning_rate": 1.088058031934592e-05,
      "loss": 0.4115,
      "num_input_tokens_seen": 18006104,
      "step": 27550
    },
    {
      "epoch": 14.44182389937107,
      "grad_norm": 0.15667742490768433,
      "learning_rate": 1.087114572476929e-05,
      "loss": 0.4726,
      "num_input_tokens_seen": 18010040,
      "step": 27555
    },
    {
      "epoch": 14.444444444444445,
      "grad_norm": 0.1332666277885437,
      "learning_rate": 1.0861714085787783e-05,
      "loss": 0.4142,
      "num_input_tokens_seen": 18013720,
      "step": 27560
    },
    {
      "epoch": 14.44706498951782,
      "grad_norm": 0.12685805559158325,
      "learning_rate": 1.0852285404374402e-05,
      "loss": 0.3602,
      "num_input_tokens_seen": 18016696,
      "step": 27565
    },
    {
      "epoch": 14.449685534591195,
      "grad_norm": 0.11559215933084488,
      "learning_rate": 1.0842859682501505e-05,
      "loss": 0.5813,
      "num_input_tokens_seen": 18020024,
      "step": 27570
    },
    {
      "epoch": 14.45230607966457,
      "grad_norm": 0.25449663400650024,
      "learning_rate": 1.0833436922140841e-05,
      "loss": 0.4054,
      "num_input_tokens_seen": 18023352,
      "step": 27575
    },
    {
      "epoch": 14.454926624737945,
      "grad_norm": 0.21538645029067993,
      "learning_rate": 1.0824017125263555e-05,
      "loss": 0.4851,
      "num_input_tokens_seen": 18026648,
      "step": 27580
    },
    {
      "epoch": 14.45754716981132,
      "grad_norm": 0.18972110748291016,
      "learning_rate": 1.0814600293840154e-05,
      "loss": 0.4006,
      "num_input_tokens_seen": 18029624,
      "step": 27585
    },
    {
      "epoch": 14.460167714884696,
      "grad_norm": 0.1615440845489502,
      "learning_rate": 1.080518642984055e-05,
      "loss": 0.6109,
      "num_input_tokens_seen": 18032376,
      "step": 27590
    },
    {
      "epoch": 14.46278825995807,
      "grad_norm": 0.14680878818035126,
      "learning_rate": 1.0795775535233988e-05,
      "loss": 0.4386,
      "num_input_tokens_seen": 18035448,
      "step": 27595
    },
    {
      "epoch": 14.465408805031446,
      "grad_norm": 0.10999786108732224,
      "learning_rate": 1.0786367611989145e-05,
      "loss": 0.5047,
      "num_input_tokens_seen": 18039576,
      "step": 27600
    },
    {
      "epoch": 14.468029350104821,
      "grad_norm": 0.20022375881671906,
      "learning_rate": 1.0776962662074031e-05,
      "loss": 0.3952,
      "num_input_tokens_seen": 18044952,
      "step": 27605
    },
    {
      "epoch": 14.470649895178196,
      "grad_norm": 0.23932099342346191,
      "learning_rate": 1.0767560687456062e-05,
      "loss": 0.4785,
      "num_input_tokens_seen": 18048472,
      "step": 27610
    },
    {
      "epoch": 14.473270440251572,
      "grad_norm": 0.14768332242965698,
      "learning_rate": 1.0758161690102036e-05,
      "loss": 0.4839,
      "num_input_tokens_seen": 18052088,
      "step": 27615
    },
    {
      "epoch": 14.475890985324948,
      "grad_norm": 0.15289665758609772,
      "learning_rate": 1.0748765671978095e-05,
      "loss": 0.4879,
      "num_input_tokens_seen": 18055000,
      "step": 27620
    },
    {
      "epoch": 14.478511530398324,
      "grad_norm": 0.11961079388856888,
      "learning_rate": 1.0739372635049797e-05,
      "loss": 0.4546,
      "num_input_tokens_seen": 18058712,
      "step": 27625
    },
    {
      "epoch": 14.481132075471699,
      "grad_norm": 0.3265731632709503,
      "learning_rate": 1.0729982581282042e-05,
      "loss": 0.3643,
      "num_input_tokens_seen": 18062616,
      "step": 27630
    },
    {
      "epoch": 14.483752620545074,
      "grad_norm": 0.2939284145832062,
      "learning_rate": 1.0720595512639129e-05,
      "loss": 0.4416,
      "num_input_tokens_seen": 18066360,
      "step": 27635
    },
    {
      "epoch": 14.48637316561845,
      "grad_norm": 0.2896718382835388,
      "learning_rate": 1.0711211431084728e-05,
      "loss": 0.5076,
      "num_input_tokens_seen": 18069848,
      "step": 27640
    },
    {
      "epoch": 14.488993710691824,
      "grad_norm": 0.13369683921337128,
      "learning_rate": 1.0701830338581887e-05,
      "loss": 0.4781,
      "num_input_tokens_seen": 18072504,
      "step": 27645
    },
    {
      "epoch": 14.4916142557652,
      "grad_norm": 0.19235087931156158,
      "learning_rate": 1.069245223709301e-05,
      "loss": 0.3921,
      "num_input_tokens_seen": 18075960,
      "step": 27650
    },
    {
      "epoch": 14.494234800838575,
      "grad_norm": 0.12989529967308044,
      "learning_rate": 1.0683077128579904e-05,
      "loss": 0.4651,
      "num_input_tokens_seen": 18079608,
      "step": 27655
    },
    {
      "epoch": 14.49685534591195,
      "grad_norm": 0.13935236632823944,
      "learning_rate": 1.0673705015003718e-05,
      "loss": 0.6298,
      "num_input_tokens_seen": 18082232,
      "step": 27660
    },
    {
      "epoch": 14.499475890985325,
      "grad_norm": 0.1300833523273468,
      "learning_rate": 1.0664335898325002e-05,
      "loss": 0.3957,
      "num_input_tokens_seen": 18085432,
      "step": 27665
    },
    {
      "epoch": 14.5020964360587,
      "grad_norm": 0.22135518491268158,
      "learning_rate": 1.065496978050367e-05,
      "loss": 0.4331,
      "num_input_tokens_seen": 18089176,
      "step": 27670
    },
    {
      "epoch": 14.504716981132075,
      "grad_norm": 0.3442683219909668,
      "learning_rate": 1.0645606663499e-05,
      "loss": 0.5587,
      "num_input_tokens_seen": 18092600,
      "step": 27675
    },
    {
      "epoch": 14.50733752620545,
      "grad_norm": 0.1814754754304886,
      "learning_rate": 1.0636246549269649e-05,
      "loss": 0.6381,
      "num_input_tokens_seen": 18096056,
      "step": 27680
    },
    {
      "epoch": 14.509958071278826,
      "grad_norm": 0.10097813606262207,
      "learning_rate": 1.062688943977366e-05,
      "loss": 0.5486,
      "num_input_tokens_seen": 18099192,
      "step": 27685
    },
    {
      "epoch": 14.5125786163522,
      "grad_norm": 0.15098364651203156,
      "learning_rate": 1.0617535336968413e-05,
      "loss": 0.5052,
      "num_input_tokens_seen": 18106744,
      "step": 27690
    },
    {
      "epoch": 14.515199161425576,
      "grad_norm": 0.1867903769016266,
      "learning_rate": 1.0608184242810698e-05,
      "loss": 0.4638,
      "num_input_tokens_seen": 18109592,
      "step": 27695
    },
    {
      "epoch": 14.517819706498951,
      "grad_norm": 0.19787046313285828,
      "learning_rate": 1.0598836159256642e-05,
      "loss": 0.4792,
      "num_input_tokens_seen": 18112632,
      "step": 27700
    },
    {
      "epoch": 14.520440251572326,
      "grad_norm": 0.1622135043144226,
      "learning_rate": 1.0589491088261763e-05,
      "loss": 0.5209,
      "num_input_tokens_seen": 18116824,
      "step": 27705
    },
    {
      "epoch": 14.523060796645701,
      "grad_norm": 0.20204158127307892,
      "learning_rate": 1.0580149031780945e-05,
      "loss": 0.6024,
      "num_input_tokens_seen": 18120504,
      "step": 27710
    },
    {
      "epoch": 14.525681341719078,
      "grad_norm": 0.3287411630153656,
      "learning_rate": 1.0570809991768447e-05,
      "loss": 0.4114,
      "num_input_tokens_seen": 18123768,
      "step": 27715
    },
    {
      "epoch": 14.528301886792454,
      "grad_norm": 0.1655583381652832,
      "learning_rate": 1.0561473970177873e-05,
      "loss": 0.4036,
      "num_input_tokens_seen": 18126584,
      "step": 27720
    },
    {
      "epoch": 14.530922431865829,
      "grad_norm": 0.11587724834680557,
      "learning_rate": 1.0552140968962232e-05,
      "loss": 0.4225,
      "num_input_tokens_seen": 18131480,
      "step": 27725
    },
    {
      "epoch": 14.533542976939204,
      "grad_norm": 0.19922219216823578,
      "learning_rate": 1.054281099007386e-05,
      "loss": 0.6131,
      "num_input_tokens_seen": 18135224,
      "step": 27730
    },
    {
      "epoch": 14.536163522012579,
      "grad_norm": 0.1746472269296646,
      "learning_rate": 1.0533484035464492e-05,
      "loss": 0.3769,
      "num_input_tokens_seen": 18138232,
      "step": 27735
    },
    {
      "epoch": 14.538784067085954,
      "grad_norm": 0.13639618456363678,
      "learning_rate": 1.0524160107085231e-05,
      "loss": 0.5462,
      "num_input_tokens_seen": 18142040,
      "step": 27740
    },
    {
      "epoch": 14.54140461215933,
      "grad_norm": 0.13924190402030945,
      "learning_rate": 1.0514839206886518e-05,
      "loss": 0.4699,
      "num_input_tokens_seen": 18144888,
      "step": 27745
    },
    {
      "epoch": 14.544025157232705,
      "grad_norm": 0.4298059046268463,
      "learning_rate": 1.0505521336818195e-05,
      "loss": 0.5364,
      "num_input_tokens_seen": 18147672,
      "step": 27750
    },
    {
      "epoch": 14.54664570230608,
      "grad_norm": 0.21898573637008667,
      "learning_rate": 1.0496206498829438e-05,
      "loss": 0.3142,
      "num_input_tokens_seen": 18150808,
      "step": 27755
    },
    {
      "epoch": 14.549266247379455,
      "grad_norm": 0.11436732858419418,
      "learning_rate": 1.0486894694868814e-05,
      "loss": 0.4372,
      "num_input_tokens_seen": 18153400,
      "step": 27760
    },
    {
      "epoch": 14.55188679245283,
      "grad_norm": 0.13252122700214386,
      "learning_rate": 1.0477585926884246e-05,
      "loss": 0.3847,
      "num_input_tokens_seen": 18157432,
      "step": 27765
    },
    {
      "epoch": 14.554507337526205,
      "grad_norm": 0.34983959794044495,
      "learning_rate": 1.046828019682303e-05,
      "loss": 0.5994,
      "num_input_tokens_seen": 18160472,
      "step": 27770
    },
    {
      "epoch": 14.55712788259958,
      "grad_norm": 0.18069098889827728,
      "learning_rate": 1.0458977506631808e-05,
      "loss": 0.4644,
      "num_input_tokens_seen": 18163608,
      "step": 27775
    },
    {
      "epoch": 14.559748427672956,
      "grad_norm": 0.15091300010681152,
      "learning_rate": 1.0449677858256593e-05,
      "loss": 0.458,
      "num_input_tokens_seen": 18167736,
      "step": 27780
    },
    {
      "epoch": 14.56236897274633,
      "grad_norm": 0.21907052397727966,
      "learning_rate": 1.0440381253642773e-05,
      "loss": 0.3862,
      "num_input_tokens_seen": 18171160,
      "step": 27785
    },
    {
      "epoch": 14.564989517819706,
      "grad_norm": 0.15195219218730927,
      "learning_rate": 1.0431087694735087e-05,
      "loss": 0.6004,
      "num_input_tokens_seen": 18174168,
      "step": 27790
    },
    {
      "epoch": 14.567610062893081,
      "grad_norm": 0.09930340200662613,
      "learning_rate": 1.0421797183477644e-05,
      "loss": 0.6195,
      "num_input_tokens_seen": 18177080,
      "step": 27795
    },
    {
      "epoch": 14.570230607966456,
      "grad_norm": 0.10615040361881256,
      "learning_rate": 1.041250972181392e-05,
      "loss": 0.4928,
      "num_input_tokens_seen": 18181976,
      "step": 27800
    },
    {
      "epoch": 14.572851153039831,
      "grad_norm": 0.1569836139678955,
      "learning_rate": 1.0403225311686743e-05,
      "loss": 0.5251,
      "num_input_tokens_seen": 18186168,
      "step": 27805
    },
    {
      "epoch": 14.575471698113208,
      "grad_norm": 0.14025428891181946,
      "learning_rate": 1.0393943955038293e-05,
      "loss": 0.4236,
      "num_input_tokens_seen": 18189656,
      "step": 27810
    },
    {
      "epoch": 14.578092243186584,
      "grad_norm": 0.16084074974060059,
      "learning_rate": 1.038466565381013e-05,
      "loss": 0.3726,
      "num_input_tokens_seen": 18191768,
      "step": 27815
    },
    {
      "epoch": 14.580712788259959,
      "grad_norm": 0.09835737943649292,
      "learning_rate": 1.0375390409943175e-05,
      "loss": 0.413,
      "num_input_tokens_seen": 18194296,
      "step": 27820
    },
    {
      "epoch": 14.583333333333334,
      "grad_norm": 0.1599222868680954,
      "learning_rate": 1.0366118225377708e-05,
      "loss": 0.5163,
      "num_input_tokens_seen": 18197528,
      "step": 27825
    },
    {
      "epoch": 14.585953878406709,
      "grad_norm": 0.15761932730674744,
      "learning_rate": 1.0356849102053353e-05,
      "loss": 0.3693,
      "num_input_tokens_seen": 18200600,
      "step": 27830
    },
    {
      "epoch": 14.588574423480084,
      "grad_norm": 0.21305470168590546,
      "learning_rate": 1.0347583041909104e-05,
      "loss": 0.4982,
      "num_input_tokens_seen": 18203864,
      "step": 27835
    },
    {
      "epoch": 14.59119496855346,
      "grad_norm": 0.17403799295425415,
      "learning_rate": 1.0338320046883315e-05,
      "loss": 0.4928,
      "num_input_tokens_seen": 18207192,
      "step": 27840
    },
    {
      "epoch": 14.593815513626835,
      "grad_norm": 0.13271157443523407,
      "learning_rate": 1.0329060118913705e-05,
      "loss": 0.4665,
      "num_input_tokens_seen": 18211352,
      "step": 27845
    },
    {
      "epoch": 14.59643605870021,
      "grad_norm": 0.18376490473747253,
      "learning_rate": 1.0319803259937353e-05,
      "loss": 0.4725,
      "num_input_tokens_seen": 18214200,
      "step": 27850
    },
    {
      "epoch": 14.599056603773585,
      "grad_norm": 0.16964150965213776,
      "learning_rate": 1.0310549471890676e-05,
      "loss": 0.4402,
      "num_input_tokens_seen": 18216920,
      "step": 27855
    },
    {
      "epoch": 14.60167714884696,
      "grad_norm": 0.20375695824623108,
      "learning_rate": 1.0301298756709454e-05,
      "loss": 0.5942,
      "num_input_tokens_seen": 18220184,
      "step": 27860
    },
    {
      "epoch": 14.604297693920335,
      "grad_norm": 0.1871718466281891,
      "learning_rate": 1.0292051116328844e-05,
      "loss": 0.3831,
      "num_input_tokens_seen": 18224280,
      "step": 27865
    },
    {
      "epoch": 14.60691823899371,
      "grad_norm": 0.2534862160682678,
      "learning_rate": 1.028280655268334e-05,
      "loss": 0.3753,
      "num_input_tokens_seen": 18226744,
      "step": 27870
    },
    {
      "epoch": 14.609538784067086,
      "grad_norm": 0.18247751891613007,
      "learning_rate": 1.0273565067706806e-05,
      "loss": 0.5576,
      "num_input_tokens_seen": 18230232,
      "step": 27875
    },
    {
      "epoch": 14.61215932914046,
      "grad_norm": 0.15842373669147491,
      "learning_rate": 1.0264326663332457e-05,
      "loss": 0.4638,
      "num_input_tokens_seen": 18234040,
      "step": 27880
    },
    {
      "epoch": 14.614779874213836,
      "grad_norm": 0.1816728413105011,
      "learning_rate": 1.0255091341492859e-05,
      "loss": 0.2274,
      "num_input_tokens_seen": 18236632,
      "step": 27885
    },
    {
      "epoch": 14.617400419287211,
      "grad_norm": 0.25142258405685425,
      "learning_rate": 1.0245859104119923e-05,
      "loss": 0.4451,
      "num_input_tokens_seen": 18239640,
      "step": 27890
    },
    {
      "epoch": 14.620020964360586,
      "grad_norm": 0.13801544904708862,
      "learning_rate": 1.0236629953144938e-05,
      "loss": 0.495,
      "num_input_tokens_seen": 18242648,
      "step": 27895
    },
    {
      "epoch": 14.622641509433961,
      "grad_norm": 0.13479217886924744,
      "learning_rate": 1.0227403890498537e-05,
      "loss": 0.4286,
      "num_input_tokens_seen": 18246584,
      "step": 27900
    },
    {
      "epoch": 14.625262054507338,
      "grad_norm": 0.23871338367462158,
      "learning_rate": 1.021818091811072e-05,
      "loss": 0.6158,
      "num_input_tokens_seen": 18249400,
      "step": 27905
    },
    {
      "epoch": 14.627882599580714,
      "grad_norm": 0.13141880929470062,
      "learning_rate": 1.0208961037910813e-05,
      "loss": 0.5953,
      "num_input_tokens_seen": 18252952,
      "step": 27910
    },
    {
      "epoch": 14.630503144654089,
      "grad_norm": 0.140572190284729,
      "learning_rate": 1.0199744251827503e-05,
      "loss": 0.5413,
      "num_input_tokens_seen": 18257176,
      "step": 27915
    },
    {
      "epoch": 14.633123689727464,
      "grad_norm": 0.1244434118270874,
      "learning_rate": 1.0190530561788845e-05,
      "loss": 0.485,
      "num_input_tokens_seen": 18260280,
      "step": 27920
    },
    {
      "epoch": 14.635744234800839,
      "grad_norm": 0.1436985433101654,
      "learning_rate": 1.0181319969722236e-05,
      "loss": 0.4961,
      "num_input_tokens_seen": 18263480,
      "step": 27925
    },
    {
      "epoch": 14.638364779874214,
      "grad_norm": 0.16595877707004547,
      "learning_rate": 1.0172112477554441e-05,
      "loss": 0.3517,
      "num_input_tokens_seen": 18267128,
      "step": 27930
    },
    {
      "epoch": 14.64098532494759,
      "grad_norm": 0.13160265982151031,
      "learning_rate": 1.016290808721154e-05,
      "loss": 0.5818,
      "num_input_tokens_seen": 18270968,
      "step": 27935
    },
    {
      "epoch": 14.643605870020965,
      "grad_norm": 0.17175893485546112,
      "learning_rate": 1.0153706800619007e-05,
      "loss": 0.5245,
      "num_input_tokens_seen": 18273720,
      "step": 27940
    },
    {
      "epoch": 14.64622641509434,
      "grad_norm": 0.13102498650550842,
      "learning_rate": 1.0144508619701623e-05,
      "loss": 0.4579,
      "num_input_tokens_seen": 18276920,
      "step": 27945
    },
    {
      "epoch": 14.648846960167715,
      "grad_norm": 0.16096261143684387,
      "learning_rate": 1.013531354638356e-05,
      "loss": 0.6012,
      "num_input_tokens_seen": 18280952,
      "step": 27950
    },
    {
      "epoch": 14.65146750524109,
      "grad_norm": 0.17839115858078003,
      "learning_rate": 1.0126121582588316e-05,
      "loss": 0.6483,
      "num_input_tokens_seen": 18284120,
      "step": 27955
    },
    {
      "epoch": 14.654088050314465,
      "grad_norm": 0.14951235055923462,
      "learning_rate": 1.0116932730238754e-05,
      "loss": 0.5314,
      "num_input_tokens_seen": 18288120,
      "step": 27960
    },
    {
      "epoch": 14.65670859538784,
      "grad_norm": 0.17426860332489014,
      "learning_rate": 1.0107746991257063e-05,
      "loss": 0.4999,
      "num_input_tokens_seen": 18291224,
      "step": 27965
    },
    {
      "epoch": 14.659329140461216,
      "grad_norm": 0.10182803869247437,
      "learning_rate": 1.0098564367564811e-05,
      "loss": 0.391,
      "num_input_tokens_seen": 18294520,
      "step": 27970
    },
    {
      "epoch": 14.66194968553459,
      "grad_norm": 0.1637389212846756,
      "learning_rate": 1.0089384861082876e-05,
      "loss": 0.4032,
      "num_input_tokens_seen": 18298296,
      "step": 27975
    },
    {
      "epoch": 14.664570230607966,
      "grad_norm": 0.18061593174934387,
      "learning_rate": 1.0080208473731523e-05,
      "loss": 0.4641,
      "num_input_tokens_seen": 18301016,
      "step": 27980
    },
    {
      "epoch": 14.667190775681341,
      "grad_norm": 0.14110521972179413,
      "learning_rate": 1.0071035207430352e-05,
      "loss": 0.4903,
      "num_input_tokens_seen": 18304728,
      "step": 27985
    },
    {
      "epoch": 14.669811320754716,
      "grad_norm": 0.16966156661510468,
      "learning_rate": 1.0061865064098285e-05,
      "loss": 0.5181,
      "num_input_tokens_seen": 18307736,
      "step": 27990
    },
    {
      "epoch": 14.672431865828091,
      "grad_norm": 0.1630447655916214,
      "learning_rate": 1.0052698045653625e-05,
      "loss": 0.4157,
      "num_input_tokens_seen": 18310744,
      "step": 27995
    },
    {
      "epoch": 14.675052410901468,
      "grad_norm": 0.3297748565673828,
      "learning_rate": 1.0043534154014015e-05,
      "loss": 0.5733,
      "num_input_tokens_seen": 18313336,
      "step": 28000
    },
    {
      "epoch": 14.677672955974844,
      "grad_norm": 0.26720112562179565,
      "learning_rate": 1.003437339109642e-05,
      "loss": 0.5971,
      "num_input_tokens_seen": 18316120,
      "step": 28005
    },
    {
      "epoch": 14.680293501048219,
      "grad_norm": 0.12333113700151443,
      "learning_rate": 1.0025215758817183e-05,
      "loss": 0.5239,
      "num_input_tokens_seen": 18319480,
      "step": 28010
    },
    {
      "epoch": 14.682914046121594,
      "grad_norm": 0.11458618193864822,
      "learning_rate": 1.0016061259091958e-05,
      "loss": 0.3656,
      "num_input_tokens_seen": 18323288,
      "step": 28015
    },
    {
      "epoch": 14.685534591194969,
      "grad_norm": 0.1544247269630432,
      "learning_rate": 1.000690989383577e-05,
      "loss": 0.436,
      "num_input_tokens_seen": 18325752,
      "step": 28020
    },
    {
      "epoch": 14.688155136268344,
      "grad_norm": 0.14964865148067474,
      "learning_rate": 9.997761664962993e-06,
      "loss": 0.4004,
      "num_input_tokens_seen": 18330424,
      "step": 28025
    },
    {
      "epoch": 14.69077568134172,
      "grad_norm": 0.31449592113494873,
      "learning_rate": 9.988616574387314e-06,
      "loss": 0.4668,
      "num_input_tokens_seen": 18333336,
      "step": 28030
    },
    {
      "epoch": 14.693396226415095,
      "grad_norm": 0.14152859151363373,
      "learning_rate": 9.979474624021787e-06,
      "loss": 0.432,
      "num_input_tokens_seen": 18336760,
      "step": 28035
    },
    {
      "epoch": 14.69601677148847,
      "grad_norm": 0.25612694025039673,
      "learning_rate": 9.970335815778815e-06,
      "loss": 0.4417,
      "num_input_tokens_seen": 18339288,
      "step": 28040
    },
    {
      "epoch": 14.698637316561845,
      "grad_norm": 0.11282684653997421,
      "learning_rate": 9.96120015157011e-06,
      "loss": 0.648,
      "num_input_tokens_seen": 18343064,
      "step": 28045
    },
    {
      "epoch": 14.70125786163522,
      "grad_norm": 0.11348342150449753,
      "learning_rate": 9.952067633306767e-06,
      "loss": 0.4847,
      "num_input_tokens_seen": 18346584,
      "step": 28050
    },
    {
      "epoch": 14.703878406708595,
      "grad_norm": 0.19381752610206604,
      "learning_rate": 9.942938262899204e-06,
      "loss": 0.404,
      "num_input_tokens_seen": 18349752,
      "step": 28055
    },
    {
      "epoch": 14.70649895178197,
      "grad_norm": 0.21954300999641418,
      "learning_rate": 9.93381204225717e-06,
      "loss": 0.3881,
      "num_input_tokens_seen": 18352632,
      "step": 28060
    },
    {
      "epoch": 14.709119496855346,
      "grad_norm": 0.17949922382831573,
      "learning_rate": 9.924688973289782e-06,
      "loss": 0.4835,
      "num_input_tokens_seen": 18356280,
      "step": 28065
    },
    {
      "epoch": 14.71174004192872,
      "grad_norm": 0.12947280704975128,
      "learning_rate": 9.915569057905463e-06,
      "loss": 0.3781,
      "num_input_tokens_seen": 18359512,
      "step": 28070
    },
    {
      "epoch": 14.714360587002096,
      "grad_norm": 0.30305713415145874,
      "learning_rate": 9.906452298012003e-06,
      "loss": 0.5373,
      "num_input_tokens_seen": 18362072,
      "step": 28075
    },
    {
      "epoch": 14.716981132075471,
      "grad_norm": 0.23192809522151947,
      "learning_rate": 9.897338695516529e-06,
      "loss": 0.4967,
      "num_input_tokens_seen": 18365400,
      "step": 28080
    },
    {
      "epoch": 14.719601677148846,
      "grad_norm": 0.10057318955659866,
      "learning_rate": 9.888228252325507e-06,
      "loss": 0.5113,
      "num_input_tokens_seen": 18369176,
      "step": 28085
    },
    {
      "epoch": 14.722222222222221,
      "grad_norm": 0.20996491611003876,
      "learning_rate": 9.87912097034473e-06,
      "loss": 0.3914,
      "num_input_tokens_seen": 18372472,
      "step": 28090
    },
    {
      "epoch": 14.724842767295598,
      "grad_norm": 0.17039617896080017,
      "learning_rate": 9.870016851479327e-06,
      "loss": 0.4952,
      "num_input_tokens_seen": 18375608,
      "step": 28095
    },
    {
      "epoch": 14.727463312368974,
      "grad_norm": 0.19220611453056335,
      "learning_rate": 9.860915897633791e-06,
      "loss": 0.5933,
      "num_input_tokens_seen": 18378424,
      "step": 28100
    },
    {
      "epoch": 14.730083857442349,
      "grad_norm": 0.1769351065158844,
      "learning_rate": 9.85181811071193e-06,
      "loss": 0.5798,
      "num_input_tokens_seen": 18383544,
      "step": 28105
    },
    {
      "epoch": 14.732704402515724,
      "grad_norm": 0.17063632607460022,
      "learning_rate": 9.842723492616913e-06,
      "loss": 0.5348,
      "num_input_tokens_seen": 18386488,
      "step": 28110
    },
    {
      "epoch": 14.735324947589099,
      "grad_norm": 0.10250205546617508,
      "learning_rate": 9.833632045251207e-06,
      "loss": 0.351,
      "num_input_tokens_seen": 18389272,
      "step": 28115
    },
    {
      "epoch": 14.737945492662474,
      "grad_norm": 0.15967793762683868,
      "learning_rate": 9.824543770516656e-06,
      "loss": 0.401,
      "num_input_tokens_seen": 18393240,
      "step": 28120
    },
    {
      "epoch": 14.74056603773585,
      "grad_norm": 0.15705245733261108,
      "learning_rate": 9.815458670314412e-06,
      "loss": 0.4849,
      "num_input_tokens_seen": 18396504,
      "step": 28125
    },
    {
      "epoch": 14.743186582809225,
      "grad_norm": 0.26751643419265747,
      "learning_rate": 9.80637674654498e-06,
      "loss": 0.4221,
      "num_input_tokens_seen": 18399096,
      "step": 28130
    },
    {
      "epoch": 14.7458071278826,
      "grad_norm": 0.17927126586437225,
      "learning_rate": 9.797298001108193e-06,
      "loss": 0.4706,
      "num_input_tokens_seen": 18402936,
      "step": 28135
    },
    {
      "epoch": 14.748427672955975,
      "grad_norm": 0.223566472530365,
      "learning_rate": 9.788222435903232e-06,
      "loss": 0.5005,
      "num_input_tokens_seen": 18406008,
      "step": 28140
    },
    {
      "epoch": 14.75104821802935,
      "grad_norm": 0.11338064819574356,
      "learning_rate": 9.779150052828596e-06,
      "loss": 0.5265,
      "num_input_tokens_seen": 18409752,
      "step": 28145
    },
    {
      "epoch": 14.753668763102725,
      "grad_norm": 0.1786912977695465,
      "learning_rate": 9.770080853782108e-06,
      "loss": 0.4216,
      "num_input_tokens_seen": 18412408,
      "step": 28150
    },
    {
      "epoch": 14.7562893081761,
      "grad_norm": 0.23041939735412598,
      "learning_rate": 9.76101484066095e-06,
      "loss": 0.6504,
      "num_input_tokens_seen": 18415672,
      "step": 28155
    },
    {
      "epoch": 14.758909853249476,
      "grad_norm": 0.1508532464504242,
      "learning_rate": 9.751952015361637e-06,
      "loss": 0.6177,
      "num_input_tokens_seen": 18418904,
      "step": 28160
    },
    {
      "epoch": 14.76153039832285,
      "grad_norm": 0.13281893730163574,
      "learning_rate": 9.742892379780003e-06,
      "loss": 0.4443,
      "num_input_tokens_seen": 18421816,
      "step": 28165
    },
    {
      "epoch": 14.764150943396226,
      "grad_norm": 0.18735897541046143,
      "learning_rate": 9.733835935811237e-06,
      "loss": 0.6664,
      "num_input_tokens_seen": 18424696,
      "step": 28170
    },
    {
      "epoch": 14.766771488469601,
      "grad_norm": 0.22706185281276703,
      "learning_rate": 9.724782685349809e-06,
      "loss": 0.4661,
      "num_input_tokens_seen": 18427864,
      "step": 28175
    },
    {
      "epoch": 14.769392033542976,
      "grad_norm": 0.2531859576702118,
      "learning_rate": 9.715732630289576e-06,
      "loss": 0.6509,
      "num_input_tokens_seen": 18430872,
      "step": 28180
    },
    {
      "epoch": 14.772012578616351,
      "grad_norm": 0.20734000205993652,
      "learning_rate": 9.706685772523701e-06,
      "loss": 0.4305,
      "num_input_tokens_seen": 18433400,
      "step": 28185
    },
    {
      "epoch": 14.774633123689728,
      "grad_norm": 0.19268128275871277,
      "learning_rate": 9.697642113944692e-06,
      "loss": 0.5672,
      "num_input_tokens_seen": 18436536,
      "step": 28190
    },
    {
      "epoch": 14.777253668763104,
      "grad_norm": 0.11057222634553909,
      "learning_rate": 9.688601656444381e-06,
      "loss": 0.379,
      "num_input_tokens_seen": 18439896,
      "step": 28195
    },
    {
      "epoch": 14.779874213836479,
      "grad_norm": 0.13360217213630676,
      "learning_rate": 9.679564401913921e-06,
      "loss": 0.4316,
      "num_input_tokens_seen": 18445112,
      "step": 28200
    },
    {
      "epoch": 14.782494758909854,
      "grad_norm": 0.2639119327068329,
      "learning_rate": 9.670530352243798e-06,
      "loss": 0.4044,
      "num_input_tokens_seen": 18448056,
      "step": 28205
    },
    {
      "epoch": 14.785115303983229,
      "grad_norm": 0.5835424065589905,
      "learning_rate": 9.661499509323835e-06,
      "loss": 0.4544,
      "num_input_tokens_seen": 18451320,
      "step": 28210
    },
    {
      "epoch": 14.787735849056604,
      "grad_norm": 0.1477695107460022,
      "learning_rate": 9.652471875043182e-06,
      "loss": 0.3731,
      "num_input_tokens_seen": 18454168,
      "step": 28215
    },
    {
      "epoch": 14.79035639412998,
      "grad_norm": 0.1590428650379181,
      "learning_rate": 9.64344745129033e-06,
      "loss": 0.5114,
      "num_input_tokens_seen": 18457144,
      "step": 28220
    },
    {
      "epoch": 14.792976939203355,
      "grad_norm": 0.1362147480249405,
      "learning_rate": 9.634426239953073e-06,
      "loss": 0.459,
      "num_input_tokens_seen": 18460088,
      "step": 28225
    },
    {
      "epoch": 14.79559748427673,
      "grad_norm": 0.30266207456588745,
      "learning_rate": 9.62540824291854e-06,
      "loss": 0.5544,
      "num_input_tokens_seen": 18463800,
      "step": 28230
    },
    {
      "epoch": 14.798218029350105,
      "grad_norm": 0.1109979972243309,
      "learning_rate": 9.616393462073197e-06,
      "loss": 0.5715,
      "num_input_tokens_seen": 18466744,
      "step": 28235
    },
    {
      "epoch": 14.80083857442348,
      "grad_norm": 0.2817226052284241,
      "learning_rate": 9.607381899302839e-06,
      "loss": 0.6426,
      "num_input_tokens_seen": 18469848,
      "step": 28240
    },
    {
      "epoch": 14.803459119496855,
      "grad_norm": 0.15302647650241852,
      "learning_rate": 9.59837355649258e-06,
      "loss": 0.5757,
      "num_input_tokens_seen": 18473496,
      "step": 28245
    },
    {
      "epoch": 14.80607966457023,
      "grad_norm": 0.17343051731586456,
      "learning_rate": 9.589368435526868e-06,
      "loss": 0.4818,
      "num_input_tokens_seen": 18477048,
      "step": 28250
    },
    {
      "epoch": 14.808700209643606,
      "grad_norm": 0.23296093940734863,
      "learning_rate": 9.580366538289468e-06,
      "loss": 0.5188,
      "num_input_tokens_seen": 18479640,
      "step": 28255
    },
    {
      "epoch": 14.81132075471698,
      "grad_norm": 0.182090163230896,
      "learning_rate": 9.57136786666346e-06,
      "loss": 0.5169,
      "num_input_tokens_seen": 18482808,
      "step": 28260
    },
    {
      "epoch": 14.813941299790356,
      "grad_norm": 0.23802374303340912,
      "learning_rate": 9.562372422531276e-06,
      "loss": 0.5798,
      "num_input_tokens_seen": 18485496,
      "step": 28265
    },
    {
      "epoch": 14.816561844863731,
      "grad_norm": 0.12318160384893417,
      "learning_rate": 9.553380207774662e-06,
      "loss": 0.4307,
      "num_input_tokens_seen": 18488568,
      "step": 28270
    },
    {
      "epoch": 14.819182389937106,
      "grad_norm": 0.17048379778862,
      "learning_rate": 9.54439122427469e-06,
      "loss": 0.4857,
      "num_input_tokens_seen": 18491256,
      "step": 28275
    },
    {
      "epoch": 14.821802935010481,
      "grad_norm": 0.20861029624938965,
      "learning_rate": 9.535405473911748e-06,
      "loss": 0.4249,
      "num_input_tokens_seen": 18494424,
      "step": 28280
    },
    {
      "epoch": 14.824423480083858,
      "grad_norm": 0.15427306294441223,
      "learning_rate": 9.526422958565546e-06,
      "loss": 0.494,
      "num_input_tokens_seen": 18497496,
      "step": 28285
    },
    {
      "epoch": 14.827044025157234,
      "grad_norm": 0.15570080280303955,
      "learning_rate": 9.517443680115124e-06,
      "loss": 0.4506,
      "num_input_tokens_seen": 18501304,
      "step": 28290
    },
    {
      "epoch": 14.829664570230609,
      "grad_norm": 0.20487163960933685,
      "learning_rate": 9.50846764043885e-06,
      "loss": 0.4943,
      "num_input_tokens_seen": 18505048,
      "step": 28295
    },
    {
      "epoch": 14.832285115303984,
      "grad_norm": 0.17827309668064117,
      "learning_rate": 9.499494841414422e-06,
      "loss": 0.511,
      "num_input_tokens_seen": 18508216,
      "step": 28300
    },
    {
      "epoch": 14.834905660377359,
      "grad_norm": 0.11797379702329636,
      "learning_rate": 9.490525284918822e-06,
      "loss": 0.4758,
      "num_input_tokens_seen": 18511192,
      "step": 28305
    },
    {
      "epoch": 14.837526205450734,
      "grad_norm": 0.2694932520389557,
      "learning_rate": 9.4815589728284e-06,
      "loss": 0.4562,
      "num_input_tokens_seen": 18514296,
      "step": 28310
    },
    {
      "epoch": 14.84014675052411,
      "grad_norm": 0.11586476862430573,
      "learning_rate": 9.472595907018788e-06,
      "loss": 0.631,
      "num_input_tokens_seen": 18517560,
      "step": 28315
    },
    {
      "epoch": 14.842767295597485,
      "grad_norm": 0.22333304584026337,
      "learning_rate": 9.463636089364968e-06,
      "loss": 0.5196,
      "num_input_tokens_seen": 18520888,
      "step": 28320
    },
    {
      "epoch": 14.84538784067086,
      "grad_norm": 0.17402002215385437,
      "learning_rate": 9.454679521741239e-06,
      "loss": 0.5086,
      "num_input_tokens_seen": 18525016,
      "step": 28325
    },
    {
      "epoch": 14.848008385744235,
      "grad_norm": 0.12097810208797455,
      "learning_rate": 9.445726206021194e-06,
      "loss": 0.5515,
      "num_input_tokens_seen": 18529112,
      "step": 28330
    },
    {
      "epoch": 14.85062893081761,
      "grad_norm": 0.11643462628126144,
      "learning_rate": 9.436776144077772e-06,
      "loss": 0.4936,
      "num_input_tokens_seen": 18532344,
      "step": 28335
    },
    {
      "epoch": 14.853249475890985,
      "grad_norm": 0.1670067310333252,
      "learning_rate": 9.427829337783237e-06,
      "loss": 0.4567,
      "num_input_tokens_seen": 18535608,
      "step": 28340
    },
    {
      "epoch": 14.85587002096436,
      "grad_norm": 0.2240627557039261,
      "learning_rate": 9.418885789009138e-06,
      "loss": 0.4623,
      "num_input_tokens_seen": 18538200,
      "step": 28345
    },
    {
      "epoch": 14.858490566037736,
      "grad_norm": 0.1800660789012909,
      "learning_rate": 9.409945499626374e-06,
      "loss": 0.3619,
      "num_input_tokens_seen": 18540600,
      "step": 28350
    },
    {
      "epoch": 14.86111111111111,
      "grad_norm": 0.1655336171388626,
      "learning_rate": 9.40100847150516e-06,
      "loss": 0.4868,
      "num_input_tokens_seen": 18543192,
      "step": 28355
    },
    {
      "epoch": 14.863731656184486,
      "grad_norm": 0.2441740334033966,
      "learning_rate": 9.392074706515003e-06,
      "loss": 0.5088,
      "num_input_tokens_seen": 18546488,
      "step": 28360
    },
    {
      "epoch": 14.866352201257861,
      "grad_norm": 0.18587543070316315,
      "learning_rate": 9.383144206524753e-06,
      "loss": 0.5256,
      "num_input_tokens_seen": 18549752,
      "step": 28365
    },
    {
      "epoch": 14.868972746331236,
      "grad_norm": 0.199893057346344,
      "learning_rate": 9.374216973402575e-06,
      "loss": 0.4829,
      "num_input_tokens_seen": 18552696,
      "step": 28370
    },
    {
      "epoch": 14.871593291404611,
      "grad_norm": 0.3007734417915344,
      "learning_rate": 9.365293009015926e-06,
      "loss": 0.5547,
      "num_input_tokens_seen": 18555416,
      "step": 28375
    },
    {
      "epoch": 14.874213836477988,
      "grad_norm": 0.1714286059141159,
      "learning_rate": 9.356372315231623e-06,
      "loss": 0.6923,
      "num_input_tokens_seen": 18558712,
      "step": 28380
    },
    {
      "epoch": 14.876834381551364,
      "grad_norm": 0.18517479300498962,
      "learning_rate": 9.347454893915749e-06,
      "loss": 0.4574,
      "num_input_tokens_seen": 18562072,
      "step": 28385
    },
    {
      "epoch": 14.879454926624739,
      "grad_norm": 0.18615245819091797,
      "learning_rate": 9.338540746933741e-06,
      "loss": 0.5602,
      "num_input_tokens_seen": 18565208,
      "step": 28390
    },
    {
      "epoch": 14.882075471698114,
      "grad_norm": 0.19590039551258087,
      "learning_rate": 9.329629876150339e-06,
      "loss": 0.5242,
      "num_input_tokens_seen": 18568568,
      "step": 28395
    },
    {
      "epoch": 14.884696016771489,
      "grad_norm": 0.19856718182563782,
      "learning_rate": 9.320722283429581e-06,
      "loss": 0.4284,
      "num_input_tokens_seen": 18571384,
      "step": 28400
    },
    {
      "epoch": 14.887316561844864,
      "grad_norm": 0.1684098243713379,
      "learning_rate": 9.311817970634854e-06,
      "loss": 0.3441,
      "num_input_tokens_seen": 18574680,
      "step": 28405
    },
    {
      "epoch": 14.88993710691824,
      "grad_norm": 0.18627502024173737,
      "learning_rate": 9.302916939628814e-06,
      "loss": 0.4019,
      "num_input_tokens_seen": 18577464,
      "step": 28410
    },
    {
      "epoch": 14.892557651991615,
      "grad_norm": 0.12763363122940063,
      "learning_rate": 9.29401919227347e-06,
      "loss": 0.4016,
      "num_input_tokens_seen": 18580024,
      "step": 28415
    },
    {
      "epoch": 14.89517819706499,
      "grad_norm": 0.24219200015068054,
      "learning_rate": 9.28512473043013e-06,
      "loss": 0.4633,
      "num_input_tokens_seen": 18582840,
      "step": 28420
    },
    {
      "epoch": 14.897798742138365,
      "grad_norm": 0.1857113391160965,
      "learning_rate": 9.276233555959419e-06,
      "loss": 0.6857,
      "num_input_tokens_seen": 18585880,
      "step": 28425
    },
    {
      "epoch": 14.90041928721174,
      "grad_norm": 0.16549068689346313,
      "learning_rate": 9.267345670721254e-06,
      "loss": 0.4821,
      "num_input_tokens_seen": 18588920,
      "step": 28430
    },
    {
      "epoch": 14.903039832285115,
      "grad_norm": 0.25929898023605347,
      "learning_rate": 9.258461076574895e-06,
      "loss": 0.434,
      "num_input_tokens_seen": 18591736,
      "step": 28435
    },
    {
      "epoch": 14.90566037735849,
      "grad_norm": 0.28678542375564575,
      "learning_rate": 9.24957977537888e-06,
      "loss": 0.5732,
      "num_input_tokens_seen": 18594392,
      "step": 28440
    },
    {
      "epoch": 14.908280922431866,
      "grad_norm": 0.09833338856697083,
      "learning_rate": 9.240701768991086e-06,
      "loss": 0.5358,
      "num_input_tokens_seen": 18598424,
      "step": 28445
    },
    {
      "epoch": 14.91090146750524,
      "grad_norm": 0.15536130964756012,
      "learning_rate": 9.231827059268692e-06,
      "loss": 0.4421,
      "num_input_tokens_seen": 18601976,
      "step": 28450
    },
    {
      "epoch": 14.913522012578616,
      "grad_norm": 0.4413043260574341,
      "learning_rate": 9.222955648068193e-06,
      "loss": 0.6736,
      "num_input_tokens_seen": 18604568,
      "step": 28455
    },
    {
      "epoch": 14.916142557651991,
      "grad_norm": 0.1674363613128662,
      "learning_rate": 9.214087537245378e-06,
      "loss": 0.4183,
      "num_input_tokens_seen": 18607544,
      "step": 28460
    },
    {
      "epoch": 14.918763102725366,
      "grad_norm": 0.1125970110297203,
      "learning_rate": 9.205222728655344e-06,
      "loss": 0.4259,
      "num_input_tokens_seen": 18611352,
      "step": 28465
    },
    {
      "epoch": 14.921383647798741,
      "grad_norm": 0.047651369124650955,
      "learning_rate": 9.196361224152522e-06,
      "loss": 0.3652,
      "num_input_tokens_seen": 18616920,
      "step": 28470
    },
    {
      "epoch": 14.924004192872118,
      "grad_norm": 0.24209682643413544,
      "learning_rate": 9.187503025590632e-06,
      "loss": 0.7387,
      "num_input_tokens_seen": 18619544,
      "step": 28475
    },
    {
      "epoch": 14.926624737945493,
      "grad_norm": 0.12084509432315826,
      "learning_rate": 9.17864813482272e-06,
      "loss": 0.3841,
      "num_input_tokens_seen": 18622072,
      "step": 28480
    },
    {
      "epoch": 14.929245283018869,
      "grad_norm": 0.1557905524969101,
      "learning_rate": 9.169796553701122e-06,
      "loss": 0.6416,
      "num_input_tokens_seen": 18624952,
      "step": 28485
    },
    {
      "epoch": 14.931865828092244,
      "grad_norm": 0.18971659243106842,
      "learning_rate": 9.160948284077473e-06,
      "loss": 0.6688,
      "num_input_tokens_seen": 18627672,
      "step": 28490
    },
    {
      "epoch": 14.934486373165619,
      "grad_norm": 0.1260416954755783,
      "learning_rate": 9.152103327802738e-06,
      "loss": 0.4473,
      "num_input_tokens_seen": 18631160,
      "step": 28495
    },
    {
      "epoch": 14.937106918238994,
      "grad_norm": 0.1882263720035553,
      "learning_rate": 9.143261686727189e-06,
      "loss": 0.46,
      "num_input_tokens_seen": 18634168,
      "step": 28500
    },
    {
      "epoch": 14.93972746331237,
      "grad_norm": 0.21439921855926514,
      "learning_rate": 9.134423362700392e-06,
      "loss": 0.4202,
      "num_input_tokens_seen": 18636984,
      "step": 28505
    },
    {
      "epoch": 14.942348008385745,
      "grad_norm": 0.16357868909835815,
      "learning_rate": 9.12558835757123e-06,
      "loss": 0.5241,
      "num_input_tokens_seen": 18639864,
      "step": 28510
    },
    {
      "epoch": 14.94496855345912,
      "grad_norm": 0.23166321218013763,
      "learning_rate": 9.116756673187878e-06,
      "loss": 0.4771,
      "num_input_tokens_seen": 18642264,
      "step": 28515
    },
    {
      "epoch": 14.947589098532495,
      "grad_norm": 0.09803608059883118,
      "learning_rate": 9.107928311397817e-06,
      "loss": 0.5404,
      "num_input_tokens_seen": 18646072,
      "step": 28520
    },
    {
      "epoch": 14.95020964360587,
      "grad_norm": 0.18964660167694092,
      "learning_rate": 9.09910327404785e-06,
      "loss": 0.4251,
      "num_input_tokens_seen": 18648696,
      "step": 28525
    },
    {
      "epoch": 14.952830188679245,
      "grad_norm": 0.11627431213855743,
      "learning_rate": 9.090281562984068e-06,
      "loss": 0.4386,
      "num_input_tokens_seen": 18651608,
      "step": 28530
    },
    {
      "epoch": 14.95545073375262,
      "grad_norm": 0.22838963568210602,
      "learning_rate": 9.081463180051883e-06,
      "loss": 0.4863,
      "num_input_tokens_seen": 18655064,
      "step": 28535
    },
    {
      "epoch": 14.958071278825996,
      "grad_norm": 0.18318058550357819,
      "learning_rate": 9.072648127095996e-06,
      "loss": 0.4098,
      "num_input_tokens_seen": 18658040,
      "step": 28540
    },
    {
      "epoch": 14.96069182389937,
      "grad_norm": 0.23717086017131805,
      "learning_rate": 9.0638364059604e-06,
      "loss": 0.444,
      "num_input_tokens_seen": 18660472,
      "step": 28545
    },
    {
      "epoch": 14.963312368972746,
      "grad_norm": 0.15203838050365448,
      "learning_rate": 9.055028018488417e-06,
      "loss": 0.4941,
      "num_input_tokens_seen": 18664216,
      "step": 28550
    },
    {
      "epoch": 14.965932914046121,
      "grad_norm": 0.10641761869192123,
      "learning_rate": 9.046222966522665e-06,
      "loss": 0.3704,
      "num_input_tokens_seen": 18667224,
      "step": 28555
    },
    {
      "epoch": 14.968553459119496,
      "grad_norm": 0.19242657721042633,
      "learning_rate": 9.037421251905059e-06,
      "loss": 0.4741,
      "num_input_tokens_seen": 18670008,
      "step": 28560
    },
    {
      "epoch": 14.971174004192871,
      "grad_norm": 0.17879894375801086,
      "learning_rate": 9.028622876476833e-06,
      "loss": 0.4231,
      "num_input_tokens_seen": 18672888,
      "step": 28565
    },
    {
      "epoch": 14.973794549266248,
      "grad_norm": 0.14508669078350067,
      "learning_rate": 9.019827842078474e-06,
      "loss": 0.4211,
      "num_input_tokens_seen": 18676440,
      "step": 28570
    },
    {
      "epoch": 14.976415094339622,
      "grad_norm": 0.1863776445388794,
      "learning_rate": 9.011036150549817e-06,
      "loss": 0.5299,
      "num_input_tokens_seen": 18679480,
      "step": 28575
    },
    {
      "epoch": 14.979035639412999,
      "grad_norm": 0.26533597707748413,
      "learning_rate": 9.002247803729987e-06,
      "loss": 0.4917,
      "num_input_tokens_seen": 18683096,
      "step": 28580
    },
    {
      "epoch": 14.981656184486374,
      "grad_norm": 0.5159878134727478,
      "learning_rate": 8.993462803457404e-06,
      "loss": 0.622,
      "num_input_tokens_seen": 18685304,
      "step": 28585
    },
    {
      "epoch": 14.984276729559749,
      "grad_norm": 0.17050357162952423,
      "learning_rate": 8.984681151569799e-06,
      "loss": 0.5655,
      "num_input_tokens_seen": 18687640,
      "step": 28590
    },
    {
      "epoch": 14.986897274633124,
      "grad_norm": 0.21476303040981293,
      "learning_rate": 8.975902849904183e-06,
      "loss": 0.3936,
      "num_input_tokens_seen": 18691832,
      "step": 28595
    },
    {
      "epoch": 14.9895178197065,
      "grad_norm": 0.18072421848773956,
      "learning_rate": 8.967127900296874e-06,
      "loss": 0.6733,
      "num_input_tokens_seen": 18694744,
      "step": 28600
    },
    {
      "epoch": 14.992138364779874,
      "grad_norm": 0.12504597008228302,
      "learning_rate": 8.95835630458349e-06,
      "loss": 0.5825,
      "num_input_tokens_seen": 18699096,
      "step": 28605
    },
    {
      "epoch": 14.99475890985325,
      "grad_norm": 0.20905639231204987,
      "learning_rate": 8.949588064598958e-06,
      "loss": 0.6017,
      "num_input_tokens_seen": 18702168,
      "step": 28610
    },
    {
      "epoch": 14.997379454926625,
      "grad_norm": 0.2088862806558609,
      "learning_rate": 8.940823182177499e-06,
      "loss": 0.5052,
      "num_input_tokens_seen": 18704920,
      "step": 28615
    },
    {
      "epoch": 15.0,
      "grad_norm": 0.18302971124649048,
      "learning_rate": 8.932061659152607e-06,
      "loss": 0.4217,
      "num_input_tokens_seen": 18706976,
      "step": 28620
    },
    {
      "epoch": 15.0,
      "eval_loss": 0.4922962486743927,
      "eval_runtime": 10.279,
      "eval_samples_per_second": 82.498,
      "eval_steps_per_second": 20.625,
      "num_input_tokens_seen": 18706976,
      "step": 28620
    },
    {
      "epoch": 15.002620545073375,
      "grad_norm": 0.16045929491519928,
      "learning_rate": 8.92330349735711e-06,
      "loss": 0.4396,
      "num_input_tokens_seen": 18710720,
      "step": 28625
    },
    {
      "epoch": 15.00524109014675,
      "grad_norm": 0.17698413133621216,
      "learning_rate": 8.914548698623104e-06,
      "loss": 0.5219,
      "num_input_tokens_seen": 18714528,
      "step": 28630
    },
    {
      "epoch": 15.007861635220126,
      "grad_norm": 0.22696074843406677,
      "learning_rate": 8.905797264781995e-06,
      "loss": 0.4504,
      "num_input_tokens_seen": 18717120,
      "step": 28635
    },
    {
      "epoch": 15.0104821802935,
      "grad_norm": 0.13429318368434906,
      "learning_rate": 8.89704919766449e-06,
      "loss": 0.484,
      "num_input_tokens_seen": 18719776,
      "step": 28640
    },
    {
      "epoch": 15.013102725366876,
      "grad_norm": 0.13663432002067566,
      "learning_rate": 8.888304499100586e-06,
      "loss": 0.4306,
      "num_input_tokens_seen": 18722976,
      "step": 28645
    },
    {
      "epoch": 15.015723270440251,
      "grad_norm": 0.19842198491096497,
      "learning_rate": 8.879563170919566e-06,
      "loss": 0.5582,
      "num_input_tokens_seen": 18726432,
      "step": 28650
    },
    {
      "epoch": 15.018343815513626,
      "grad_norm": 0.17330318689346313,
      "learning_rate": 8.870825214950016e-06,
      "loss": 0.4461,
      "num_input_tokens_seen": 18729440,
      "step": 28655
    },
    {
      "epoch": 15.020964360587001,
      "grad_norm": 0.179080069065094,
      "learning_rate": 8.862090633019818e-06,
      "loss": 0.4263,
      "num_input_tokens_seen": 18732256,
      "step": 28660
    },
    {
      "epoch": 15.023584905660377,
      "grad_norm": 0.207890123128891,
      "learning_rate": 8.853359426956146e-06,
      "loss": 0.4993,
      "num_input_tokens_seen": 18736480,
      "step": 28665
    },
    {
      "epoch": 15.026205450733753,
      "grad_norm": 0.5291009545326233,
      "learning_rate": 8.844631598585482e-06,
      "loss": 0.4484,
      "num_input_tokens_seen": 18739200,
      "step": 28670
    },
    {
      "epoch": 15.028825995807129,
      "grad_norm": 0.2065230756998062,
      "learning_rate": 8.835907149733569e-06,
      "loss": 0.3501,
      "num_input_tokens_seen": 18741536,
      "step": 28675
    },
    {
      "epoch": 15.031446540880504,
      "grad_norm": 0.13828855752944946,
      "learning_rate": 8.827186082225477e-06,
      "loss": 0.6292,
      "num_input_tokens_seen": 18744672,
      "step": 28680
    },
    {
      "epoch": 15.034067085953879,
      "grad_norm": 0.2846357822418213,
      "learning_rate": 8.81846839788554e-06,
      "loss": 0.6201,
      "num_input_tokens_seen": 18747904,
      "step": 28685
    },
    {
      "epoch": 15.036687631027254,
      "grad_norm": 0.10659713298082352,
      "learning_rate": 8.809754098537407e-06,
      "loss": 0.4899,
      "num_input_tokens_seen": 18752000,
      "step": 28690
    },
    {
      "epoch": 15.03930817610063,
      "grad_norm": 0.16450878977775574,
      "learning_rate": 8.801043186004016e-06,
      "loss": 0.4731,
      "num_input_tokens_seen": 18755616,
      "step": 28695
    },
    {
      "epoch": 15.041928721174004,
      "grad_norm": 0.2685394287109375,
      "learning_rate": 8.792335662107575e-06,
      "loss": 0.5627,
      "num_input_tokens_seen": 18758240,
      "step": 28700
    },
    {
      "epoch": 15.04454926624738,
      "grad_norm": 0.11705180257558823,
      "learning_rate": 8.783631528669605e-06,
      "loss": 0.3186,
      "num_input_tokens_seen": 18762368,
      "step": 28705
    },
    {
      "epoch": 15.047169811320755,
      "grad_norm": 0.18871568143367767,
      "learning_rate": 8.774930787510924e-06,
      "loss": 0.4696,
      "num_input_tokens_seen": 18765152,
      "step": 28710
    },
    {
      "epoch": 15.04979035639413,
      "grad_norm": 0.13284432888031006,
      "learning_rate": 8.766233440451608e-06,
      "loss": 0.6872,
      "num_input_tokens_seen": 18768928,
      "step": 28715
    },
    {
      "epoch": 15.052410901467505,
      "grad_norm": 0.20735831558704376,
      "learning_rate": 8.75753948931105e-06,
      "loss": 0.3689,
      "num_input_tokens_seen": 18771488,
      "step": 28720
    },
    {
      "epoch": 15.05503144654088,
      "grad_norm": 0.2009425163269043,
      "learning_rate": 8.748848935907936e-06,
      "loss": 0.557,
      "num_input_tokens_seen": 18774272,
      "step": 28725
    },
    {
      "epoch": 15.057651991614255,
      "grad_norm": 0.2893683612346649,
      "learning_rate": 8.740161782060214e-06,
      "loss": 0.5817,
      "num_input_tokens_seen": 18776704,
      "step": 28730
    },
    {
      "epoch": 15.06027253668763,
      "grad_norm": 0.2683747410774231,
      "learning_rate": 8.731478029585144e-06,
      "loss": 0.5285,
      "num_input_tokens_seen": 18779520,
      "step": 28735
    },
    {
      "epoch": 15.062893081761006,
      "grad_norm": 0.14196422696113586,
      "learning_rate": 8.722797680299278e-06,
      "loss": 0.4845,
      "num_input_tokens_seen": 18782336,
      "step": 28740
    },
    {
      "epoch": 15.065513626834381,
      "grad_norm": 0.12432560324668884,
      "learning_rate": 8.71412073601843e-06,
      "loss": 0.5035,
      "num_input_tokens_seen": 18785504,
      "step": 28745
    },
    {
      "epoch": 15.068134171907756,
      "grad_norm": 0.13489753007888794,
      "learning_rate": 8.705447198557731e-06,
      "loss": 0.3543,
      "num_input_tokens_seen": 18788480,
      "step": 28750
    },
    {
      "epoch": 15.070754716981131,
      "grad_norm": 0.19639770686626434,
      "learning_rate": 8.696777069731574e-06,
      "loss": 0.5441,
      "num_input_tokens_seen": 18794144,
      "step": 28755
    },
    {
      "epoch": 15.073375262054507,
      "grad_norm": 0.14907442033290863,
      "learning_rate": 8.688110351353654e-06,
      "loss": 0.4675,
      "num_input_tokens_seen": 18797376,
      "step": 28760
    },
    {
      "epoch": 15.075995807127883,
      "grad_norm": 0.13231134414672852,
      "learning_rate": 8.679447045236962e-06,
      "loss": 0.5341,
      "num_input_tokens_seen": 18800672,
      "step": 28765
    },
    {
      "epoch": 15.078616352201259,
      "grad_norm": 0.1184251457452774,
      "learning_rate": 8.670787153193746e-06,
      "loss": 0.4557,
      "num_input_tokens_seen": 18803648,
      "step": 28770
    },
    {
      "epoch": 15.081236897274634,
      "grad_norm": 0.2424207627773285,
      "learning_rate": 8.662130677035574e-06,
      "loss": 0.4649,
      "num_input_tokens_seen": 18806144,
      "step": 28775
    },
    {
      "epoch": 15.083857442348009,
      "grad_norm": 0.13718655705451965,
      "learning_rate": 8.65347761857326e-06,
      "loss": 0.4652,
      "num_input_tokens_seen": 18809152,
      "step": 28780
    },
    {
      "epoch": 15.086477987421384,
      "grad_norm": 0.27022549510002136,
      "learning_rate": 8.64482797961694e-06,
      "loss": 0.4768,
      "num_input_tokens_seen": 18812192,
      "step": 28785
    },
    {
      "epoch": 15.08909853249476,
      "grad_norm": 0.19044427573680878,
      "learning_rate": 8.636181761976016e-06,
      "loss": 0.5191,
      "num_input_tokens_seen": 18815168,
      "step": 28790
    },
    {
      "epoch": 15.091719077568134,
      "grad_norm": 0.19873158633708954,
      "learning_rate": 8.62753896745919e-06,
      "loss": 0.412,
      "num_input_tokens_seen": 18819072,
      "step": 28795
    },
    {
      "epoch": 15.09433962264151,
      "grad_norm": 0.23238436877727509,
      "learning_rate": 8.618899597874413e-06,
      "loss": 0.6617,
      "num_input_tokens_seen": 18821728,
      "step": 28800
    },
    {
      "epoch": 15.096960167714885,
      "grad_norm": 0.17162573337554932,
      "learning_rate": 8.610263655028964e-06,
      "loss": 0.5403,
      "num_input_tokens_seen": 18824544,
      "step": 28805
    },
    {
      "epoch": 15.09958071278826,
      "grad_norm": 0.1356554925441742,
      "learning_rate": 8.601631140729366e-06,
      "loss": 0.359,
      "num_input_tokens_seen": 18828032,
      "step": 28810
    },
    {
      "epoch": 15.102201257861635,
      "grad_norm": 0.1797749400138855,
      "learning_rate": 8.593002056781451e-06,
      "loss": 0.4892,
      "num_input_tokens_seen": 18830688,
      "step": 28815
    },
    {
      "epoch": 15.10482180293501,
      "grad_norm": 0.15845397114753723,
      "learning_rate": 8.584376404990326e-06,
      "loss": 0.3482,
      "num_input_tokens_seen": 18833664,
      "step": 28820
    },
    {
      "epoch": 15.107442348008385,
      "grad_norm": 0.08474709838628769,
      "learning_rate": 8.575754187160384e-06,
      "loss": 0.3796,
      "num_input_tokens_seen": 18837536,
      "step": 28825
    },
    {
      "epoch": 15.11006289308176,
      "grad_norm": 0.163369819521904,
      "learning_rate": 8.567135405095291e-06,
      "loss": 0.4459,
      "num_input_tokens_seen": 18840000,
      "step": 28830
    },
    {
      "epoch": 15.112683438155136,
      "grad_norm": 0.10543754696846008,
      "learning_rate": 8.558520060597985e-06,
      "loss": 0.5117,
      "num_input_tokens_seen": 18844224,
      "step": 28835
    },
    {
      "epoch": 15.115303983228511,
      "grad_norm": 0.10762208700180054,
      "learning_rate": 8.54990815547071e-06,
      "loss": 0.3844,
      "num_input_tokens_seen": 18847104,
      "step": 28840
    },
    {
      "epoch": 15.117924528301886,
      "grad_norm": 0.2525423467159271,
      "learning_rate": 8.541299691514974e-06,
      "loss": 0.6292,
      "num_input_tokens_seen": 18849536,
      "step": 28845
    },
    {
      "epoch": 15.120545073375261,
      "grad_norm": 0.197297140955925,
      "learning_rate": 8.53269467053158e-06,
      "loss": 0.445,
      "num_input_tokens_seen": 18854528,
      "step": 28850
    },
    {
      "epoch": 15.123165618448636,
      "grad_norm": 0.2958197295665741,
      "learning_rate": 8.524093094320593e-06,
      "loss": 0.4803,
      "num_input_tokens_seen": 18857344,
      "step": 28855
    },
    {
      "epoch": 15.125786163522013,
      "grad_norm": 0.332190603017807,
      "learning_rate": 8.515494964681357e-06,
      "loss": 0.5119,
      "num_input_tokens_seen": 18859872,
      "step": 28860
    },
    {
      "epoch": 15.128406708595389,
      "grad_norm": 0.2965421676635742,
      "learning_rate": 8.506900283412506e-06,
      "loss": 0.3454,
      "num_input_tokens_seen": 18862240,
      "step": 28865
    },
    {
      "epoch": 15.131027253668764,
      "grad_norm": 0.15008646249771118,
      "learning_rate": 8.498309052311953e-06,
      "loss": 0.3943,
      "num_input_tokens_seen": 18865472,
      "step": 28870
    },
    {
      "epoch": 15.133647798742139,
      "grad_norm": 0.13466346263885498,
      "learning_rate": 8.489721273176887e-06,
      "loss": 0.5541,
      "num_input_tokens_seen": 18868960,
      "step": 28875
    },
    {
      "epoch": 15.136268343815514,
      "grad_norm": 0.2236967831850052,
      "learning_rate": 8.481136947803777e-06,
      "loss": 0.4822,
      "num_input_tokens_seen": 18871616,
      "step": 28880
    },
    {
      "epoch": 15.13888888888889,
      "grad_norm": 0.14217768609523773,
      "learning_rate": 8.472556077988362e-06,
      "loss": 0.5644,
      "num_input_tokens_seen": 18875040,
      "step": 28885
    },
    {
      "epoch": 15.141509433962264,
      "grad_norm": 0.1635511815547943,
      "learning_rate": 8.46397866552565e-06,
      "loss": 0.5342,
      "num_input_tokens_seen": 18877632,
      "step": 28890
    },
    {
      "epoch": 15.14412997903564,
      "grad_norm": 0.17534291744232178,
      "learning_rate": 8.455404712209952e-06,
      "loss": 0.5157,
      "num_input_tokens_seen": 18881056,
      "step": 28895
    },
    {
      "epoch": 15.146750524109015,
      "grad_norm": 0.5688650012016296,
      "learning_rate": 8.446834219834836e-06,
      "loss": 0.5517,
      "num_input_tokens_seen": 18883616,
      "step": 28900
    },
    {
      "epoch": 15.14937106918239,
      "grad_norm": 0.1531732976436615,
      "learning_rate": 8.43826719019316e-06,
      "loss": 0.4396,
      "num_input_tokens_seen": 18886592,
      "step": 28905
    },
    {
      "epoch": 15.151991614255765,
      "grad_norm": 0.12457000464200974,
      "learning_rate": 8.429703625077043e-06,
      "loss": 0.4464,
      "num_input_tokens_seen": 18891104,
      "step": 28910
    },
    {
      "epoch": 15.15461215932914,
      "grad_norm": 0.16002322733402252,
      "learning_rate": 8.421143526277875e-06,
      "loss": 0.3721,
      "num_input_tokens_seen": 18894528,
      "step": 28915
    },
    {
      "epoch": 15.157232704402515,
      "grad_norm": 0.14328034222126007,
      "learning_rate": 8.41258689558634e-06,
      "loss": 0.5385,
      "num_input_tokens_seen": 18897760,
      "step": 28920
    },
    {
      "epoch": 15.15985324947589,
      "grad_norm": 0.11869090050458908,
      "learning_rate": 8.404033734792386e-06,
      "loss": 0.4521,
      "num_input_tokens_seen": 18900800,
      "step": 28925
    },
    {
      "epoch": 15.162473794549266,
      "grad_norm": 0.16682866215705872,
      "learning_rate": 8.395484045685248e-06,
      "loss": 0.4601,
      "num_input_tokens_seen": 18903680,
      "step": 28930
    },
    {
      "epoch": 15.165094339622641,
      "grad_norm": 0.14559869468212128,
      "learning_rate": 8.386937830053412e-06,
      "loss": 0.4804,
      "num_input_tokens_seen": 18906336,
      "step": 28935
    },
    {
      "epoch": 15.167714884696016,
      "grad_norm": 0.18655453622341156,
      "learning_rate": 8.37839508968464e-06,
      "loss": 0.496,
      "num_input_tokens_seen": 18910112,
      "step": 28940
    },
    {
      "epoch": 15.170335429769391,
      "grad_norm": 0.23603296279907227,
      "learning_rate": 8.369855826365988e-06,
      "loss": 0.4974,
      "num_input_tokens_seen": 18912928,
      "step": 28945
    },
    {
      "epoch": 15.172955974842766,
      "grad_norm": 0.14298391342163086,
      "learning_rate": 8.361320041883772e-06,
      "loss": 0.5074,
      "num_input_tokens_seen": 18916416,
      "step": 28950
    },
    {
      "epoch": 15.175576519916143,
      "grad_norm": 0.1931624859571457,
      "learning_rate": 8.352787738023576e-06,
      "loss": 0.6064,
      "num_input_tokens_seen": 18919968,
      "step": 28955
    },
    {
      "epoch": 15.178197064989519,
      "grad_norm": 0.1415848582983017,
      "learning_rate": 8.344258916570274e-06,
      "loss": 0.3912,
      "num_input_tokens_seen": 18923392,
      "step": 28960
    },
    {
      "epoch": 15.180817610062894,
      "grad_norm": 0.11232009530067444,
      "learning_rate": 8.335733579307988e-06,
      "loss": 0.462,
      "num_input_tokens_seen": 18926528,
      "step": 28965
    },
    {
      "epoch": 15.183438155136269,
      "grad_norm": 0.12736187875270844,
      "learning_rate": 8.327211728020113e-06,
      "loss": 0.3575,
      "num_input_tokens_seen": 18929696,
      "step": 28970
    },
    {
      "epoch": 15.186058700209644,
      "grad_norm": 0.14210449159145355,
      "learning_rate": 8.31869336448933e-06,
      "loss": 0.4528,
      "num_input_tokens_seen": 18932384,
      "step": 28975
    },
    {
      "epoch": 15.18867924528302,
      "grad_norm": 0.13114626705646515,
      "learning_rate": 8.310178490497586e-06,
      "loss": 0.4867,
      "num_input_tokens_seen": 18934880,
      "step": 28980
    },
    {
      "epoch": 15.191299790356394,
      "grad_norm": 0.24591174721717834,
      "learning_rate": 8.301667107826103e-06,
      "loss": 0.3869,
      "num_input_tokens_seen": 18938816,
      "step": 28985
    },
    {
      "epoch": 15.19392033542977,
      "grad_norm": 0.2737949788570404,
      "learning_rate": 8.293159218255345e-06,
      "loss": 0.4563,
      "num_input_tokens_seen": 18941824,
      "step": 28990
    },
    {
      "epoch": 15.196540880503145,
      "grad_norm": 0.2036719173192978,
      "learning_rate": 8.284654823565088e-06,
      "loss": 0.4719,
      "num_input_tokens_seen": 18945056,
      "step": 28995
    },
    {
      "epoch": 15.19916142557652,
      "grad_norm": 0.18462905287742615,
      "learning_rate": 8.276153925534333e-06,
      "loss": 0.5365,
      "num_input_tokens_seen": 18948256,
      "step": 29000
    },
    {
      "epoch": 15.201781970649895,
      "grad_norm": 0.2671254873275757,
      "learning_rate": 8.267656525941383e-06,
      "loss": 0.549,
      "num_input_tokens_seen": 18951008,
      "step": 29005
    },
    {
      "epoch": 15.20440251572327,
      "grad_norm": 0.14594462513923645,
      "learning_rate": 8.259162626563801e-06,
      "loss": 0.4443,
      "num_input_tokens_seen": 18954464,
      "step": 29010
    },
    {
      "epoch": 15.207023060796645,
      "grad_norm": 0.099685437977314,
      "learning_rate": 8.250672229178402e-06,
      "loss": 0.4492,
      "num_input_tokens_seen": 18958464,
      "step": 29015
    },
    {
      "epoch": 15.20964360587002,
      "grad_norm": 0.14016181230545044,
      "learning_rate": 8.242185335561295e-06,
      "loss": 0.4709,
      "num_input_tokens_seen": 18961152,
      "step": 29020
    },
    {
      "epoch": 15.212264150943396,
      "grad_norm": 0.13166877627372742,
      "learning_rate": 8.23370194748782e-06,
      "loss": 0.3402,
      "num_input_tokens_seen": 18964640,
      "step": 29025
    },
    {
      "epoch": 15.214884696016771,
      "grad_norm": 0.11364465206861496,
      "learning_rate": 8.22522206673262e-06,
      "loss": 0.4482,
      "num_input_tokens_seen": 18967616,
      "step": 29030
    },
    {
      "epoch": 15.217505241090146,
      "grad_norm": 0.15828636288642883,
      "learning_rate": 8.216745695069589e-06,
      "loss": 0.5188,
      "num_input_tokens_seen": 18971136,
      "step": 29035
    },
    {
      "epoch": 15.220125786163521,
      "grad_norm": 0.14759375154972076,
      "learning_rate": 8.208272834271894e-06,
      "loss": 0.6249,
      "num_input_tokens_seen": 18974208,
      "step": 29040
    },
    {
      "epoch": 15.222746331236896,
      "grad_norm": 0.1822466254234314,
      "learning_rate": 8.19980348611194e-06,
      "loss": 0.4816,
      "num_input_tokens_seen": 18977408,
      "step": 29045
    },
    {
      "epoch": 15.225366876310272,
      "grad_norm": 0.23922096192836761,
      "learning_rate": 8.191337652361439e-06,
      "loss": 0.5275,
      "num_input_tokens_seen": 18980288,
      "step": 29050
    },
    {
      "epoch": 15.227987421383649,
      "grad_norm": 0.1541508138179779,
      "learning_rate": 8.182875334791332e-06,
      "loss": 0.3793,
      "num_input_tokens_seen": 18984096,
      "step": 29055
    },
    {
      "epoch": 15.230607966457024,
      "grad_norm": 0.35380998253822327,
      "learning_rate": 8.174416535171841e-06,
      "loss": 0.4922,
      "num_input_tokens_seen": 18987648,
      "step": 29060
    },
    {
      "epoch": 15.233228511530399,
      "grad_norm": 0.09791067987680435,
      "learning_rate": 8.165961255272467e-06,
      "loss": 0.43,
      "num_input_tokens_seen": 18990208,
      "step": 29065
    },
    {
      "epoch": 15.235849056603774,
      "grad_norm": 0.2505693733692169,
      "learning_rate": 8.157509496861935e-06,
      "loss": 0.4572,
      "num_input_tokens_seen": 18993440,
      "step": 29070
    },
    {
      "epoch": 15.23846960167715,
      "grad_norm": 0.13445858657360077,
      "learning_rate": 8.149061261708266e-06,
      "loss": 0.4514,
      "num_input_tokens_seen": 18997024,
      "step": 29075
    },
    {
      "epoch": 15.241090146750524,
      "grad_norm": 0.34511271119117737,
      "learning_rate": 8.140616551578745e-06,
      "loss": 0.4863,
      "num_input_tokens_seen": 19000864,
      "step": 29080
    },
    {
      "epoch": 15.2437106918239,
      "grad_norm": 0.1472836434841156,
      "learning_rate": 8.13217536823989e-06,
      "loss": 0.5057,
      "num_input_tokens_seen": 19004576,
      "step": 29085
    },
    {
      "epoch": 15.246331236897275,
      "grad_norm": 0.2474534958600998,
      "learning_rate": 8.12373771345752e-06,
      "loss": 0.4449,
      "num_input_tokens_seen": 19008128,
      "step": 29090
    },
    {
      "epoch": 15.24895178197065,
      "grad_norm": 0.16220712661743164,
      "learning_rate": 8.115303588996676e-06,
      "loss": 0.5469,
      "num_input_tokens_seen": 19013952,
      "step": 29095
    },
    {
      "epoch": 15.251572327044025,
      "grad_norm": 0.16140545904636383,
      "learning_rate": 8.10687299662169e-06,
      "loss": 0.5614,
      "num_input_tokens_seen": 19017216,
      "step": 29100
    },
    {
      "epoch": 15.2541928721174,
      "grad_norm": 0.13902918994426727,
      "learning_rate": 8.098445938096147e-06,
      "loss": 0.4096,
      "num_input_tokens_seen": 19020096,
      "step": 29105
    },
    {
      "epoch": 15.256813417190775,
      "grad_norm": 0.19025461375713348,
      "learning_rate": 8.090022415182898e-06,
      "loss": 0.4488,
      "num_input_tokens_seen": 19023456,
      "step": 29110
    },
    {
      "epoch": 15.25943396226415,
      "grad_norm": 0.26619529724121094,
      "learning_rate": 8.081602429644033e-06,
      "loss": 0.5129,
      "num_input_tokens_seen": 19027424,
      "step": 29115
    },
    {
      "epoch": 15.262054507337526,
      "grad_norm": 0.24077709019184113,
      "learning_rate": 8.073185983240932e-06,
      "loss": 0.5672,
      "num_input_tokens_seen": 19030976,
      "step": 29120
    },
    {
      "epoch": 15.264675052410901,
      "grad_norm": 0.20164787769317627,
      "learning_rate": 8.064773077734206e-06,
      "loss": 0.4663,
      "num_input_tokens_seen": 19034976,
      "step": 29125
    },
    {
      "epoch": 15.267295597484276,
      "grad_norm": 0.1525619775056839,
      "learning_rate": 8.05636371488374e-06,
      "loss": 0.4829,
      "num_input_tokens_seen": 19038752,
      "step": 29130
    },
    {
      "epoch": 15.269916142557651,
      "grad_norm": 0.2901260554790497,
      "learning_rate": 8.047957896448696e-06,
      "loss": 0.5011,
      "num_input_tokens_seen": 19041536,
      "step": 29135
    },
    {
      "epoch": 15.272536687631026,
      "grad_norm": 0.2393587827682495,
      "learning_rate": 8.039555624187451e-06,
      "loss": 0.4927,
      "num_input_tokens_seen": 19044448,
      "step": 29140
    },
    {
      "epoch": 15.275157232704402,
      "grad_norm": 0.21673241257667542,
      "learning_rate": 8.031156899857681e-06,
      "loss": 0.5549,
      "num_input_tokens_seen": 19047904,
      "step": 29145
    },
    {
      "epoch": 15.277777777777779,
      "grad_norm": 0.09358938038349152,
      "learning_rate": 8.022761725216288e-06,
      "loss": 0.5345,
      "num_input_tokens_seen": 19052864,
      "step": 29150
    },
    {
      "epoch": 15.280398322851154,
      "grad_norm": 0.13225558400154114,
      "learning_rate": 8.014370102019456e-06,
      "loss": 0.6296,
      "num_input_tokens_seen": 19055456,
      "step": 29155
    },
    {
      "epoch": 15.283018867924529,
      "grad_norm": 0.1876051276922226,
      "learning_rate": 8.005982032022616e-06,
      "loss": 0.4226,
      "num_input_tokens_seen": 19058240,
      "step": 29160
    },
    {
      "epoch": 15.285639412997904,
      "grad_norm": 0.14090827107429504,
      "learning_rate": 7.997597516980467e-06,
      "loss": 0.4927,
      "num_input_tokens_seen": 19062336,
      "step": 29165
    },
    {
      "epoch": 15.28825995807128,
      "grad_norm": 0.12229380011558533,
      "learning_rate": 7.989216558646942e-06,
      "loss": 0.3192,
      "num_input_tokens_seen": 19066592,
      "step": 29170
    },
    {
      "epoch": 15.290880503144654,
      "grad_norm": 0.1297973394393921,
      "learning_rate": 7.980839158775235e-06,
      "loss": 0.4419,
      "num_input_tokens_seen": 19069888,
      "step": 29175
    },
    {
      "epoch": 15.29350104821803,
      "grad_norm": 0.14153672754764557,
      "learning_rate": 7.972465319117814e-06,
      "loss": 0.3859,
      "num_input_tokens_seen": 19073152,
      "step": 29180
    },
    {
      "epoch": 15.296121593291405,
      "grad_norm": 0.14859439432621002,
      "learning_rate": 7.964095041426386e-06,
      "loss": 0.3996,
      "num_input_tokens_seen": 19075904,
      "step": 29185
    },
    {
      "epoch": 15.29874213836478,
      "grad_norm": 0.1037951409816742,
      "learning_rate": 7.95572832745192e-06,
      "loss": 0.6341,
      "num_input_tokens_seen": 19079584,
      "step": 29190
    },
    {
      "epoch": 15.301362683438155,
      "grad_norm": 0.1398431807756424,
      "learning_rate": 7.947365178944643e-06,
      "loss": 0.4046,
      "num_input_tokens_seen": 19082656,
      "step": 29195
    },
    {
      "epoch": 15.30398322851153,
      "grad_norm": 0.11770959198474884,
      "learning_rate": 7.939005597654025e-06,
      "loss": 0.5001,
      "num_input_tokens_seen": 19086208,
      "step": 29200
    },
    {
      "epoch": 15.306603773584905,
      "grad_norm": 0.1641090363264084,
      "learning_rate": 7.930649585328787e-06,
      "loss": 0.4382,
      "num_input_tokens_seen": 19089184,
      "step": 29205
    },
    {
      "epoch": 15.30922431865828,
      "grad_norm": 0.18674194812774658,
      "learning_rate": 7.922297143716918e-06,
      "loss": 0.471,
      "num_input_tokens_seen": 19093664,
      "step": 29210
    },
    {
      "epoch": 15.311844863731656,
      "grad_norm": 0.10413993149995804,
      "learning_rate": 7.913948274565652e-06,
      "loss": 0.3999,
      "num_input_tokens_seen": 19098048,
      "step": 29215
    },
    {
      "epoch": 15.314465408805031,
      "grad_norm": 0.16798464953899384,
      "learning_rate": 7.905602979621491e-06,
      "loss": 0.4655,
      "num_input_tokens_seen": 19101760,
      "step": 29220
    },
    {
      "epoch": 15.317085953878406,
      "grad_norm": 0.24214418232440948,
      "learning_rate": 7.897261260630159e-06,
      "loss": 0.42,
      "num_input_tokens_seen": 19106912,
      "step": 29225
    },
    {
      "epoch": 15.319706498951781,
      "grad_norm": 0.15355750918388367,
      "learning_rate": 7.888923119336647e-06,
      "loss": 0.5354,
      "num_input_tokens_seen": 19110272,
      "step": 29230
    },
    {
      "epoch": 15.322327044025156,
      "grad_norm": 0.14994587004184723,
      "learning_rate": 7.880588557485203e-06,
      "loss": 0.4828,
      "num_input_tokens_seen": 19113408,
      "step": 29235
    },
    {
      "epoch": 15.324947589098532,
      "grad_norm": 0.16129690408706665,
      "learning_rate": 7.872257576819325e-06,
      "loss": 0.5738,
      "num_input_tokens_seen": 19116864,
      "step": 29240
    },
    {
      "epoch": 15.327568134171909,
      "grad_norm": 0.12115903943777084,
      "learning_rate": 7.863930179081769e-06,
      "loss": 0.4323,
      "num_input_tokens_seen": 19120576,
      "step": 29245
    },
    {
      "epoch": 15.330188679245284,
      "grad_norm": 0.24485377967357635,
      "learning_rate": 7.855606366014517e-06,
      "loss": 0.4545,
      "num_input_tokens_seen": 19123872,
      "step": 29250
    },
    {
      "epoch": 15.332809224318659,
      "grad_norm": 0.1707865446805954,
      "learning_rate": 7.847286139358814e-06,
      "loss": 0.4506,
      "num_input_tokens_seen": 19126688,
      "step": 29255
    },
    {
      "epoch": 15.335429769392034,
      "grad_norm": 0.13095958530902863,
      "learning_rate": 7.838969500855162e-06,
      "loss": 0.3803,
      "num_input_tokens_seen": 19129504,
      "step": 29260
    },
    {
      "epoch": 15.33805031446541,
      "grad_norm": 0.22864268720149994,
      "learning_rate": 7.830656452243307e-06,
      "loss": 0.6836,
      "num_input_tokens_seen": 19132064,
      "step": 29265
    },
    {
      "epoch": 15.340670859538784,
      "grad_norm": 0.12718455493450165,
      "learning_rate": 7.822346995262241e-06,
      "loss": 0.349,
      "num_input_tokens_seen": 19134848,
      "step": 29270
    },
    {
      "epoch": 15.34329140461216,
      "grad_norm": 0.1441870629787445,
      "learning_rate": 7.814041131650223e-06,
      "loss": 0.6911,
      "num_input_tokens_seen": 19137632,
      "step": 29275
    },
    {
      "epoch": 15.345911949685535,
      "grad_norm": 0.1597907692193985,
      "learning_rate": 7.805738863144731e-06,
      "loss": 0.4518,
      "num_input_tokens_seen": 19140672,
      "step": 29280
    },
    {
      "epoch": 15.34853249475891,
      "grad_norm": 0.09830059856176376,
      "learning_rate": 7.797440191482496e-06,
      "loss": 0.5,
      "num_input_tokens_seen": 19144192,
      "step": 29285
    },
    {
      "epoch": 15.351153039832285,
      "grad_norm": 0.2051810324192047,
      "learning_rate": 7.789145118399518e-06,
      "loss": 0.5032,
      "num_input_tokens_seen": 19146880,
      "step": 29290
    },
    {
      "epoch": 15.35377358490566,
      "grad_norm": 0.11284211277961731,
      "learning_rate": 7.780853645631032e-06,
      "loss": 0.4428,
      "num_input_tokens_seen": 19150400,
      "step": 29295
    },
    {
      "epoch": 15.356394129979035,
      "grad_norm": 0.20055606961250305,
      "learning_rate": 7.772565774911522e-06,
      "loss": 0.4252,
      "num_input_tokens_seen": 19153504,
      "step": 29300
    },
    {
      "epoch": 15.35901467505241,
      "grad_norm": 0.1126517802476883,
      "learning_rate": 7.76428150797471e-06,
      "loss": 0.4157,
      "num_input_tokens_seen": 19156960,
      "step": 29305
    },
    {
      "epoch": 15.361635220125786,
      "grad_norm": 0.24268728494644165,
      "learning_rate": 7.756000846553562e-06,
      "loss": 0.7278,
      "num_input_tokens_seen": 19159744,
      "step": 29310
    },
    {
      "epoch": 15.364255765199161,
      "grad_norm": 0.15041354298591614,
      "learning_rate": 7.74772379238031e-06,
      "loss": 0.4379,
      "num_input_tokens_seen": 19163808,
      "step": 29315
    },
    {
      "epoch": 15.366876310272536,
      "grad_norm": 0.2786502540111542,
      "learning_rate": 7.739450347186417e-06,
      "loss": 0.4437,
      "num_input_tokens_seen": 19166304,
      "step": 29320
    },
    {
      "epoch": 15.369496855345911,
      "grad_norm": 0.15692485868930817,
      "learning_rate": 7.731180512702599e-06,
      "loss": 0.456,
      "num_input_tokens_seen": 19169792,
      "step": 29325
    },
    {
      "epoch": 15.372117400419286,
      "grad_norm": 0.15524360537528992,
      "learning_rate": 7.722914290658795e-06,
      "loss": 0.5097,
      "num_input_tokens_seen": 19172448,
      "step": 29330
    },
    {
      "epoch": 15.374737945492662,
      "grad_norm": 0.16478072106838226,
      "learning_rate": 7.714651682784224e-06,
      "loss": 0.4329,
      "num_input_tokens_seen": 19176352,
      "step": 29335
    },
    {
      "epoch": 15.377358490566039,
      "grad_norm": 0.39506515860557556,
      "learning_rate": 7.706392690807313e-06,
      "loss": 0.6242,
      "num_input_tokens_seen": 19179808,
      "step": 29340
    },
    {
      "epoch": 15.379979035639414,
      "grad_norm": 0.22926297783851624,
      "learning_rate": 7.69813731645575e-06,
      "loss": 0.4702,
      "num_input_tokens_seen": 19183872,
      "step": 29345
    },
    {
      "epoch": 15.382599580712789,
      "grad_norm": 0.14004147052764893,
      "learning_rate": 7.689885561456475e-06,
      "loss": 0.4071,
      "num_input_tokens_seen": 19187584,
      "step": 29350
    },
    {
      "epoch": 15.385220125786164,
      "grad_norm": 0.12902314960956573,
      "learning_rate": 7.681637427535663e-06,
      "loss": 0.5064,
      "num_input_tokens_seen": 19192320,
      "step": 29355
    },
    {
      "epoch": 15.38784067085954,
      "grad_norm": 0.15549245476722717,
      "learning_rate": 7.673392916418715e-06,
      "loss": 0.4535,
      "num_input_tokens_seen": 19195712,
      "step": 29360
    },
    {
      "epoch": 15.390461215932914,
      "grad_norm": 0.13183782994747162,
      "learning_rate": 7.665152029830303e-06,
      "loss": 0.595,
      "num_input_tokens_seen": 19199328,
      "step": 29365
    },
    {
      "epoch": 15.39308176100629,
      "grad_norm": 0.195990189909935,
      "learning_rate": 7.656914769494314e-06,
      "loss": 0.4895,
      "num_input_tokens_seen": 19202368,
      "step": 29370
    },
    {
      "epoch": 15.395702306079665,
      "grad_norm": 0.1332482546567917,
      "learning_rate": 7.648681137133892e-06,
      "loss": 0.5268,
      "num_input_tokens_seen": 19205728,
      "step": 29375
    },
    {
      "epoch": 15.39832285115304,
      "grad_norm": 0.21878531575202942,
      "learning_rate": 7.640451134471432e-06,
      "loss": 0.3857,
      "num_input_tokens_seen": 19208768,
      "step": 29380
    },
    {
      "epoch": 15.400943396226415,
      "grad_norm": 0.16408361494541168,
      "learning_rate": 7.632224763228538e-06,
      "loss": 0.5783,
      "num_input_tokens_seen": 19213152,
      "step": 29385
    },
    {
      "epoch": 15.40356394129979,
      "grad_norm": 0.1524103432893753,
      "learning_rate": 7.6240020251260896e-06,
      "loss": 0.5463,
      "num_input_tokens_seen": 19215392,
      "step": 29390
    },
    {
      "epoch": 15.406184486373165,
      "grad_norm": 0.11674787104129791,
      "learning_rate": 7.615782921884174e-06,
      "loss": 0.4759,
      "num_input_tokens_seen": 19218944,
      "step": 29395
    },
    {
      "epoch": 15.40880503144654,
      "grad_norm": 0.12087468057870865,
      "learning_rate": 7.607567455222142e-06,
      "loss": 0.6023,
      "num_input_tokens_seen": 19221984,
      "step": 29400
    },
    {
      "epoch": 15.411425576519916,
      "grad_norm": 0.1150742918252945,
      "learning_rate": 7.599355626858582e-06,
      "loss": 0.6296,
      "num_input_tokens_seen": 19225376,
      "step": 29405
    },
    {
      "epoch": 15.414046121593291,
      "grad_norm": 0.16013072431087494,
      "learning_rate": 7.591147438511298e-06,
      "loss": 0.5105,
      "num_input_tokens_seen": 19228448,
      "step": 29410
    },
    {
      "epoch": 15.416666666666666,
      "grad_norm": 0.15907274186611176,
      "learning_rate": 7.5829428918973625e-06,
      "loss": 0.8093,
      "num_input_tokens_seen": 19231712,
      "step": 29415
    },
    {
      "epoch": 15.419287211740041,
      "grad_norm": 0.12870632112026215,
      "learning_rate": 7.574741988733075e-06,
      "loss": 0.5489,
      "num_input_tokens_seen": 19234144,
      "step": 29420
    },
    {
      "epoch": 15.421907756813416,
      "grad_norm": 0.1624319851398468,
      "learning_rate": 7.56654473073396e-06,
      "loss": 0.545,
      "num_input_tokens_seen": 19237056,
      "step": 29425
    },
    {
      "epoch": 15.424528301886792,
      "grad_norm": 0.10103825479745865,
      "learning_rate": 7.558351119614796e-06,
      "loss": 0.4576,
      "num_input_tokens_seen": 19239904,
      "step": 29430
    },
    {
      "epoch": 15.427148846960169,
      "grad_norm": 0.13285817205905914,
      "learning_rate": 7.5501611570896e-06,
      "loss": 0.4877,
      "num_input_tokens_seen": 19243552,
      "step": 29435
    },
    {
      "epoch": 15.429769392033544,
      "grad_norm": 0.19149310886859894,
      "learning_rate": 7.541974844871602e-06,
      "loss": 0.5014,
      "num_input_tokens_seen": 19246560,
      "step": 29440
    },
    {
      "epoch": 15.432389937106919,
      "grad_norm": 0.09263833612203598,
      "learning_rate": 7.5337921846732965e-06,
      "loss": 0.4778,
      "num_input_tokens_seen": 19249632,
      "step": 29445
    },
    {
      "epoch": 15.435010482180294,
      "grad_norm": 0.23932866752147675,
      "learning_rate": 7.525613178206409e-06,
      "loss": 0.4827,
      "num_input_tokens_seen": 19252736,
      "step": 29450
    },
    {
      "epoch": 15.43763102725367,
      "grad_norm": 0.14746186137199402,
      "learning_rate": 7.517437827181878e-06,
      "loss": 0.4714,
      "num_input_tokens_seen": 19255968,
      "step": 29455
    },
    {
      "epoch": 15.440251572327044,
      "grad_norm": 0.15967465937137604,
      "learning_rate": 7.509266133309908e-06,
      "loss": 0.5181,
      "num_input_tokens_seen": 19259680,
      "step": 29460
    },
    {
      "epoch": 15.44287211740042,
      "grad_norm": 0.19683465361595154,
      "learning_rate": 7.501098098299914e-06,
      "loss": 0.4394,
      "num_input_tokens_seen": 19262848,
      "step": 29465
    },
    {
      "epoch": 15.445492662473795,
      "grad_norm": 0.17486456036567688,
      "learning_rate": 7.492933723860557e-06,
      "loss": 0.446,
      "num_input_tokens_seen": 19265536,
      "step": 29470
    },
    {
      "epoch": 15.44811320754717,
      "grad_norm": 0.19655631482601166,
      "learning_rate": 7.4847730116997334e-06,
      "loss": 0.3793,
      "num_input_tokens_seen": 19268384,
      "step": 29475
    },
    {
      "epoch": 15.450733752620545,
      "grad_norm": 0.09217974543571472,
      "learning_rate": 7.476615963524583e-06,
      "loss": 0.469,
      "num_input_tokens_seen": 19272000,
      "step": 29480
    },
    {
      "epoch": 15.45335429769392,
      "grad_norm": 0.14154760539531708,
      "learning_rate": 7.468462581041452e-06,
      "loss": 0.5648,
      "num_input_tokens_seen": 19275360,
      "step": 29485
    },
    {
      "epoch": 15.455974842767295,
      "grad_norm": 0.22061307728290558,
      "learning_rate": 7.460312865955932e-06,
      "loss": 0.4755,
      "num_input_tokens_seen": 19279488,
      "step": 29490
    },
    {
      "epoch": 15.45859538784067,
      "grad_norm": 0.38737034797668457,
      "learning_rate": 7.4521668199728584e-06,
      "loss": 0.4904,
      "num_input_tokens_seen": 19282176,
      "step": 29495
    },
    {
      "epoch": 15.461215932914046,
      "grad_norm": 0.2034190595149994,
      "learning_rate": 7.4440244447962884e-06,
      "loss": 0.5621,
      "num_input_tokens_seen": 19285056,
      "step": 29500
    },
    {
      "epoch": 15.463836477987421,
      "grad_norm": 0.1576596349477768,
      "learning_rate": 7.435885742129523e-06,
      "loss": 0.4426,
      "num_input_tokens_seen": 19288704,
      "step": 29505
    },
    {
      "epoch": 15.466457023060796,
      "grad_norm": 0.2066907286643982,
      "learning_rate": 7.427750713675071e-06,
      "loss": 0.7354,
      "num_input_tokens_seen": 19291584,
      "step": 29510
    },
    {
      "epoch": 15.469077568134171,
      "grad_norm": 0.22900837659835815,
      "learning_rate": 7.419619361134702e-06,
      "loss": 0.4123,
      "num_input_tokens_seen": 19295072,
      "step": 29515
    },
    {
      "epoch": 15.471698113207546,
      "grad_norm": 0.17389191687107086,
      "learning_rate": 7.411491686209387e-06,
      "loss": 0.4698,
      "num_input_tokens_seen": 19298144,
      "step": 29520
    },
    {
      "epoch": 15.474318658280922,
      "grad_norm": 0.1801394522190094,
      "learning_rate": 7.4033676905993525e-06,
      "loss": 0.6493,
      "num_input_tokens_seen": 19301472,
      "step": 29525
    },
    {
      "epoch": 15.476939203354299,
      "grad_norm": 0.23816782236099243,
      "learning_rate": 7.3952473760040444e-06,
      "loss": 0.4755,
      "num_input_tokens_seen": 19303936,
      "step": 29530
    },
    {
      "epoch": 15.479559748427674,
      "grad_norm": 0.11844202876091003,
      "learning_rate": 7.387130744122148e-06,
      "loss": 0.509,
      "num_input_tokens_seen": 19308416,
      "step": 29535
    },
    {
      "epoch": 15.482180293501049,
      "grad_norm": 0.24965472519397736,
      "learning_rate": 7.379017796651558e-06,
      "loss": 0.5289,
      "num_input_tokens_seen": 19312992,
      "step": 29540
    },
    {
      "epoch": 15.484800838574424,
      "grad_norm": 0.33828791975975037,
      "learning_rate": 7.3709085352894085e-06,
      "loss": 0.3682,
      "num_input_tokens_seen": 19317792,
      "step": 29545
    },
    {
      "epoch": 15.4874213836478,
      "grad_norm": 0.1902296543121338,
      "learning_rate": 7.362802961732071e-06,
      "loss": 0.4164,
      "num_input_tokens_seen": 19320928,
      "step": 29550
    },
    {
      "epoch": 15.490041928721174,
      "grad_norm": 0.1621943712234497,
      "learning_rate": 7.354701077675136e-06,
      "loss": 0.5252,
      "num_input_tokens_seen": 19323584,
      "step": 29555
    },
    {
      "epoch": 15.49266247379455,
      "grad_norm": 0.1355486363172531,
      "learning_rate": 7.346602884813439e-06,
      "loss": 0.4834,
      "num_input_tokens_seen": 19326112,
      "step": 29560
    },
    {
      "epoch": 15.495283018867925,
      "grad_norm": 0.1344204694032669,
      "learning_rate": 7.338508384841014e-06,
      "loss": 0.4566,
      "num_input_tokens_seen": 19329824,
      "step": 29565
    },
    {
      "epoch": 15.4979035639413,
      "grad_norm": 0.17644527554512024,
      "learning_rate": 7.3304175794511345e-06,
      "loss": 0.4901,
      "num_input_tokens_seen": 19332544,
      "step": 29570
    },
    {
      "epoch": 15.500524109014675,
      "grad_norm": 0.23687389492988586,
      "learning_rate": 7.3223304703363135e-06,
      "loss": 0.4365,
      "num_input_tokens_seen": 19335744,
      "step": 29575
    },
    {
      "epoch": 15.50314465408805,
      "grad_norm": 0.17129331827163696,
      "learning_rate": 7.314247059188281e-06,
      "loss": 0.6352,
      "num_input_tokens_seen": 19339456,
      "step": 29580
    },
    {
      "epoch": 15.505765199161425,
      "grad_norm": 0.19396907091140747,
      "learning_rate": 7.306167347697992e-06,
      "loss": 0.6067,
      "num_input_tokens_seen": 19342752,
      "step": 29585
    },
    {
      "epoch": 15.5083857442348,
      "grad_norm": 0.13678622245788574,
      "learning_rate": 7.29809133755564e-06,
      "loss": 0.4751,
      "num_input_tokens_seen": 19345536,
      "step": 29590
    },
    {
      "epoch": 15.511006289308176,
      "grad_norm": 0.1940038949251175,
      "learning_rate": 7.290019030450629e-06,
      "loss": 0.4492,
      "num_input_tokens_seen": 19348768,
      "step": 29595
    },
    {
      "epoch": 15.51362683438155,
      "grad_norm": 0.201769158244133,
      "learning_rate": 7.2819504280715794e-06,
      "loss": 0.421,
      "num_input_tokens_seen": 19352128,
      "step": 29600
    },
    {
      "epoch": 15.516247379454926,
      "grad_norm": 0.1604640781879425,
      "learning_rate": 7.273885532106364e-06,
      "loss": 0.5657,
      "num_input_tokens_seen": 19354784,
      "step": 29605
    },
    {
      "epoch": 15.518867924528301,
      "grad_norm": 0.1485796570777893,
      "learning_rate": 7.265824344242064e-06,
      "loss": 0.5651,
      "num_input_tokens_seen": 19358528,
      "step": 29610
    },
    {
      "epoch": 15.521488469601676,
      "grad_norm": 0.19269098341464996,
      "learning_rate": 7.257766866165e-06,
      "loss": 0.4272,
      "num_input_tokens_seen": 19361760,
      "step": 29615
    },
    {
      "epoch": 15.524109014675052,
      "grad_norm": 0.12386975437402725,
      "learning_rate": 7.24971309956069e-06,
      "loss": 0.442,
      "num_input_tokens_seen": 19365504,
      "step": 29620
    },
    {
      "epoch": 15.526729559748428,
      "grad_norm": 0.182892307639122,
      "learning_rate": 7.241663046113887e-06,
      "loss": 0.3362,
      "num_input_tokens_seen": 19369344,
      "step": 29625
    },
    {
      "epoch": 15.529350104821804,
      "grad_norm": 0.2257661372423172,
      "learning_rate": 7.233616707508576e-06,
      "loss": 0.457,
      "num_input_tokens_seen": 19372224,
      "step": 29630
    },
    {
      "epoch": 15.531970649895179,
      "grad_norm": 0.20540010929107666,
      "learning_rate": 7.225574085427961e-06,
      "loss": 0.4687,
      "num_input_tokens_seen": 19374976,
      "step": 29635
    },
    {
      "epoch": 15.534591194968554,
      "grad_norm": 0.17783845961093903,
      "learning_rate": 7.217535181554474e-06,
      "loss": 0.3828,
      "num_input_tokens_seen": 19377440,
      "step": 29640
    },
    {
      "epoch": 15.53721174004193,
      "grad_norm": 0.29547783732414246,
      "learning_rate": 7.209499997569747e-06,
      "loss": 0.5085,
      "num_input_tokens_seen": 19380032,
      "step": 29645
    },
    {
      "epoch": 15.539832285115304,
      "grad_norm": 0.13899634778499603,
      "learning_rate": 7.201468535154663e-06,
      "loss": 0.3685,
      "num_input_tokens_seen": 19383488,
      "step": 29650
    },
    {
      "epoch": 15.54245283018868,
      "grad_norm": 0.24393588304519653,
      "learning_rate": 7.193440795989295e-06,
      "loss": 0.5468,
      "num_input_tokens_seen": 19386560,
      "step": 29655
    },
    {
      "epoch": 15.545073375262055,
      "grad_norm": 0.19403110444545746,
      "learning_rate": 7.185416781752965e-06,
      "loss": 0.5407,
      "num_input_tokens_seen": 19389216,
      "step": 29660
    },
    {
      "epoch": 15.54769392033543,
      "grad_norm": 0.19419549405574799,
      "learning_rate": 7.177396494124206e-06,
      "loss": 0.5276,
      "num_input_tokens_seen": 19393056,
      "step": 29665
    },
    {
      "epoch": 15.550314465408805,
      "grad_norm": 0.21202421188354492,
      "learning_rate": 7.169379934780779e-06,
      "loss": 0.5836,
      "num_input_tokens_seen": 19396672,
      "step": 29670
    },
    {
      "epoch": 15.55293501048218,
      "grad_norm": 0.21177008748054504,
      "learning_rate": 7.161367105399644e-06,
      "loss": 0.6324,
      "num_input_tokens_seen": 19399168,
      "step": 29675
    },
    {
      "epoch": 15.555555555555555,
      "grad_norm": 0.0430024117231369,
      "learning_rate": 7.153358007656991e-06,
      "loss": 0.4021,
      "num_input_tokens_seen": 19405248,
      "step": 29680
    },
    {
      "epoch": 15.55817610062893,
      "grad_norm": 0.20599955320358276,
      "learning_rate": 7.145352643228237e-06,
      "loss": 0.4986,
      "num_input_tokens_seen": 19407712,
      "step": 29685
    },
    {
      "epoch": 15.560796645702306,
      "grad_norm": 0.16764254868030548,
      "learning_rate": 7.137351013788013e-06,
      "loss": 0.3906,
      "num_input_tokens_seen": 19410816,
      "step": 29690
    },
    {
      "epoch": 15.56341719077568,
      "grad_norm": 0.1390874683856964,
      "learning_rate": 7.129353121010177e-06,
      "loss": 0.4399,
      "num_input_tokens_seen": 19413888,
      "step": 29695
    },
    {
      "epoch": 15.566037735849056,
      "grad_norm": 0.1885666847229004,
      "learning_rate": 7.121358966567779e-06,
      "loss": 0.5399,
      "num_input_tokens_seen": 19417344,
      "step": 29700
    },
    {
      "epoch": 15.568658280922431,
      "grad_norm": 0.15905380249023438,
      "learning_rate": 7.113368552133126e-06,
      "loss": 0.3914,
      "num_input_tokens_seen": 19420512,
      "step": 29705
    },
    {
      "epoch": 15.571278825995806,
      "grad_norm": 0.11577579379081726,
      "learning_rate": 7.105381879377701e-06,
      "loss": 0.3792,
      "num_input_tokens_seen": 19423328,
      "step": 29710
    },
    {
      "epoch": 15.573899371069182,
      "grad_norm": 0.18797901272773743,
      "learning_rate": 7.097398949972234e-06,
      "loss": 0.4793,
      "num_input_tokens_seen": 19427136,
      "step": 29715
    },
    {
      "epoch": 15.576519916142558,
      "grad_norm": 0.14667575061321259,
      "learning_rate": 7.089419765586672e-06,
      "loss": 0.4781,
      "num_input_tokens_seen": 19430496,
      "step": 29720
    },
    {
      "epoch": 15.579140461215934,
      "grad_norm": 0.2605992555618286,
      "learning_rate": 7.081444327890152e-06,
      "loss": 0.4897,
      "num_input_tokens_seen": 19433792,
      "step": 29725
    },
    {
      "epoch": 15.581761006289309,
      "grad_norm": 0.18639689683914185,
      "learning_rate": 7.073472638551054e-06,
      "loss": 0.4082,
      "num_input_tokens_seen": 19436384,
      "step": 29730
    },
    {
      "epoch": 15.584381551362684,
      "grad_norm": 0.2126515954732895,
      "learning_rate": 7.065504699236969e-06,
      "loss": 0.5306,
      "num_input_tokens_seen": 19439200,
      "step": 29735
    },
    {
      "epoch": 15.58700209643606,
      "grad_norm": 0.13622795045375824,
      "learning_rate": 7.057540511614688e-06,
      "loss": 0.4452,
      "num_input_tokens_seen": 19442400,
      "step": 29740
    },
    {
      "epoch": 15.589622641509434,
      "grad_norm": 0.17056190967559814,
      "learning_rate": 7.049580077350232e-06,
      "loss": 0.4983,
      "num_input_tokens_seen": 19446464,
      "step": 29745
    },
    {
      "epoch": 15.59224318658281,
      "grad_norm": 0.22512996196746826,
      "learning_rate": 7.041623398108843e-06,
      "loss": 0.3436,
      "num_input_tokens_seen": 19449600,
      "step": 29750
    },
    {
      "epoch": 15.594863731656185,
      "grad_norm": 0.17749828100204468,
      "learning_rate": 7.033670475554949e-06,
      "loss": 0.3902,
      "num_input_tokens_seen": 19452608,
      "step": 29755
    },
    {
      "epoch": 15.59748427672956,
      "grad_norm": 0.10482729226350784,
      "learning_rate": 7.025721311352224e-06,
      "loss": 0.4181,
      "num_input_tokens_seen": 19456832,
      "step": 29760
    },
    {
      "epoch": 15.600104821802935,
      "grad_norm": 0.15210264921188354,
      "learning_rate": 7.017775907163546e-06,
      "loss": 0.4365,
      "num_input_tokens_seen": 19459136,
      "step": 29765
    },
    {
      "epoch": 15.60272536687631,
      "grad_norm": 0.1668158769607544,
      "learning_rate": 7.009834264650989e-06,
      "loss": 0.3939,
      "num_input_tokens_seen": 19463360,
      "step": 29770
    },
    {
      "epoch": 15.605345911949685,
      "grad_norm": 0.20780688524246216,
      "learning_rate": 7.001896385475867e-06,
      "loss": 0.3933,
      "num_input_tokens_seen": 19466432,
      "step": 29775
    },
    {
      "epoch": 15.60796645702306,
      "grad_norm": 0.11692659556865692,
      "learning_rate": 6.993962271298682e-06,
      "loss": 0.3843,
      "num_input_tokens_seen": 19470592,
      "step": 29780
    },
    {
      "epoch": 15.610587002096436,
      "grad_norm": 0.1215558722615242,
      "learning_rate": 6.986031923779166e-06,
      "loss": 0.377,
      "num_input_tokens_seen": 19474496,
      "step": 29785
    },
    {
      "epoch": 15.61320754716981,
      "grad_norm": 0.11186295002698898,
      "learning_rate": 6.978105344576264e-06,
      "loss": 0.4438,
      "num_input_tokens_seen": 19477792,
      "step": 29790
    },
    {
      "epoch": 15.615828092243186,
      "grad_norm": 0.17622016370296478,
      "learning_rate": 6.970182535348111e-06,
      "loss": 0.4682,
      "num_input_tokens_seen": 19480736,
      "step": 29795
    },
    {
      "epoch": 15.618448637316561,
      "grad_norm": 0.09956519305706024,
      "learning_rate": 6.962263497752086e-06,
      "loss": 0.5061,
      "num_input_tokens_seen": 19485312,
      "step": 29800
    },
    {
      "epoch": 15.621069182389936,
      "grad_norm": 0.17189498245716095,
      "learning_rate": 6.9543482334447436e-06,
      "loss": 0.3973,
      "num_input_tokens_seen": 19488544,
      "step": 29805
    },
    {
      "epoch": 15.623689727463312,
      "grad_norm": 0.18339498341083527,
      "learning_rate": 6.946436744081875e-06,
      "loss": 0.3758,
      "num_input_tokens_seen": 19491744,
      "step": 29810
    },
    {
      "epoch": 15.626310272536688,
      "grad_norm": 0.13523711264133453,
      "learning_rate": 6.938529031318472e-06,
      "loss": 0.4946,
      "num_input_tokens_seen": 19495136,
      "step": 29815
    },
    {
      "epoch": 15.628930817610064,
      "grad_norm": 0.17486721277236938,
      "learning_rate": 6.9306250968087485e-06,
      "loss": 0.4741,
      "num_input_tokens_seen": 19498272,
      "step": 29820
    },
    {
      "epoch": 15.631551362683439,
      "grad_norm": 0.33445823192596436,
      "learning_rate": 6.922724942206102e-06,
      "loss": 0.4647,
      "num_input_tokens_seen": 19501792,
      "step": 29825
    },
    {
      "epoch": 15.634171907756814,
      "grad_norm": 0.24349145591259003,
      "learning_rate": 6.914828569163167e-06,
      "loss": 0.6202,
      "num_input_tokens_seen": 19504640,
      "step": 29830
    },
    {
      "epoch": 15.63679245283019,
      "grad_norm": 0.12504488229751587,
      "learning_rate": 6.906935979331763e-06,
      "loss": 0.4082,
      "num_input_tokens_seen": 19508032,
      "step": 29835
    },
    {
      "epoch": 15.639412997903564,
      "grad_norm": 0.1256273090839386,
      "learning_rate": 6.8990471743629356e-06,
      "loss": 0.3747,
      "num_input_tokens_seen": 19511008,
      "step": 29840
    },
    {
      "epoch": 15.64203354297694,
      "grad_norm": 0.13543707132339478,
      "learning_rate": 6.89116215590693e-06,
      "loss": 0.6028,
      "num_input_tokens_seen": 19514240,
      "step": 29845
    },
    {
      "epoch": 15.644654088050315,
      "grad_norm": 0.20918630063533783,
      "learning_rate": 6.8832809256132146e-06,
      "loss": 0.4133,
      "num_input_tokens_seen": 19517056,
      "step": 29850
    },
    {
      "epoch": 15.64727463312369,
      "grad_norm": 0.3853319585323334,
      "learning_rate": 6.875403485130444e-06,
      "loss": 0.4334,
      "num_input_tokens_seen": 19520256,
      "step": 29855
    },
    {
      "epoch": 15.649895178197065,
      "grad_norm": 0.27225324511528015,
      "learning_rate": 6.867529836106479e-06,
      "loss": 0.4908,
      "num_input_tokens_seen": 19522976,
      "step": 29860
    },
    {
      "epoch": 15.65251572327044,
      "grad_norm": 0.1835269033908844,
      "learning_rate": 6.8596599801884045e-06,
      "loss": 0.4269,
      "num_input_tokens_seen": 19526784,
      "step": 29865
    },
    {
      "epoch": 15.655136268343815,
      "grad_norm": 0.15151885151863098,
      "learning_rate": 6.851793919022509e-06,
      "loss": 0.3862,
      "num_input_tokens_seen": 19529920,
      "step": 29870
    },
    {
      "epoch": 15.65775681341719,
      "grad_norm": 0.18735648691654205,
      "learning_rate": 6.843931654254285e-06,
      "loss": 0.5014,
      "num_input_tokens_seen": 19532608,
      "step": 29875
    },
    {
      "epoch": 15.660377358490566,
      "grad_norm": 0.33389732241630554,
      "learning_rate": 6.836073187528425e-06,
      "loss": 0.4443,
      "num_input_tokens_seen": 19535680,
      "step": 29880
    },
    {
      "epoch": 15.66299790356394,
      "grad_norm": 0.21729087829589844,
      "learning_rate": 6.828218520488821e-06,
      "loss": 0.4315,
      "num_input_tokens_seen": 19538144,
      "step": 29885
    },
    {
      "epoch": 15.665618448637316,
      "grad_norm": 0.14654669165611267,
      "learning_rate": 6.820367654778589e-06,
      "loss": 0.3078,
      "num_input_tokens_seen": 19541344,
      "step": 29890
    },
    {
      "epoch": 15.668238993710691,
      "grad_norm": 0.12545233964920044,
      "learning_rate": 6.812520592040039e-06,
      "loss": 0.517,
      "num_input_tokens_seen": 19544640,
      "step": 29895
    },
    {
      "epoch": 15.670859538784066,
      "grad_norm": 0.15196141600608826,
      "learning_rate": 6.804677333914689e-06,
      "loss": 0.3485,
      "num_input_tokens_seen": 19547872,
      "step": 29900
    },
    {
      "epoch": 15.673480083857442,
      "grad_norm": 0.14228610694408417,
      "learning_rate": 6.796837882043261e-06,
      "loss": 0.4037,
      "num_input_tokens_seen": 19550880,
      "step": 29905
    },
    {
      "epoch": 15.676100628930818,
      "grad_norm": 0.191222682595253,
      "learning_rate": 6.7890022380656785e-06,
      "loss": 0.656,
      "num_input_tokens_seen": 19554112,
      "step": 29910
    },
    {
      "epoch": 15.678721174004194,
      "grad_norm": 0.17406386137008667,
      "learning_rate": 6.781170403621056e-06,
      "loss": 0.4228,
      "num_input_tokens_seen": 19557760,
      "step": 29915
    },
    {
      "epoch": 15.681341719077569,
      "grad_norm": 0.313753604888916,
      "learning_rate": 6.773342380347736e-06,
      "loss": 0.5008,
      "num_input_tokens_seen": 19561312,
      "step": 29920
    },
    {
      "epoch": 15.683962264150944,
      "grad_norm": 0.1482924520969391,
      "learning_rate": 6.765518169883248e-06,
      "loss": 0.6225,
      "num_input_tokens_seen": 19565248,
      "step": 29925
    },
    {
      "epoch": 15.68658280922432,
      "grad_norm": 0.25306975841522217,
      "learning_rate": 6.757697773864338e-06,
      "loss": 0.5445,
      "num_input_tokens_seen": 19567808,
      "step": 29930
    },
    {
      "epoch": 15.689203354297694,
      "grad_norm": 0.12845325469970703,
      "learning_rate": 6.749881193926932e-06,
      "loss": 0.5874,
      "num_input_tokens_seen": 19570880,
      "step": 29935
    },
    {
      "epoch": 15.69182389937107,
      "grad_norm": 0.2059239149093628,
      "learning_rate": 6.742068431706167e-06,
      "loss": 0.4716,
      "num_input_tokens_seen": 19573536,
      "step": 29940
    },
    {
      "epoch": 15.694444444444445,
      "grad_norm": 0.20368511974811554,
      "learning_rate": 6.734259488836386e-06,
      "loss": 0.4824,
      "num_input_tokens_seen": 19577408,
      "step": 29945
    },
    {
      "epoch": 15.69706498951782,
      "grad_norm": 0.14164476096630096,
      "learning_rate": 6.726454366951135e-06,
      "loss": 0.4552,
      "num_input_tokens_seen": 19580960,
      "step": 29950
    },
    {
      "epoch": 15.699685534591195,
      "grad_norm": 0.19249776005744934,
      "learning_rate": 6.7186530676831616e-06,
      "loss": 0.7082,
      "num_input_tokens_seen": 19584448,
      "step": 29955
    },
    {
      "epoch": 15.70230607966457,
      "grad_norm": 0.14348585903644562,
      "learning_rate": 6.710855592664403e-06,
      "loss": 0.4763,
      "num_input_tokens_seen": 19587392,
      "step": 29960
    },
    {
      "epoch": 15.704926624737945,
      "grad_norm": 0.21348081529140472,
      "learning_rate": 6.703061943525993e-06,
      "loss": 0.4029,
      "num_input_tokens_seen": 19590240,
      "step": 29965
    },
    {
      "epoch": 15.70754716981132,
      "grad_norm": 0.1484224647283554,
      "learning_rate": 6.69527212189828e-06,
      "loss": 0.442,
      "num_input_tokens_seen": 19593056,
      "step": 29970
    },
    {
      "epoch": 15.710167714884696,
      "grad_norm": 0.10290328413248062,
      "learning_rate": 6.687486129410811e-06,
      "loss": 0.4278,
      "num_input_tokens_seen": 19596160,
      "step": 29975
    },
    {
      "epoch": 15.71278825995807,
      "grad_norm": 0.24055631458759308,
      "learning_rate": 6.679703967692322e-06,
      "loss": 0.3443,
      "num_input_tokens_seen": 19601184,
      "step": 29980
    },
    {
      "epoch": 15.715408805031446,
      "grad_norm": 0.14101265370845795,
      "learning_rate": 6.67192563837076e-06,
      "loss": 0.4209,
      "num_input_tokens_seen": 19604160,
      "step": 29985
    },
    {
      "epoch": 15.718029350104821,
      "grad_norm": 0.09578943252563477,
      "learning_rate": 6.664151143073258e-06,
      "loss": 0.4034,
      "num_input_tokens_seen": 19607872,
      "step": 29990
    },
    {
      "epoch": 15.720649895178196,
      "grad_norm": 0.10711199790239334,
      "learning_rate": 6.656380483426141e-06,
      "loss": 0.3458,
      "num_input_tokens_seen": 19612096,
      "step": 29995
    },
    {
      "epoch": 15.723270440251572,
      "grad_norm": 0.1610538214445114,
      "learning_rate": 6.648613661054956e-06,
      "loss": 0.4923,
      "num_input_tokens_seen": 19614976,
      "step": 30000
    },
    {
      "epoch": 15.725890985324948,
      "grad_norm": 0.16561047732830048,
      "learning_rate": 6.6408506775844256e-06,
      "loss": 0.5191,
      "num_input_tokens_seen": 19618624,
      "step": 30005
    },
    {
      "epoch": 15.728511530398324,
      "grad_norm": 0.29266101121902466,
      "learning_rate": 6.633091534638492e-06,
      "loss": 0.5779,
      "num_input_tokens_seen": 19621920,
      "step": 30010
    },
    {
      "epoch": 15.731132075471699,
      "grad_norm": 0.13867099583148956,
      "learning_rate": 6.625336233840257e-06,
      "loss": 0.4168,
      "num_input_tokens_seen": 19624352,
      "step": 30015
    },
    {
      "epoch": 15.733752620545074,
      "grad_norm": 0.1390789896249771,
      "learning_rate": 6.617584776812064e-06,
      "loss": 0.3943,
      "num_input_tokens_seen": 19631744,
      "step": 30020
    },
    {
      "epoch": 15.73637316561845,
      "grad_norm": 0.19069579243659973,
      "learning_rate": 6.6098371651754085e-06,
      "loss": 0.432,
      "num_input_tokens_seen": 19634592,
      "step": 30025
    },
    {
      "epoch": 15.738993710691824,
      "grad_norm": 0.14570379257202148,
      "learning_rate": 6.602093400551012e-06,
      "loss": 0.4803,
      "num_input_tokens_seen": 19637472,
      "step": 30030
    },
    {
      "epoch": 15.7416142557652,
      "grad_norm": 0.12564896047115326,
      "learning_rate": 6.59435348455879e-06,
      "loss": 0.5751,
      "num_input_tokens_seen": 19641344,
      "step": 30035
    },
    {
      "epoch": 15.744234800838575,
      "grad_norm": 0.20479047298431396,
      "learning_rate": 6.586617418817828e-06,
      "loss": 0.4205,
      "num_input_tokens_seen": 19644320,
      "step": 30040
    },
    {
      "epoch": 15.74685534591195,
      "grad_norm": 0.12852750718593597,
      "learning_rate": 6.578885204946439e-06,
      "loss": 0.3766,
      "num_input_tokens_seen": 19648512,
      "step": 30045
    },
    {
      "epoch": 15.749475890985325,
      "grad_norm": 0.19624388217926025,
      "learning_rate": 6.571156844562098e-06,
      "loss": 0.5729,
      "num_input_tokens_seen": 19651296,
      "step": 30050
    },
    {
      "epoch": 15.7520964360587,
      "grad_norm": 0.2089843451976776,
      "learning_rate": 6.563432339281497e-06,
      "loss": 0.5105,
      "num_input_tokens_seen": 19654080,
      "step": 30055
    },
    {
      "epoch": 15.754716981132075,
      "grad_norm": 0.15771697461605072,
      "learning_rate": 6.555711690720517e-06,
      "loss": 0.5096,
      "num_input_tokens_seen": 19658240,
      "step": 30060
    },
    {
      "epoch": 15.75733752620545,
      "grad_norm": 0.15681414306163788,
      "learning_rate": 6.547994900494234e-06,
      "loss": 0.6184,
      "num_input_tokens_seen": 19661792,
      "step": 30065
    },
    {
      "epoch": 15.759958071278826,
      "grad_norm": 0.34679973125457764,
      "learning_rate": 6.540281970216899e-06,
      "loss": 0.5228,
      "num_input_tokens_seen": 19664288,
      "step": 30070
    },
    {
      "epoch": 15.7625786163522,
      "grad_norm": 0.20883047580718994,
      "learning_rate": 6.5325729015019845e-06,
      "loss": 0.5071,
      "num_input_tokens_seen": 19670656,
      "step": 30075
    },
    {
      "epoch": 15.765199161425576,
      "grad_norm": 0.1632656455039978,
      "learning_rate": 6.524867695962122e-06,
      "loss": 0.518,
      "num_input_tokens_seen": 19674048,
      "step": 30080
    },
    {
      "epoch": 15.767819706498951,
      "grad_norm": 0.18776024878025055,
      "learning_rate": 6.517166355209165e-06,
      "loss": 0.385,
      "num_input_tokens_seen": 19676896,
      "step": 30085
    },
    {
      "epoch": 15.770440251572326,
      "grad_norm": 0.09141217172145844,
      "learning_rate": 6.50946888085415e-06,
      "loss": 0.4565,
      "num_input_tokens_seen": 19679392,
      "step": 30090
    },
    {
      "epoch": 15.773060796645701,
      "grad_norm": 0.17335103452205658,
      "learning_rate": 6.5017752745072855e-06,
      "loss": 0.4954,
      "num_input_tokens_seen": 19682560,
      "step": 30095
    },
    {
      "epoch": 15.775681341719078,
      "grad_norm": 0.24109280109405518,
      "learning_rate": 6.494085537777994e-06,
      "loss": 0.4753,
      "num_input_tokens_seen": 19685312,
      "step": 30100
    },
    {
      "epoch": 15.778301886792454,
      "grad_norm": 0.32456323504447937,
      "learning_rate": 6.48639967227489e-06,
      "loss": 0.4418,
      "num_input_tokens_seen": 19688064,
      "step": 30105
    },
    {
      "epoch": 15.780922431865829,
      "grad_norm": 0.13579046726226807,
      "learning_rate": 6.47871767960575e-06,
      "loss": 0.5022,
      "num_input_tokens_seen": 19690976,
      "step": 30110
    },
    {
      "epoch": 15.783542976939204,
      "grad_norm": 0.14620338380336761,
      "learning_rate": 6.471039561377581e-06,
      "loss": 0.5156,
      "num_input_tokens_seen": 19693728,
      "step": 30115
    },
    {
      "epoch": 15.786163522012579,
      "grad_norm": 0.21023835241794586,
      "learning_rate": 6.463365319196538e-06,
      "loss": 0.5381,
      "num_input_tokens_seen": 19700864,
      "step": 30120
    },
    {
      "epoch": 15.788784067085954,
      "grad_norm": 0.21623243391513824,
      "learning_rate": 6.4556949546679905e-06,
      "loss": 0.3957,
      "num_input_tokens_seen": 19703488,
      "step": 30125
    },
    {
      "epoch": 15.79140461215933,
      "grad_norm": 0.13103018701076508,
      "learning_rate": 6.448028469396497e-06,
      "loss": 0.3915,
      "num_input_tokens_seen": 19706944,
      "step": 30130
    },
    {
      "epoch": 15.794025157232705,
      "grad_norm": 0.10851925611495972,
      "learning_rate": 6.440365864985801e-06,
      "loss": 0.4523,
      "num_input_tokens_seen": 19710400,
      "step": 30135
    },
    {
      "epoch": 15.79664570230608,
      "grad_norm": 0.17315901815891266,
      "learning_rate": 6.432707143038818e-06,
      "loss": 0.4826,
      "num_input_tokens_seen": 19713056,
      "step": 30140
    },
    {
      "epoch": 15.799266247379455,
      "grad_norm": 0.11901304125785828,
      "learning_rate": 6.4250523051576834e-06,
      "loss": 0.4792,
      "num_input_tokens_seen": 19717248,
      "step": 30145
    },
    {
      "epoch": 15.80188679245283,
      "grad_norm": 0.29578539729118347,
      "learning_rate": 6.417401352943686e-06,
      "loss": 0.54,
      "num_input_tokens_seen": 19719552,
      "step": 30150
    },
    {
      "epoch": 15.804507337526205,
      "grad_norm": 0.20647752285003662,
      "learning_rate": 6.409754287997322e-06,
      "loss": 0.523,
      "num_input_tokens_seen": 19722304,
      "step": 30155
    },
    {
      "epoch": 15.80712788259958,
      "grad_norm": 0.1610206961631775,
      "learning_rate": 6.40211111191828e-06,
      "loss": 0.4939,
      "num_input_tokens_seen": 19725440,
      "step": 30160
    },
    {
      "epoch": 15.809748427672956,
      "grad_norm": 0.2516213059425354,
      "learning_rate": 6.394471826305409e-06,
      "loss": 0.5051,
      "num_input_tokens_seen": 19728864,
      "step": 30165
    },
    {
      "epoch": 15.81236897274633,
      "grad_norm": 0.3640362024307251,
      "learning_rate": 6.3868364327567795e-06,
      "loss": 0.4956,
      "num_input_tokens_seen": 19732192,
      "step": 30170
    },
    {
      "epoch": 15.814989517819706,
      "grad_norm": 0.1365659087896347,
      "learning_rate": 6.379204932869606e-06,
      "loss": 0.4837,
      "num_input_tokens_seen": 19735040,
      "step": 30175
    },
    {
      "epoch": 15.817610062893081,
      "grad_norm": 0.2497345209121704,
      "learning_rate": 6.371577328240327e-06,
      "loss": 0.4638,
      "num_input_tokens_seen": 19738368,
      "step": 30180
    },
    {
      "epoch": 15.820230607966456,
      "grad_norm": 0.2038443237543106,
      "learning_rate": 6.363953620464547e-06,
      "loss": 0.3787,
      "num_input_tokens_seen": 19742176,
      "step": 30185
    },
    {
      "epoch": 15.822851153039831,
      "grad_norm": 0.14877255260944366,
      "learning_rate": 6.356333811137064e-06,
      "loss": 0.4546,
      "num_input_tokens_seen": 19745760,
      "step": 30190
    },
    {
      "epoch": 15.825471698113208,
      "grad_norm": 0.12048286199569702,
      "learning_rate": 6.34871790185185e-06,
      "loss": 0.437,
      "num_input_tokens_seen": 19748832,
      "step": 30195
    },
    {
      "epoch": 15.828092243186584,
      "grad_norm": 0.17538239061832428,
      "learning_rate": 6.341105894202057e-06,
      "loss": 0.4715,
      "num_input_tokens_seen": 19752000,
      "step": 30200
    },
    {
      "epoch": 15.830712788259959,
      "grad_norm": 0.12777917087078094,
      "learning_rate": 6.333497789780041e-06,
      "loss": 0.5127,
      "num_input_tokens_seen": 19755744,
      "step": 30205
    },
    {
      "epoch": 15.833333333333334,
      "grad_norm": 0.16296647489070892,
      "learning_rate": 6.325893590177329e-06,
      "loss": 0.5183,
      "num_input_tokens_seen": 19759072,
      "step": 30210
    },
    {
      "epoch": 15.835953878406709,
      "grad_norm": 0.10622256994247437,
      "learning_rate": 6.318293296984631e-06,
      "loss": 0.5557,
      "num_input_tokens_seen": 19762432,
      "step": 30215
    },
    {
      "epoch": 15.838574423480084,
      "grad_norm": 0.18986976146697998,
      "learning_rate": 6.3106969117918495e-06,
      "loss": 0.4631,
      "num_input_tokens_seen": 19765248,
      "step": 30220
    },
    {
      "epoch": 15.84119496855346,
      "grad_norm": 0.1608695685863495,
      "learning_rate": 6.303104436188057e-06,
      "loss": 0.4598,
      "num_input_tokens_seen": 19768224,
      "step": 30225
    },
    {
      "epoch": 15.843815513626835,
      "grad_norm": 0.1394812911748886,
      "learning_rate": 6.2955158717615036e-06,
      "loss": 0.462,
      "num_input_tokens_seen": 19771232,
      "step": 30230
    },
    {
      "epoch": 15.84643605870021,
      "grad_norm": 0.14279593527317047,
      "learning_rate": 6.287931220099638e-06,
      "loss": 0.4672,
      "num_input_tokens_seen": 19774016,
      "step": 30235
    },
    {
      "epoch": 15.849056603773585,
      "grad_norm": 0.156614750623703,
      "learning_rate": 6.280350482789082e-06,
      "loss": 0.6261,
      "num_input_tokens_seen": 19777568,
      "step": 30240
    },
    {
      "epoch": 15.85167714884696,
      "grad_norm": 0.28495103120803833,
      "learning_rate": 6.272773661415645e-06,
      "loss": 0.4077,
      "num_input_tokens_seen": 19780448,
      "step": 30245
    },
    {
      "epoch": 15.854297693920335,
      "grad_norm": 0.20087410509586334,
      "learning_rate": 6.26520075756431e-06,
      "loss": 0.4057,
      "num_input_tokens_seen": 19783808,
      "step": 30250
    },
    {
      "epoch": 15.85691823899371,
      "grad_norm": 0.1360318809747696,
      "learning_rate": 6.2576317728192304e-06,
      "loss": 0.4281,
      "num_input_tokens_seen": 19787136,
      "step": 30255
    },
    {
      "epoch": 15.859538784067086,
      "grad_norm": 0.18718022108078003,
      "learning_rate": 6.250066708763761e-06,
      "loss": 0.5466,
      "num_input_tokens_seen": 19789920,
      "step": 30260
    },
    {
      "epoch": 15.86215932914046,
      "grad_norm": 0.12004192173480988,
      "learning_rate": 6.242505566980422e-06,
      "loss": 0.4955,
      "num_input_tokens_seen": 19793024,
      "step": 30265
    },
    {
      "epoch": 15.864779874213836,
      "grad_norm": 0.2806258797645569,
      "learning_rate": 6.234948349050931e-06,
      "loss": 0.487,
      "num_input_tokens_seen": 19796672,
      "step": 30270
    },
    {
      "epoch": 15.867400419287211,
      "grad_norm": 0.19650566577911377,
      "learning_rate": 6.227395056556162e-06,
      "loss": 0.4199,
      "num_input_tokens_seen": 19798944,
      "step": 30275
    },
    {
      "epoch": 15.870020964360586,
      "grad_norm": 0.6830571293830872,
      "learning_rate": 6.219845691076173e-06,
      "loss": 0.547,
      "num_input_tokens_seen": 19801664,
      "step": 30280
    },
    {
      "epoch": 15.872641509433961,
      "grad_norm": 0.15798227488994598,
      "learning_rate": 6.212300254190206e-06,
      "loss": 0.5884,
      "num_input_tokens_seen": 19804384,
      "step": 30285
    },
    {
      "epoch": 15.875262054507338,
      "grad_norm": 0.15200933814048767,
      "learning_rate": 6.204758747476688e-06,
      "loss": 0.5319,
      "num_input_tokens_seen": 19807744,
      "step": 30290
    },
    {
      "epoch": 15.877882599580714,
      "grad_norm": 0.15582150220870972,
      "learning_rate": 6.1972211725132095e-06,
      "loss": 0.3281,
      "num_input_tokens_seen": 19810336,
      "step": 30295
    },
    {
      "epoch": 15.880503144654089,
      "grad_norm": 0.12665611505508423,
      "learning_rate": 6.189687530876559e-06,
      "loss": 0.435,
      "num_input_tokens_seen": 19813440,
      "step": 30300
    },
    {
      "epoch": 15.883123689727464,
      "grad_norm": 0.1983945220708847,
      "learning_rate": 6.182157824142676e-06,
      "loss": 0.6135,
      "num_input_tokens_seen": 19816128,
      "step": 30305
    },
    {
      "epoch": 15.885744234800839,
      "grad_norm": 0.2188902348279953,
      "learning_rate": 6.174632053886681e-06,
      "loss": 0.3694,
      "num_input_tokens_seen": 19818656,
      "step": 30310
    },
    {
      "epoch": 15.888364779874214,
      "grad_norm": 0.1322951316833496,
      "learning_rate": 6.167110221682893e-06,
      "loss": 0.492,
      "num_input_tokens_seen": 19821856,
      "step": 30315
    },
    {
      "epoch": 15.89098532494759,
      "grad_norm": 0.1828390210866928,
      "learning_rate": 6.159592329104788e-06,
      "loss": 0.5032,
      "num_input_tokens_seen": 19824864,
      "step": 30320
    },
    {
      "epoch": 15.893605870020965,
      "grad_norm": 0.16351404786109924,
      "learning_rate": 6.152078377725032e-06,
      "loss": 0.3919,
      "num_input_tokens_seen": 19828352,
      "step": 30325
    },
    {
      "epoch": 15.89622641509434,
      "grad_norm": 0.1879964917898178,
      "learning_rate": 6.144568369115453e-06,
      "loss": 0.5473,
      "num_input_tokens_seen": 19830848,
      "step": 30330
    },
    {
      "epoch": 15.898846960167715,
      "grad_norm": 0.13012468814849854,
      "learning_rate": 6.137062304847046e-06,
      "loss": 0.4984,
      "num_input_tokens_seen": 19836448,
      "step": 30335
    },
    {
      "epoch": 15.90146750524109,
      "grad_norm": 0.1957559734582901,
      "learning_rate": 6.129560186490008e-06,
      "loss": 0.3661,
      "num_input_tokens_seen": 19839296,
      "step": 30340
    },
    {
      "epoch": 15.904088050314465,
      "grad_norm": 0.22911544144153595,
      "learning_rate": 6.122062015613694e-06,
      "loss": 0.4592,
      "num_input_tokens_seen": 19842176,
      "step": 30345
    },
    {
      "epoch": 15.90670859538784,
      "grad_norm": 0.16894659399986267,
      "learning_rate": 6.114567793786641e-06,
      "loss": 0.3889,
      "num_input_tokens_seen": 19845216,
      "step": 30350
    },
    {
      "epoch": 15.909329140461216,
      "grad_norm": 0.11898379772901535,
      "learning_rate": 6.107077522576543e-06,
      "loss": 0.5569,
      "num_input_tokens_seen": 19848064,
      "step": 30355
    },
    {
      "epoch": 15.91194968553459,
      "grad_norm": 0.1848626583814621,
      "learning_rate": 6.099591203550292e-06,
      "loss": 0.4791,
      "num_input_tokens_seen": 19850912,
      "step": 30360
    },
    {
      "epoch": 15.914570230607966,
      "grad_norm": 0.17918844521045685,
      "learning_rate": 6.092108838273927e-06,
      "loss": 0.517,
      "num_input_tokens_seen": 19854688,
      "step": 30365
    },
    {
      "epoch": 15.917190775681341,
      "grad_norm": 0.21305564045906067,
      "learning_rate": 6.08463042831268e-06,
      "loss": 0.5037,
      "num_input_tokens_seen": 19857792,
      "step": 30370
    },
    {
      "epoch": 15.919811320754716,
      "grad_norm": 0.1604437530040741,
      "learning_rate": 6.0771559752309496e-06,
      "loss": 0.4877,
      "num_input_tokens_seen": 19860672,
      "step": 30375
    },
    {
      "epoch": 15.922431865828091,
      "grad_norm": 0.10651437938213348,
      "learning_rate": 6.069685480592313e-06,
      "loss": 0.4514,
      "num_input_tokens_seen": 19863584,
      "step": 30380
    },
    {
      "epoch": 15.925052410901468,
      "grad_norm": 0.22554267942905426,
      "learning_rate": 6.062218945959497e-06,
      "loss": 0.5387,
      "num_input_tokens_seen": 19866912,
      "step": 30385
    },
    {
      "epoch": 15.927672955974844,
      "grad_norm": 0.15164193511009216,
      "learning_rate": 6.054756372894435e-06,
      "loss": 0.6624,
      "num_input_tokens_seen": 19869920,
      "step": 30390
    },
    {
      "epoch": 15.930293501048219,
      "grad_norm": 0.17674191296100616,
      "learning_rate": 6.04729776295819e-06,
      "loss": 0.6556,
      "num_input_tokens_seen": 19872736,
      "step": 30395
    },
    {
      "epoch": 15.932914046121594,
      "grad_norm": 0.17118103802204132,
      "learning_rate": 6.0398431177110306e-06,
      "loss": 0.5808,
      "num_input_tokens_seen": 19875584,
      "step": 30400
    },
    {
      "epoch": 15.935534591194969,
      "grad_norm": 0.15410920977592468,
      "learning_rate": 6.032392438712389e-06,
      "loss": 0.5067,
      "num_input_tokens_seen": 19879168,
      "step": 30405
    },
    {
      "epoch": 15.938155136268344,
      "grad_norm": 0.21616996824741364,
      "learning_rate": 6.024945727520847e-06,
      "loss": 0.6233,
      "num_input_tokens_seen": 19882528,
      "step": 30410
    },
    {
      "epoch": 15.94077568134172,
      "grad_norm": 0.18673373758792877,
      "learning_rate": 6.01750298569419e-06,
      "loss": 0.4457,
      "num_input_tokens_seen": 19885440,
      "step": 30415
    },
    {
      "epoch": 15.943396226415095,
      "grad_norm": 0.25913164019584656,
      "learning_rate": 6.010064214789335e-06,
      "loss": 0.395,
      "num_input_tokens_seen": 19888320,
      "step": 30420
    },
    {
      "epoch": 15.94601677148847,
      "grad_norm": 0.132792130112648,
      "learning_rate": 6.002629416362399e-06,
      "loss": 0.6121,
      "num_input_tokens_seen": 19891520,
      "step": 30425
    },
    {
      "epoch": 15.948637316561845,
      "grad_norm": 0.2045900523662567,
      "learning_rate": 5.995198591968662e-06,
      "loss": 0.3648,
      "num_input_tokens_seen": 19894688,
      "step": 30430
    },
    {
      "epoch": 15.95125786163522,
      "grad_norm": 0.3293905556201935,
      "learning_rate": 5.987771743162554e-06,
      "loss": 0.5123,
      "num_input_tokens_seen": 19897408,
      "step": 30435
    },
    {
      "epoch": 15.953878406708595,
      "grad_norm": 0.18924696743488312,
      "learning_rate": 5.9803488714976955e-06,
      "loss": 0.6041,
      "num_input_tokens_seen": 19901056,
      "step": 30440
    },
    {
      "epoch": 15.95649895178197,
      "grad_norm": 0.08356214314699173,
      "learning_rate": 5.972929978526872e-06,
      "loss": 0.3249,
      "num_input_tokens_seen": 19905056,
      "step": 30445
    },
    {
      "epoch": 15.959119496855346,
      "grad_norm": 0.12984247505664825,
      "learning_rate": 5.965515065802019e-06,
      "loss": 0.464,
      "num_input_tokens_seen": 19908160,
      "step": 30450
    },
    {
      "epoch": 15.96174004192872,
      "grad_norm": 0.17151780426502228,
      "learning_rate": 5.958104134874254e-06,
      "loss": 0.4342,
      "num_input_tokens_seen": 19911040,
      "step": 30455
    },
    {
      "epoch": 15.964360587002096,
      "grad_norm": 0.1415376514196396,
      "learning_rate": 5.950697187293872e-06,
      "loss": 0.4842,
      "num_input_tokens_seen": 19914720,
      "step": 30460
    },
    {
      "epoch": 15.966981132075471,
      "grad_norm": 0.22823522984981537,
      "learning_rate": 5.943294224610305e-06,
      "loss": 0.5445,
      "num_input_tokens_seen": 19917312,
      "step": 30465
    },
    {
      "epoch": 15.969601677148846,
      "grad_norm": 0.1595769226551056,
      "learning_rate": 5.935895248372175e-06,
      "loss": 0.5784,
      "num_input_tokens_seen": 19920480,
      "step": 30470
    },
    {
      "epoch": 15.972222222222221,
      "grad_norm": 0.2328779548406601,
      "learning_rate": 5.928500260127273e-06,
      "loss": 0.4303,
      "num_input_tokens_seen": 19923104,
      "step": 30475
    },
    {
      "epoch": 15.974842767295598,
      "grad_norm": 0.11722105741500854,
      "learning_rate": 5.921109261422531e-06,
      "loss": 0.5077,
      "num_input_tokens_seen": 19926688,
      "step": 30480
    },
    {
      "epoch": 15.977463312368974,
      "grad_norm": 0.1661914587020874,
      "learning_rate": 5.913722253804071e-06,
      "loss": 0.3742,
      "num_input_tokens_seen": 19930624,
      "step": 30485
    },
    {
      "epoch": 15.980083857442349,
      "grad_norm": 0.09552443027496338,
      "learning_rate": 5.906339238817163e-06,
      "loss": 0.5309,
      "num_input_tokens_seen": 19933984,
      "step": 30490
    },
    {
      "epoch": 15.982704402515724,
      "grad_norm": 0.13437482714653015,
      "learning_rate": 5.8989602180062554e-06,
      "loss": 0.4073,
      "num_input_tokens_seen": 19936288,
      "step": 30495
    },
    {
      "epoch": 15.985324947589099,
      "grad_norm": 0.21180607378482819,
      "learning_rate": 5.891585192914953e-06,
      "loss": 0.5133,
      "num_input_tokens_seen": 19939648,
      "step": 30500
    },
    {
      "epoch": 15.987945492662474,
      "grad_norm": 0.154703289270401,
      "learning_rate": 5.884214165086036e-06,
      "loss": 0.5356,
      "num_input_tokens_seen": 19942848,
      "step": 30505
    },
    {
      "epoch": 15.99056603773585,
      "grad_norm": 0.1754412055015564,
      "learning_rate": 5.876847136061428e-06,
      "loss": 0.54,
      "num_input_tokens_seen": 19946176,
      "step": 30510
    },
    {
      "epoch": 15.993186582809225,
      "grad_norm": 0.2024642676115036,
      "learning_rate": 5.869484107382228e-06,
      "loss": 0.3817,
      "num_input_tokens_seen": 19948800,
      "step": 30515
    },
    {
      "epoch": 15.9958071278826,
      "grad_norm": 0.33536970615386963,
      "learning_rate": 5.862125080588696e-06,
      "loss": 0.4828,
      "num_input_tokens_seen": 19952288,
      "step": 30520
    },
    {
      "epoch": 15.998427672955975,
      "grad_norm": 0.23186013102531433,
      "learning_rate": 5.854770057220263e-06,
      "loss": 0.4192,
      "num_input_tokens_seen": 19955424,
      "step": 30525
    },
    {
      "epoch": 16.0,
      "eval_loss": 0.49204808473587036,
      "eval_runtime": 10.1554,
      "eval_samples_per_second": 83.502,
      "eval_steps_per_second": 20.876,
      "num_input_tokens_seen": 19956544,
      "step": 30528
    },
    {
      "epoch": 16.00104821802935,
      "grad_norm": 0.2683218717575073,
      "learning_rate": 5.8474190388155216e-06,
      "loss": 0.5025,
      "num_input_tokens_seen": 19957376,
      "step": 30530
    },
    {
      "epoch": 16.003668763102727,
      "grad_norm": 0.13922420144081116,
      "learning_rate": 5.840072026912205e-06,
      "loss": 0.5614,
      "num_input_tokens_seen": 19960768,
      "step": 30535
    },
    {
      "epoch": 16.0062893081761,
      "grad_norm": 0.16094055771827698,
      "learning_rate": 5.83272902304724e-06,
      "loss": 0.3894,
      "num_input_tokens_seen": 19964416,
      "step": 30540
    },
    {
      "epoch": 16.008909853249477,
      "grad_norm": 0.20510277152061462,
      "learning_rate": 5.825390028756683e-06,
      "loss": 0.5285,
      "num_input_tokens_seen": 19967904,
      "step": 30545
    },
    {
      "epoch": 16.01153039832285,
      "grad_norm": 0.273510217666626,
      "learning_rate": 5.818055045575776e-06,
      "loss": 0.6519,
      "num_input_tokens_seen": 19971872,
      "step": 30550
    },
    {
      "epoch": 16.014150943396228,
      "grad_norm": 0.19354186952114105,
      "learning_rate": 5.810724075038917e-06,
      "loss": 0.3968,
      "num_input_tokens_seen": 19975072,
      "step": 30555
    },
    {
      "epoch": 16.0167714884696,
      "grad_norm": 0.5659545660018921,
      "learning_rate": 5.803397118679665e-06,
      "loss": 0.6371,
      "num_input_tokens_seen": 19978144,
      "step": 30560
    },
    {
      "epoch": 16.019392033542978,
      "grad_norm": 0.12665529549121857,
      "learning_rate": 5.796074178030727e-06,
      "loss": 0.4125,
      "num_input_tokens_seen": 19981216,
      "step": 30565
    },
    {
      "epoch": 16.02201257861635,
      "grad_norm": 0.18914470076560974,
      "learning_rate": 5.788755254623976e-06,
      "loss": 0.6402,
      "num_input_tokens_seen": 19984896,
      "step": 30570
    },
    {
      "epoch": 16.02463312368973,
      "grad_norm": 0.11435147374868393,
      "learning_rate": 5.7814403499904525e-06,
      "loss": 0.4513,
      "num_input_tokens_seen": 19988256,
      "step": 30575
    },
    {
      "epoch": 16.0272536687631,
      "grad_norm": 0.26278263330459595,
      "learning_rate": 5.774129465660352e-06,
      "loss": 0.4984,
      "num_input_tokens_seen": 19991744,
      "step": 30580
    },
    {
      "epoch": 16.02987421383648,
      "grad_norm": 0.14731957018375397,
      "learning_rate": 5.7668226031630246e-06,
      "loss": 0.4408,
      "num_input_tokens_seen": 19995104,
      "step": 30585
    },
    {
      "epoch": 16.032494758909852,
      "grad_norm": 0.19300997257232666,
      "learning_rate": 5.759519764026999e-06,
      "loss": 0.4718,
      "num_input_tokens_seen": 19997472,
      "step": 30590
    },
    {
      "epoch": 16.03511530398323,
      "grad_norm": 0.24072322249412537,
      "learning_rate": 5.7522209497799185e-06,
      "loss": 0.4679,
      "num_input_tokens_seen": 19999776,
      "step": 30595
    },
    {
      "epoch": 16.037735849056602,
      "grad_norm": 0.22348842024803162,
      "learning_rate": 5.7449261619486246e-06,
      "loss": 0.5561,
      "num_input_tokens_seen": 20002848,
      "step": 30600
    },
    {
      "epoch": 16.04035639412998,
      "grad_norm": 0.11935862898826599,
      "learning_rate": 5.737635402059105e-06,
      "loss": 0.3333,
      "num_input_tokens_seen": 20006176,
      "step": 30605
    },
    {
      "epoch": 16.042976939203353,
      "grad_norm": 0.2120298594236374,
      "learning_rate": 5.730348671636501e-06,
      "loss": 0.4336,
      "num_input_tokens_seen": 20009088,
      "step": 30610
    },
    {
      "epoch": 16.04559748427673,
      "grad_norm": 0.2168256789445877,
      "learning_rate": 5.723065972205124e-06,
      "loss": 0.3954,
      "num_input_tokens_seen": 20012000,
      "step": 30615
    },
    {
      "epoch": 16.048218029350103,
      "grad_norm": 0.17743003368377686,
      "learning_rate": 5.715787305288423e-06,
      "loss": 0.6344,
      "num_input_tokens_seen": 20016416,
      "step": 30620
    },
    {
      "epoch": 16.05083857442348,
      "grad_norm": 0.2003060132265091,
      "learning_rate": 5.7085126724090024e-06,
      "loss": 0.5018,
      "num_input_tokens_seen": 20020608,
      "step": 30625
    },
    {
      "epoch": 16.053459119496857,
      "grad_norm": 0.17378608882427216,
      "learning_rate": 5.7012420750886445e-06,
      "loss": 0.6044,
      "num_input_tokens_seen": 20023360,
      "step": 30630
    },
    {
      "epoch": 16.05607966457023,
      "grad_norm": 0.09953340888023376,
      "learning_rate": 5.693975514848271e-06,
      "loss": 0.4109,
      "num_input_tokens_seen": 20026240,
      "step": 30635
    },
    {
      "epoch": 16.058700209643607,
      "grad_norm": 0.13816653192043304,
      "learning_rate": 5.686712993207971e-06,
      "loss": 0.5591,
      "num_input_tokens_seen": 20029568,
      "step": 30640
    },
    {
      "epoch": 16.06132075471698,
      "grad_norm": 0.18468531966209412,
      "learning_rate": 5.679454511686977e-06,
      "loss": 0.6051,
      "num_input_tokens_seen": 20032288,
      "step": 30645
    },
    {
      "epoch": 16.063941299790358,
      "grad_norm": 0.15149998664855957,
      "learning_rate": 5.672200071803671e-06,
      "loss": 0.4188,
      "num_input_tokens_seen": 20035584,
      "step": 30650
    },
    {
      "epoch": 16.06656184486373,
      "grad_norm": 0.19047591090202332,
      "learning_rate": 5.664949675075607e-06,
      "loss": 0.3938,
      "num_input_tokens_seen": 20038912,
      "step": 30655
    },
    {
      "epoch": 16.069182389937108,
      "grad_norm": 0.26314789056777954,
      "learning_rate": 5.657703323019486e-06,
      "loss": 0.6093,
      "num_input_tokens_seen": 20042656,
      "step": 30660
    },
    {
      "epoch": 16.07180293501048,
      "grad_norm": 0.16485978662967682,
      "learning_rate": 5.65046101715116e-06,
      "loss": 0.5781,
      "num_input_tokens_seen": 20046528,
      "step": 30665
    },
    {
      "epoch": 16.07442348008386,
      "grad_norm": 0.2837487459182739,
      "learning_rate": 5.643222758985644e-06,
      "loss": 0.4165,
      "num_input_tokens_seen": 20049600,
      "step": 30670
    },
    {
      "epoch": 16.07704402515723,
      "grad_norm": 0.144491508603096,
      "learning_rate": 5.635988550037094e-06,
      "loss": 0.5464,
      "num_input_tokens_seen": 20052480,
      "step": 30675
    },
    {
      "epoch": 16.07966457023061,
      "grad_norm": 0.19284799695014954,
      "learning_rate": 5.628758391818817e-06,
      "loss": 0.5979,
      "num_input_tokens_seen": 20056128,
      "step": 30680
    },
    {
      "epoch": 16.082285115303982,
      "grad_norm": 0.10475711524486542,
      "learning_rate": 5.621532285843287e-06,
      "loss": 0.4424,
      "num_input_tokens_seen": 20059264,
      "step": 30685
    },
    {
      "epoch": 16.08490566037736,
      "grad_norm": 0.11802835762500763,
      "learning_rate": 5.614310233622119e-06,
      "loss": 0.4667,
      "num_input_tokens_seen": 20062112,
      "step": 30690
    },
    {
      "epoch": 16.087526205450732,
      "grad_norm": 0.19821317493915558,
      "learning_rate": 5.607092236666093e-06,
      "loss": 0.6156,
      "num_input_tokens_seen": 20064896,
      "step": 30695
    },
    {
      "epoch": 16.09014675052411,
      "grad_norm": 0.19314327836036682,
      "learning_rate": 5.599878296485125e-06,
      "loss": 0.4992,
      "num_input_tokens_seen": 20067776,
      "step": 30700
    },
    {
      "epoch": 16.092767295597483,
      "grad_norm": 0.16523733735084534,
      "learning_rate": 5.592668414588284e-06,
      "loss": 0.4906,
      "num_input_tokens_seen": 20071072,
      "step": 30705
    },
    {
      "epoch": 16.09538784067086,
      "grad_norm": 0.2053290456533432,
      "learning_rate": 5.585462592483801e-06,
      "loss": 0.4793,
      "num_input_tokens_seen": 20073920,
      "step": 30710
    },
    {
      "epoch": 16.098008385744233,
      "grad_norm": 0.12416775524616241,
      "learning_rate": 5.578260831679047e-06,
      "loss": 0.5483,
      "num_input_tokens_seen": 20077952,
      "step": 30715
    },
    {
      "epoch": 16.10062893081761,
      "grad_norm": 0.16837149858474731,
      "learning_rate": 5.571063133680562e-06,
      "loss": 0.568,
      "num_input_tokens_seen": 20081472,
      "step": 30720
    },
    {
      "epoch": 16.103249475890987,
      "grad_norm": 0.1127442941069603,
      "learning_rate": 5.563869499994004e-06,
      "loss": 0.4096,
      "num_input_tokens_seen": 20084128,
      "step": 30725
    },
    {
      "epoch": 16.10587002096436,
      "grad_norm": 0.16619881987571716,
      "learning_rate": 5.556679932124215e-06,
      "loss": 0.5316,
      "num_input_tokens_seen": 20087488,
      "step": 30730
    },
    {
      "epoch": 16.108490566037737,
      "grad_norm": 0.2159409075975418,
      "learning_rate": 5.549494431575159e-06,
      "loss": 0.5475,
      "num_input_tokens_seen": 20090656,
      "step": 30735
    },
    {
      "epoch": 16.11111111111111,
      "grad_norm": 0.29522302746772766,
      "learning_rate": 5.542312999849961e-06,
      "loss": 0.4619,
      "num_input_tokens_seen": 20093632,
      "step": 30740
    },
    {
      "epoch": 16.113731656184488,
      "grad_norm": 0.1353537142276764,
      "learning_rate": 5.5351356384509e-06,
      "loss": 0.3712,
      "num_input_tokens_seen": 20097312,
      "step": 30745
    },
    {
      "epoch": 16.11635220125786,
      "grad_norm": 0.15307089686393738,
      "learning_rate": 5.527962348879401e-06,
      "loss": 0.7093,
      "num_input_tokens_seen": 20100832,
      "step": 30750
    },
    {
      "epoch": 16.118972746331238,
      "grad_norm": 0.22695784270763397,
      "learning_rate": 5.520793132636027e-06,
      "loss": 0.5985,
      "num_input_tokens_seen": 20104800,
      "step": 30755
    },
    {
      "epoch": 16.12159329140461,
      "grad_norm": 0.09505252540111542,
      "learning_rate": 5.513627991220502e-06,
      "loss": 0.4423,
      "num_input_tokens_seen": 20108544,
      "step": 30760
    },
    {
      "epoch": 16.12421383647799,
      "grad_norm": 0.1239548921585083,
      "learning_rate": 5.506466926131684e-06,
      "loss": 0.2877,
      "num_input_tokens_seen": 20111680,
      "step": 30765
    },
    {
      "epoch": 16.12683438155136,
      "grad_norm": 0.1587217152118683,
      "learning_rate": 5.4993099388675916e-06,
      "loss": 0.374,
      "num_input_tokens_seen": 20115328,
      "step": 30770
    },
    {
      "epoch": 16.12945492662474,
      "grad_norm": 0.10950341075658798,
      "learning_rate": 5.49215703092539e-06,
      "loss": 0.3884,
      "num_input_tokens_seen": 20118752,
      "step": 30775
    },
    {
      "epoch": 16.132075471698112,
      "grad_norm": 0.17670193314552307,
      "learning_rate": 5.4850082038013755e-06,
      "loss": 0.3738,
      "num_input_tokens_seen": 20121504,
      "step": 30780
    },
    {
      "epoch": 16.13469601677149,
      "grad_norm": 0.11562714725732803,
      "learning_rate": 5.4778634589910115e-06,
      "loss": 0.5375,
      "num_input_tokens_seen": 20125440,
      "step": 30785
    },
    {
      "epoch": 16.137316561844862,
      "grad_norm": 0.17030911147594452,
      "learning_rate": 5.4707227979888834e-06,
      "loss": 0.5067,
      "num_input_tokens_seen": 20128288,
      "step": 30790
    },
    {
      "epoch": 16.13993710691824,
      "grad_norm": 0.17983952164649963,
      "learning_rate": 5.463586222288747e-06,
      "loss": 0.4896,
      "num_input_tokens_seen": 20131328,
      "step": 30795
    },
    {
      "epoch": 16.142557651991613,
      "grad_norm": 0.14458225667476654,
      "learning_rate": 5.456453733383493e-06,
      "loss": 0.298,
      "num_input_tokens_seen": 20135616,
      "step": 30800
    },
    {
      "epoch": 16.14517819706499,
      "grad_norm": 0.20088757574558258,
      "learning_rate": 5.449325332765151e-06,
      "loss": 0.4647,
      "num_input_tokens_seen": 20139040,
      "step": 30805
    },
    {
      "epoch": 16.147798742138363,
      "grad_norm": 0.15713676810264587,
      "learning_rate": 5.4422010219249e-06,
      "loss": 0.5339,
      "num_input_tokens_seen": 20141984,
      "step": 30810
    },
    {
      "epoch": 16.15041928721174,
      "grad_norm": 0.13114725053310394,
      "learning_rate": 5.435080802353077e-06,
      "loss": 0.4462,
      "num_input_tokens_seen": 20144448,
      "step": 30815
    },
    {
      "epoch": 16.153039832285117,
      "grad_norm": 0.19564968347549438,
      "learning_rate": 5.4279646755391325e-06,
      "loss": 0.4757,
      "num_input_tokens_seen": 20150048,
      "step": 30820
    },
    {
      "epoch": 16.15566037735849,
      "grad_norm": 0.1888575404882431,
      "learning_rate": 5.420852642971691e-06,
      "loss": 0.4322,
      "num_input_tokens_seen": 20153600,
      "step": 30825
    },
    {
      "epoch": 16.158280922431867,
      "grad_norm": 0.16748937964439392,
      "learning_rate": 5.41374470613851e-06,
      "loss": 0.4186,
      "num_input_tokens_seen": 20156448,
      "step": 30830
    },
    {
      "epoch": 16.16090146750524,
      "grad_norm": 0.30316585302352905,
      "learning_rate": 5.40664086652648e-06,
      "loss": 0.4762,
      "num_input_tokens_seen": 20160256,
      "step": 30835
    },
    {
      "epoch": 16.163522012578618,
      "grad_norm": 0.2547560930252075,
      "learning_rate": 5.399541125621646e-06,
      "loss": 0.5274,
      "num_input_tokens_seen": 20163168,
      "step": 30840
    },
    {
      "epoch": 16.16614255765199,
      "grad_norm": 0.15416987240314484,
      "learning_rate": 5.392445484909203e-06,
      "loss": 0.3768,
      "num_input_tokens_seen": 20166048,
      "step": 30845
    },
    {
      "epoch": 16.168763102725368,
      "grad_norm": 0.1525043547153473,
      "learning_rate": 5.385353945873464e-06,
      "loss": 0.4463,
      "num_input_tokens_seen": 20169632,
      "step": 30850
    },
    {
      "epoch": 16.17138364779874,
      "grad_norm": 0.13673362135887146,
      "learning_rate": 5.378266509997909e-06,
      "loss": 0.4936,
      "num_input_tokens_seen": 20173920,
      "step": 30855
    },
    {
      "epoch": 16.17400419287212,
      "grad_norm": 0.20077139139175415,
      "learning_rate": 5.371183178765135e-06,
      "loss": 0.605,
      "num_input_tokens_seen": 20176448,
      "step": 30860
    },
    {
      "epoch": 16.17662473794549,
      "grad_norm": 0.17038367688655853,
      "learning_rate": 5.364103953656904e-06,
      "loss": 0.5239,
      "num_input_tokens_seen": 20180448,
      "step": 30865
    },
    {
      "epoch": 16.17924528301887,
      "grad_norm": 0.1908324509859085,
      "learning_rate": 5.35702883615411e-06,
      "loss": 0.5013,
      "num_input_tokens_seen": 20183232,
      "step": 30870
    },
    {
      "epoch": 16.181865828092242,
      "grad_norm": 0.2230510115623474,
      "learning_rate": 5.34995782773679e-06,
      "loss": 0.4692,
      "num_input_tokens_seen": 20186240,
      "step": 30875
    },
    {
      "epoch": 16.18448637316562,
      "grad_norm": 0.13531377911567688,
      "learning_rate": 5.3428909298841115e-06,
      "loss": 0.401,
      "num_input_tokens_seen": 20189888,
      "step": 30880
    },
    {
      "epoch": 16.187106918238992,
      "grad_norm": 0.14107061922550201,
      "learning_rate": 5.335828144074384e-06,
      "loss": 0.4772,
      "num_input_tokens_seen": 20193440,
      "step": 30885
    },
    {
      "epoch": 16.18972746331237,
      "grad_norm": 0.12984877824783325,
      "learning_rate": 5.3287694717850695e-06,
      "loss": 0.4618,
      "num_input_tokens_seen": 20197088,
      "step": 30890
    },
    {
      "epoch": 16.192348008385743,
      "grad_norm": 0.2278796285390854,
      "learning_rate": 5.321714914492759e-06,
      "loss": 0.5034,
      "num_input_tokens_seen": 20200736,
      "step": 30895
    },
    {
      "epoch": 16.19496855345912,
      "grad_norm": 0.1548575758934021,
      "learning_rate": 5.314664473673195e-06,
      "loss": 0.4461,
      "num_input_tokens_seen": 20203712,
      "step": 30900
    },
    {
      "epoch": 16.197589098532493,
      "grad_norm": 0.16305506229400635,
      "learning_rate": 5.307618150801239e-06,
      "loss": 0.5299,
      "num_input_tokens_seen": 20206688,
      "step": 30905
    },
    {
      "epoch": 16.20020964360587,
      "grad_norm": 0.17001104354858398,
      "learning_rate": 5.300575947350897e-06,
      "loss": 0.511,
      "num_input_tokens_seen": 20210112,
      "step": 30910
    },
    {
      "epoch": 16.202830188679247,
      "grad_norm": 0.15005113184452057,
      "learning_rate": 5.293537864795325e-06,
      "loss": 0.5123,
      "num_input_tokens_seen": 20212768,
      "step": 30915
    },
    {
      "epoch": 16.20545073375262,
      "grad_norm": 0.14622832834720612,
      "learning_rate": 5.2865039046068075e-06,
      "loss": 0.5983,
      "num_input_tokens_seen": 20216928,
      "step": 30920
    },
    {
      "epoch": 16.208071278825997,
      "grad_norm": 0.15783840417861938,
      "learning_rate": 5.279474068256767e-06,
      "loss": 0.4164,
      "num_input_tokens_seen": 20219488,
      "step": 30925
    },
    {
      "epoch": 16.21069182389937,
      "grad_norm": 0.20035378634929657,
      "learning_rate": 5.272448357215779e-06,
      "loss": 0.3818,
      "num_input_tokens_seen": 20222592,
      "step": 30930
    },
    {
      "epoch": 16.213312368972748,
      "grad_norm": 0.13685102760791779,
      "learning_rate": 5.26542677295353e-06,
      "loss": 0.6089,
      "num_input_tokens_seen": 20227616,
      "step": 30935
    },
    {
      "epoch": 16.21593291404612,
      "grad_norm": 0.38537830114364624,
      "learning_rate": 5.258409316938848e-06,
      "loss": 0.5228,
      "num_input_tokens_seen": 20230240,
      "step": 30940
    },
    {
      "epoch": 16.218553459119498,
      "grad_norm": 0.30859631299972534,
      "learning_rate": 5.251395990639713e-06,
      "loss": 0.5281,
      "num_input_tokens_seen": 20233120,
      "step": 30945
    },
    {
      "epoch": 16.22117400419287,
      "grad_norm": 0.08872199803590775,
      "learning_rate": 5.244386795523235e-06,
      "loss": 0.5369,
      "num_input_tokens_seen": 20236512,
      "step": 30950
    },
    {
      "epoch": 16.22379454926625,
      "grad_norm": 0.12633918225765228,
      "learning_rate": 5.23738173305566e-06,
      "loss": 0.361,
      "num_input_tokens_seen": 20241024,
      "step": 30955
    },
    {
      "epoch": 16.22641509433962,
      "grad_norm": 0.16603565216064453,
      "learning_rate": 5.230380804702362e-06,
      "loss": 0.3818,
      "num_input_tokens_seen": 20244800,
      "step": 30960
    },
    {
      "epoch": 16.229035639413,
      "grad_norm": 0.11325641721487045,
      "learning_rate": 5.223384011927851e-06,
      "loss": 0.5187,
      "num_input_tokens_seen": 20249152,
      "step": 30965
    },
    {
      "epoch": 16.231656184486372,
      "grad_norm": 0.24154111742973328,
      "learning_rate": 5.216391356195779e-06,
      "loss": 0.5465,
      "num_input_tokens_seen": 20252512,
      "step": 30970
    },
    {
      "epoch": 16.23427672955975,
      "grad_norm": 0.2865656018257141,
      "learning_rate": 5.209402838968935e-06,
      "loss": 0.4373,
      "num_input_tokens_seen": 20256224,
      "step": 30975
    },
    {
      "epoch": 16.236897274633122,
      "grad_norm": 0.22666677832603455,
      "learning_rate": 5.202418461709233e-06,
      "loss": 0.6026,
      "num_input_tokens_seen": 20258656,
      "step": 30980
    },
    {
      "epoch": 16.2395178197065,
      "grad_norm": 0.12136593461036682,
      "learning_rate": 5.195438225877738e-06,
      "loss": 0.3752,
      "num_input_tokens_seen": 20262528,
      "step": 30985
    },
    {
      "epoch": 16.242138364779873,
      "grad_norm": 0.10904145985841751,
      "learning_rate": 5.18846213293461e-06,
      "loss": 0.4854,
      "num_input_tokens_seen": 20266176,
      "step": 30990
    },
    {
      "epoch": 16.24475890985325,
      "grad_norm": 0.22034132480621338,
      "learning_rate": 5.181490184339185e-06,
      "loss": 0.3945,
      "num_input_tokens_seen": 20269088,
      "step": 30995
    },
    {
      "epoch": 16.247379454926623,
      "grad_norm": 0.2999689280986786,
      "learning_rate": 5.17452238154991e-06,
      "loss": 0.4092,
      "num_input_tokens_seen": 20272384,
      "step": 31000
    },
    {
      "epoch": 16.25,
      "grad_norm": 0.1268121749162674,
      "learning_rate": 5.1675587260243715e-06,
      "loss": 0.4473,
      "num_input_tokens_seen": 20276416,
      "step": 31005
    },
    {
      "epoch": 16.252620545073377,
      "grad_norm": 0.22224339842796326,
      "learning_rate": 5.160599219219295e-06,
      "loss": 0.4523,
      "num_input_tokens_seen": 20279904,
      "step": 31010
    },
    {
      "epoch": 16.25524109014675,
      "grad_norm": 0.15229541063308716,
      "learning_rate": 5.1536438625905185e-06,
      "loss": 0.4529,
      "num_input_tokens_seen": 20283200,
      "step": 31015
    },
    {
      "epoch": 16.257861635220127,
      "grad_norm": 0.27778416872024536,
      "learning_rate": 5.146692657593025e-06,
      "loss": 0.4132,
      "num_input_tokens_seen": 20286336,
      "step": 31020
    },
    {
      "epoch": 16.2604821802935,
      "grad_norm": 0.2902076244354248,
      "learning_rate": 5.139745605680926e-06,
      "loss": 0.4867,
      "num_input_tokens_seen": 20289024,
      "step": 31025
    },
    {
      "epoch": 16.263102725366878,
      "grad_norm": 0.1701793670654297,
      "learning_rate": 5.132802708307469e-06,
      "loss": 0.5612,
      "num_input_tokens_seen": 20291968,
      "step": 31030
    },
    {
      "epoch": 16.26572327044025,
      "grad_norm": 0.1781141310930252,
      "learning_rate": 5.125863966925035e-06,
      "loss": 0.4563,
      "num_input_tokens_seen": 20295072,
      "step": 31035
    },
    {
      "epoch": 16.268343815513628,
      "grad_norm": 0.12785418331623077,
      "learning_rate": 5.118929382985116e-06,
      "loss": 0.3616,
      "num_input_tokens_seen": 20298112,
      "step": 31040
    },
    {
      "epoch": 16.270964360587,
      "grad_norm": 0.13184690475463867,
      "learning_rate": 5.111998957938363e-06,
      "loss": 0.5653,
      "num_input_tokens_seen": 20301664,
      "step": 31045
    },
    {
      "epoch": 16.27358490566038,
      "grad_norm": 0.1823980212211609,
      "learning_rate": 5.105072693234525e-06,
      "loss": 0.4792,
      "num_input_tokens_seen": 20304608,
      "step": 31050
    },
    {
      "epoch": 16.27620545073375,
      "grad_norm": 0.19748225808143616,
      "learning_rate": 5.098150590322506e-06,
      "loss": 0.5044,
      "num_input_tokens_seen": 20307840,
      "step": 31055
    },
    {
      "epoch": 16.27882599580713,
      "grad_norm": 0.1455896496772766,
      "learning_rate": 5.091232650650335e-06,
      "loss": 0.4055,
      "num_input_tokens_seen": 20310784,
      "step": 31060
    },
    {
      "epoch": 16.281446540880502,
      "grad_norm": 0.11945386230945587,
      "learning_rate": 5.0843188756651645e-06,
      "loss": 0.3062,
      "num_input_tokens_seen": 20313344,
      "step": 31065
    },
    {
      "epoch": 16.28406708595388,
      "grad_norm": 0.26799991726875305,
      "learning_rate": 5.077409266813276e-06,
      "loss": 0.4257,
      "num_input_tokens_seen": 20316256,
      "step": 31070
    },
    {
      "epoch": 16.286687631027252,
      "grad_norm": 0.14906395971775055,
      "learning_rate": 5.0705038255400765e-06,
      "loss": 0.4601,
      "num_input_tokens_seen": 20320672,
      "step": 31075
    },
    {
      "epoch": 16.28930817610063,
      "grad_norm": 0.07940781116485596,
      "learning_rate": 5.063602553290106e-06,
      "loss": 0.4621,
      "num_input_tokens_seen": 20324128,
      "step": 31080
    },
    {
      "epoch": 16.291928721174003,
      "grad_norm": 0.18210609257221222,
      "learning_rate": 5.056705451507035e-06,
      "loss": 0.4709,
      "num_input_tokens_seen": 20331520,
      "step": 31085
    },
    {
      "epoch": 16.29454926624738,
      "grad_norm": 0.14745844900608063,
      "learning_rate": 5.049812521633668e-06,
      "loss": 0.4356,
      "num_input_tokens_seen": 20334560,
      "step": 31090
    },
    {
      "epoch": 16.297169811320753,
      "grad_norm": 0.17210061848163605,
      "learning_rate": 5.042923765111907e-06,
      "loss": 0.493,
      "num_input_tokens_seen": 20337856,
      "step": 31095
    },
    {
      "epoch": 16.29979035639413,
      "grad_norm": 0.13362734019756317,
      "learning_rate": 5.036039183382821e-06,
      "loss": 0.3759,
      "num_input_tokens_seen": 20340224,
      "step": 31100
    },
    {
      "epoch": 16.302410901467507,
      "grad_norm": 0.18616607785224915,
      "learning_rate": 5.02915877788657e-06,
      "loss": 0.485,
      "num_input_tokens_seen": 20343968,
      "step": 31105
    },
    {
      "epoch": 16.30503144654088,
      "grad_norm": 0.11701547354459763,
      "learning_rate": 5.0222825500624625e-06,
      "loss": 0.4304,
      "num_input_tokens_seen": 20348096,
      "step": 31110
    },
    {
      "epoch": 16.307651991614257,
      "grad_norm": 0.2039836347103119,
      "learning_rate": 5.015410501348935e-06,
      "loss": 0.5593,
      "num_input_tokens_seen": 20351072,
      "step": 31115
    },
    {
      "epoch": 16.31027253668763,
      "grad_norm": 0.12456894665956497,
      "learning_rate": 5.008542633183527e-06,
      "loss": 0.4561,
      "num_input_tokens_seen": 20354528,
      "step": 31120
    },
    {
      "epoch": 16.312893081761008,
      "grad_norm": 0.1438521444797516,
      "learning_rate": 5.001678947002927e-06,
      "loss": 0.4534,
      "num_input_tokens_seen": 20357664,
      "step": 31125
    },
    {
      "epoch": 16.31551362683438,
      "grad_norm": 0.21277368068695068,
      "learning_rate": 4.994819444242943e-06,
      "loss": 0.4306,
      "num_input_tokens_seen": 20361056,
      "step": 31130
    },
    {
      "epoch": 16.318134171907758,
      "grad_norm": 0.18329712748527527,
      "learning_rate": 4.9879641263384956e-06,
      "loss": 0.3597,
      "num_input_tokens_seen": 20364160,
      "step": 31135
    },
    {
      "epoch": 16.32075471698113,
      "grad_norm": 0.15200626850128174,
      "learning_rate": 4.981112994723644e-06,
      "loss": 0.5747,
      "num_input_tokens_seen": 20367072,
      "step": 31140
    },
    {
      "epoch": 16.32337526205451,
      "grad_norm": 0.20108042657375336,
      "learning_rate": 4.974266050831572e-06,
      "loss": 0.4428,
      "num_input_tokens_seen": 20369600,
      "step": 31145
    },
    {
      "epoch": 16.32599580712788,
      "grad_norm": 0.12778912484645844,
      "learning_rate": 4.967423296094567e-06,
      "loss": 0.4686,
      "num_input_tokens_seen": 20372768,
      "step": 31150
    },
    {
      "epoch": 16.32861635220126,
      "grad_norm": 0.18089544773101807,
      "learning_rate": 4.960584731944076e-06,
      "loss": 0.5352,
      "num_input_tokens_seen": 20375680,
      "step": 31155
    },
    {
      "epoch": 16.331236897274632,
      "grad_norm": 0.16626951098442078,
      "learning_rate": 4.953750359810627e-06,
      "loss": 0.4506,
      "num_input_tokens_seen": 20378272,
      "step": 31160
    },
    {
      "epoch": 16.33385744234801,
      "grad_norm": 0.09807069599628448,
      "learning_rate": 4.946920181123904e-06,
      "loss": 0.459,
      "num_input_tokens_seen": 20381888,
      "step": 31165
    },
    {
      "epoch": 16.336477987421382,
      "grad_norm": 0.13336235284805298,
      "learning_rate": 4.940094197312709e-06,
      "loss": 0.324,
      "num_input_tokens_seen": 20385856,
      "step": 31170
    },
    {
      "epoch": 16.33909853249476,
      "grad_norm": 0.2508552372455597,
      "learning_rate": 4.933272409804946e-06,
      "loss": 0.5181,
      "num_input_tokens_seen": 20389120,
      "step": 31175
    },
    {
      "epoch": 16.341719077568133,
      "grad_norm": 0.25172901153564453,
      "learning_rate": 4.926454820027663e-06,
      "loss": 0.4473,
      "num_input_tokens_seen": 20392672,
      "step": 31180
    },
    {
      "epoch": 16.34433962264151,
      "grad_norm": 0.17909644544124603,
      "learning_rate": 4.919641429407024e-06,
      "loss": 0.4522,
      "num_input_tokens_seen": 20395232,
      "step": 31185
    },
    {
      "epoch": 16.346960167714883,
      "grad_norm": 0.13451629877090454,
      "learning_rate": 4.912832239368304e-06,
      "loss": 0.4224,
      "num_input_tokens_seen": 20398368,
      "step": 31190
    },
    {
      "epoch": 16.34958071278826,
      "grad_norm": 0.24174436926841736,
      "learning_rate": 4.906027251335918e-06,
      "loss": 0.3982,
      "num_input_tokens_seen": 20400768,
      "step": 31195
    },
    {
      "epoch": 16.352201257861637,
      "grad_norm": 0.11524783074855804,
      "learning_rate": 4.899226466733381e-06,
      "loss": 0.5457,
      "num_input_tokens_seen": 20404448,
      "step": 31200
    },
    {
      "epoch": 16.35482180293501,
      "grad_norm": 0.1720929592847824,
      "learning_rate": 4.892429886983346e-06,
      "loss": 0.5431,
      "num_input_tokens_seen": 20407264,
      "step": 31205
    },
    {
      "epoch": 16.357442348008387,
      "grad_norm": 0.1382119059562683,
      "learning_rate": 4.88563751350758e-06,
      "loss": 0.3506,
      "num_input_tokens_seen": 20410272,
      "step": 31210
    },
    {
      "epoch": 16.36006289308176,
      "grad_norm": 0.09938270598649979,
      "learning_rate": 4.878849347726977e-06,
      "loss": 0.5168,
      "num_input_tokens_seen": 20413280,
      "step": 31215
    },
    {
      "epoch": 16.362683438155138,
      "grad_norm": 0.17383794486522675,
      "learning_rate": 4.872065391061526e-06,
      "loss": 0.5208,
      "num_input_tokens_seen": 20415776,
      "step": 31220
    },
    {
      "epoch": 16.36530398322851,
      "grad_norm": 0.3440277874469757,
      "learning_rate": 4.865285644930373e-06,
      "loss": 0.5386,
      "num_input_tokens_seen": 20418944,
      "step": 31225
    },
    {
      "epoch": 16.367924528301888,
      "grad_norm": 0.17158138751983643,
      "learning_rate": 4.858510110751746e-06,
      "loss": 0.6484,
      "num_input_tokens_seen": 20422112,
      "step": 31230
    },
    {
      "epoch": 16.37054507337526,
      "grad_norm": 0.18245640397071838,
      "learning_rate": 4.851738789943019e-06,
      "loss": 0.3969,
      "num_input_tokens_seen": 20424736,
      "step": 31235
    },
    {
      "epoch": 16.37316561844864,
      "grad_norm": 0.20575344562530518,
      "learning_rate": 4.844971683920674e-06,
      "loss": 0.4492,
      "num_input_tokens_seen": 20428288,
      "step": 31240
    },
    {
      "epoch": 16.37578616352201,
      "grad_norm": 0.14994287490844727,
      "learning_rate": 4.838208794100316e-06,
      "loss": 0.5032,
      "num_input_tokens_seen": 20432320,
      "step": 31245
    },
    {
      "epoch": 16.37840670859539,
      "grad_norm": 0.2638530433177948,
      "learning_rate": 4.831450121896661e-06,
      "loss": 0.4248,
      "num_input_tokens_seen": 20434976,
      "step": 31250
    },
    {
      "epoch": 16.381027253668762,
      "grad_norm": 0.19149982929229736,
      "learning_rate": 4.824695668723542e-06,
      "loss": 0.4831,
      "num_input_tokens_seen": 20437664,
      "step": 31255
    },
    {
      "epoch": 16.38364779874214,
      "grad_norm": 0.12166422605514526,
      "learning_rate": 4.817945435993915e-06,
      "loss": 0.4792,
      "num_input_tokens_seen": 20440896,
      "step": 31260
    },
    {
      "epoch": 16.386268343815512,
      "grad_norm": 0.125807985663414,
      "learning_rate": 4.811199425119853e-06,
      "loss": 0.4135,
      "num_input_tokens_seen": 20444064,
      "step": 31265
    },
    {
      "epoch": 16.38888888888889,
      "grad_norm": 0.2753680944442749,
      "learning_rate": 4.804457637512552e-06,
      "loss": 0.4479,
      "num_input_tokens_seen": 20446944,
      "step": 31270
    },
    {
      "epoch": 16.391509433962263,
      "grad_norm": 0.17681214213371277,
      "learning_rate": 4.79772007458231e-06,
      "loss": 0.5108,
      "num_input_tokens_seen": 20449888,
      "step": 31275
    },
    {
      "epoch": 16.39412997903564,
      "grad_norm": 0.2124709039926529,
      "learning_rate": 4.7909867377385424e-06,
      "loss": 0.4604,
      "num_input_tokens_seen": 20453824,
      "step": 31280
    },
    {
      "epoch": 16.396750524109013,
      "grad_norm": 0.16333742439746857,
      "learning_rate": 4.784257628389794e-06,
      "loss": 0.6441,
      "num_input_tokens_seen": 20457184,
      "step": 31285
    },
    {
      "epoch": 16.39937106918239,
      "grad_norm": 0.15342013537883759,
      "learning_rate": 4.7775327479437166e-06,
      "loss": 0.4839,
      "num_input_tokens_seen": 20460864,
      "step": 31290
    },
    {
      "epoch": 16.401991614255767,
      "grad_norm": 0.12769146263599396,
      "learning_rate": 4.770812097807076e-06,
      "loss": 0.4023,
      "num_input_tokens_seen": 20463392,
      "step": 31295
    },
    {
      "epoch": 16.40461215932914,
      "grad_norm": 0.2528645396232605,
      "learning_rate": 4.764095679385766e-06,
      "loss": 0.438,
      "num_input_tokens_seen": 20466432,
      "step": 31300
    },
    {
      "epoch": 16.407232704402517,
      "grad_norm": 0.1343441903591156,
      "learning_rate": 4.7573834940847745e-06,
      "loss": 0.5223,
      "num_input_tokens_seen": 20469312,
      "step": 31305
    },
    {
      "epoch": 16.40985324947589,
      "grad_norm": 0.16256208717823029,
      "learning_rate": 4.750675543308212e-06,
      "loss": 0.5628,
      "num_input_tokens_seen": 20471936,
      "step": 31310
    },
    {
      "epoch": 16.412473794549268,
      "grad_norm": 0.13244514167308807,
      "learning_rate": 4.743971828459307e-06,
      "loss": 0.314,
      "num_input_tokens_seen": 20474720,
      "step": 31315
    },
    {
      "epoch": 16.41509433962264,
      "grad_norm": 0.16257750988006592,
      "learning_rate": 4.7372723509404025e-06,
      "loss": 0.5866,
      "num_input_tokens_seen": 20478784,
      "step": 31320
    },
    {
      "epoch": 16.417714884696018,
      "grad_norm": 0.1870707869529724,
      "learning_rate": 4.7305771121529615e-06,
      "loss": 0.3766,
      "num_input_tokens_seen": 20481824,
      "step": 31325
    },
    {
      "epoch": 16.42033542976939,
      "grad_norm": 0.2710503935813904,
      "learning_rate": 4.723886113497542e-06,
      "loss": 0.4001,
      "num_input_tokens_seen": 20484704,
      "step": 31330
    },
    {
      "epoch": 16.42295597484277,
      "grad_norm": 0.23390744626522064,
      "learning_rate": 4.7171993563738195e-06,
      "loss": 0.5553,
      "num_input_tokens_seen": 20487936,
      "step": 31335
    },
    {
      "epoch": 16.42557651991614,
      "grad_norm": 0.3391329348087311,
      "learning_rate": 4.7105168421805924e-06,
      "loss": 0.5616,
      "num_input_tokens_seen": 20491264,
      "step": 31340
    },
    {
      "epoch": 16.42819706498952,
      "grad_norm": 0.20518779754638672,
      "learning_rate": 4.703838572315767e-06,
      "loss": 0.4846,
      "num_input_tokens_seen": 20494112,
      "step": 31345
    },
    {
      "epoch": 16.430817610062892,
      "grad_norm": 0.2092701941728592,
      "learning_rate": 4.697164548176367e-06,
      "loss": 0.4725,
      "num_input_tokens_seen": 20497184,
      "step": 31350
    },
    {
      "epoch": 16.43343815513627,
      "grad_norm": 0.17771029472351074,
      "learning_rate": 4.690494771158519e-06,
      "loss": 0.3858,
      "num_input_tokens_seen": 20500864,
      "step": 31355
    },
    {
      "epoch": 16.436058700209642,
      "grad_norm": 0.2117460072040558,
      "learning_rate": 4.683829242657453e-06,
      "loss": 0.4639,
      "num_input_tokens_seen": 20504928,
      "step": 31360
    },
    {
      "epoch": 16.43867924528302,
      "grad_norm": 0.2381017655134201,
      "learning_rate": 4.677167964067533e-06,
      "loss": 0.5359,
      "num_input_tokens_seen": 20507776,
      "step": 31365
    },
    {
      "epoch": 16.441299790356393,
      "grad_norm": 0.17893236875534058,
      "learning_rate": 4.670510936782216e-06,
      "loss": 0.5388,
      "num_input_tokens_seen": 20510944,
      "step": 31370
    },
    {
      "epoch": 16.44392033542977,
      "grad_norm": 0.16346445679664612,
      "learning_rate": 4.6638581621940815e-06,
      "loss": 0.4719,
      "num_input_tokens_seen": 20514784,
      "step": 31375
    },
    {
      "epoch": 16.446540880503143,
      "grad_norm": 0.1872468888759613,
      "learning_rate": 4.657209641694821e-06,
      "loss": 0.4142,
      "num_input_tokens_seen": 20517248,
      "step": 31380
    },
    {
      "epoch": 16.44916142557652,
      "grad_norm": 0.16990068554878235,
      "learning_rate": 4.65056537667522e-06,
      "loss": 0.4573,
      "num_input_tokens_seen": 20520512,
      "step": 31385
    },
    {
      "epoch": 16.451781970649897,
      "grad_norm": 0.16711503267288208,
      "learning_rate": 4.643925368525176e-06,
      "loss": 0.4841,
      "num_input_tokens_seen": 20523328,
      "step": 31390
    },
    {
      "epoch": 16.45440251572327,
      "grad_norm": 0.1481838971376419,
      "learning_rate": 4.637289618633711e-06,
      "loss": 0.4057,
      "num_input_tokens_seen": 20527296,
      "step": 31395
    },
    {
      "epoch": 16.457023060796647,
      "grad_norm": 0.26454243063926697,
      "learning_rate": 4.630658128388948e-06,
      "loss": 0.6071,
      "num_input_tokens_seen": 20529952,
      "step": 31400
    },
    {
      "epoch": 16.45964360587002,
      "grad_norm": 0.14823377132415771,
      "learning_rate": 4.624030899178125e-06,
      "loss": 0.5386,
      "num_input_tokens_seen": 20533760,
      "step": 31405
    },
    {
      "epoch": 16.462264150943398,
      "grad_norm": 0.205144464969635,
      "learning_rate": 4.617407932387569e-06,
      "loss": 0.4237,
      "num_input_tokens_seen": 20536384,
      "step": 31410
    },
    {
      "epoch": 16.46488469601677,
      "grad_norm": 0.13343064486980438,
      "learning_rate": 4.610789229402743e-06,
      "loss": 0.2827,
      "num_input_tokens_seen": 20539744,
      "step": 31415
    },
    {
      "epoch": 16.467505241090148,
      "grad_norm": 0.15428583323955536,
      "learning_rate": 4.604174791608193e-06,
      "loss": 0.4416,
      "num_input_tokens_seen": 20543904,
      "step": 31420
    },
    {
      "epoch": 16.47012578616352,
      "grad_norm": 0.1262700855731964,
      "learning_rate": 4.597564620387587e-06,
      "loss": 0.3544,
      "num_input_tokens_seen": 20547552,
      "step": 31425
    },
    {
      "epoch": 16.4727463312369,
      "grad_norm": 0.16054180264472961,
      "learning_rate": 4.590958717123708e-06,
      "loss": 0.3802,
      "num_input_tokens_seen": 20549984,
      "step": 31430
    },
    {
      "epoch": 16.47536687631027,
      "grad_norm": 0.15882708132266998,
      "learning_rate": 4.5843570831984174e-06,
      "loss": 0.4744,
      "num_input_tokens_seen": 20553280,
      "step": 31435
    },
    {
      "epoch": 16.47798742138365,
      "grad_norm": 0.10128266364336014,
      "learning_rate": 4.577759719992719e-06,
      "loss": 0.5225,
      "num_input_tokens_seen": 20557376,
      "step": 31440
    },
    {
      "epoch": 16.480607966457022,
      "grad_norm": 0.18642130494117737,
      "learning_rate": 4.571166628886692e-06,
      "loss": 0.5726,
      "num_input_tokens_seen": 20560288,
      "step": 31445
    },
    {
      "epoch": 16.4832285115304,
      "grad_norm": 0.09024818241596222,
      "learning_rate": 4.564577811259541e-06,
      "loss": 0.4991,
      "num_input_tokens_seen": 20563392,
      "step": 31450
    },
    {
      "epoch": 16.485849056603772,
      "grad_norm": 0.16696497797966003,
      "learning_rate": 4.557993268489574e-06,
      "loss": 0.5251,
      "num_input_tokens_seen": 20566272,
      "step": 31455
    },
    {
      "epoch": 16.48846960167715,
      "grad_norm": 0.17198172211647034,
      "learning_rate": 4.551413001954205e-06,
      "loss": 0.4666,
      "num_input_tokens_seen": 20568672,
      "step": 31460
    },
    {
      "epoch": 16.491090146750523,
      "grad_norm": 0.153442844748497,
      "learning_rate": 4.54483701302994e-06,
      "loss": 0.5571,
      "num_input_tokens_seen": 20572096,
      "step": 31465
    },
    {
      "epoch": 16.4937106918239,
      "grad_norm": 0.126547709107399,
      "learning_rate": 4.538265303092415e-06,
      "loss": 0.387,
      "num_input_tokens_seen": 20575008,
      "step": 31470
    },
    {
      "epoch": 16.496331236897273,
      "grad_norm": 0.17506496608257294,
      "learning_rate": 4.531697873516338e-06,
      "loss": 0.6184,
      "num_input_tokens_seen": 20578304,
      "step": 31475
    },
    {
      "epoch": 16.49895178197065,
      "grad_norm": 0.21043509244918823,
      "learning_rate": 4.525134725675556e-06,
      "loss": 0.5424,
      "num_input_tokens_seen": 20581248,
      "step": 31480
    },
    {
      "epoch": 16.501572327044027,
      "grad_norm": 0.18618065118789673,
      "learning_rate": 4.518575860943003e-06,
      "loss": 0.4365,
      "num_input_tokens_seen": 20584032,
      "step": 31485
    },
    {
      "epoch": 16.5041928721174,
      "grad_norm": 0.14734593033790588,
      "learning_rate": 4.512021280690712e-06,
      "loss": 0.5437,
      "num_input_tokens_seen": 20587872,
      "step": 31490
    },
    {
      "epoch": 16.506813417190777,
      "grad_norm": 0.1721504181623459,
      "learning_rate": 4.505470986289828e-06,
      "loss": 0.458,
      "num_input_tokens_seen": 20590880,
      "step": 31495
    },
    {
      "epoch": 16.50943396226415,
      "grad_norm": 0.25660407543182373,
      "learning_rate": 4.498924979110605e-06,
      "loss": 0.338,
      "num_input_tokens_seen": 20593792,
      "step": 31500
    },
    {
      "epoch": 16.512054507337528,
      "grad_norm": 0.13167217373847961,
      "learning_rate": 4.492383260522384e-06,
      "loss": 0.4259,
      "num_input_tokens_seen": 20596992,
      "step": 31505
    },
    {
      "epoch": 16.5146750524109,
      "grad_norm": 0.1947692185640335,
      "learning_rate": 4.485845831893626e-06,
      "loss": 0.5403,
      "num_input_tokens_seen": 20599808,
      "step": 31510
    },
    {
      "epoch": 16.517295597484278,
      "grad_norm": 0.11880382150411606,
      "learning_rate": 4.479312694591878e-06,
      "loss": 0.5159,
      "num_input_tokens_seen": 20603136,
      "step": 31515
    },
    {
      "epoch": 16.51991614255765,
      "grad_norm": 0.17827121913433075,
      "learning_rate": 4.4727838499838e-06,
      "loss": 0.4283,
      "num_input_tokens_seen": 20605792,
      "step": 31520
    },
    {
      "epoch": 16.52253668763103,
      "grad_norm": 0.28028106689453125,
      "learning_rate": 4.4662592994351545e-06,
      "loss": 0.5673,
      "num_input_tokens_seen": 20608928,
      "step": 31525
    },
    {
      "epoch": 16.5251572327044,
      "grad_norm": 0.18706290423870087,
      "learning_rate": 4.4597390443108065e-06,
      "loss": 0.5108,
      "num_input_tokens_seen": 20611680,
      "step": 31530
    },
    {
      "epoch": 16.52777777777778,
      "grad_norm": 0.18740776181221008,
      "learning_rate": 4.453223085974712e-06,
      "loss": 0.3592,
      "num_input_tokens_seen": 20614528,
      "step": 31535
    },
    {
      "epoch": 16.530398322851152,
      "grad_norm": 0.1679849624633789,
      "learning_rate": 4.4467114257899425e-06,
      "loss": 0.4685,
      "num_input_tokens_seen": 20617984,
      "step": 31540
    },
    {
      "epoch": 16.53301886792453,
      "grad_norm": 0.14729978144168854,
      "learning_rate": 4.44020406511865e-06,
      "loss": 0.3618,
      "num_input_tokens_seen": 20621952,
      "step": 31545
    },
    {
      "epoch": 16.535639412997902,
      "grad_norm": 0.3755266070365906,
      "learning_rate": 4.43370100532211e-06,
      "loss": 0.5899,
      "num_input_tokens_seen": 20624416,
      "step": 31550
    },
    {
      "epoch": 16.53825995807128,
      "grad_norm": 0.11567986756563187,
      "learning_rate": 4.427202247760695e-06,
      "loss": 0.3678,
      "num_input_tokens_seen": 20627456,
      "step": 31555
    },
    {
      "epoch": 16.540880503144653,
      "grad_norm": 0.16614292562007904,
      "learning_rate": 4.420707793793854e-06,
      "loss": 0.4411,
      "num_input_tokens_seen": 20630688,
      "step": 31560
    },
    {
      "epoch": 16.54350104821803,
      "grad_norm": 0.14071232080459595,
      "learning_rate": 4.414217644780172e-06,
      "loss": 0.5224,
      "num_input_tokens_seen": 20635552,
      "step": 31565
    },
    {
      "epoch": 16.546121593291403,
      "grad_norm": 0.4179965853691101,
      "learning_rate": 4.407731802077294e-06,
      "loss": 0.5296,
      "num_input_tokens_seen": 20638272,
      "step": 31570
    },
    {
      "epoch": 16.54874213836478,
      "grad_norm": 0.18404394388198853,
      "learning_rate": 4.401250267041996e-06,
      "loss": 0.568,
      "num_input_tokens_seen": 20641536,
      "step": 31575
    },
    {
      "epoch": 16.551362683438157,
      "grad_norm": 0.18990884721279144,
      "learning_rate": 4.394773041030137e-06,
      "loss": 0.4034,
      "num_input_tokens_seen": 20644864,
      "step": 31580
    },
    {
      "epoch": 16.55398322851153,
      "grad_norm": 0.14606569707393646,
      "learning_rate": 4.388300125396685e-06,
      "loss": 0.5473,
      "num_input_tokens_seen": 20648224,
      "step": 31585
    },
    {
      "epoch": 16.556603773584907,
      "grad_norm": 0.15587425231933594,
      "learning_rate": 4.381831521495699e-06,
      "loss": 0.4755,
      "num_input_tokens_seen": 20652800,
      "step": 31590
    },
    {
      "epoch": 16.55922431865828,
      "grad_norm": 0.1741257607936859,
      "learning_rate": 4.37536723068033e-06,
      "loss": 0.5795,
      "num_input_tokens_seen": 20655776,
      "step": 31595
    },
    {
      "epoch": 16.561844863731658,
      "grad_norm": 0.14885619282722473,
      "learning_rate": 4.368907254302837e-06,
      "loss": 0.6059,
      "num_input_tokens_seen": 20659040,
      "step": 31600
    },
    {
      "epoch": 16.56446540880503,
      "grad_norm": 0.15428099036216736,
      "learning_rate": 4.3624515937145725e-06,
      "loss": 0.4362,
      "num_input_tokens_seen": 20662560,
      "step": 31605
    },
    {
      "epoch": 16.567085953878408,
      "grad_norm": 0.17484068870544434,
      "learning_rate": 4.3560002502659905e-06,
      "loss": 0.3512,
      "num_input_tokens_seen": 20664992,
      "step": 31610
    },
    {
      "epoch": 16.56970649895178,
      "grad_norm": 0.12777113914489746,
      "learning_rate": 4.349553225306644e-06,
      "loss": 0.4868,
      "num_input_tokens_seen": 20667488,
      "step": 31615
    },
    {
      "epoch": 16.572327044025158,
      "grad_norm": 0.32502633333206177,
      "learning_rate": 4.343110520185167e-06,
      "loss": 0.5309,
      "num_input_tokens_seen": 20670304,
      "step": 31620
    },
    {
      "epoch": 16.57494758909853,
      "grad_norm": 0.21586184203624725,
      "learning_rate": 4.336672136249298e-06,
      "loss": 0.4406,
      "num_input_tokens_seen": 20674080,
      "step": 31625
    },
    {
      "epoch": 16.57756813417191,
      "grad_norm": 0.184454083442688,
      "learning_rate": 4.33023807484588e-06,
      "loss": 0.4142,
      "num_input_tokens_seen": 20676928,
      "step": 31630
    },
    {
      "epoch": 16.580188679245282,
      "grad_norm": 0.1469159573316574,
      "learning_rate": 4.323808337320842e-06,
      "loss": 0.4036,
      "num_input_tokens_seen": 20680096,
      "step": 31635
    },
    {
      "epoch": 16.58280922431866,
      "grad_norm": 0.16712909936904907,
      "learning_rate": 4.31738292501922e-06,
      "loss": 0.5385,
      "num_input_tokens_seen": 20683360,
      "step": 31640
    },
    {
      "epoch": 16.585429769392032,
      "grad_norm": 0.2345161736011505,
      "learning_rate": 4.31096183928513e-06,
      "loss": 0.4892,
      "num_input_tokens_seen": 20685920,
      "step": 31645
    },
    {
      "epoch": 16.58805031446541,
      "grad_norm": 0.19150495529174805,
      "learning_rate": 4.304545081461783e-06,
      "loss": 0.4025,
      "num_input_tokens_seen": 20688576,
      "step": 31650
    },
    {
      "epoch": 16.590670859538783,
      "grad_norm": 0.43720147013664246,
      "learning_rate": 4.298132652891501e-06,
      "loss": 0.4318,
      "num_input_tokens_seen": 20691424,
      "step": 31655
    },
    {
      "epoch": 16.59329140461216,
      "grad_norm": 0.1635892540216446,
      "learning_rate": 4.291724554915688e-06,
      "loss": 0.4788,
      "num_input_tokens_seen": 20694432,
      "step": 31660
    },
    {
      "epoch": 16.595911949685533,
      "grad_norm": 0.18186689913272858,
      "learning_rate": 4.285320788874853e-06,
      "loss": 0.4301,
      "num_input_tokens_seen": 20697856,
      "step": 31665
    },
    {
      "epoch": 16.59853249475891,
      "grad_norm": 0.1737651228904724,
      "learning_rate": 4.278921356108581e-06,
      "loss": 0.5798,
      "num_input_tokens_seen": 20701504,
      "step": 31670
    },
    {
      "epoch": 16.601153039832283,
      "grad_norm": 0.11972217261791229,
      "learning_rate": 4.272526257955556e-06,
      "loss": 0.4627,
      "num_input_tokens_seen": 20705056,
      "step": 31675
    },
    {
      "epoch": 16.60377358490566,
      "grad_norm": 0.16649597883224487,
      "learning_rate": 4.266135495753567e-06,
      "loss": 0.4513,
      "num_input_tokens_seen": 20707872,
      "step": 31680
    },
    {
      "epoch": 16.606394129979037,
      "grad_norm": 0.15181250870227814,
      "learning_rate": 4.259749070839491e-06,
      "loss": 0.4875,
      "num_input_tokens_seen": 20710560,
      "step": 31685
    },
    {
      "epoch": 16.60901467505241,
      "grad_norm": 0.1550874263048172,
      "learning_rate": 4.253366984549289e-06,
      "loss": 0.4896,
      "num_input_tokens_seen": 20714016,
      "step": 31690
    },
    {
      "epoch": 16.611635220125788,
      "grad_norm": 0.3486098051071167,
      "learning_rate": 4.24698923821803e-06,
      "loss": 0.4458,
      "num_input_tokens_seen": 20717120,
      "step": 31695
    },
    {
      "epoch": 16.61425576519916,
      "grad_norm": 0.1384591907262802,
      "learning_rate": 4.240615833179861e-06,
      "loss": 0.3956,
      "num_input_tokens_seen": 20720160,
      "step": 31700
    },
    {
      "epoch": 16.616876310272538,
      "grad_norm": 0.21250052750110626,
      "learning_rate": 4.234246770768016e-06,
      "loss": 0.5092,
      "num_input_tokens_seen": 20723328,
      "step": 31705
    },
    {
      "epoch": 16.61949685534591,
      "grad_norm": 0.20415344834327698,
      "learning_rate": 4.227882052314841e-06,
      "loss": 0.4621,
      "num_input_tokens_seen": 20725632,
      "step": 31710
    },
    {
      "epoch": 16.622117400419288,
      "grad_norm": 0.2426723688840866,
      "learning_rate": 4.2215216791517575e-06,
      "loss": 0.6103,
      "num_input_tokens_seen": 20729600,
      "step": 31715
    },
    {
      "epoch": 16.62473794549266,
      "grad_norm": 0.16820932924747467,
      "learning_rate": 4.215165652609293e-06,
      "loss": 0.5277,
      "num_input_tokens_seen": 20732512,
      "step": 31720
    },
    {
      "epoch": 16.62735849056604,
      "grad_norm": 0.2620726227760315,
      "learning_rate": 4.208813974017048e-06,
      "loss": 0.5141,
      "num_input_tokens_seen": 20736544,
      "step": 31725
    },
    {
      "epoch": 16.629979035639412,
      "grad_norm": 0.34040403366088867,
      "learning_rate": 4.202466644703712e-06,
      "loss": 0.5562,
      "num_input_tokens_seen": 20739968,
      "step": 31730
    },
    {
      "epoch": 16.63259958071279,
      "grad_norm": 0.3047727942466736,
      "learning_rate": 4.196123665997087e-06,
      "loss": 0.5006,
      "num_input_tokens_seen": 20743296,
      "step": 31735
    },
    {
      "epoch": 16.635220125786162,
      "grad_norm": 0.14221087098121643,
      "learning_rate": 4.189785039224045e-06,
      "loss": 0.5412,
      "num_input_tokens_seen": 20745696,
      "step": 31740
    },
    {
      "epoch": 16.63784067085954,
      "grad_norm": 0.10958564281463623,
      "learning_rate": 4.183450765710567e-06,
      "loss": 0.621,
      "num_input_tokens_seen": 20750272,
      "step": 31745
    },
    {
      "epoch": 16.640461215932913,
      "grad_norm": 0.19388698041439056,
      "learning_rate": 4.177120846781693e-06,
      "loss": 0.4849,
      "num_input_tokens_seen": 20754112,
      "step": 31750
    },
    {
      "epoch": 16.64308176100629,
      "grad_norm": 0.12059912085533142,
      "learning_rate": 4.170795283761583e-06,
      "loss": 0.4221,
      "num_input_tokens_seen": 20756928,
      "step": 31755
    },
    {
      "epoch": 16.645702306079663,
      "grad_norm": 0.18046998977661133,
      "learning_rate": 4.164474077973462e-06,
      "loss": 0.5538,
      "num_input_tokens_seen": 20759968,
      "step": 31760
    },
    {
      "epoch": 16.64832285115304,
      "grad_norm": 0.1619407683610916,
      "learning_rate": 4.158157230739659e-06,
      "loss": 0.4747,
      "num_input_tokens_seen": 20762560,
      "step": 31765
    },
    {
      "epoch": 16.650943396226417,
      "grad_norm": 0.22622326016426086,
      "learning_rate": 4.151844743381589e-06,
      "loss": 0.4569,
      "num_input_tokens_seen": 20765120,
      "step": 31770
    },
    {
      "epoch": 16.65356394129979,
      "grad_norm": 0.19173076748847961,
      "learning_rate": 4.145536617219756e-06,
      "loss": 0.4349,
      "num_input_tokens_seen": 20767840,
      "step": 31775
    },
    {
      "epoch": 16.656184486373167,
      "grad_norm": 0.16052919626235962,
      "learning_rate": 4.139232853573735e-06,
      "loss": 0.491,
      "num_input_tokens_seen": 20771200,
      "step": 31780
    },
    {
      "epoch": 16.65880503144654,
      "grad_norm": 0.23639819025993347,
      "learning_rate": 4.132933453762214e-06,
      "loss": 0.5049,
      "num_input_tokens_seen": 20774336,
      "step": 31785
    },
    {
      "epoch": 16.661425576519918,
      "grad_norm": 0.15725946426391602,
      "learning_rate": 4.126638419102946e-06,
      "loss": 0.5334,
      "num_input_tokens_seen": 20777248,
      "step": 31790
    },
    {
      "epoch": 16.66404612159329,
      "grad_norm": 0.17440083622932434,
      "learning_rate": 4.120347750912784e-06,
      "loss": 0.4613,
      "num_input_tokens_seen": 20780288,
      "step": 31795
    },
    {
      "epoch": 16.666666666666668,
      "grad_norm": 0.2963395416736603,
      "learning_rate": 4.1140614505076724e-06,
      "loss": 0.6201,
      "num_input_tokens_seen": 20782848,
      "step": 31800
    },
    {
      "epoch": 16.66928721174004,
      "grad_norm": 0.12828844785690308,
      "learning_rate": 4.10777951920262e-06,
      "loss": 0.4988,
      "num_input_tokens_seen": 20787168,
      "step": 31805
    },
    {
      "epoch": 16.671907756813418,
      "grad_norm": 0.14918994903564453,
      "learning_rate": 4.1015019583117455e-06,
      "loss": 0.5938,
      "num_input_tokens_seen": 20790176,
      "step": 31810
    },
    {
      "epoch": 16.67452830188679,
      "grad_norm": 0.13652361929416656,
      "learning_rate": 4.095228769148232e-06,
      "loss": 0.4374,
      "num_input_tokens_seen": 20792864,
      "step": 31815
    },
    {
      "epoch": 16.67714884696017,
      "grad_norm": 0.14069753885269165,
      "learning_rate": 4.088959953024368e-06,
      "loss": 0.4491,
      "num_input_tokens_seen": 20797408,
      "step": 31820
    },
    {
      "epoch": 16.679769392033542,
      "grad_norm": 0.22264015674591064,
      "learning_rate": 4.082695511251522e-06,
      "loss": 0.4298,
      "num_input_tokens_seen": 20800128,
      "step": 31825
    },
    {
      "epoch": 16.68238993710692,
      "grad_norm": 0.2205701768398285,
      "learning_rate": 4.076435445140131e-06,
      "loss": 0.5282,
      "num_input_tokens_seen": 20803008,
      "step": 31830
    },
    {
      "epoch": 16.685010482180292,
      "grad_norm": 0.16240687668323517,
      "learning_rate": 4.070179755999737e-06,
      "loss": 0.5777,
      "num_input_tokens_seen": 20806144,
      "step": 31835
    },
    {
      "epoch": 16.68763102725367,
      "grad_norm": 0.25733765959739685,
      "learning_rate": 4.063928445138962e-06,
      "loss": 0.4262,
      "num_input_tokens_seen": 20809472,
      "step": 31840
    },
    {
      "epoch": 16.690251572327043,
      "grad_norm": 0.19378000497817993,
      "learning_rate": 4.057681513865499e-06,
      "loss": 0.3982,
      "num_input_tokens_seen": 20812288,
      "step": 31845
    },
    {
      "epoch": 16.69287211740042,
      "grad_norm": 0.22385092079639435,
      "learning_rate": 4.0514389634861426e-06,
      "loss": 0.5102,
      "num_input_tokens_seen": 20815296,
      "step": 31850
    },
    {
      "epoch": 16.695492662473793,
      "grad_norm": 0.21470403671264648,
      "learning_rate": 4.045200795306767e-06,
      "loss": 0.5324,
      "num_input_tokens_seen": 20822656,
      "step": 31855
    },
    {
      "epoch": 16.69811320754717,
      "grad_norm": 0.12706446647644043,
      "learning_rate": 4.038967010632314e-06,
      "loss": 0.5207,
      "num_input_tokens_seen": 20826016,
      "step": 31860
    },
    {
      "epoch": 16.700733752620543,
      "grad_norm": 0.13486497104167938,
      "learning_rate": 4.032737610766826e-06,
      "loss": 0.4309,
      "num_input_tokens_seen": 20829824,
      "step": 31865
    },
    {
      "epoch": 16.70335429769392,
      "grad_norm": 0.1430911272764206,
      "learning_rate": 4.026512597013429e-06,
      "loss": 0.4287,
      "num_input_tokens_seen": 20832512,
      "step": 31870
    },
    {
      "epoch": 16.705974842767297,
      "grad_norm": 0.1598624587059021,
      "learning_rate": 4.020291970674312e-06,
      "loss": 0.4536,
      "num_input_tokens_seen": 20836000,
      "step": 31875
    },
    {
      "epoch": 16.70859538784067,
      "grad_norm": 0.27723413705825806,
      "learning_rate": 4.0140757330507705e-06,
      "loss": 0.3179,
      "num_input_tokens_seen": 20838976,
      "step": 31880
    },
    {
      "epoch": 16.711215932914047,
      "grad_norm": 0.2207682579755783,
      "learning_rate": 4.007863885443161e-06,
      "loss": 0.582,
      "num_input_tokens_seen": 20842208,
      "step": 31885
    },
    {
      "epoch": 16.71383647798742,
      "grad_norm": 0.238201305270195,
      "learning_rate": 4.001656429150935e-06,
      "loss": 0.5199,
      "num_input_tokens_seen": 20846272,
      "step": 31890
    },
    {
      "epoch": 16.716457023060798,
      "grad_norm": 0.21452437341213226,
      "learning_rate": 3.9954533654726235e-06,
      "loss": 0.5552,
      "num_input_tokens_seen": 20848896,
      "step": 31895
    },
    {
      "epoch": 16.71907756813417,
      "grad_norm": 0.17331194877624512,
      "learning_rate": 3.989254695705841e-06,
      "loss": 0.4916,
      "num_input_tokens_seen": 20852288,
      "step": 31900
    },
    {
      "epoch": 16.721698113207548,
      "grad_norm": 0.09884653240442276,
      "learning_rate": 3.983060421147272e-06,
      "loss": 0.5234,
      "num_input_tokens_seen": 20855136,
      "step": 31905
    },
    {
      "epoch": 16.72431865828092,
      "grad_norm": 0.13162925839424133,
      "learning_rate": 3.976870543092684e-06,
      "loss": 0.4049,
      "num_input_tokens_seen": 20858432,
      "step": 31910
    },
    {
      "epoch": 16.7269392033543,
      "grad_norm": 0.1258552074432373,
      "learning_rate": 3.970685062836932e-06,
      "loss": 0.3973,
      "num_input_tokens_seen": 20861280,
      "step": 31915
    },
    {
      "epoch": 16.729559748427672,
      "grad_norm": 0.2862514853477478,
      "learning_rate": 3.96450398167395e-06,
      "loss": 0.4317,
      "num_input_tokens_seen": 20863968,
      "step": 31920
    },
    {
      "epoch": 16.73218029350105,
      "grad_norm": 0.22624057531356812,
      "learning_rate": 3.958327300896758e-06,
      "loss": 0.5906,
      "num_input_tokens_seen": 20866112,
      "step": 31925
    },
    {
      "epoch": 16.734800838574422,
      "grad_norm": 0.13855311274528503,
      "learning_rate": 3.952155021797429e-06,
      "loss": 0.4145,
      "num_input_tokens_seen": 20869856,
      "step": 31930
    },
    {
      "epoch": 16.7374213836478,
      "grad_norm": 0.13777939975261688,
      "learning_rate": 3.945987145667151e-06,
      "loss": 0.4482,
      "num_input_tokens_seen": 20872448,
      "step": 31935
    },
    {
      "epoch": 16.740041928721173,
      "grad_norm": 0.137624591588974,
      "learning_rate": 3.939823673796162e-06,
      "loss": 0.3652,
      "num_input_tokens_seen": 20875008,
      "step": 31940
    },
    {
      "epoch": 16.74266247379455,
      "grad_norm": 0.15284310281276703,
      "learning_rate": 3.933664607473789e-06,
      "loss": 0.6055,
      "num_input_tokens_seen": 20878144,
      "step": 31945
    },
    {
      "epoch": 16.745283018867923,
      "grad_norm": 0.0999169573187828,
      "learning_rate": 3.927509947988445e-06,
      "loss": 0.4766,
      "num_input_tokens_seen": 20881216,
      "step": 31950
    },
    {
      "epoch": 16.7479035639413,
      "grad_norm": 0.16066312789916992,
      "learning_rate": 3.9213596966276164e-06,
      "loss": 0.4065,
      "num_input_tokens_seen": 20885120,
      "step": 31955
    },
    {
      "epoch": 16.750524109014677,
      "grad_norm": 0.13816618919372559,
      "learning_rate": 3.9152138546778625e-06,
      "loss": 0.5902,
      "num_input_tokens_seen": 20887808,
      "step": 31960
    },
    {
      "epoch": 16.75314465408805,
      "grad_norm": 0.143878772854805,
      "learning_rate": 3.9090724234248175e-06,
      "loss": 0.4592,
      "num_input_tokens_seen": 20890720,
      "step": 31965
    },
    {
      "epoch": 16.755765199161427,
      "grad_norm": 0.1565684825181961,
      "learning_rate": 3.902935404153202e-06,
      "loss": 0.4343,
      "num_input_tokens_seen": 20893440,
      "step": 31970
    },
    {
      "epoch": 16.7583857442348,
      "grad_norm": 0.12862376868724823,
      "learning_rate": 3.896802798146809e-06,
      "loss": 0.4241,
      "num_input_tokens_seen": 20897792,
      "step": 31975
    },
    {
      "epoch": 16.761006289308177,
      "grad_norm": 0.18504644930362701,
      "learning_rate": 3.890674606688524e-06,
      "loss": 0.5147,
      "num_input_tokens_seen": 20899936,
      "step": 31980
    },
    {
      "epoch": 16.76362683438155,
      "grad_norm": 0.18240058422088623,
      "learning_rate": 3.88455083106028e-06,
      "loss": 0.4263,
      "num_input_tokens_seen": 20905728,
      "step": 31985
    },
    {
      "epoch": 16.766247379454928,
      "grad_norm": 0.17509187757968903,
      "learning_rate": 3.878431472543098e-06,
      "loss": 0.5096,
      "num_input_tokens_seen": 20909376,
      "step": 31990
    },
    {
      "epoch": 16.7688679245283,
      "grad_norm": 0.16013766825199127,
      "learning_rate": 3.872316532417086e-06,
      "loss": 0.5021,
      "num_input_tokens_seen": 20913088,
      "step": 31995
    },
    {
      "epoch": 16.771488469601678,
      "grad_norm": 0.15375132858753204,
      "learning_rate": 3.866206011961418e-06,
      "loss": 0.5878,
      "num_input_tokens_seen": 20915808,
      "step": 32000
    },
    {
      "epoch": 16.77410901467505,
      "grad_norm": 0.13230858743190765,
      "learning_rate": 3.860099912454346e-06,
      "loss": 0.5745,
      "num_input_tokens_seen": 20920032,
      "step": 32005
    },
    {
      "epoch": 16.77672955974843,
      "grad_norm": 0.15123580396175385,
      "learning_rate": 3.853998235173201e-06,
      "loss": 0.4396,
      "num_input_tokens_seen": 20922912,
      "step": 32010
    },
    {
      "epoch": 16.779350104821802,
      "grad_norm": 0.2336423099040985,
      "learning_rate": 3.847900981394379e-06,
      "loss": 0.5432,
      "num_input_tokens_seen": 20926176,
      "step": 32015
    },
    {
      "epoch": 16.78197064989518,
      "grad_norm": 0.18585020303726196,
      "learning_rate": 3.841808152393348e-06,
      "loss": 0.5003,
      "num_input_tokens_seen": 20928896,
      "step": 32020
    },
    {
      "epoch": 16.784591194968552,
      "grad_norm": 0.17280049622058868,
      "learning_rate": 3.835719749444666e-06,
      "loss": 0.4192,
      "num_input_tokens_seen": 20931968,
      "step": 32025
    },
    {
      "epoch": 16.78721174004193,
      "grad_norm": 0.19038762152194977,
      "learning_rate": 3.8296357738219606e-06,
      "loss": 0.493,
      "num_input_tokens_seen": 20935168,
      "step": 32030
    },
    {
      "epoch": 16.789832285115303,
      "grad_norm": 0.11433825641870499,
      "learning_rate": 3.823556226797931e-06,
      "loss": 0.5189,
      "num_input_tokens_seen": 20937856,
      "step": 32035
    },
    {
      "epoch": 16.79245283018868,
      "grad_norm": 0.1359403282403946,
      "learning_rate": 3.817481109644344e-06,
      "loss": 0.4795,
      "num_input_tokens_seen": 20941120,
      "step": 32040
    },
    {
      "epoch": 16.795073375262053,
      "grad_norm": 0.1731661558151245,
      "learning_rate": 3.811410423632042e-06,
      "loss": 0.5434,
      "num_input_tokens_seen": 20944416,
      "step": 32045
    },
    {
      "epoch": 16.79769392033543,
      "grad_norm": 0.18205732107162476,
      "learning_rate": 3.8053441700309473e-06,
      "loss": 0.4728,
      "num_input_tokens_seen": 20948384,
      "step": 32050
    },
    {
      "epoch": 16.800314465408803,
      "grad_norm": 0.18315178155899048,
      "learning_rate": 3.7992823501100507e-06,
      "loss": 0.4931,
      "num_input_tokens_seen": 20951328,
      "step": 32055
    },
    {
      "epoch": 16.80293501048218,
      "grad_norm": 0.2279156893491745,
      "learning_rate": 3.7932249651374207e-06,
      "loss": 0.5231,
      "num_input_tokens_seen": 20954304,
      "step": 32060
    },
    {
      "epoch": 16.805555555555557,
      "grad_norm": 0.14368809759616852,
      "learning_rate": 3.787172016380183e-06,
      "loss": 0.4533,
      "num_input_tokens_seen": 20957600,
      "step": 32065
    },
    {
      "epoch": 16.80817610062893,
      "grad_norm": 0.10974244028329849,
      "learning_rate": 3.7811235051045558e-06,
      "loss": 0.4504,
      "num_input_tokens_seen": 20961248,
      "step": 32070
    },
    {
      "epoch": 16.810796645702307,
      "grad_norm": 0.15689292550086975,
      "learning_rate": 3.775079432575809e-06,
      "loss": 0.3975,
      "num_input_tokens_seen": 20964864,
      "step": 32075
    },
    {
      "epoch": 16.81341719077568,
      "grad_norm": 0.10183970630168915,
      "learning_rate": 3.7690398000582964e-06,
      "loss": 0.4481,
      "num_input_tokens_seen": 20968160,
      "step": 32080
    },
    {
      "epoch": 16.816037735849058,
      "grad_norm": 0.11400634795427322,
      "learning_rate": 3.763004608815443e-06,
      "loss": 0.4864,
      "num_input_tokens_seen": 20971072,
      "step": 32085
    },
    {
      "epoch": 16.81865828092243,
      "grad_norm": 0.16238075494766235,
      "learning_rate": 3.756973860109747e-06,
      "loss": 0.5067,
      "num_input_tokens_seen": 20974016,
      "step": 32090
    },
    {
      "epoch": 16.821278825995808,
      "grad_norm": 0.16896983981132507,
      "learning_rate": 3.7509475552027663e-06,
      "loss": 0.3989,
      "num_input_tokens_seen": 20977280,
      "step": 32095
    },
    {
      "epoch": 16.82389937106918,
      "grad_norm": 0.2373257577419281,
      "learning_rate": 3.744925695355131e-06,
      "loss": 0.6263,
      "num_input_tokens_seen": 20980704,
      "step": 32100
    },
    {
      "epoch": 16.82651991614256,
      "grad_norm": 0.1473025679588318,
      "learning_rate": 3.7389082818265476e-06,
      "loss": 0.7032,
      "num_input_tokens_seen": 20983552,
      "step": 32105
    },
    {
      "epoch": 16.829140461215932,
      "grad_norm": 0.15746742486953735,
      "learning_rate": 3.732895315875792e-06,
      "loss": 0.4325,
      "num_input_tokens_seen": 20986464,
      "step": 32110
    },
    {
      "epoch": 16.83176100628931,
      "grad_norm": 0.16685129702091217,
      "learning_rate": 3.726886798760715e-06,
      "loss": 0.5121,
      "num_input_tokens_seen": 20990112,
      "step": 32115
    },
    {
      "epoch": 16.834381551362682,
      "grad_norm": 0.154524028301239,
      "learning_rate": 3.720882731738215e-06,
      "loss": 0.4714,
      "num_input_tokens_seen": 20993184,
      "step": 32120
    },
    {
      "epoch": 16.83700209643606,
      "grad_norm": 0.26673397421836853,
      "learning_rate": 3.714883116064291e-06,
      "loss": 0.5733,
      "num_input_tokens_seen": 20995936,
      "step": 32125
    },
    {
      "epoch": 16.839622641509433,
      "grad_norm": 0.39361271262168884,
      "learning_rate": 3.708887952993978e-06,
      "loss": 0.5654,
      "num_input_tokens_seen": 20999008,
      "step": 32130
    },
    {
      "epoch": 16.84224318658281,
      "grad_norm": 0.1389070451259613,
      "learning_rate": 3.7028972437814057e-06,
      "loss": 0.4575,
      "num_input_tokens_seen": 21002304,
      "step": 32135
    },
    {
      "epoch": 16.844863731656183,
      "grad_norm": 0.1180969625711441,
      "learning_rate": 3.696910989679764e-06,
      "loss": 0.4222,
      "num_input_tokens_seen": 21005344,
      "step": 32140
    },
    {
      "epoch": 16.84748427672956,
      "grad_norm": 0.10926049947738647,
      "learning_rate": 3.6909291919413003e-06,
      "loss": 0.4155,
      "num_input_tokens_seen": 21009312,
      "step": 32145
    },
    {
      "epoch": 16.850104821802937,
      "grad_norm": 0.23308008909225464,
      "learning_rate": 3.684951851817342e-06,
      "loss": 0.386,
      "num_input_tokens_seen": 21012544,
      "step": 32150
    },
    {
      "epoch": 16.85272536687631,
      "grad_norm": 0.14862756431102753,
      "learning_rate": 3.678978970558286e-06,
      "loss": 0.4295,
      "num_input_tokens_seen": 21015200,
      "step": 32155
    },
    {
      "epoch": 16.855345911949687,
      "grad_norm": 0.3655952215194702,
      "learning_rate": 3.673010549413583e-06,
      "loss": 0.3818,
      "num_input_tokens_seen": 21017696,
      "step": 32160
    },
    {
      "epoch": 16.85796645702306,
      "grad_norm": 0.1586896777153015,
      "learning_rate": 3.6670465896317592e-06,
      "loss": 0.433,
      "num_input_tokens_seen": 21020800,
      "step": 32165
    },
    {
      "epoch": 16.860587002096437,
      "grad_norm": 0.16501963138580322,
      "learning_rate": 3.661087092460419e-06,
      "loss": 0.4167,
      "num_input_tokens_seen": 21023552,
      "step": 32170
    },
    {
      "epoch": 16.86320754716981,
      "grad_norm": 0.1546645164489746,
      "learning_rate": 3.655132059146202e-06,
      "loss": 0.5026,
      "num_input_tokens_seen": 21026656,
      "step": 32175
    },
    {
      "epoch": 16.865828092243188,
      "grad_norm": 0.15258944034576416,
      "learning_rate": 3.6491814909348526e-06,
      "loss": 0.5502,
      "num_input_tokens_seen": 21030464,
      "step": 32180
    },
    {
      "epoch": 16.86844863731656,
      "grad_norm": 0.1316206008195877,
      "learning_rate": 3.6432353890711424e-06,
      "loss": 0.5428,
      "num_input_tokens_seen": 21033472,
      "step": 32185
    },
    {
      "epoch": 16.871069182389938,
      "grad_norm": 0.10506828874349594,
      "learning_rate": 3.6372937547989425e-06,
      "loss": 0.403,
      "num_input_tokens_seen": 21036512,
      "step": 32190
    },
    {
      "epoch": 16.87368972746331,
      "grad_norm": 0.172610342502594,
      "learning_rate": 3.631356589361173e-06,
      "loss": 0.463,
      "num_input_tokens_seen": 21039936,
      "step": 32195
    },
    {
      "epoch": 16.87631027253669,
      "grad_norm": 0.18345974385738373,
      "learning_rate": 3.625423893999813e-06,
      "loss": 0.6433,
      "num_input_tokens_seen": 21043360,
      "step": 32200
    },
    {
      "epoch": 16.878930817610062,
      "grad_norm": 0.20681804418563843,
      "learning_rate": 3.6194956699559217e-06,
      "loss": 0.5029,
      "num_input_tokens_seen": 21046880,
      "step": 32205
    },
    {
      "epoch": 16.88155136268344,
      "grad_norm": 0.44475266337394714,
      "learning_rate": 3.6135719184696188e-06,
      "loss": 0.3268,
      "num_input_tokens_seen": 21049664,
      "step": 32210
    },
    {
      "epoch": 16.884171907756812,
      "grad_norm": 0.17901895940303802,
      "learning_rate": 3.6076526407800766e-06,
      "loss": 0.5249,
      "num_input_tokens_seen": 21052832,
      "step": 32215
    },
    {
      "epoch": 16.88679245283019,
      "grad_norm": 0.165954127907753,
      "learning_rate": 3.601737838125552e-06,
      "loss": 0.4302,
      "num_input_tokens_seen": 21057120,
      "step": 32220
    },
    {
      "epoch": 16.889412997903563,
      "grad_norm": 0.20941531658172607,
      "learning_rate": 3.595827511743341e-06,
      "loss": 0.5876,
      "num_input_tokens_seen": 21060512,
      "step": 32225
    },
    {
      "epoch": 16.89203354297694,
      "grad_norm": 0.19373305141925812,
      "learning_rate": 3.5899216628698217e-06,
      "loss": 0.4882,
      "num_input_tokens_seen": 21062720,
      "step": 32230
    },
    {
      "epoch": 16.894654088050313,
      "grad_norm": 0.15803037583827972,
      "learning_rate": 3.584020292740431e-06,
      "loss": 0.4108,
      "num_input_tokens_seen": 21065504,
      "step": 32235
    },
    {
      "epoch": 16.89727463312369,
      "grad_norm": 0.2836763560771942,
      "learning_rate": 3.578123402589675e-06,
      "loss": 0.3523,
      "num_input_tokens_seen": 21069472,
      "step": 32240
    },
    {
      "epoch": 16.899895178197063,
      "grad_norm": 0.22926156222820282,
      "learning_rate": 3.5722309936511044e-06,
      "loss": 0.4986,
      "num_input_tokens_seen": 21072192,
      "step": 32245
    },
    {
      "epoch": 16.90251572327044,
      "grad_norm": 0.145721897482872,
      "learning_rate": 3.5663430671573572e-06,
      "loss": 0.4505,
      "num_input_tokens_seen": 21075264,
      "step": 32250
    },
    {
      "epoch": 16.905136268343817,
      "grad_norm": 0.169371098279953,
      "learning_rate": 3.5604596243401038e-06,
      "loss": 0.4223,
      "num_input_tokens_seen": 21079008,
      "step": 32255
    },
    {
      "epoch": 16.90775681341719,
      "grad_norm": 0.1401350349187851,
      "learning_rate": 3.554580666430102e-06,
      "loss": 0.4293,
      "num_input_tokens_seen": 21081888,
      "step": 32260
    },
    {
      "epoch": 16.910377358490567,
      "grad_norm": 0.28614693880081177,
      "learning_rate": 3.5487061946571614e-06,
      "loss": 0.5112,
      "num_input_tokens_seen": 21084608,
      "step": 32265
    },
    {
      "epoch": 16.91299790356394,
      "grad_norm": 0.1315387487411499,
      "learning_rate": 3.542836210250161e-06,
      "loss": 0.6393,
      "num_input_tokens_seen": 21087680,
      "step": 32270
    },
    {
      "epoch": 16.915618448637318,
      "grad_norm": 0.13323234021663666,
      "learning_rate": 3.536970714437032e-06,
      "loss": 0.5357,
      "num_input_tokens_seen": 21090944,
      "step": 32275
    },
    {
      "epoch": 16.91823899371069,
      "grad_norm": 0.13957887887954712,
      "learning_rate": 3.5311097084447575e-06,
      "loss": 0.3378,
      "num_input_tokens_seen": 21094272,
      "step": 32280
    },
    {
      "epoch": 16.920859538784068,
      "grad_norm": 0.19888538122177124,
      "learning_rate": 3.5252531934994033e-06,
      "loss": 0.45,
      "num_input_tokens_seen": 21097952,
      "step": 32285
    },
    {
      "epoch": 16.92348008385744,
      "grad_norm": 0.14938127994537354,
      "learning_rate": 3.5194011708260866e-06,
      "loss": 0.4971,
      "num_input_tokens_seen": 21101408,
      "step": 32290
    },
    {
      "epoch": 16.92610062893082,
      "grad_norm": 0.19561035931110382,
      "learning_rate": 3.513553641648984e-06,
      "loss": 0.542,
      "num_input_tokens_seen": 21105888,
      "step": 32295
    },
    {
      "epoch": 16.928721174004192,
      "grad_norm": 0.2353934645652771,
      "learning_rate": 3.5077106071913307e-06,
      "loss": 0.4271,
      "num_input_tokens_seen": 21108992,
      "step": 32300
    },
    {
      "epoch": 16.93134171907757,
      "grad_norm": 0.18511226773262024,
      "learning_rate": 3.5018720686754133e-06,
      "loss": 0.3458,
      "num_input_tokens_seen": 21111648,
      "step": 32305
    },
    {
      "epoch": 16.933962264150942,
      "grad_norm": 0.1807551234960556,
      "learning_rate": 3.4960380273225985e-06,
      "loss": 0.5364,
      "num_input_tokens_seen": 21115872,
      "step": 32310
    },
    {
      "epoch": 16.93658280922432,
      "grad_norm": 0.17210158705711365,
      "learning_rate": 3.490208484353297e-06,
      "loss": 0.3643,
      "num_input_tokens_seen": 21118880,
      "step": 32315
    },
    {
      "epoch": 16.939203354297693,
      "grad_norm": 0.1279861330986023,
      "learning_rate": 3.4843834409869824e-06,
      "loss": 0.411,
      "num_input_tokens_seen": 21121792,
      "step": 32320
    },
    {
      "epoch": 16.94182389937107,
      "grad_norm": 0.13454575836658478,
      "learning_rate": 3.4785628984421977e-06,
      "loss": 0.8134,
      "num_input_tokens_seen": 21125024,
      "step": 32325
    },
    {
      "epoch": 16.944444444444443,
      "grad_norm": 0.18861964344978333,
      "learning_rate": 3.4727468579365235e-06,
      "loss": 0.5524,
      "num_input_tokens_seen": 21128000,
      "step": 32330
    },
    {
      "epoch": 16.94706498951782,
      "grad_norm": 0.11616441607475281,
      "learning_rate": 3.466935320686604e-06,
      "loss": 0.5868,
      "num_input_tokens_seen": 21131808,
      "step": 32335
    },
    {
      "epoch": 16.949685534591197,
      "grad_norm": 0.21764497458934784,
      "learning_rate": 3.4611282879081553e-06,
      "loss": 0.5675,
      "num_input_tokens_seen": 21135488,
      "step": 32340
    },
    {
      "epoch": 16.95230607966457,
      "grad_norm": 0.1417093724012375,
      "learning_rate": 3.455325760815939e-06,
      "loss": 0.4757,
      "num_input_tokens_seen": 21138656,
      "step": 32345
    },
    {
      "epoch": 16.954926624737947,
      "grad_norm": 0.28560468554496765,
      "learning_rate": 3.4495277406237843e-06,
      "loss": 0.4196,
      "num_input_tokens_seen": 21141184,
      "step": 32350
    },
    {
      "epoch": 16.95754716981132,
      "grad_norm": 0.1620294153690338,
      "learning_rate": 3.443734228544565e-06,
      "loss": 0.4888,
      "num_input_tokens_seen": 21144288,
      "step": 32355
    },
    {
      "epoch": 16.960167714884697,
      "grad_norm": 0.20037034153938293,
      "learning_rate": 3.437945225790215e-06,
      "loss": 0.5592,
      "num_input_tokens_seen": 21146912,
      "step": 32360
    },
    {
      "epoch": 16.96278825995807,
      "grad_norm": 0.16064344346523285,
      "learning_rate": 3.432160733571729e-06,
      "loss": 0.3855,
      "num_input_tokens_seen": 21150304,
      "step": 32365
    },
    {
      "epoch": 16.965408805031448,
      "grad_norm": 0.1073603481054306,
      "learning_rate": 3.4263807530991595e-06,
      "loss": 0.5026,
      "num_input_tokens_seen": 21153760,
      "step": 32370
    },
    {
      "epoch": 16.96802935010482,
      "grad_norm": 0.10181494057178497,
      "learning_rate": 3.4206052855816155e-06,
      "loss": 0.4836,
      "num_input_tokens_seen": 21156960,
      "step": 32375
    },
    {
      "epoch": 16.970649895178198,
      "grad_norm": 0.35460904240608215,
      "learning_rate": 3.4148343322272567e-06,
      "loss": 0.5067,
      "num_input_tokens_seen": 21159744,
      "step": 32380
    },
    {
      "epoch": 16.97327044025157,
      "grad_norm": 0.1317535936832428,
      "learning_rate": 3.4090678942432923e-06,
      "loss": 0.411,
      "num_input_tokens_seen": 21166304,
      "step": 32385
    },
    {
      "epoch": 16.97589098532495,
      "grad_norm": 0.17759913206100464,
      "learning_rate": 3.4033059728360018e-06,
      "loss": 0.4485,
      "num_input_tokens_seen": 21169440,
      "step": 32390
    },
    {
      "epoch": 16.978511530398322,
      "grad_norm": 0.20853252708911896,
      "learning_rate": 3.397548569210715e-06,
      "loss": 0.4613,
      "num_input_tokens_seen": 21173696,
      "step": 32395
    },
    {
      "epoch": 16.9811320754717,
      "grad_norm": 0.1867535561323166,
      "learning_rate": 3.3917956845718136e-06,
      "loss": 0.4676,
      "num_input_tokens_seen": 21176672,
      "step": 32400
    },
    {
      "epoch": 16.983752620545072,
      "grad_norm": 0.28619706630706787,
      "learning_rate": 3.38604732012274e-06,
      "loss": 0.4573,
      "num_input_tokens_seen": 21180064,
      "step": 32405
    },
    {
      "epoch": 16.98637316561845,
      "grad_norm": 0.28021040558815,
      "learning_rate": 3.3803034770659825e-06,
      "loss": 0.6989,
      "num_input_tokens_seen": 21183680,
      "step": 32410
    },
    {
      "epoch": 16.988993710691823,
      "grad_norm": 0.2011776864528656,
      "learning_rate": 3.3745641566030826e-06,
      "loss": 0.4474,
      "num_input_tokens_seen": 21187072,
      "step": 32415
    },
    {
      "epoch": 16.9916142557652,
      "grad_norm": 0.1970411092042923,
      "learning_rate": 3.3688293599346442e-06,
      "loss": 0.5786,
      "num_input_tokens_seen": 21191776,
      "step": 32420
    },
    {
      "epoch": 16.994234800838573,
      "grad_norm": 0.14073042571544647,
      "learning_rate": 3.3630990882603213e-06,
      "loss": 0.606,
      "num_input_tokens_seen": 21196288,
      "step": 32425
    },
    {
      "epoch": 16.99685534591195,
      "grad_norm": 0.2217145711183548,
      "learning_rate": 3.35737334277883e-06,
      "loss": 0.4409,
      "num_input_tokens_seen": 21200992,
      "step": 32430
    },
    {
      "epoch": 16.999475890985323,
      "grad_norm": 0.18866629898548126,
      "learning_rate": 3.351652124687918e-06,
      "loss": 0.4315,
      "num_input_tokens_seen": 21204320,
      "step": 32435
    },
    {
      "epoch": 17.0,
      "eval_loss": 0.49190816283226013,
      "eval_runtime": 10.1631,
      "eval_samples_per_second": 83.439,
      "eval_steps_per_second": 20.86,
      "num_input_tokens_seen": 21204416,
      "step": 32436
    },
    {
      "epoch": 17.0020964360587,
      "grad_norm": 0.16485804319381714,
      "learning_rate": 3.345935435184408e-06,
      "loss": 0.4696,
      "num_input_tokens_seen": 21206752,
      "step": 32440
    },
    {
      "epoch": 17.004716981132077,
      "grad_norm": 0.1434231400489807,
      "learning_rate": 3.3402232754641576e-06,
      "loss": 0.4283,
      "num_input_tokens_seen": 21210656,
      "step": 32445
    },
    {
      "epoch": 17.00733752620545,
      "grad_norm": 0.2216811627149582,
      "learning_rate": 3.334515646722089e-06,
      "loss": 0.5781,
      "num_input_tokens_seen": 21213216,
      "step": 32450
    },
    {
      "epoch": 17.009958071278827,
      "grad_norm": 0.15302933752536774,
      "learning_rate": 3.32881255015218e-06,
      "loss": 0.4302,
      "num_input_tokens_seen": 21216416,
      "step": 32455
    },
    {
      "epoch": 17.0125786163522,
      "grad_norm": 0.1401742398738861,
      "learning_rate": 3.3231139869474433e-06,
      "loss": 0.3646,
      "num_input_tokens_seen": 21220352,
      "step": 32460
    },
    {
      "epoch": 17.015199161425578,
      "grad_norm": 0.19341593980789185,
      "learning_rate": 3.317419958299961e-06,
      "loss": 0.4024,
      "num_input_tokens_seen": 21223296,
      "step": 32465
    },
    {
      "epoch": 17.01781970649895,
      "grad_norm": 0.18863050639629364,
      "learning_rate": 3.31173046540085e-06,
      "loss": 0.4278,
      "num_input_tokens_seen": 21225824,
      "step": 32470
    },
    {
      "epoch": 17.020440251572328,
      "grad_norm": 0.1811840683221817,
      "learning_rate": 3.306045509440295e-06,
      "loss": 0.3945,
      "num_input_tokens_seen": 21228704,
      "step": 32475
    },
    {
      "epoch": 17.0230607966457,
      "grad_norm": 0.21355612576007843,
      "learning_rate": 3.300365091607521e-06,
      "loss": 0.4242,
      "num_input_tokens_seen": 21231648,
      "step": 32480
    },
    {
      "epoch": 17.02568134171908,
      "grad_norm": 0.18015679717063904,
      "learning_rate": 3.2946892130908127e-06,
      "loss": 0.5068,
      "num_input_tokens_seen": 21234048,
      "step": 32485
    },
    {
      "epoch": 17.028301886792452,
      "grad_norm": 0.2610754668712616,
      "learning_rate": 3.289017875077488e-06,
      "loss": 0.548,
      "num_input_tokens_seen": 21237568,
      "step": 32490
    },
    {
      "epoch": 17.03092243186583,
      "grad_norm": 0.3288052976131439,
      "learning_rate": 3.283351078753938e-06,
      "loss": 0.4638,
      "num_input_tokens_seen": 21240672,
      "step": 32495
    },
    {
      "epoch": 17.033542976939202,
      "grad_norm": 0.15736506879329681,
      "learning_rate": 3.2776888253055777e-06,
      "loss": 0.4385,
      "num_input_tokens_seen": 21243200,
      "step": 32500
    },
    {
      "epoch": 17.03616352201258,
      "grad_norm": 0.1971014142036438,
      "learning_rate": 3.272031115916896e-06,
      "loss": 0.3679,
      "num_input_tokens_seen": 21247936,
      "step": 32505
    },
    {
      "epoch": 17.038784067085953,
      "grad_norm": 0.21375598013401031,
      "learning_rate": 3.2663779517714238e-06,
      "loss": 0.4777,
      "num_input_tokens_seen": 21250560,
      "step": 32510
    },
    {
      "epoch": 17.04140461215933,
      "grad_norm": 0.10054485499858856,
      "learning_rate": 3.260729334051729e-06,
      "loss": 0.514,
      "num_input_tokens_seen": 21254432,
      "step": 32515
    },
    {
      "epoch": 17.044025157232703,
      "grad_norm": 0.1605594903230667,
      "learning_rate": 3.2550852639394446e-06,
      "loss": 0.474,
      "num_input_tokens_seen": 21257568,
      "step": 32520
    },
    {
      "epoch": 17.04664570230608,
      "grad_norm": 0.14165836572647095,
      "learning_rate": 3.2494457426152452e-06,
      "loss": 0.4941,
      "num_input_tokens_seen": 21260992,
      "step": 32525
    },
    {
      "epoch": 17.049266247379453,
      "grad_norm": 0.1793237179517746,
      "learning_rate": 3.243810771258851e-06,
      "loss": 0.4613,
      "num_input_tokens_seen": 21263328,
      "step": 32530
    },
    {
      "epoch": 17.05188679245283,
      "grad_norm": 0.19749359786510468,
      "learning_rate": 3.2381803510490387e-06,
      "loss": 0.4828,
      "num_input_tokens_seen": 21266752,
      "step": 32535
    },
    {
      "epoch": 17.054507337526207,
      "grad_norm": 0.17740419507026672,
      "learning_rate": 3.2325544831636217e-06,
      "loss": 0.5059,
      "num_input_tokens_seen": 21269728,
      "step": 32540
    },
    {
      "epoch": 17.05712788259958,
      "grad_norm": 0.20196233689785004,
      "learning_rate": 3.22693316877947e-06,
      "loss": 0.5087,
      "num_input_tokens_seen": 21272864,
      "step": 32545
    },
    {
      "epoch": 17.059748427672957,
      "grad_norm": 0.21146787703037262,
      "learning_rate": 3.2213164090725057e-06,
      "loss": 0.4126,
      "num_input_tokens_seen": 21276320,
      "step": 32550
    },
    {
      "epoch": 17.06236897274633,
      "grad_norm": 0.195911705493927,
      "learning_rate": 3.2157042052176763e-06,
      "loss": 0.4936,
      "num_input_tokens_seen": 21279360,
      "step": 32555
    },
    {
      "epoch": 17.064989517819708,
      "grad_norm": 0.34717264771461487,
      "learning_rate": 3.2100965583890003e-06,
      "loss": 0.5381,
      "num_input_tokens_seen": 21282112,
      "step": 32560
    },
    {
      "epoch": 17.06761006289308,
      "grad_norm": 0.19265761971473694,
      "learning_rate": 3.204493469759537e-06,
      "loss": 0.6066,
      "num_input_tokens_seen": 21284832,
      "step": 32565
    },
    {
      "epoch": 17.070230607966458,
      "grad_norm": 0.23981338739395142,
      "learning_rate": 3.1988949405013764e-06,
      "loss": 0.4722,
      "num_input_tokens_seen": 21288256,
      "step": 32570
    },
    {
      "epoch": 17.07285115303983,
      "grad_norm": 0.18984124064445496,
      "learning_rate": 3.193300971785676e-06,
      "loss": 0.4376,
      "num_input_tokens_seen": 21291616,
      "step": 32575
    },
    {
      "epoch": 17.07547169811321,
      "grad_norm": 0.2772868275642395,
      "learning_rate": 3.1877115647826335e-06,
      "loss": 0.5637,
      "num_input_tokens_seen": 21294720,
      "step": 32580
    },
    {
      "epoch": 17.078092243186582,
      "grad_norm": 0.112557552754879,
      "learning_rate": 3.1821267206614797e-06,
      "loss": 0.5027,
      "num_input_tokens_seen": 21298400,
      "step": 32585
    },
    {
      "epoch": 17.08071278825996,
      "grad_norm": 0.19886831939220428,
      "learning_rate": 3.176546440590508e-06,
      "loss": 0.5184,
      "num_input_tokens_seen": 21301376,
      "step": 32590
    },
    {
      "epoch": 17.083333333333332,
      "grad_norm": 0.1711912602186203,
      "learning_rate": 3.170970725737041e-06,
      "loss": 0.4209,
      "num_input_tokens_seen": 21303840,
      "step": 32595
    },
    {
      "epoch": 17.08595387840671,
      "grad_norm": 0.2860611081123352,
      "learning_rate": 3.165399577267461e-06,
      "loss": 0.4373,
      "num_input_tokens_seen": 21307136,
      "step": 32600
    },
    {
      "epoch": 17.088574423480082,
      "grad_norm": 0.17467299103736877,
      "learning_rate": 3.159832996347184e-06,
      "loss": 0.4061,
      "num_input_tokens_seen": 21309824,
      "step": 32605
    },
    {
      "epoch": 17.09119496855346,
      "grad_norm": 0.12191525101661682,
      "learning_rate": 3.1542709841406865e-06,
      "loss": 0.4139,
      "num_input_tokens_seen": 21312448,
      "step": 32610
    },
    {
      "epoch": 17.093815513626833,
      "grad_norm": 0.17677919566631317,
      "learning_rate": 3.148713541811468e-06,
      "loss": 0.3844,
      "num_input_tokens_seen": 21315040,
      "step": 32615
    },
    {
      "epoch": 17.09643605870021,
      "grad_norm": 0.19200509786605835,
      "learning_rate": 3.1431606705220803e-06,
      "loss": 0.5379,
      "num_input_tokens_seen": 21317696,
      "step": 32620
    },
    {
      "epoch": 17.099056603773583,
      "grad_norm": 0.22253632545471191,
      "learning_rate": 3.1376123714341234e-06,
      "loss": 0.565,
      "num_input_tokens_seen": 21320320,
      "step": 32625
    },
    {
      "epoch": 17.10167714884696,
      "grad_norm": 0.13981302082538605,
      "learning_rate": 3.1320686457082387e-06,
      "loss": 0.2986,
      "num_input_tokens_seen": 21323136,
      "step": 32630
    },
    {
      "epoch": 17.104297693920337,
      "grad_norm": 0.11552302539348602,
      "learning_rate": 3.126529494504113e-06,
      "loss": 0.4989,
      "num_input_tokens_seen": 21325664,
      "step": 32635
    },
    {
      "epoch": 17.10691823899371,
      "grad_norm": 0.12651190161705017,
      "learning_rate": 3.1209949189804744e-06,
      "loss": 0.446,
      "num_input_tokens_seen": 21328352,
      "step": 32640
    },
    {
      "epoch": 17.109538784067087,
      "grad_norm": 0.20774595439434052,
      "learning_rate": 3.115464920295089e-06,
      "loss": 0.4555,
      "num_input_tokens_seen": 21332704,
      "step": 32645
    },
    {
      "epoch": 17.11215932914046,
      "grad_norm": 0.1298418641090393,
      "learning_rate": 3.1099394996047666e-06,
      "loss": 0.5477,
      "num_input_tokens_seen": 21336352,
      "step": 32650
    },
    {
      "epoch": 17.114779874213838,
      "grad_norm": 0.09455883502960205,
      "learning_rate": 3.104418658065367e-06,
      "loss": 0.3965,
      "num_input_tokens_seen": 21340288,
      "step": 32655
    },
    {
      "epoch": 17.11740041928721,
      "grad_norm": 0.3415176570415497,
      "learning_rate": 3.098902396831782e-06,
      "loss": 0.3182,
      "num_input_tokens_seen": 21344128,
      "step": 32660
    },
    {
      "epoch": 17.120020964360588,
      "grad_norm": 0.14668917655944824,
      "learning_rate": 3.0933907170579646e-06,
      "loss": 0.4952,
      "num_input_tokens_seen": 21347296,
      "step": 32665
    },
    {
      "epoch": 17.12264150943396,
      "grad_norm": 0.17945507168769836,
      "learning_rate": 3.0878836198968835e-06,
      "loss": 0.4265,
      "num_input_tokens_seen": 21350240,
      "step": 32670
    },
    {
      "epoch": 17.12526205450734,
      "grad_norm": 0.24869485199451447,
      "learning_rate": 3.0823811065005564e-06,
      "loss": 0.4681,
      "num_input_tokens_seen": 21352704,
      "step": 32675
    },
    {
      "epoch": 17.127882599580712,
      "grad_norm": 0.258003294467926,
      "learning_rate": 3.0768831780200542e-06,
      "loss": 0.425,
      "num_input_tokens_seen": 21355904,
      "step": 32680
    },
    {
      "epoch": 17.13050314465409,
      "grad_norm": 0.14550724625587463,
      "learning_rate": 3.071389835605476e-06,
      "loss": 0.4657,
      "num_input_tokens_seen": 21359136,
      "step": 32685
    },
    {
      "epoch": 17.133123689727462,
      "grad_norm": 0.1422029733657837,
      "learning_rate": 3.065901080405978e-06,
      "loss": 0.5424,
      "num_input_tokens_seen": 21362496,
      "step": 32690
    },
    {
      "epoch": 17.13574423480084,
      "grad_norm": 0.16320686042308807,
      "learning_rate": 3.060416913569733e-06,
      "loss": 0.4361,
      "num_input_tokens_seen": 21364928,
      "step": 32695
    },
    {
      "epoch": 17.138364779874212,
      "grad_norm": 0.1163744404911995,
      "learning_rate": 3.054937336243968e-06,
      "loss": 0.5201,
      "num_input_tokens_seen": 21367808,
      "step": 32700
    },
    {
      "epoch": 17.14098532494759,
      "grad_norm": 0.21634499728679657,
      "learning_rate": 3.049462349574947e-06,
      "loss": 0.3601,
      "num_input_tokens_seen": 21371040,
      "step": 32705
    },
    {
      "epoch": 17.143605870020963,
      "grad_norm": 0.16780368983745575,
      "learning_rate": 3.043991954707981e-06,
      "loss": 0.4858,
      "num_input_tokens_seen": 21374528,
      "step": 32710
    },
    {
      "epoch": 17.14622641509434,
      "grad_norm": 0.24697619676589966,
      "learning_rate": 3.0385261527874075e-06,
      "loss": 0.4202,
      "num_input_tokens_seen": 21376864,
      "step": 32715
    },
    {
      "epoch": 17.148846960167713,
      "grad_norm": 0.20359279215335846,
      "learning_rate": 3.0330649449566205e-06,
      "loss": 0.5385,
      "num_input_tokens_seen": 21379840,
      "step": 32720
    },
    {
      "epoch": 17.15146750524109,
      "grad_norm": 0.1234729066491127,
      "learning_rate": 3.027608332358034e-06,
      "loss": 0.5969,
      "num_input_tokens_seen": 21383072,
      "step": 32725
    },
    {
      "epoch": 17.154088050314467,
      "grad_norm": 0.17059791088104248,
      "learning_rate": 3.022156316133104e-06,
      "loss": 0.5241,
      "num_input_tokens_seen": 21386080,
      "step": 32730
    },
    {
      "epoch": 17.15670859538784,
      "grad_norm": 0.19347906112670898,
      "learning_rate": 3.016708897422335e-06,
      "loss": 0.5296,
      "num_input_tokens_seen": 21388640,
      "step": 32735
    },
    {
      "epoch": 17.159329140461217,
      "grad_norm": 0.17899982631206512,
      "learning_rate": 3.0112660773652685e-06,
      "loss": 0.4636,
      "num_input_tokens_seen": 21391552,
      "step": 32740
    },
    {
      "epoch": 17.16194968553459,
      "grad_norm": 0.16398008167743683,
      "learning_rate": 3.005827857100482e-06,
      "loss": 0.5531,
      "num_input_tokens_seen": 21395040,
      "step": 32745
    },
    {
      "epoch": 17.164570230607968,
      "grad_norm": 0.2028597742319107,
      "learning_rate": 3.000394237765586e-06,
      "loss": 0.4899,
      "num_input_tokens_seen": 21398688,
      "step": 32750
    },
    {
      "epoch": 17.16719077568134,
      "grad_norm": 0.20381174981594086,
      "learning_rate": 2.9949652204972254e-06,
      "loss": 0.5707,
      "num_input_tokens_seen": 21402144,
      "step": 32755
    },
    {
      "epoch": 17.169811320754718,
      "grad_norm": 0.1925162971019745,
      "learning_rate": 2.989540806431093e-06,
      "loss": 0.6766,
      "num_input_tokens_seen": 21404352,
      "step": 32760
    },
    {
      "epoch": 17.17243186582809,
      "grad_norm": 0.1354042887687683,
      "learning_rate": 2.984120996701917e-06,
      "loss": 0.3535,
      "num_input_tokens_seen": 21407808,
      "step": 32765
    },
    {
      "epoch": 17.17505241090147,
      "grad_norm": 0.16055597364902496,
      "learning_rate": 2.9787057924434618e-06,
      "loss": 0.5286,
      "num_input_tokens_seen": 21410976,
      "step": 32770
    },
    {
      "epoch": 17.177672955974842,
      "grad_norm": 0.12785640358924866,
      "learning_rate": 2.9732951947885166e-06,
      "loss": 0.4896,
      "num_input_tokens_seen": 21414976,
      "step": 32775
    },
    {
      "epoch": 17.18029350104822,
      "grad_norm": 0.13314682245254517,
      "learning_rate": 2.967889204868929e-06,
      "loss": 0.3574,
      "num_input_tokens_seen": 21418368,
      "step": 32780
    },
    {
      "epoch": 17.182914046121592,
      "grad_norm": 0.1706882119178772,
      "learning_rate": 2.96248782381556e-06,
      "loss": 0.3266,
      "num_input_tokens_seen": 21421664,
      "step": 32785
    },
    {
      "epoch": 17.18553459119497,
      "grad_norm": 0.1400708109140396,
      "learning_rate": 2.957091052758321e-06,
      "loss": 0.4988,
      "num_input_tokens_seen": 21428544,
      "step": 32790
    },
    {
      "epoch": 17.188155136268342,
      "grad_norm": 0.11669822782278061,
      "learning_rate": 2.9516988928261517e-06,
      "loss": 0.446,
      "num_input_tokens_seen": 21431520,
      "step": 32795
    },
    {
      "epoch": 17.19077568134172,
      "grad_norm": 0.1538921296596527,
      "learning_rate": 2.9463113451470403e-06,
      "loss": 0.4334,
      "num_input_tokens_seen": 21435104,
      "step": 32800
    },
    {
      "epoch": 17.193396226415093,
      "grad_norm": 0.19280508160591125,
      "learning_rate": 2.94092841084799e-06,
      "loss": 0.5216,
      "num_input_tokens_seen": 21437888,
      "step": 32805
    },
    {
      "epoch": 17.19601677148847,
      "grad_norm": 0.12810085713863373,
      "learning_rate": 2.9355500910550586e-06,
      "loss": 0.4293,
      "num_input_tokens_seen": 21441280,
      "step": 32810
    },
    {
      "epoch": 17.198637316561843,
      "grad_norm": 0.1356944739818573,
      "learning_rate": 2.9301763868933157e-06,
      "loss": 0.3385,
      "num_input_tokens_seen": 21445472,
      "step": 32815
    },
    {
      "epoch": 17.20125786163522,
      "grad_norm": 0.18136627972126007,
      "learning_rate": 2.9248072994868896e-06,
      "loss": 0.5265,
      "num_input_tokens_seen": 21447872,
      "step": 32820
    },
    {
      "epoch": 17.203878406708597,
      "grad_norm": 0.13183794915676117,
      "learning_rate": 2.919442829958932e-06,
      "loss": 0.4704,
      "num_input_tokens_seen": 21451456,
      "step": 32825
    },
    {
      "epoch": 17.20649895178197,
      "grad_norm": 0.1302786022424698,
      "learning_rate": 2.9140829794316234e-06,
      "loss": 0.3648,
      "num_input_tokens_seen": 21454560,
      "step": 32830
    },
    {
      "epoch": 17.209119496855347,
      "grad_norm": 0.12627415359020233,
      "learning_rate": 2.908727749026191e-06,
      "loss": 0.4601,
      "num_input_tokens_seen": 21457248,
      "step": 32835
    },
    {
      "epoch": 17.21174004192872,
      "grad_norm": 0.22037523984909058,
      "learning_rate": 2.9033771398628756e-06,
      "loss": 0.4481,
      "num_input_tokens_seen": 21461056,
      "step": 32840
    },
    {
      "epoch": 17.214360587002098,
      "grad_norm": 0.10349477827548981,
      "learning_rate": 2.898031153060973e-06,
      "loss": 0.5712,
      "num_input_tokens_seen": 21466176,
      "step": 32845
    },
    {
      "epoch": 17.21698113207547,
      "grad_norm": 0.3127022385597229,
      "learning_rate": 2.8926897897388033e-06,
      "loss": 0.4098,
      "num_input_tokens_seen": 21469824,
      "step": 32850
    },
    {
      "epoch": 17.219601677148848,
      "grad_norm": 0.1800146847963333,
      "learning_rate": 2.8873530510137136e-06,
      "loss": 0.462,
      "num_input_tokens_seen": 21473472,
      "step": 32855
    },
    {
      "epoch": 17.22222222222222,
      "grad_norm": 0.14478863775730133,
      "learning_rate": 2.8820209380020892e-06,
      "loss": 0.4144,
      "num_input_tokens_seen": 21476640,
      "step": 32860
    },
    {
      "epoch": 17.2248427672956,
      "grad_norm": 0.17027468979358673,
      "learning_rate": 2.8766934518193566e-06,
      "loss": 0.4856,
      "num_input_tokens_seen": 21479712,
      "step": 32865
    },
    {
      "epoch": 17.22746331236897,
      "grad_norm": 0.14223051071166992,
      "learning_rate": 2.87137059357995e-06,
      "loss": 0.5239,
      "num_input_tokens_seen": 21483104,
      "step": 32870
    },
    {
      "epoch": 17.23008385744235,
      "grad_norm": 0.18270422518253326,
      "learning_rate": 2.8660523643973618e-06,
      "loss": 0.4378,
      "num_input_tokens_seen": 21486080,
      "step": 32875
    },
    {
      "epoch": 17.232704402515722,
      "grad_norm": 0.15347817540168762,
      "learning_rate": 2.860738765384105e-06,
      "loss": 0.4336,
      "num_input_tokens_seen": 21489792,
      "step": 32880
    },
    {
      "epoch": 17.2353249475891,
      "grad_norm": 0.15720930695533752,
      "learning_rate": 2.8554297976517154e-06,
      "loss": 0.4499,
      "num_input_tokens_seen": 21492704,
      "step": 32885
    },
    {
      "epoch": 17.237945492662472,
      "grad_norm": 0.09503760188817978,
      "learning_rate": 2.8501254623107707e-06,
      "loss": 0.4869,
      "num_input_tokens_seen": 21496992,
      "step": 32890
    },
    {
      "epoch": 17.24056603773585,
      "grad_norm": 0.2480195313692093,
      "learning_rate": 2.84482576047089e-06,
      "loss": 0.4626,
      "num_input_tokens_seen": 21500320,
      "step": 32895
    },
    {
      "epoch": 17.243186582809223,
      "grad_norm": 0.19332103431224823,
      "learning_rate": 2.839530693240694e-06,
      "loss": 0.5347,
      "num_input_tokens_seen": 21503584,
      "step": 32900
    },
    {
      "epoch": 17.2458071278826,
      "grad_norm": 0.2304137498140335,
      "learning_rate": 2.8342402617278606e-06,
      "loss": 0.5425,
      "num_input_tokens_seen": 21506240,
      "step": 32905
    },
    {
      "epoch": 17.248427672955973,
      "grad_norm": 0.19885867834091187,
      "learning_rate": 2.82895446703908e-06,
      "loss": 0.4096,
      "num_input_tokens_seen": 21510208,
      "step": 32910
    },
    {
      "epoch": 17.25104821802935,
      "grad_norm": 0.13031376898288727,
      "learning_rate": 2.823673310280084e-06,
      "loss": 0.3799,
      "num_input_tokens_seen": 21513344,
      "step": 32915
    },
    {
      "epoch": 17.253668763102727,
      "grad_norm": 0.22826631367206573,
      "learning_rate": 2.8183967925556305e-06,
      "loss": 0.4893,
      "num_input_tokens_seen": 21515968,
      "step": 32920
    },
    {
      "epoch": 17.2562893081761,
      "grad_norm": 0.16122207045555115,
      "learning_rate": 2.8131249149695115e-06,
      "loss": 0.4792,
      "num_input_tokens_seen": 21519200,
      "step": 32925
    },
    {
      "epoch": 17.258909853249477,
      "grad_norm": 0.17720140516757965,
      "learning_rate": 2.8078576786245397e-06,
      "loss": 0.5109,
      "num_input_tokens_seen": 21522368,
      "step": 32930
    },
    {
      "epoch": 17.26153039832285,
      "grad_norm": 0.16691285371780396,
      "learning_rate": 2.8025950846225557e-06,
      "loss": 0.4486,
      "num_input_tokens_seen": 21525280,
      "step": 32935
    },
    {
      "epoch": 17.264150943396228,
      "grad_norm": 0.260875940322876,
      "learning_rate": 2.7973371340644373e-06,
      "loss": 0.3889,
      "num_input_tokens_seen": 21528736,
      "step": 32940
    },
    {
      "epoch": 17.2667714884696,
      "grad_norm": 0.31854838132858276,
      "learning_rate": 2.7920838280500907e-06,
      "loss": 0.5568,
      "num_input_tokens_seen": 21531904,
      "step": 32945
    },
    {
      "epoch": 17.269392033542978,
      "grad_norm": 0.20102934539318085,
      "learning_rate": 2.7868351676784504e-06,
      "loss": 0.4567,
      "num_input_tokens_seen": 21534592,
      "step": 32950
    },
    {
      "epoch": 17.27201257861635,
      "grad_norm": 0.17190773785114288,
      "learning_rate": 2.7815911540474665e-06,
      "loss": 0.5439,
      "num_input_tokens_seen": 21538880,
      "step": 32955
    },
    {
      "epoch": 17.27463312368973,
      "grad_norm": 0.11841466277837753,
      "learning_rate": 2.7763517882541364e-06,
      "loss": 0.4235,
      "num_input_tokens_seen": 21542688,
      "step": 32960
    },
    {
      "epoch": 17.2772536687631,
      "grad_norm": 0.1777888834476471,
      "learning_rate": 2.771117071394466e-06,
      "loss": 0.4516,
      "num_input_tokens_seen": 21545760,
      "step": 32965
    },
    {
      "epoch": 17.27987421383648,
      "grad_norm": 0.1966712921857834,
      "learning_rate": 2.7658870045635055e-06,
      "loss": 0.4941,
      "num_input_tokens_seen": 21548736,
      "step": 32970
    },
    {
      "epoch": 17.282494758909852,
      "grad_norm": 0.24334816634655,
      "learning_rate": 2.7606615888553234e-06,
      "loss": 0.5846,
      "num_input_tokens_seen": 21551936,
      "step": 32975
    },
    {
      "epoch": 17.28511530398323,
      "grad_norm": 0.20189625024795532,
      "learning_rate": 2.755440825363023e-06,
      "loss": 0.5773,
      "num_input_tokens_seen": 21555296,
      "step": 32980
    },
    {
      "epoch": 17.287735849056602,
      "grad_norm": 0.16416232287883759,
      "learning_rate": 2.75022471517872e-06,
      "loss": 0.4815,
      "num_input_tokens_seen": 21558144,
      "step": 32985
    },
    {
      "epoch": 17.29035639412998,
      "grad_norm": 0.28293731808662415,
      "learning_rate": 2.745013259393564e-06,
      "loss": 0.6441,
      "num_input_tokens_seen": 21562048,
      "step": 32990
    },
    {
      "epoch": 17.292976939203353,
      "grad_norm": 0.13594332337379456,
      "learning_rate": 2.739806459097735e-06,
      "loss": 0.4903,
      "num_input_tokens_seen": 21565312,
      "step": 32995
    },
    {
      "epoch": 17.29559748427673,
      "grad_norm": 0.16347672045230865,
      "learning_rate": 2.7346043153804403e-06,
      "loss": 0.4721,
      "num_input_tokens_seen": 21568960,
      "step": 33000
    },
    {
      "epoch": 17.298218029350103,
      "grad_norm": 0.10509294271469116,
      "learning_rate": 2.729406829329903e-06,
      "loss": 0.4433,
      "num_input_tokens_seen": 21572064,
      "step": 33005
    },
    {
      "epoch": 17.30083857442348,
      "grad_norm": 0.17516610026359558,
      "learning_rate": 2.72421400203339e-06,
      "loss": 0.4787,
      "num_input_tokens_seen": 21575328,
      "step": 33010
    },
    {
      "epoch": 17.303459119496857,
      "grad_norm": 0.1263948380947113,
      "learning_rate": 2.719025834577163e-06,
      "loss": 0.4665,
      "num_input_tokens_seen": 21578912,
      "step": 33015
    },
    {
      "epoch": 17.30607966457023,
      "grad_norm": 0.17477737367153168,
      "learning_rate": 2.713842328046537e-06,
      "loss": 0.4681,
      "num_input_tokens_seen": 21582560,
      "step": 33020
    },
    {
      "epoch": 17.308700209643607,
      "grad_norm": 0.1987924575805664,
      "learning_rate": 2.7086634835258416e-06,
      "loss": 0.4775,
      "num_input_tokens_seen": 21586912,
      "step": 33025
    },
    {
      "epoch": 17.31132075471698,
      "grad_norm": 0.2622120678424835,
      "learning_rate": 2.7034893020984333e-06,
      "loss": 0.5389,
      "num_input_tokens_seen": 21589920,
      "step": 33030
    },
    {
      "epoch": 17.313941299790358,
      "grad_norm": 0.17201486229896545,
      "learning_rate": 2.6983197848466952e-06,
      "loss": 0.4423,
      "num_input_tokens_seen": 21592416,
      "step": 33035
    },
    {
      "epoch": 17.31656184486373,
      "grad_norm": 0.28882354497909546,
      "learning_rate": 2.6931549328520265e-06,
      "loss": 0.5494,
      "num_input_tokens_seen": 21596448,
      "step": 33040
    },
    {
      "epoch": 17.319182389937108,
      "grad_norm": 0.12786419689655304,
      "learning_rate": 2.6879947471948523e-06,
      "loss": 0.4804,
      "num_input_tokens_seen": 21599968,
      "step": 33045
    },
    {
      "epoch": 17.32180293501048,
      "grad_norm": 0.10930803418159485,
      "learning_rate": 2.6828392289546285e-06,
      "loss": 0.381,
      "num_input_tokens_seen": 21603168,
      "step": 33050
    },
    {
      "epoch": 17.32442348008386,
      "grad_norm": 0.11710986495018005,
      "learning_rate": 2.677688379209828e-06,
      "loss": 0.5817,
      "num_input_tokens_seen": 21606336,
      "step": 33055
    },
    {
      "epoch": 17.32704402515723,
      "grad_norm": 0.19008077681064606,
      "learning_rate": 2.6725421990379586e-06,
      "loss": 0.5118,
      "num_input_tokens_seen": 21609728,
      "step": 33060
    },
    {
      "epoch": 17.32966457023061,
      "grad_norm": 0.1449241042137146,
      "learning_rate": 2.667400689515537e-06,
      "loss": 0.5247,
      "num_input_tokens_seen": 21612448,
      "step": 33065
    },
    {
      "epoch": 17.332285115303982,
      "grad_norm": 0.15102753043174744,
      "learning_rate": 2.6622638517181e-06,
      "loss": 0.4222,
      "num_input_tokens_seen": 21615424,
      "step": 33070
    },
    {
      "epoch": 17.33490566037736,
      "grad_norm": 0.21424390375614166,
      "learning_rate": 2.657131686720224e-06,
      "loss": 0.4397,
      "num_input_tokens_seen": 21618400,
      "step": 33075
    },
    {
      "epoch": 17.337526205450732,
      "grad_norm": 0.22523747384548187,
      "learning_rate": 2.652004195595498e-06,
      "loss": 0.5158,
      "num_input_tokens_seen": 21622688,
      "step": 33080
    },
    {
      "epoch": 17.34014675052411,
      "grad_norm": 0.23059067130088806,
      "learning_rate": 2.6468813794165356e-06,
      "loss": 0.5493,
      "num_input_tokens_seen": 21626176,
      "step": 33085
    },
    {
      "epoch": 17.342767295597483,
      "grad_norm": 0.8535956740379333,
      "learning_rate": 2.641763239254977e-06,
      "loss": 0.3607,
      "num_input_tokens_seen": 21629440,
      "step": 33090
    },
    {
      "epoch": 17.34538784067086,
      "grad_norm": 0.14570093154907227,
      "learning_rate": 2.6366497761814713e-06,
      "loss": 0.6678,
      "num_input_tokens_seen": 21632000,
      "step": 33095
    },
    {
      "epoch": 17.348008385744233,
      "grad_norm": 0.10014166682958603,
      "learning_rate": 2.631540991265691e-06,
      "loss": 0.4573,
      "num_input_tokens_seen": 21635104,
      "step": 33100
    },
    {
      "epoch": 17.35062893081761,
      "grad_norm": 0.11041896045207977,
      "learning_rate": 2.6264368855763443e-06,
      "loss": 0.5907,
      "num_input_tokens_seen": 21638624,
      "step": 33105
    },
    {
      "epoch": 17.353249475890987,
      "grad_norm": 0.15860599279403687,
      "learning_rate": 2.6213374601811504e-06,
      "loss": 0.4224,
      "num_input_tokens_seen": 21641664,
      "step": 33110
    },
    {
      "epoch": 17.35587002096436,
      "grad_norm": 0.18896766006946564,
      "learning_rate": 2.616242716146855e-06,
      "loss": 0.5221,
      "num_input_tokens_seen": 21644672,
      "step": 33115
    },
    {
      "epoch": 17.358490566037737,
      "grad_norm": 0.24885781109333038,
      "learning_rate": 2.6111526545392145e-06,
      "loss": 0.455,
      "num_input_tokens_seen": 21647328,
      "step": 33120
    },
    {
      "epoch": 17.36111111111111,
      "grad_norm": 0.1288330852985382,
      "learning_rate": 2.6060672764230106e-06,
      "loss": 0.4158,
      "num_input_tokens_seen": 21650368,
      "step": 33125
    },
    {
      "epoch": 17.363731656184488,
      "grad_norm": 0.17148318886756897,
      "learning_rate": 2.600986582862047e-06,
      "loss": 0.6642,
      "num_input_tokens_seen": 21654400,
      "step": 33130
    },
    {
      "epoch": 17.36635220125786,
      "grad_norm": 0.2195356786251068,
      "learning_rate": 2.5959105749191498e-06,
      "loss": 0.4349,
      "num_input_tokens_seen": 21657856,
      "step": 33135
    },
    {
      "epoch": 17.368972746331238,
      "grad_norm": 0.12998712062835693,
      "learning_rate": 2.590839253656166e-06,
      "loss": 0.3442,
      "num_input_tokens_seen": 21661088,
      "step": 33140
    },
    {
      "epoch": 17.37159329140461,
      "grad_norm": 0.12142357230186462,
      "learning_rate": 2.5857726201339473e-06,
      "loss": 0.3612,
      "num_input_tokens_seen": 21666144,
      "step": 33145
    },
    {
      "epoch": 17.37421383647799,
      "grad_norm": 0.11427928507328033,
      "learning_rate": 2.5807106754123873e-06,
      "loss": 0.4527,
      "num_input_tokens_seen": 21669152,
      "step": 33150
    },
    {
      "epoch": 17.37683438155136,
      "grad_norm": 0.16084539890289307,
      "learning_rate": 2.575653420550378e-06,
      "loss": 0.4825,
      "num_input_tokens_seen": 21671840,
      "step": 33155
    },
    {
      "epoch": 17.37945492662474,
      "grad_norm": 0.16798502206802368,
      "learning_rate": 2.57060085660584e-06,
      "loss": 0.371,
      "num_input_tokens_seen": 21674400,
      "step": 33160
    },
    {
      "epoch": 17.382075471698112,
      "grad_norm": 0.220657616853714,
      "learning_rate": 2.565552984635716e-06,
      "loss": 0.4061,
      "num_input_tokens_seen": 21677280,
      "step": 33165
    },
    {
      "epoch": 17.38469601677149,
      "grad_norm": 0.2003779262304306,
      "learning_rate": 2.5605098056959652e-06,
      "loss": 0.6554,
      "num_input_tokens_seen": 21680352,
      "step": 33170
    },
    {
      "epoch": 17.387316561844862,
      "grad_norm": 0.12047620117664337,
      "learning_rate": 2.555471320841557e-06,
      "loss": 0.4853,
      "num_input_tokens_seen": 21684768,
      "step": 33175
    },
    {
      "epoch": 17.38993710691824,
      "grad_norm": 0.24589575827121735,
      "learning_rate": 2.55043753112649e-06,
      "loss": 0.6554,
      "num_input_tokens_seen": 21687776,
      "step": 33180
    },
    {
      "epoch": 17.392557651991613,
      "grad_norm": 0.14151091873645782,
      "learning_rate": 2.5454084376037717e-06,
      "loss": 0.3943,
      "num_input_tokens_seen": 21691392,
      "step": 33185
    },
    {
      "epoch": 17.39517819706499,
      "grad_norm": 0.2548291087150574,
      "learning_rate": 2.5403840413254304e-06,
      "loss": 0.6562,
      "num_input_tokens_seen": 21694016,
      "step": 33190
    },
    {
      "epoch": 17.397798742138363,
      "grad_norm": 0.1593877524137497,
      "learning_rate": 2.5353643433425196e-06,
      "loss": 0.5104,
      "num_input_tokens_seen": 21697152,
      "step": 33195
    },
    {
      "epoch": 17.40041928721174,
      "grad_norm": 0.09748691320419312,
      "learning_rate": 2.5303493447050965e-06,
      "loss": 0.5661,
      "num_input_tokens_seen": 21701504,
      "step": 33200
    },
    {
      "epoch": 17.403039832285117,
      "grad_norm": 0.18077223002910614,
      "learning_rate": 2.525339046462244e-06,
      "loss": 0.6784,
      "num_input_tokens_seen": 21704640,
      "step": 33205
    },
    {
      "epoch": 17.40566037735849,
      "grad_norm": 0.11752818524837494,
      "learning_rate": 2.5203334496620544e-06,
      "loss": 0.3857,
      "num_input_tokens_seen": 21708160,
      "step": 33210
    },
    {
      "epoch": 17.408280922431867,
      "grad_norm": 0.1903447061777115,
      "learning_rate": 2.515332555351646e-06,
      "loss": 0.4904,
      "num_input_tokens_seen": 21710976,
      "step": 33215
    },
    {
      "epoch": 17.41090146750524,
      "grad_norm": 0.12484502792358398,
      "learning_rate": 2.510336364577154e-06,
      "loss": 0.7005,
      "num_input_tokens_seen": 21714368,
      "step": 33220
    },
    {
      "epoch": 17.413522012578618,
      "grad_norm": 0.14572420716285706,
      "learning_rate": 2.5053448783837143e-06,
      "loss": 0.5839,
      "num_input_tokens_seen": 21716896,
      "step": 33225
    },
    {
      "epoch": 17.41614255765199,
      "grad_norm": 0.16519233584403992,
      "learning_rate": 2.5003580978154917e-06,
      "loss": 0.4919,
      "num_input_tokens_seen": 21719488,
      "step": 33230
    },
    {
      "epoch": 17.418763102725368,
      "grad_norm": 0.195540651679039,
      "learning_rate": 2.4953760239156716e-06,
      "loss": 0.4855,
      "num_input_tokens_seen": 21724000,
      "step": 33235
    },
    {
      "epoch": 17.42138364779874,
      "grad_norm": 0.1460888385772705,
      "learning_rate": 2.4903986577264366e-06,
      "loss": 0.4564,
      "num_input_tokens_seen": 21727424,
      "step": 33240
    },
    {
      "epoch": 17.42400419287212,
      "grad_norm": 0.15318144857883453,
      "learning_rate": 2.4854260002889986e-06,
      "loss": 0.4191,
      "num_input_tokens_seen": 21732032,
      "step": 33245
    },
    {
      "epoch": 17.42662473794549,
      "grad_norm": 0.12296661734580994,
      "learning_rate": 2.480458052643586e-06,
      "loss": 0.6613,
      "num_input_tokens_seen": 21735488,
      "step": 33250
    },
    {
      "epoch": 17.42924528301887,
      "grad_norm": 0.2091701626777649,
      "learning_rate": 2.4754948158294293e-06,
      "loss": 0.4978,
      "num_input_tokens_seen": 21737984,
      "step": 33255
    },
    {
      "epoch": 17.431865828092242,
      "grad_norm": 0.15894171595573425,
      "learning_rate": 2.4705362908847814e-06,
      "loss": 0.5138,
      "num_input_tokens_seen": 21741088,
      "step": 33260
    },
    {
      "epoch": 17.43448637316562,
      "grad_norm": 0.1018497496843338,
      "learning_rate": 2.4655824788469172e-06,
      "loss": 0.5625,
      "num_input_tokens_seen": 21744928,
      "step": 33265
    },
    {
      "epoch": 17.437106918238992,
      "grad_norm": 0.2915990650653839,
      "learning_rate": 2.4606333807521087e-06,
      "loss": 0.5891,
      "num_input_tokens_seen": 21748288,
      "step": 33270
    },
    {
      "epoch": 17.43972746331237,
      "grad_norm": 0.21268314123153687,
      "learning_rate": 2.455688997635658e-06,
      "loss": 0.4728,
      "num_input_tokens_seen": 21751968,
      "step": 33275
    },
    {
      "epoch": 17.442348008385743,
      "grad_norm": 0.15131288766860962,
      "learning_rate": 2.4507493305318652e-06,
      "loss": 0.4543,
      "num_input_tokens_seen": 21755648,
      "step": 33280
    },
    {
      "epoch": 17.44496855345912,
      "grad_norm": 0.2311946153640747,
      "learning_rate": 2.445814380474057e-06,
      "loss": 0.4621,
      "num_input_tokens_seen": 21759040,
      "step": 33285
    },
    {
      "epoch": 17.447589098532493,
      "grad_norm": 0.16643647849559784,
      "learning_rate": 2.4408841484945684e-06,
      "loss": 0.4374,
      "num_input_tokens_seen": 21762272,
      "step": 33290
    },
    {
      "epoch": 17.45020964360587,
      "grad_norm": 0.2908034324645996,
      "learning_rate": 2.435958635624755e-06,
      "loss": 0.6092,
      "num_input_tokens_seen": 21765600,
      "step": 33295
    },
    {
      "epoch": 17.452830188679247,
      "grad_norm": 0.23106276988983154,
      "learning_rate": 2.4310378428949678e-06,
      "loss": 0.4813,
      "num_input_tokens_seen": 21768032,
      "step": 33300
    },
    {
      "epoch": 17.45545073375262,
      "grad_norm": 0.2924659550189972,
      "learning_rate": 2.4261217713345804e-06,
      "loss": 0.5672,
      "num_input_tokens_seen": 21770304,
      "step": 33305
    },
    {
      "epoch": 17.458071278825997,
      "grad_norm": 0.1416982114315033,
      "learning_rate": 2.421210421971981e-06,
      "loss": 0.4495,
      "num_input_tokens_seen": 21774016,
      "step": 33310
    },
    {
      "epoch": 17.46069182389937,
      "grad_norm": 0.14517278969287872,
      "learning_rate": 2.416303795834571e-06,
      "loss": 0.5814,
      "num_input_tokens_seen": 21778048,
      "step": 33315
    },
    {
      "epoch": 17.463312368972748,
      "grad_norm": 0.21075539290905,
      "learning_rate": 2.4114018939487615e-06,
      "loss": 0.3059,
      "num_input_tokens_seen": 21780736,
      "step": 33320
    },
    {
      "epoch": 17.46593291404612,
      "grad_norm": 0.17625048756599426,
      "learning_rate": 2.4065047173399662e-06,
      "loss": 0.5256,
      "num_input_tokens_seen": 21784704,
      "step": 33325
    },
    {
      "epoch": 17.468553459119498,
      "grad_norm": 0.10506114363670349,
      "learning_rate": 2.4016122670326295e-06,
      "loss": 0.3706,
      "num_input_tokens_seen": 21788096,
      "step": 33330
    },
    {
      "epoch": 17.47117400419287,
      "grad_norm": 0.15324538946151733,
      "learning_rate": 2.396724544050186e-06,
      "loss": 0.378,
      "num_input_tokens_seen": 21791808,
      "step": 33335
    },
    {
      "epoch": 17.47379454926625,
      "grad_norm": 0.16632084548473358,
      "learning_rate": 2.3918415494150944e-06,
      "loss": 0.6394,
      "num_input_tokens_seen": 21795072,
      "step": 33340
    },
    {
      "epoch": 17.47641509433962,
      "grad_norm": 0.16031226515769958,
      "learning_rate": 2.3869632841488245e-06,
      "loss": 0.4658,
      "num_input_tokens_seen": 21798016,
      "step": 33345
    },
    {
      "epoch": 17.479035639413,
      "grad_norm": 0.1275802105665207,
      "learning_rate": 2.382089749271857e-06,
      "loss": 0.4243,
      "num_input_tokens_seen": 21801632,
      "step": 33350
    },
    {
      "epoch": 17.481656184486372,
      "grad_norm": 0.2524811029434204,
      "learning_rate": 2.3772209458036737e-06,
      "loss": 0.3859,
      "num_input_tokens_seen": 21804416,
      "step": 33355
    },
    {
      "epoch": 17.48427672955975,
      "grad_norm": 0.2027893215417862,
      "learning_rate": 2.372356874762771e-06,
      "loss": 0.5499,
      "num_input_tokens_seen": 21808160,
      "step": 33360
    },
    {
      "epoch": 17.486897274633122,
      "grad_norm": 0.32321691513061523,
      "learning_rate": 2.36749753716666e-06,
      "loss": 0.5526,
      "num_input_tokens_seen": 21810976,
      "step": 33365
    },
    {
      "epoch": 17.4895178197065,
      "grad_norm": 0.12373635172843933,
      "learning_rate": 2.3626429340318613e-06,
      "loss": 0.4733,
      "num_input_tokens_seen": 21814880,
      "step": 33370
    },
    {
      "epoch": 17.492138364779873,
      "grad_norm": 0.2002386599779129,
      "learning_rate": 2.357793066373901e-06,
      "loss": 0.4166,
      "num_input_tokens_seen": 21817792,
      "step": 33375
    },
    {
      "epoch": 17.49475890985325,
      "grad_norm": 0.15733328461647034,
      "learning_rate": 2.352947935207317e-06,
      "loss": 0.369,
      "num_input_tokens_seen": 21821312,
      "step": 33380
    },
    {
      "epoch": 17.497379454926623,
      "grad_norm": 0.11073945462703705,
      "learning_rate": 2.3481075415456495e-06,
      "loss": 0.4441,
      "num_input_tokens_seen": 21825664,
      "step": 33385
    },
    {
      "epoch": 17.5,
      "grad_norm": 0.15068408846855164,
      "learning_rate": 2.3432718864014576e-06,
      "loss": 0.5175,
      "num_input_tokens_seen": 21829088,
      "step": 33390
    },
    {
      "epoch": 17.502620545073377,
      "grad_norm": 0.15733593702316284,
      "learning_rate": 2.338440970786304e-06,
      "loss": 0.4857,
      "num_input_tokens_seen": 21832000,
      "step": 33395
    },
    {
      "epoch": 17.50524109014675,
      "grad_norm": 0.13297392427921295,
      "learning_rate": 2.3336147957107644e-06,
      "loss": 0.4178,
      "num_input_tokens_seen": 21834816,
      "step": 33400
    },
    {
      "epoch": 17.507861635220127,
      "grad_norm": 0.13888223469257355,
      "learning_rate": 2.3287933621844227e-06,
      "loss": 0.474,
      "num_input_tokens_seen": 21837184,
      "step": 33405
    },
    {
      "epoch": 17.5104821802935,
      "grad_norm": 0.1410173922777176,
      "learning_rate": 2.3239766712158633e-06,
      "loss": 0.489,
      "num_input_tokens_seen": 21842080,
      "step": 33410
    },
    {
      "epoch": 17.513102725366878,
      "grad_norm": 0.20150205492973328,
      "learning_rate": 2.319164723812675e-06,
      "loss": 0.4773,
      "num_input_tokens_seen": 21844736,
      "step": 33415
    },
    {
      "epoch": 17.51572327044025,
      "grad_norm": 0.12171954661607742,
      "learning_rate": 2.314357520981472e-06,
      "loss": 0.4947,
      "num_input_tokens_seen": 21848736,
      "step": 33420
    },
    {
      "epoch": 17.518343815513628,
      "grad_norm": 0.2452259659767151,
      "learning_rate": 2.309555063727864e-06,
      "loss": 0.4042,
      "num_input_tokens_seen": 21851808,
      "step": 33425
    },
    {
      "epoch": 17.520964360587,
      "grad_norm": 0.15441347658634186,
      "learning_rate": 2.304757353056475e-06,
      "loss": 0.4557,
      "num_input_tokens_seen": 21854752,
      "step": 33430
    },
    {
      "epoch": 17.52358490566038,
      "grad_norm": 0.2053976058959961,
      "learning_rate": 2.2999643899709267e-06,
      "loss": 0.4038,
      "num_input_tokens_seen": 21858016,
      "step": 33435
    },
    {
      "epoch": 17.52620545073375,
      "grad_norm": 0.11343531310558319,
      "learning_rate": 2.2951761754738456e-06,
      "loss": 0.351,
      "num_input_tokens_seen": 21860928,
      "step": 33440
    },
    {
      "epoch": 17.52882599580713,
      "grad_norm": 0.20066866278648376,
      "learning_rate": 2.29039271056688e-06,
      "loss": 0.3791,
      "num_input_tokens_seen": 21864416,
      "step": 33445
    },
    {
      "epoch": 17.531446540880502,
      "grad_norm": 0.11152440309524536,
      "learning_rate": 2.285613996250674e-06,
      "loss": 0.4728,
      "num_input_tokens_seen": 21867808,
      "step": 33450
    },
    {
      "epoch": 17.53406708595388,
      "grad_norm": 0.11258520185947418,
      "learning_rate": 2.2808400335248865e-06,
      "loss": 0.5408,
      "num_input_tokens_seen": 21871008,
      "step": 33455
    },
    {
      "epoch": 17.536687631027252,
      "grad_norm": 0.2766547203063965,
      "learning_rate": 2.2760708233881654e-06,
      "loss": 0.4338,
      "num_input_tokens_seen": 21873984,
      "step": 33460
    },
    {
      "epoch": 17.53930817610063,
      "grad_norm": 0.19967293739318848,
      "learning_rate": 2.271306366838183e-06,
      "loss": 0.4266,
      "num_input_tokens_seen": 21877216,
      "step": 33465
    },
    {
      "epoch": 17.541928721174003,
      "grad_norm": 0.17263388633728027,
      "learning_rate": 2.266546664871602e-06,
      "loss": 0.3247,
      "num_input_tokens_seen": 21880544,
      "step": 33470
    },
    {
      "epoch": 17.54454926624738,
      "grad_norm": 0.2969203591346741,
      "learning_rate": 2.261791718484102e-06,
      "loss": 0.7291,
      "num_input_tokens_seen": 21883296,
      "step": 33475
    },
    {
      "epoch": 17.547169811320753,
      "grad_norm": 0.2000511884689331,
      "learning_rate": 2.257041528670362e-06,
      "loss": 0.5139,
      "num_input_tokens_seen": 21885888,
      "step": 33480
    },
    {
      "epoch": 17.54979035639413,
      "grad_norm": 0.14426343142986298,
      "learning_rate": 2.2522960964240753e-06,
      "loss": 0.4818,
      "num_input_tokens_seen": 21888736,
      "step": 33485
    },
    {
      "epoch": 17.552410901467507,
      "grad_norm": 0.2296229600906372,
      "learning_rate": 2.247555422737924e-06,
      "loss": 0.6203,
      "num_input_tokens_seen": 21891264,
      "step": 33490
    },
    {
      "epoch": 17.55503144654088,
      "grad_norm": 0.142706036567688,
      "learning_rate": 2.2428195086036024e-06,
      "loss": 0.7623,
      "num_input_tokens_seen": 21893760,
      "step": 33495
    },
    {
      "epoch": 17.557651991614257,
      "grad_norm": 0.1412694901227951,
      "learning_rate": 2.238088355011811e-06,
      "loss": 0.4887,
      "num_input_tokens_seen": 21897280,
      "step": 33500
    },
    {
      "epoch": 17.56027253668763,
      "grad_norm": 0.13324761390686035,
      "learning_rate": 2.2333619629522518e-06,
      "loss": 0.4553,
      "num_input_tokens_seen": 21900640,
      "step": 33505
    },
    {
      "epoch": 17.562893081761008,
      "grad_norm": 0.18127226829528809,
      "learning_rate": 2.2286403334136424e-06,
      "loss": 0.4552,
      "num_input_tokens_seen": 21903744,
      "step": 33510
    },
    {
      "epoch": 17.56551362683438,
      "grad_norm": 0.2286238670349121,
      "learning_rate": 2.223923467383679e-06,
      "loss": 0.4123,
      "num_input_tokens_seen": 21905984,
      "step": 33515
    },
    {
      "epoch": 17.568134171907758,
      "grad_norm": 0.1010180413722992,
      "learning_rate": 2.2192113658490886e-06,
      "loss": 0.496,
      "num_input_tokens_seen": 21910272,
      "step": 33520
    },
    {
      "epoch": 17.57075471698113,
      "grad_norm": 0.1919795572757721,
      "learning_rate": 2.2145040297955776e-06,
      "loss": 0.6434,
      "num_input_tokens_seen": 21913216,
      "step": 33525
    },
    {
      "epoch": 17.57337526205451,
      "grad_norm": 0.22669905424118042,
      "learning_rate": 2.209801460207875e-06,
      "loss": 0.3496,
      "num_input_tokens_seen": 21916128,
      "step": 33530
    },
    {
      "epoch": 17.57599580712788,
      "grad_norm": 0.20668266713619232,
      "learning_rate": 2.2051036580697042e-06,
      "loss": 0.366,
      "num_input_tokens_seen": 21918592,
      "step": 33535
    },
    {
      "epoch": 17.57861635220126,
      "grad_norm": 0.16661126911640167,
      "learning_rate": 2.2004106243637885e-06,
      "loss": 0.4995,
      "num_input_tokens_seen": 21921216,
      "step": 33540
    },
    {
      "epoch": 17.581236897274632,
      "grad_norm": 0.13472990691661835,
      "learning_rate": 2.1957223600718587e-06,
      "loss": 0.468,
      "num_input_tokens_seen": 21925056,
      "step": 33545
    },
    {
      "epoch": 17.58385744234801,
      "grad_norm": 0.16862082481384277,
      "learning_rate": 2.1910388661746493e-06,
      "loss": 0.4657,
      "num_input_tokens_seen": 21928320,
      "step": 33550
    },
    {
      "epoch": 17.586477987421382,
      "grad_norm": 0.205536350607872,
      "learning_rate": 2.186360143651886e-06,
      "loss": 0.5029,
      "num_input_tokens_seen": 21931968,
      "step": 33555
    },
    {
      "epoch": 17.58909853249476,
      "grad_norm": 0.14326289296150208,
      "learning_rate": 2.1816861934823103e-06,
      "loss": 0.427,
      "num_input_tokens_seen": 21935008,
      "step": 33560
    },
    {
      "epoch": 17.591719077568133,
      "grad_norm": 0.18250468373298645,
      "learning_rate": 2.1770170166436595e-06,
      "loss": 0.4944,
      "num_input_tokens_seen": 21937344,
      "step": 33565
    },
    {
      "epoch": 17.59433962264151,
      "grad_norm": 0.23608708381652832,
      "learning_rate": 2.172352614112669e-06,
      "loss": 0.4722,
      "num_input_tokens_seen": 21939936,
      "step": 33570
    },
    {
      "epoch": 17.596960167714883,
      "grad_norm": 0.2685841917991638,
      "learning_rate": 2.167692986865083e-06,
      "loss": 0.4534,
      "num_input_tokens_seen": 21942400,
      "step": 33575
    },
    {
      "epoch": 17.59958071278826,
      "grad_norm": 0.375129371881485,
      "learning_rate": 2.1630381358756362e-06,
      "loss": 0.4488,
      "num_input_tokens_seen": 21945632,
      "step": 33580
    },
    {
      "epoch": 17.602201257861637,
      "grad_norm": 0.29337093234062195,
      "learning_rate": 2.158388062118072e-06,
      "loss": 0.5944,
      "num_input_tokens_seen": 21949312,
      "step": 33585
    },
    {
      "epoch": 17.60482180293501,
      "grad_norm": 0.11205004900693893,
      "learning_rate": 2.1537427665651393e-06,
      "loss": 0.4123,
      "num_input_tokens_seen": 21955872,
      "step": 33590
    },
    {
      "epoch": 17.607442348008387,
      "grad_norm": 0.13745400309562683,
      "learning_rate": 2.1491022501885726e-06,
      "loss": 0.52,
      "num_input_tokens_seen": 21959712,
      "step": 33595
    },
    {
      "epoch": 17.61006289308176,
      "grad_norm": 0.10875708609819412,
      "learning_rate": 2.1444665139591174e-06,
      "loss": 0.6043,
      "num_input_tokens_seen": 21963136,
      "step": 33600
    },
    {
      "epoch": 17.612683438155138,
      "grad_norm": 0.13708893954753876,
      "learning_rate": 2.139835558846523e-06,
      "loss": 0.609,
      "num_input_tokens_seen": 21966912,
      "step": 33605
    },
    {
      "epoch": 17.61530398322851,
      "grad_norm": 0.17589008808135986,
      "learning_rate": 2.1352093858195225e-06,
      "loss": 0.558,
      "num_input_tokens_seen": 21969152,
      "step": 33610
    },
    {
      "epoch": 17.617924528301888,
      "grad_norm": 0.33651238679885864,
      "learning_rate": 2.1305879958458676e-06,
      "loss": 0.5714,
      "num_input_tokens_seen": 21971744,
      "step": 33615
    },
    {
      "epoch": 17.62054507337526,
      "grad_norm": 0.22936321794986725,
      "learning_rate": 2.1259713898922894e-06,
      "loss": 0.3779,
      "num_input_tokens_seen": 21974592,
      "step": 33620
    },
    {
      "epoch": 17.62316561844864,
      "grad_norm": 0.187221959233284,
      "learning_rate": 2.1213595689245386e-06,
      "loss": 0.4646,
      "num_input_tokens_seen": 21978528,
      "step": 33625
    },
    {
      "epoch": 17.62578616352201,
      "grad_norm": 0.2291717380285263,
      "learning_rate": 2.1167525339073513e-06,
      "loss": 0.4973,
      "num_input_tokens_seen": 21981792,
      "step": 33630
    },
    {
      "epoch": 17.62840670859539,
      "grad_norm": 0.2218627631664276,
      "learning_rate": 2.112150285804473e-06,
      "loss": 0.3939,
      "num_input_tokens_seen": 21985344,
      "step": 33635
    },
    {
      "epoch": 17.631027253668762,
      "grad_norm": 0.12091799825429916,
      "learning_rate": 2.107552825578632e-06,
      "loss": 0.5353,
      "num_input_tokens_seen": 21988352,
      "step": 33640
    },
    {
      "epoch": 17.63364779874214,
      "grad_norm": 0.14854782819747925,
      "learning_rate": 2.102960154191572e-06,
      "loss": 0.451,
      "num_input_tokens_seen": 21992608,
      "step": 33645
    },
    {
      "epoch": 17.636268343815512,
      "grad_norm": 0.09119328111410141,
      "learning_rate": 2.098372272604021e-06,
      "loss": 0.4208,
      "num_input_tokens_seen": 21996160,
      "step": 33650
    },
    {
      "epoch": 17.63888888888889,
      "grad_norm": 0.1126469224691391,
      "learning_rate": 2.0937891817757155e-06,
      "loss": 0.4436,
      "num_input_tokens_seen": 21999712,
      "step": 33655
    },
    {
      "epoch": 17.641509433962263,
      "grad_norm": 0.1537744402885437,
      "learning_rate": 2.089210882665385e-06,
      "loss": 0.4269,
      "num_input_tokens_seen": 22002432,
      "step": 33660
    },
    {
      "epoch": 17.64412997903564,
      "grad_norm": 0.25898444652557373,
      "learning_rate": 2.0846373762307635e-06,
      "loss": 0.6336,
      "num_input_tokens_seen": 22004992,
      "step": 33665
    },
    {
      "epoch": 17.646750524109013,
      "grad_norm": 0.11981324851512909,
      "learning_rate": 2.080068663428569e-06,
      "loss": 0.4428,
      "num_input_tokens_seen": 22008480,
      "step": 33670
    },
    {
      "epoch": 17.64937106918239,
      "grad_norm": 0.16288791596889496,
      "learning_rate": 2.075504745214521e-06,
      "loss": 0.485,
      "num_input_tokens_seen": 22011968,
      "step": 33675
    },
    {
      "epoch": 17.651991614255767,
      "grad_norm": 0.22926466166973114,
      "learning_rate": 2.0709456225433437e-06,
      "loss": 0.6323,
      "num_input_tokens_seen": 22015424,
      "step": 33680
    },
    {
      "epoch": 17.65461215932914,
      "grad_norm": 0.1040513664484024,
      "learning_rate": 2.0663912963687554e-06,
      "loss": 0.4138,
      "num_input_tokens_seen": 22019776,
      "step": 33685
    },
    {
      "epoch": 17.657232704402517,
      "grad_norm": 0.21705982089042664,
      "learning_rate": 2.0618417676434713e-06,
      "loss": 0.5052,
      "num_input_tokens_seen": 22023072,
      "step": 33690
    },
    {
      "epoch": 17.65985324947589,
      "grad_norm": 0.2581614553928375,
      "learning_rate": 2.0572970373191975e-06,
      "loss": 0.4578,
      "num_input_tokens_seen": 22026272,
      "step": 33695
    },
    {
      "epoch": 17.662473794549268,
      "grad_norm": 0.16462832689285278,
      "learning_rate": 2.052757106346634e-06,
      "loss": 0.5485,
      "num_input_tokens_seen": 22029216,
      "step": 33700
    },
    {
      "epoch": 17.66509433962264,
      "grad_norm": 0.134621262550354,
      "learning_rate": 2.048221975675488e-06,
      "loss": 0.477,
      "num_input_tokens_seen": 22032416,
      "step": 33705
    },
    {
      "epoch": 17.667714884696018,
      "grad_norm": 0.17292365431785583,
      "learning_rate": 2.043691646254456e-06,
      "loss": 0.3551,
      "num_input_tokens_seen": 22035488,
      "step": 33710
    },
    {
      "epoch": 17.67033542976939,
      "grad_norm": 0.16752055287361145,
      "learning_rate": 2.039166119031233e-06,
      "loss": 0.5662,
      "num_input_tokens_seen": 22039232,
      "step": 33715
    },
    {
      "epoch": 17.67295597484277,
      "grad_norm": 0.22251485288143158,
      "learning_rate": 2.034645394952511e-06,
      "loss": 0.5262,
      "num_input_tokens_seen": 22041888,
      "step": 33720
    },
    {
      "epoch": 17.67557651991614,
      "grad_norm": 0.15354272723197937,
      "learning_rate": 2.030129474963971e-06,
      "loss": 0.5126,
      "num_input_tokens_seen": 22044416,
      "step": 33725
    },
    {
      "epoch": 17.67819706498952,
      "grad_norm": 0.07318726181983948,
      "learning_rate": 2.025618360010284e-06,
      "loss": 0.3897,
      "num_input_tokens_seen": 22047552,
      "step": 33730
    },
    {
      "epoch": 17.680817610062892,
      "grad_norm": 0.21177077293395996,
      "learning_rate": 2.0211120510351317e-06,
      "loss": 0.5015,
      "num_input_tokens_seen": 22049856,
      "step": 33735
    },
    {
      "epoch": 17.68343815513627,
      "grad_norm": 0.23233607411384583,
      "learning_rate": 2.0166105489811822e-06,
      "loss": 0.5007,
      "num_input_tokens_seen": 22052672,
      "step": 33740
    },
    {
      "epoch": 17.686058700209642,
      "grad_norm": 0.12823227047920227,
      "learning_rate": 2.0121138547901016e-06,
      "loss": 0.5849,
      "num_input_tokens_seen": 22055392,
      "step": 33745
    },
    {
      "epoch": 17.68867924528302,
      "grad_norm": 0.17525920271873474,
      "learning_rate": 2.0076219694025427e-06,
      "loss": 0.4959,
      "num_input_tokens_seen": 22058176,
      "step": 33750
    },
    {
      "epoch": 17.691299790356393,
      "grad_norm": 0.2487252801656723,
      "learning_rate": 2.003134893758152e-06,
      "loss": 0.4039,
      "num_input_tokens_seen": 22061152,
      "step": 33755
    },
    {
      "epoch": 17.69392033542977,
      "grad_norm": 0.16523000597953796,
      "learning_rate": 1.9986526287955805e-06,
      "loss": 0.3961,
      "num_input_tokens_seen": 22064288,
      "step": 33760
    },
    {
      "epoch": 17.696540880503143,
      "grad_norm": 0.18768151104450226,
      "learning_rate": 1.9941751754524644e-06,
      "loss": 0.5216,
      "num_input_tokens_seen": 22067008,
      "step": 33765
    },
    {
      "epoch": 17.69916142557652,
      "grad_norm": 0.22159169614315033,
      "learning_rate": 1.9897025346654412e-06,
      "loss": 0.4136,
      "num_input_tokens_seen": 22070400,
      "step": 33770
    },
    {
      "epoch": 17.701781970649897,
      "grad_norm": 0.10737475007772446,
      "learning_rate": 1.985234707370134e-06,
      "loss": 0.5664,
      "num_input_tokens_seen": 22073312,
      "step": 33775
    },
    {
      "epoch": 17.70440251572327,
      "grad_norm": 0.25236308574676514,
      "learning_rate": 1.9807716945011544e-06,
      "loss": 0.5016,
      "num_input_tokens_seen": 22076480,
      "step": 33780
    },
    {
      "epoch": 17.707023060796647,
      "grad_norm": 0.15833452343940735,
      "learning_rate": 1.9763134969921215e-06,
      "loss": 0.416,
      "num_input_tokens_seen": 22079968,
      "step": 33785
    },
    {
      "epoch": 17.70964360587002,
      "grad_norm": 0.13696512579917908,
      "learning_rate": 1.971860115775634e-06,
      "loss": 0.481,
      "num_input_tokens_seen": 22083136,
      "step": 33790
    },
    {
      "epoch": 17.712264150943398,
      "grad_norm": 0.1374780684709549,
      "learning_rate": 1.967411551783294e-06,
      "loss": 0.5111,
      "num_input_tokens_seen": 22086400,
      "step": 33795
    },
    {
      "epoch": 17.71488469601677,
      "grad_norm": 0.16204078495502472,
      "learning_rate": 1.9629678059456895e-06,
      "loss": 0.5711,
      "num_input_tokens_seen": 22089632,
      "step": 33800
    },
    {
      "epoch": 17.717505241090148,
      "grad_norm": 0.20611700415611267,
      "learning_rate": 1.9585288791924004e-06,
      "loss": 0.4692,
      "num_input_tokens_seen": 22092576,
      "step": 33805
    },
    {
      "epoch": 17.72012578616352,
      "grad_norm": 0.1650405079126358,
      "learning_rate": 1.9540947724519947e-06,
      "loss": 0.4107,
      "num_input_tokens_seen": 22096768,
      "step": 33810
    },
    {
      "epoch": 17.7227463312369,
      "grad_norm": 0.3581060469150543,
      "learning_rate": 1.9496654866520414e-06,
      "loss": 0.5898,
      "num_input_tokens_seen": 22099296,
      "step": 33815
    },
    {
      "epoch": 17.72536687631027,
      "grad_norm": 0.19193188846111298,
      "learning_rate": 1.9452410227190937e-06,
      "loss": 0.4676,
      "num_input_tokens_seen": 22101888,
      "step": 33820
    },
    {
      "epoch": 17.72798742138365,
      "grad_norm": 0.17535124719142914,
      "learning_rate": 1.9408213815787087e-06,
      "loss": 0.5974,
      "num_input_tokens_seen": 22105568,
      "step": 33825
    },
    {
      "epoch": 17.730607966457022,
      "grad_norm": 0.20257268846035004,
      "learning_rate": 1.936406564155413e-06,
      "loss": 0.5121,
      "num_input_tokens_seen": 22109248,
      "step": 33830
    },
    {
      "epoch": 17.7332285115304,
      "grad_norm": 0.0939941480755806,
      "learning_rate": 1.9319965713727435e-06,
      "loss": 0.4287,
      "num_input_tokens_seen": 22112352,
      "step": 33835
    },
    {
      "epoch": 17.735849056603772,
      "grad_norm": 0.13477297127246857,
      "learning_rate": 1.9275914041532144e-06,
      "loss": 0.5039,
      "num_input_tokens_seen": 22115200,
      "step": 33840
    },
    {
      "epoch": 17.73846960167715,
      "grad_norm": 0.20422466099262238,
      "learning_rate": 1.9231910634183394e-06,
      "loss": 0.5308,
      "num_input_tokens_seen": 22118272,
      "step": 33845
    },
    {
      "epoch": 17.741090146750523,
      "grad_norm": 0.14847803115844727,
      "learning_rate": 1.918795550088623e-06,
      "loss": 0.4752,
      "num_input_tokens_seen": 22121248,
      "step": 33850
    },
    {
      "epoch": 17.7437106918239,
      "grad_norm": 0.17626655101776123,
      "learning_rate": 1.9144048650835496e-06,
      "loss": 0.5144,
      "num_input_tokens_seen": 22124416,
      "step": 33855
    },
    {
      "epoch": 17.746331236897273,
      "grad_norm": 0.13367237150669098,
      "learning_rate": 1.9100190093216093e-06,
      "loss": 0.4157,
      "num_input_tokens_seen": 22127264,
      "step": 33860
    },
    {
      "epoch": 17.74895178197065,
      "grad_norm": 0.1427737921476364,
      "learning_rate": 1.9056379837202627e-06,
      "loss": 0.4757,
      "num_input_tokens_seen": 22130816,
      "step": 33865
    },
    {
      "epoch": 17.751572327044027,
      "grad_norm": 0.1327209621667862,
      "learning_rate": 1.9012617891959766e-06,
      "loss": 0.5071,
      "num_input_tokens_seen": 22135200,
      "step": 33870
    },
    {
      "epoch": 17.7541928721174,
      "grad_norm": 0.17067267000675201,
      "learning_rate": 1.8968904266641996e-06,
      "loss": 0.5913,
      "num_input_tokens_seen": 22138304,
      "step": 33875
    },
    {
      "epoch": 17.756813417190777,
      "grad_norm": 0.17746418714523315,
      "learning_rate": 1.892523897039372e-06,
      "loss": 0.347,
      "num_input_tokens_seen": 22141024,
      "step": 33880
    },
    {
      "epoch": 17.75943396226415,
      "grad_norm": 0.32264718413352966,
      "learning_rate": 1.888162201234922e-06,
      "loss": 0.4952,
      "num_input_tokens_seen": 22144800,
      "step": 33885
    },
    {
      "epoch": 17.762054507337528,
      "grad_norm": 0.22527629137039185,
      "learning_rate": 1.8838053401632666e-06,
      "loss": 0.694,
      "num_input_tokens_seen": 22148256,
      "step": 33890
    },
    {
      "epoch": 17.7646750524109,
      "grad_norm": 0.09080968052148819,
      "learning_rate": 1.8794533147358074e-06,
      "loss": 0.5342,
      "num_input_tokens_seen": 22151840,
      "step": 33895
    },
    {
      "epoch": 17.767295597484278,
      "grad_norm": 0.1350204199552536,
      "learning_rate": 1.875106125862941e-06,
      "loss": 0.3014,
      "num_input_tokens_seen": 22155616,
      "step": 33900
    },
    {
      "epoch": 17.76991614255765,
      "grad_norm": 0.2499210387468338,
      "learning_rate": 1.870763774454057e-06,
      "loss": 0.4806,
      "num_input_tokens_seen": 22158656,
      "step": 33905
    },
    {
      "epoch": 17.77253668763103,
      "grad_norm": 0.18675287067890167,
      "learning_rate": 1.8664262614175115e-06,
      "loss": 0.5302,
      "num_input_tokens_seen": 22161152,
      "step": 33910
    },
    {
      "epoch": 17.7751572327044,
      "grad_norm": 0.09516496956348419,
      "learning_rate": 1.862093587660671e-06,
      "loss": 0.4945,
      "num_input_tokens_seen": 22164544,
      "step": 33915
    },
    {
      "epoch": 17.77777777777778,
      "grad_norm": 0.22555486857891083,
      "learning_rate": 1.8577657540898851e-06,
      "loss": 0.6123,
      "num_input_tokens_seen": 22167712,
      "step": 33920
    },
    {
      "epoch": 17.780398322851152,
      "grad_norm": 0.11643590778112411,
      "learning_rate": 1.8534427616104798e-06,
      "loss": 0.5363,
      "num_input_tokens_seen": 22171648,
      "step": 33925
    },
    {
      "epoch": 17.78301886792453,
      "grad_norm": 0.15092796087265015,
      "learning_rate": 1.8491246111267846e-06,
      "loss": 0.3954,
      "num_input_tokens_seen": 22178912,
      "step": 33930
    },
    {
      "epoch": 17.785639412997902,
      "grad_norm": 0.28723278641700745,
      "learning_rate": 1.8448113035420934e-06,
      "loss": 0.6684,
      "num_input_tokens_seen": 22181568,
      "step": 33935
    },
    {
      "epoch": 17.78825995807128,
      "grad_norm": 0.1573181003332138,
      "learning_rate": 1.840502839758712e-06,
      "loss": 0.4416,
      "num_input_tokens_seen": 22185408,
      "step": 33940
    },
    {
      "epoch": 17.790880503144653,
      "grad_norm": 0.17262700200080872,
      "learning_rate": 1.8361992206779226e-06,
      "loss": 0.4711,
      "num_input_tokens_seen": 22188224,
      "step": 33945
    },
    {
      "epoch": 17.79350104821803,
      "grad_norm": 0.19419310986995697,
      "learning_rate": 1.8319004471999856e-06,
      "loss": 0.5004,
      "num_input_tokens_seen": 22190688,
      "step": 33950
    },
    {
      "epoch": 17.796121593291403,
      "grad_norm": 0.28026366233825684,
      "learning_rate": 1.8276065202241592e-06,
      "loss": 0.4539,
      "num_input_tokens_seen": 22193184,
      "step": 33955
    },
    {
      "epoch": 17.79874213836478,
      "grad_norm": 0.38935405015945435,
      "learning_rate": 1.8233174406486863e-06,
      "loss": 0.3936,
      "num_input_tokens_seen": 22195520,
      "step": 33960
    },
    {
      "epoch": 17.801362683438157,
      "grad_norm": 0.13952936232089996,
      "learning_rate": 1.8190332093707907e-06,
      "loss": 0.498,
      "num_input_tokens_seen": 22198208,
      "step": 33965
    },
    {
      "epoch": 17.80398322851153,
      "grad_norm": 0.10957899689674377,
      "learning_rate": 1.8147538272866833e-06,
      "loss": 0.4914,
      "num_input_tokens_seen": 22201792,
      "step": 33970
    },
    {
      "epoch": 17.806603773584907,
      "grad_norm": 0.16178491711616516,
      "learning_rate": 1.8104792952915672e-06,
      "loss": 0.4508,
      "num_input_tokens_seen": 22205632,
      "step": 33975
    },
    {
      "epoch": 17.80922431865828,
      "grad_norm": 0.2093130350112915,
      "learning_rate": 1.8062096142796192e-06,
      "loss": 0.385,
      "num_input_tokens_seen": 22208192,
      "step": 33980
    },
    {
      "epoch": 17.811844863731658,
      "grad_norm": 0.15649692714214325,
      "learning_rate": 1.8019447851440163e-06,
      "loss": 0.4273,
      "num_input_tokens_seen": 22211584,
      "step": 33985
    },
    {
      "epoch": 17.81446540880503,
      "grad_norm": 0.256321519613266,
      "learning_rate": 1.7976848087769032e-06,
      "loss": 0.6212,
      "num_input_tokens_seen": 22215360,
      "step": 33990
    },
    {
      "epoch": 17.817085953878408,
      "grad_norm": 0.16221889853477478,
      "learning_rate": 1.79342968606942e-06,
      "loss": 0.5902,
      "num_input_tokens_seen": 22218848,
      "step": 33995
    },
    {
      "epoch": 17.81970649895178,
      "grad_norm": 0.1294744610786438,
      "learning_rate": 1.7891794179116933e-06,
      "loss": 0.3993,
      "num_input_tokens_seen": 22222880,
      "step": 34000
    },
    {
      "epoch": 17.822327044025158,
      "grad_norm": 0.19161640107631683,
      "learning_rate": 1.7849340051928316e-06,
      "loss": 0.5704,
      "num_input_tokens_seen": 22229120,
      "step": 34005
    },
    {
      "epoch": 17.82494758909853,
      "grad_norm": 0.13454686105251312,
      "learning_rate": 1.780693448800927e-06,
      "loss": 0.4167,
      "num_input_tokens_seen": 22232352,
      "step": 34010
    },
    {
      "epoch": 17.82756813417191,
      "grad_norm": 0.10730200260877609,
      "learning_rate": 1.7764577496230478e-06,
      "loss": 0.3886,
      "num_input_tokens_seen": 22237856,
      "step": 34015
    },
    {
      "epoch": 17.830188679245282,
      "grad_norm": 0.24625913798809052,
      "learning_rate": 1.7722269085452602e-06,
      "loss": 0.539,
      "num_input_tokens_seen": 22240992,
      "step": 34020
    },
    {
      "epoch": 17.83280922431866,
      "grad_norm": 0.17576934397220612,
      "learning_rate": 1.7680009264526038e-06,
      "loss": 0.4377,
      "num_input_tokens_seen": 22243776,
      "step": 34025
    },
    {
      "epoch": 17.835429769392032,
      "grad_norm": 0.18731911480426788,
      "learning_rate": 1.7637798042291126e-06,
      "loss": 0.3943,
      "num_input_tokens_seen": 22246592,
      "step": 34030
    },
    {
      "epoch": 17.83805031446541,
      "grad_norm": 0.12076672911643982,
      "learning_rate": 1.7595635427577944e-06,
      "loss": 0.39,
      "num_input_tokens_seen": 22249664,
      "step": 34035
    },
    {
      "epoch": 17.840670859538783,
      "grad_norm": 0.22195778787136078,
      "learning_rate": 1.7553521429206465e-06,
      "loss": 0.4363,
      "num_input_tokens_seen": 22253312,
      "step": 34040
    },
    {
      "epoch": 17.84329140461216,
      "grad_norm": 0.1576872617006302,
      "learning_rate": 1.7511456055986335e-06,
      "loss": 0.553,
      "num_input_tokens_seen": 22256768,
      "step": 34045
    },
    {
      "epoch": 17.845911949685533,
      "grad_norm": 0.21078673005104065,
      "learning_rate": 1.746943931671724e-06,
      "loss": 0.4838,
      "num_input_tokens_seen": 22260224,
      "step": 34050
    },
    {
      "epoch": 17.84853249475891,
      "grad_norm": 0.13579769432544708,
      "learning_rate": 1.742747122018859e-06,
      "loss": 0.4124,
      "num_input_tokens_seen": 22262912,
      "step": 34055
    },
    {
      "epoch": 17.851153039832283,
      "grad_norm": 0.15393082797527313,
      "learning_rate": 1.738555177517967e-06,
      "loss": 0.3209,
      "num_input_tokens_seen": 22266240,
      "step": 34060
    },
    {
      "epoch": 17.85377358490566,
      "grad_norm": 0.12594954669475555,
      "learning_rate": 1.734368099045952e-06,
      "loss": 0.53,
      "num_input_tokens_seen": 22269472,
      "step": 34065
    },
    {
      "epoch": 17.856394129979037,
      "grad_norm": 0.21937023103237152,
      "learning_rate": 1.730185887478697e-06,
      "loss": 0.4427,
      "num_input_tokens_seen": 22273184,
      "step": 34070
    },
    {
      "epoch": 17.85901467505241,
      "grad_norm": 0.2095436155796051,
      "learning_rate": 1.7260085436910794e-06,
      "loss": 0.4347,
      "num_input_tokens_seen": 22280032,
      "step": 34075
    },
    {
      "epoch": 17.861635220125788,
      "grad_norm": 0.09687194973230362,
      "learning_rate": 1.7218360685569506e-06,
      "loss": 0.6282,
      "num_input_tokens_seen": 22284128,
      "step": 34080
    },
    {
      "epoch": 17.86425576519916,
      "grad_norm": 0.17334991693496704,
      "learning_rate": 1.7176684629491513e-06,
      "loss": 0.4842,
      "num_input_tokens_seen": 22286880,
      "step": 34085
    },
    {
      "epoch": 17.866876310272538,
      "grad_norm": 0.12262134999036789,
      "learning_rate": 1.7135057277394868e-06,
      "loss": 0.3773,
      "num_input_tokens_seen": 22290016,
      "step": 34090
    },
    {
      "epoch": 17.86949685534591,
      "grad_norm": 0.11765409260988235,
      "learning_rate": 1.7093478637987548e-06,
      "loss": 0.5534,
      "num_input_tokens_seen": 22294688,
      "step": 34095
    },
    {
      "epoch": 17.872117400419288,
      "grad_norm": 0.21287041902542114,
      "learning_rate": 1.7051948719967375e-06,
      "loss": 0.453,
      "num_input_tokens_seen": 22297856,
      "step": 34100
    },
    {
      "epoch": 17.87473794549266,
      "grad_norm": 0.2252545803785324,
      "learning_rate": 1.7010467532021901e-06,
      "loss": 0.5311,
      "num_input_tokens_seen": 22301024,
      "step": 34105
    },
    {
      "epoch": 17.87735849056604,
      "grad_norm": 0.12333763390779495,
      "learning_rate": 1.6969035082828544e-06,
      "loss": 0.5425,
      "num_input_tokens_seen": 22303456,
      "step": 34110
    },
    {
      "epoch": 17.879979035639412,
      "grad_norm": 0.18726950883865356,
      "learning_rate": 1.692765138105451e-06,
      "loss": 0.4157,
      "num_input_tokens_seen": 22306528,
      "step": 34115
    },
    {
      "epoch": 17.88259958071279,
      "grad_norm": 0.18565888702869415,
      "learning_rate": 1.6886316435356792e-06,
      "loss": 0.4617,
      "num_input_tokens_seen": 22309824,
      "step": 34120
    },
    {
      "epoch": 17.885220125786162,
      "grad_norm": 0.38843798637390137,
      "learning_rate": 1.6845030254382138e-06,
      "loss": 0.5444,
      "num_input_tokens_seen": 22313024,
      "step": 34125
    },
    {
      "epoch": 17.88784067085954,
      "grad_norm": 0.2104455530643463,
      "learning_rate": 1.6803792846767197e-06,
      "loss": 0.4988,
      "num_input_tokens_seen": 22315584,
      "step": 34130
    },
    {
      "epoch": 17.890461215932913,
      "grad_norm": 0.18098291754722595,
      "learning_rate": 1.6762604221138316e-06,
      "loss": 0.502,
      "num_input_tokens_seen": 22318144,
      "step": 34135
    },
    {
      "epoch": 17.89308176100629,
      "grad_norm": 0.41370609402656555,
      "learning_rate": 1.6721464386111774e-06,
      "loss": 0.4441,
      "num_input_tokens_seen": 22320800,
      "step": 34140
    },
    {
      "epoch": 17.895702306079663,
      "grad_norm": 0.17585264146327972,
      "learning_rate": 1.6680373350293516e-06,
      "loss": 0.6321,
      "num_input_tokens_seen": 22324064,
      "step": 34145
    },
    {
      "epoch": 17.89832285115304,
      "grad_norm": 0.3332182765007019,
      "learning_rate": 1.6639331122279223e-06,
      "loss": 0.3982,
      "num_input_tokens_seen": 22327104,
      "step": 34150
    },
    {
      "epoch": 17.900943396226417,
      "grad_norm": 0.16055196523666382,
      "learning_rate": 1.6598337710654583e-06,
      "loss": 0.5582,
      "num_input_tokens_seen": 22330528,
      "step": 34155
    },
    {
      "epoch": 17.90356394129979,
      "grad_norm": 0.2353968322277069,
      "learning_rate": 1.6557393123994874e-06,
      "loss": 0.4324,
      "num_input_tokens_seen": 22333472,
      "step": 34160
    },
    {
      "epoch": 17.906184486373167,
      "grad_norm": 0.21625885367393494,
      "learning_rate": 1.651649737086533e-06,
      "loss": 0.5199,
      "num_input_tokens_seen": 22336192,
      "step": 34165
    },
    {
      "epoch": 17.90880503144654,
      "grad_norm": 0.1376776099205017,
      "learning_rate": 1.6475650459820774e-06,
      "loss": 0.5259,
      "num_input_tokens_seen": 22338720,
      "step": 34170
    },
    {
      "epoch": 17.911425576519918,
      "grad_norm": 0.13638894259929657,
      "learning_rate": 1.643485239940598e-06,
      "loss": 0.3252,
      "num_input_tokens_seen": 22341152,
      "step": 34175
    },
    {
      "epoch": 17.91404612159329,
      "grad_norm": 0.1495407670736313,
      "learning_rate": 1.6394103198155375e-06,
      "loss": 0.4835,
      "num_input_tokens_seen": 22344352,
      "step": 34180
    },
    {
      "epoch": 17.916666666666668,
      "grad_norm": 0.16156615316867828,
      "learning_rate": 1.6353402864593276e-06,
      "loss": 0.3108,
      "num_input_tokens_seen": 22346944,
      "step": 34185
    },
    {
      "epoch": 17.91928721174004,
      "grad_norm": 0.12720294296741486,
      "learning_rate": 1.6312751407233706e-06,
      "loss": 0.4916,
      "num_input_tokens_seen": 22351040,
      "step": 34190
    },
    {
      "epoch": 17.921907756813418,
      "grad_norm": 0.16952788829803467,
      "learning_rate": 1.6272148834580559e-06,
      "loss": 0.5113,
      "num_input_tokens_seen": 22353728,
      "step": 34195
    },
    {
      "epoch": 17.92452830188679,
      "grad_norm": 0.20924273133277893,
      "learning_rate": 1.6231595155127317e-06,
      "loss": 0.5218,
      "num_input_tokens_seen": 22356928,
      "step": 34200
    },
    {
      "epoch": 17.92714884696017,
      "grad_norm": 0.17841379344463348,
      "learning_rate": 1.6191090377357414e-06,
      "loss": 0.456,
      "num_input_tokens_seen": 22359168,
      "step": 34205
    },
    {
      "epoch": 17.929769392033542,
      "grad_norm": 0.19629044830799103,
      "learning_rate": 1.6150634509743939e-06,
      "loss": 0.4456,
      "num_input_tokens_seen": 22362400,
      "step": 34210
    },
    {
      "epoch": 17.93238993710692,
      "grad_norm": 0.18677593767642975,
      "learning_rate": 1.611022756074984e-06,
      "loss": 0.4527,
      "num_input_tokens_seen": 22365408,
      "step": 34215
    },
    {
      "epoch": 17.935010482180292,
      "grad_norm": 0.09630515426397324,
      "learning_rate": 1.606986953882783e-06,
      "loss": 0.3944,
      "num_input_tokens_seen": 22368768,
      "step": 34220
    },
    {
      "epoch": 17.93763102725367,
      "grad_norm": 0.13178202509880066,
      "learning_rate": 1.6029560452420238e-06,
      "loss": 0.5244,
      "num_input_tokens_seen": 22372128,
      "step": 34225
    },
    {
      "epoch": 17.940251572327043,
      "grad_norm": 0.11992631107568741,
      "learning_rate": 1.5989300309959377e-06,
      "loss": 0.4837,
      "num_input_tokens_seen": 22375520,
      "step": 34230
    },
    {
      "epoch": 17.94287211740042,
      "grad_norm": 0.16542115807533264,
      "learning_rate": 1.5949089119867116e-06,
      "loss": 0.7052,
      "num_input_tokens_seen": 22378656,
      "step": 34235
    },
    {
      "epoch": 17.945492662473793,
      "grad_norm": 0.2392617017030716,
      "learning_rate": 1.5908926890555204e-06,
      "loss": 0.5516,
      "num_input_tokens_seen": 22381888,
      "step": 34240
    },
    {
      "epoch": 17.94811320754717,
      "grad_norm": 0.22115232050418854,
      "learning_rate": 1.5868813630425167e-06,
      "loss": 0.5137,
      "num_input_tokens_seen": 22384928,
      "step": 34245
    },
    {
      "epoch": 17.950733752620543,
      "grad_norm": 0.12013617902994156,
      "learning_rate": 1.5828749347868183e-06,
      "loss": 0.3699,
      "num_input_tokens_seen": 22388096,
      "step": 34250
    },
    {
      "epoch": 17.95335429769392,
      "grad_norm": 0.1854897439479828,
      "learning_rate": 1.5788734051265268e-06,
      "loss": 0.3167,
      "num_input_tokens_seen": 22393632,
      "step": 34255
    },
    {
      "epoch": 17.955974842767297,
      "grad_norm": 0.12374775856733322,
      "learning_rate": 1.5748767748987231e-06,
      "loss": 0.497,
      "num_input_tokens_seen": 22396768,
      "step": 34260
    },
    {
      "epoch": 17.95859538784067,
      "grad_norm": 0.1833111047744751,
      "learning_rate": 1.5708850449394436e-06,
      "loss": 0.423,
      "num_input_tokens_seen": 22399776,
      "step": 34265
    },
    {
      "epoch": 17.961215932914047,
      "grad_norm": 0.17518815398216248,
      "learning_rate": 1.5668982160837176e-06,
      "loss": 0.6008,
      "num_input_tokens_seen": 22403104,
      "step": 34270
    },
    {
      "epoch": 17.96383647798742,
      "grad_norm": 0.1936296969652176,
      "learning_rate": 1.562916289165553e-06,
      "loss": 0.4049,
      "num_input_tokens_seen": 22406592,
      "step": 34275
    },
    {
      "epoch": 17.966457023060798,
      "grad_norm": 0.1152893528342247,
      "learning_rate": 1.5589392650179114e-06,
      "loss": 0.4244,
      "num_input_tokens_seen": 22411264,
      "step": 34280
    },
    {
      "epoch": 17.96907756813417,
      "grad_norm": 0.16222336888313293,
      "learning_rate": 1.5549671444727437e-06,
      "loss": 0.4509,
      "num_input_tokens_seen": 22414976,
      "step": 34285
    },
    {
      "epoch": 17.971698113207548,
      "grad_norm": 0.13960523903369904,
      "learning_rate": 1.55099992836098e-06,
      "loss": 0.5863,
      "num_input_tokens_seen": 22418016,
      "step": 34290
    },
    {
      "epoch": 17.97431865828092,
      "grad_norm": 0.14850062131881714,
      "learning_rate": 1.5470376175125061e-06,
      "loss": 0.5727,
      "num_input_tokens_seen": 22421408,
      "step": 34295
    },
    {
      "epoch": 17.9769392033543,
      "grad_norm": 0.1223003938794136,
      "learning_rate": 1.5430802127562038e-06,
      "loss": 0.6852,
      "num_input_tokens_seen": 22423744,
      "step": 34300
    },
    {
      "epoch": 17.979559748427672,
      "grad_norm": 0.20044519007205963,
      "learning_rate": 1.5391277149199023e-06,
      "loss": 0.4564,
      "num_input_tokens_seen": 22428672,
      "step": 34305
    },
    {
      "epoch": 17.98218029350105,
      "grad_norm": 0.10975077003240585,
      "learning_rate": 1.5351801248304292e-06,
      "loss": 0.4947,
      "num_input_tokens_seen": 22431648,
      "step": 34310
    },
    {
      "epoch": 17.984800838574422,
      "grad_norm": 0.2328856736421585,
      "learning_rate": 1.5312374433135767e-06,
      "loss": 0.5025,
      "num_input_tokens_seen": 22434496,
      "step": 34315
    },
    {
      "epoch": 17.9874213836478,
      "grad_norm": 0.12138654291629791,
      "learning_rate": 1.5272996711941018e-06,
      "loss": 0.4475,
      "num_input_tokens_seen": 22438432,
      "step": 34320
    },
    {
      "epoch": 17.990041928721173,
      "grad_norm": 0.145201176404953,
      "learning_rate": 1.5233668092957482e-06,
      "loss": 0.4294,
      "num_input_tokens_seen": 22441600,
      "step": 34325
    },
    {
      "epoch": 17.99266247379455,
      "grad_norm": 0.12658676505088806,
      "learning_rate": 1.5194388584412217e-06,
      "loss": 0.4459,
      "num_input_tokens_seen": 22444160,
      "step": 34330
    },
    {
      "epoch": 17.995283018867923,
      "grad_norm": 0.2182757407426834,
      "learning_rate": 1.5155158194522013e-06,
      "loss": 0.52,
      "num_input_tokens_seen": 22447328,
      "step": 34335
    },
    {
      "epoch": 17.9979035639413,
      "grad_norm": 0.1287880837917328,
      "learning_rate": 1.5115976931493498e-06,
      "loss": 0.4564,
      "num_input_tokens_seen": 22450112,
      "step": 34340
    },
    {
      "epoch": 18.0,
      "eval_loss": 0.4917658567428589,
      "eval_runtime": 10.1585,
      "eval_samples_per_second": 83.477,
      "eval_steps_per_second": 20.869,
      "num_input_tokens_seen": 22451928,
      "step": 34344
    },
    {
      "epoch": 18.000524109014677,
      "grad_norm": 0.27370142936706543,
      "learning_rate": 1.5076844803522922e-06,
      "loss": 0.3954,
      "num_input_tokens_seen": 22452408,
      "step": 34345
    },
    {
      "epoch": 18.00314465408805,
      "grad_norm": 0.2884422242641449,
      "learning_rate": 1.5037761818796236e-06,
      "loss": 0.4604,
      "num_input_tokens_seen": 22456184,
      "step": 34350
    },
    {
      "epoch": 18.005765199161427,
      "grad_norm": 0.24629011750221252,
      "learning_rate": 1.4998727985489235e-06,
      "loss": 0.5645,
      "num_input_tokens_seen": 22458968,
      "step": 34355
    },
    {
      "epoch": 18.0083857442348,
      "grad_norm": 0.1281844526529312,
      "learning_rate": 1.4959743311767249e-06,
      "loss": 0.3859,
      "num_input_tokens_seen": 22462392,
      "step": 34360
    },
    {
      "epoch": 18.011006289308177,
      "grad_norm": 0.15749341249465942,
      "learning_rate": 1.4920807805785502e-06,
      "loss": 0.447,
      "num_input_tokens_seen": 22466360,
      "step": 34365
    },
    {
      "epoch": 18.01362683438155,
      "grad_norm": 0.08411683887243271,
      "learning_rate": 1.4881921475688815e-06,
      "loss": 0.475,
      "num_input_tokens_seen": 22470104,
      "step": 34370
    },
    {
      "epoch": 18.016247379454928,
      "grad_norm": 0.17763155698776245,
      "learning_rate": 1.4843084329611844e-06,
      "loss": 0.4503,
      "num_input_tokens_seen": 22473144,
      "step": 34375
    },
    {
      "epoch": 18.0188679245283,
      "grad_norm": 0.15998633205890656,
      "learning_rate": 1.480429637567879e-06,
      "loss": 0.5231,
      "num_input_tokens_seen": 22475640,
      "step": 34380
    },
    {
      "epoch": 18.021488469601678,
      "grad_norm": 0.33008113503456116,
      "learning_rate": 1.4765557622003684e-06,
      "loss": 0.4557,
      "num_input_tokens_seen": 22478712,
      "step": 34385
    },
    {
      "epoch": 18.02410901467505,
      "grad_norm": 0.13277028501033783,
      "learning_rate": 1.4726868076690186e-06,
      "loss": 0.5749,
      "num_input_tokens_seen": 22481624,
      "step": 34390
    },
    {
      "epoch": 18.02672955974843,
      "grad_norm": 0.12570619583129883,
      "learning_rate": 1.4688227747831767e-06,
      "loss": 0.4637,
      "num_input_tokens_seen": 22484664,
      "step": 34395
    },
    {
      "epoch": 18.029350104821802,
      "grad_norm": 0.12393802404403687,
      "learning_rate": 1.4649636643511573e-06,
      "loss": 0.4283,
      "num_input_tokens_seen": 22487128,
      "step": 34400
    },
    {
      "epoch": 18.03197064989518,
      "grad_norm": 0.25348809361457825,
      "learning_rate": 1.4611094771802364e-06,
      "loss": 0.5233,
      "num_input_tokens_seen": 22489528,
      "step": 34405
    },
    {
      "epoch": 18.034591194968552,
      "grad_norm": 0.10706218332052231,
      "learning_rate": 1.457260214076664e-06,
      "loss": 0.5359,
      "num_input_tokens_seen": 22493560,
      "step": 34410
    },
    {
      "epoch": 18.03721174004193,
      "grad_norm": 0.2169712781906128,
      "learning_rate": 1.4534158758456623e-06,
      "loss": 0.4641,
      "num_input_tokens_seen": 22496408,
      "step": 34415
    },
    {
      "epoch": 18.039832285115303,
      "grad_norm": 0.20058543980121613,
      "learning_rate": 1.4495764632914271e-06,
      "loss": 0.4087,
      "num_input_tokens_seen": 22500440,
      "step": 34420
    },
    {
      "epoch": 18.04245283018868,
      "grad_norm": 0.25430864095687866,
      "learning_rate": 1.4457419772171215e-06,
      "loss": 0.5347,
      "num_input_tokens_seen": 22503640,
      "step": 34425
    },
    {
      "epoch": 18.045073375262053,
      "grad_norm": 0.18060047924518585,
      "learning_rate": 1.4419124184248733e-06,
      "loss": 0.552,
      "num_input_tokens_seen": 22507768,
      "step": 34430
    },
    {
      "epoch": 18.04769392033543,
      "grad_norm": 0.16496731340885162,
      "learning_rate": 1.4380877877157834e-06,
      "loss": 0.4797,
      "num_input_tokens_seen": 22511256,
      "step": 34435
    },
    {
      "epoch": 18.050314465408807,
      "grad_norm": 0.16919739544391632,
      "learning_rate": 1.4342680858899143e-06,
      "loss": 0.5307,
      "num_input_tokens_seen": 22514872,
      "step": 34440
    },
    {
      "epoch": 18.05293501048218,
      "grad_norm": 0.15140800178050995,
      "learning_rate": 1.4304533137463134e-06,
      "loss": 0.38,
      "num_input_tokens_seen": 22518008,
      "step": 34445
    },
    {
      "epoch": 18.055555555555557,
      "grad_norm": 0.25425413250923157,
      "learning_rate": 1.4266434720829808e-06,
      "loss": 0.4295,
      "num_input_tokens_seen": 22520664,
      "step": 34450
    },
    {
      "epoch": 18.05817610062893,
      "grad_norm": 0.21090660989284515,
      "learning_rate": 1.4228385616968987e-06,
      "loss": 0.3669,
      "num_input_tokens_seen": 22523096,
      "step": 34455
    },
    {
      "epoch": 18.060796645702307,
      "grad_norm": 0.2533620595932007,
      "learning_rate": 1.419038583384008e-06,
      "loss": 0.5175,
      "num_input_tokens_seen": 22526616,
      "step": 34460
    },
    {
      "epoch": 18.06341719077568,
      "grad_norm": 0.1797899454832077,
      "learning_rate": 1.4152435379392148e-06,
      "loss": 0.4732,
      "num_input_tokens_seen": 22529560,
      "step": 34465
    },
    {
      "epoch": 18.066037735849058,
      "grad_norm": 0.15116694569587708,
      "learning_rate": 1.411453426156406e-06,
      "loss": 0.4618,
      "num_input_tokens_seen": 22533016,
      "step": 34470
    },
    {
      "epoch": 18.06865828092243,
      "grad_norm": 0.2811131477355957,
      "learning_rate": 1.4076682488284282e-06,
      "loss": 0.5546,
      "num_input_tokens_seen": 22535416,
      "step": 34475
    },
    {
      "epoch": 18.071278825995808,
      "grad_norm": 0.1481540948152542,
      "learning_rate": 1.4038880067470978e-06,
      "loss": 0.3588,
      "num_input_tokens_seen": 22538936,
      "step": 34480
    },
    {
      "epoch": 18.07389937106918,
      "grad_norm": 0.2045188695192337,
      "learning_rate": 1.400112700703196e-06,
      "loss": 0.457,
      "num_input_tokens_seen": 22541944,
      "step": 34485
    },
    {
      "epoch": 18.07651991614256,
      "grad_norm": 0.20489633083343506,
      "learning_rate": 1.3963423314864805e-06,
      "loss": 0.545,
      "num_input_tokens_seen": 22544568,
      "step": 34490
    },
    {
      "epoch": 18.079140461215932,
      "grad_norm": 0.15517257153987885,
      "learning_rate": 1.3925768998856587e-06,
      "loss": 0.4808,
      "num_input_tokens_seen": 22547768,
      "step": 34495
    },
    {
      "epoch": 18.08176100628931,
      "grad_norm": 0.17748884856700897,
      "learning_rate": 1.388816406688423e-06,
      "loss": 0.467,
      "num_input_tokens_seen": 22550776,
      "step": 34500
    },
    {
      "epoch": 18.084381551362682,
      "grad_norm": 0.10636338591575623,
      "learning_rate": 1.3850608526814246e-06,
      "loss": 0.3857,
      "num_input_tokens_seen": 22554392,
      "step": 34505
    },
    {
      "epoch": 18.08700209643606,
      "grad_norm": 0.13955417275428772,
      "learning_rate": 1.3813102386502853e-06,
      "loss": 0.5382,
      "num_input_tokens_seen": 22557304,
      "step": 34510
    },
    {
      "epoch": 18.089622641509433,
      "grad_norm": 0.1905469447374344,
      "learning_rate": 1.3775645653795881e-06,
      "loss": 0.4809,
      "num_input_tokens_seen": 22560536,
      "step": 34515
    },
    {
      "epoch": 18.09224318658281,
      "grad_norm": 0.12556368112564087,
      "learning_rate": 1.373823833652882e-06,
      "loss": 0.3985,
      "num_input_tokens_seen": 22563832,
      "step": 34520
    },
    {
      "epoch": 18.094863731656183,
      "grad_norm": 0.164108544588089,
      "learning_rate": 1.3700880442526876e-06,
      "loss": 0.5053,
      "num_input_tokens_seen": 22566616,
      "step": 34525
    },
    {
      "epoch": 18.09748427672956,
      "grad_norm": 0.09912601113319397,
      "learning_rate": 1.3663571979604884e-06,
      "loss": 0.5143,
      "num_input_tokens_seen": 22569976,
      "step": 34530
    },
    {
      "epoch": 18.100104821802937,
      "grad_norm": 0.2040097564458847,
      "learning_rate": 1.3626312955567406e-06,
      "loss": 0.5951,
      "num_input_tokens_seen": 22573112,
      "step": 34535
    },
    {
      "epoch": 18.10272536687631,
      "grad_norm": 0.1803688257932663,
      "learning_rate": 1.358910337820854e-06,
      "loss": 0.4664,
      "num_input_tokens_seen": 22576152,
      "step": 34540
    },
    {
      "epoch": 18.105345911949687,
      "grad_norm": 0.14737696945667267,
      "learning_rate": 1.3551943255312172e-06,
      "loss": 0.4879,
      "num_input_tokens_seen": 22579672,
      "step": 34545
    },
    {
      "epoch": 18.10796645702306,
      "grad_norm": 0.1538366675376892,
      "learning_rate": 1.3514832594651694e-06,
      "loss": 0.465,
      "num_input_tokens_seen": 22583096,
      "step": 34550
    },
    {
      "epoch": 18.110587002096437,
      "grad_norm": 0.18259172141551971,
      "learning_rate": 1.347777140399026e-06,
      "loss": 0.5961,
      "num_input_tokens_seen": 22587256,
      "step": 34555
    },
    {
      "epoch": 18.11320754716981,
      "grad_norm": 0.17876793444156647,
      "learning_rate": 1.344075969108069e-06,
      "loss": 0.6077,
      "num_input_tokens_seen": 22591352,
      "step": 34560
    },
    {
      "epoch": 18.115828092243188,
      "grad_norm": 0.22182312607765198,
      "learning_rate": 1.3403797463665352e-06,
      "loss": 0.4883,
      "num_input_tokens_seen": 22594040,
      "step": 34565
    },
    {
      "epoch": 18.11844863731656,
      "grad_norm": 0.17814312875270844,
      "learning_rate": 1.3366884729476365e-06,
      "loss": 0.5717,
      "num_input_tokens_seen": 22597208,
      "step": 34570
    },
    {
      "epoch": 18.121069182389938,
      "grad_norm": 0.15258577466011047,
      "learning_rate": 1.3330021496235467e-06,
      "loss": 0.3564,
      "num_input_tokens_seen": 22600536,
      "step": 34575
    },
    {
      "epoch": 18.12368972746331,
      "grad_norm": 0.1617054045200348,
      "learning_rate": 1.3293207771653965e-06,
      "loss": 0.5264,
      "num_input_tokens_seen": 22603160,
      "step": 34580
    },
    {
      "epoch": 18.12631027253669,
      "grad_norm": 0.17945890128612518,
      "learning_rate": 1.3256443563432891e-06,
      "loss": 0.4511,
      "num_input_tokens_seen": 22606328,
      "step": 34585
    },
    {
      "epoch": 18.128930817610062,
      "grad_norm": 0.14116550981998444,
      "learning_rate": 1.3219728879262956e-06,
      "loss": 0.4379,
      "num_input_tokens_seen": 22609624,
      "step": 34590
    },
    {
      "epoch": 18.13155136268344,
      "grad_norm": 0.18952199816703796,
      "learning_rate": 1.3183063726824375e-06,
      "loss": 0.563,
      "num_input_tokens_seen": 22612728,
      "step": 34595
    },
    {
      "epoch": 18.134171907756812,
      "grad_norm": 0.10670539736747742,
      "learning_rate": 1.3146448113787151e-06,
      "loss": 0.5285,
      "num_input_tokens_seen": 22615640,
      "step": 34600
    },
    {
      "epoch": 18.13679245283019,
      "grad_norm": 0.13878530263900757,
      "learning_rate": 1.3109882047810773e-06,
      "loss": 0.5527,
      "num_input_tokens_seen": 22618360,
      "step": 34605
    },
    {
      "epoch": 18.139412997903563,
      "grad_norm": 0.1351933628320694,
      "learning_rate": 1.307336553654448e-06,
      "loss": 0.5968,
      "num_input_tokens_seen": 22621976,
      "step": 34610
    },
    {
      "epoch": 18.14203354297694,
      "grad_norm": 0.12484265118837357,
      "learning_rate": 1.303689858762716e-06,
      "loss": 0.5633,
      "num_input_tokens_seen": 22626008,
      "step": 34615
    },
    {
      "epoch": 18.144654088050313,
      "grad_norm": 0.2430598884820938,
      "learning_rate": 1.3000481208687188e-06,
      "loss": 0.4219,
      "num_input_tokens_seen": 22629368,
      "step": 34620
    },
    {
      "epoch": 18.14727463312369,
      "grad_norm": 0.10045954585075378,
      "learning_rate": 1.2964113407342743e-06,
      "loss": 0.4171,
      "num_input_tokens_seen": 22632440,
      "step": 34625
    },
    {
      "epoch": 18.149895178197063,
      "grad_norm": 0.2010328471660614,
      "learning_rate": 1.292779519120152e-06,
      "loss": 0.4555,
      "num_input_tokens_seen": 22635160,
      "step": 34630
    },
    {
      "epoch": 18.15251572327044,
      "grad_norm": 0.11802151054143906,
      "learning_rate": 1.2891526567860856e-06,
      "loss": 0.5536,
      "num_input_tokens_seen": 22639896,
      "step": 34635
    },
    {
      "epoch": 18.155136268343817,
      "grad_norm": 0.21194769442081451,
      "learning_rate": 1.2855307544907825e-06,
      "loss": 0.4848,
      "num_input_tokens_seen": 22642968,
      "step": 34640
    },
    {
      "epoch": 18.15775681341719,
      "grad_norm": 0.10842292010784149,
      "learning_rate": 1.281913812991889e-06,
      "loss": 0.3258,
      "num_input_tokens_seen": 22646456,
      "step": 34645
    },
    {
      "epoch": 18.160377358490567,
      "grad_norm": 0.12000108510255814,
      "learning_rate": 1.2783018330460389e-06,
      "loss": 0.5136,
      "num_input_tokens_seen": 22650648,
      "step": 34650
    },
    {
      "epoch": 18.16299790356394,
      "grad_norm": 0.13018129765987396,
      "learning_rate": 1.2746948154088111e-06,
      "loss": 0.3634,
      "num_input_tokens_seen": 22654296,
      "step": 34655
    },
    {
      "epoch": 18.165618448637318,
      "grad_norm": 0.1381502002477646,
      "learning_rate": 1.2710927608347605e-06,
      "loss": 0.3615,
      "num_input_tokens_seen": 22658072,
      "step": 34660
    },
    {
      "epoch": 18.16823899371069,
      "grad_norm": 0.24417148530483246,
      "learning_rate": 1.2674956700773872e-06,
      "loss": 0.3587,
      "num_input_tokens_seen": 22661752,
      "step": 34665
    },
    {
      "epoch": 18.170859538784068,
      "grad_norm": 0.19174791872501373,
      "learning_rate": 1.2639035438891666e-06,
      "loss": 0.6807,
      "num_input_tokens_seen": 22665112,
      "step": 34670
    },
    {
      "epoch": 18.17348008385744,
      "grad_norm": 0.2320191115140915,
      "learning_rate": 1.260316383021526e-06,
      "loss": 0.4012,
      "num_input_tokens_seen": 22667640,
      "step": 34675
    },
    {
      "epoch": 18.17610062893082,
      "grad_norm": 0.13825391232967377,
      "learning_rate": 1.2567341882248645e-06,
      "loss": 0.478,
      "num_input_tokens_seen": 22670840,
      "step": 34680
    },
    {
      "epoch": 18.178721174004192,
      "grad_norm": 0.20365303754806519,
      "learning_rate": 1.2531569602485305e-06,
      "loss": 0.493,
      "num_input_tokens_seen": 22673720,
      "step": 34685
    },
    {
      "epoch": 18.18134171907757,
      "grad_norm": 0.22468121349811554,
      "learning_rate": 1.2495846998408445e-06,
      "loss": 0.5115,
      "num_input_tokens_seen": 22676984,
      "step": 34690
    },
    {
      "epoch": 18.183962264150942,
      "grad_norm": 0.12617002427577972,
      "learning_rate": 1.246017407749084e-06,
      "loss": 0.4559,
      "num_input_tokens_seen": 22680216,
      "step": 34695
    },
    {
      "epoch": 18.18658280922432,
      "grad_norm": 0.11261862516403198,
      "learning_rate": 1.2424550847194739e-06,
      "loss": 0.4473,
      "num_input_tokens_seen": 22683448,
      "step": 34700
    },
    {
      "epoch": 18.189203354297693,
      "grad_norm": 0.26189664006233215,
      "learning_rate": 1.2388977314972238e-06,
      "loss": 0.45,
      "num_input_tokens_seen": 22689912,
      "step": 34705
    },
    {
      "epoch": 18.19182389937107,
      "grad_norm": 0.145995631814003,
      "learning_rate": 1.235345348826486e-06,
      "loss": 0.3264,
      "num_input_tokens_seen": 22692696,
      "step": 34710
    },
    {
      "epoch": 18.194444444444443,
      "grad_norm": 0.1612364798784256,
      "learning_rate": 1.2317979374503825e-06,
      "loss": 0.4693,
      "num_input_tokens_seen": 22695032,
      "step": 34715
    },
    {
      "epoch": 18.19706498951782,
      "grad_norm": 0.19138281047344208,
      "learning_rate": 1.228255498110989e-06,
      "loss": 0.4799,
      "num_input_tokens_seen": 22698712,
      "step": 34720
    },
    {
      "epoch": 18.199685534591197,
      "grad_norm": 0.25306740403175354,
      "learning_rate": 1.224718031549338e-06,
      "loss": 0.5588,
      "num_input_tokens_seen": 22702200,
      "step": 34725
    },
    {
      "epoch": 18.20230607966457,
      "grad_norm": 0.17543677985668182,
      "learning_rate": 1.2211855385054344e-06,
      "loss": 0.4509,
      "num_input_tokens_seen": 22704888,
      "step": 34730
    },
    {
      "epoch": 18.204926624737947,
      "grad_norm": 0.16491790115833282,
      "learning_rate": 1.217658019718232e-06,
      "loss": 0.3856,
      "num_input_tokens_seen": 22708280,
      "step": 34735
    },
    {
      "epoch": 18.20754716981132,
      "grad_norm": 0.5066659450531006,
      "learning_rate": 1.2141354759256484e-06,
      "loss": 0.5577,
      "num_input_tokens_seen": 22711384,
      "step": 34740
    },
    {
      "epoch": 18.210167714884697,
      "grad_norm": 0.10222769528627396,
      "learning_rate": 1.2106179078645641e-06,
      "loss": 0.4163,
      "num_input_tokens_seen": 22714712,
      "step": 34745
    },
    {
      "epoch": 18.21278825995807,
      "grad_norm": 0.20625735819339752,
      "learning_rate": 1.2071053162708096e-06,
      "loss": 0.5246,
      "num_input_tokens_seen": 22717432,
      "step": 34750
    },
    {
      "epoch": 18.215408805031448,
      "grad_norm": 0.15227270126342773,
      "learning_rate": 1.203597701879175e-06,
      "loss": 0.4705,
      "num_input_tokens_seen": 22720312,
      "step": 34755
    },
    {
      "epoch": 18.21802935010482,
      "grad_norm": 0.08568425476551056,
      "learning_rate": 1.2000950654234177e-06,
      "loss": 0.395,
      "num_input_tokens_seen": 22727928,
      "step": 34760
    },
    {
      "epoch": 18.220649895178198,
      "grad_norm": 0.14421962201595306,
      "learning_rate": 1.1965974076362491e-06,
      "loss": 0.4351,
      "num_input_tokens_seen": 22730456,
      "step": 34765
    },
    {
      "epoch": 18.22327044025157,
      "grad_norm": 0.1411057412624359,
      "learning_rate": 1.193104729249342e-06,
      "loss": 0.4725,
      "num_input_tokens_seen": 22733240,
      "step": 34770
    },
    {
      "epoch": 18.22589098532495,
      "grad_norm": 0.14799968898296356,
      "learning_rate": 1.189617030993323e-06,
      "loss": 0.4182,
      "num_input_tokens_seen": 22736344,
      "step": 34775
    },
    {
      "epoch": 18.228511530398322,
      "grad_norm": 0.20476914942264557,
      "learning_rate": 1.1861343135977753e-06,
      "loss": 0.5031,
      "num_input_tokens_seen": 22739320,
      "step": 34780
    },
    {
      "epoch": 18.2311320754717,
      "grad_norm": 0.18314474821090698,
      "learning_rate": 1.182656577791244e-06,
      "loss": 0.4186,
      "num_input_tokens_seen": 22743000,
      "step": 34785
    },
    {
      "epoch": 18.233752620545072,
      "grad_norm": 0.17884525656700134,
      "learning_rate": 1.1791838243012359e-06,
      "loss": 0.5279,
      "num_input_tokens_seen": 22745752,
      "step": 34790
    },
    {
      "epoch": 18.23637316561845,
      "grad_norm": 0.15513059496879578,
      "learning_rate": 1.1757160538542117e-06,
      "loss": 0.6417,
      "num_input_tokens_seen": 22748504,
      "step": 34795
    },
    {
      "epoch": 18.238993710691823,
      "grad_norm": 0.1530883014202118,
      "learning_rate": 1.1722532671755882e-06,
      "loss": 0.4735,
      "num_input_tokens_seen": 22751704,
      "step": 34800
    },
    {
      "epoch": 18.2416142557652,
      "grad_norm": 0.13387836515903473,
      "learning_rate": 1.168795464989736e-06,
      "loss": 0.4426,
      "num_input_tokens_seen": 22755256,
      "step": 34805
    },
    {
      "epoch": 18.244234800838573,
      "grad_norm": 0.1422923058271408,
      "learning_rate": 1.165342648019993e-06,
      "loss": 0.6527,
      "num_input_tokens_seen": 22758616,
      "step": 34810
    },
    {
      "epoch": 18.24685534591195,
      "grad_norm": 0.24427121877670288,
      "learning_rate": 1.1618948169886457e-06,
      "loss": 0.3536,
      "num_input_tokens_seen": 22761432,
      "step": 34815
    },
    {
      "epoch": 18.249475890985323,
      "grad_norm": 0.10454709082841873,
      "learning_rate": 1.1584519726169446e-06,
      "loss": 0.5857,
      "num_input_tokens_seen": 22764376,
      "step": 34820
    },
    {
      "epoch": 18.2520964360587,
      "grad_norm": 0.14844636619091034,
      "learning_rate": 1.1550141156250942e-06,
      "loss": 0.4826,
      "num_input_tokens_seen": 22768056,
      "step": 34825
    },
    {
      "epoch": 18.254716981132077,
      "grad_norm": 0.1821037083864212,
      "learning_rate": 1.1515812467322556e-06,
      "loss": 0.6138,
      "num_input_tokens_seen": 22771128,
      "step": 34830
    },
    {
      "epoch": 18.25733752620545,
      "grad_norm": 0.2565770447254181,
      "learning_rate": 1.1481533666565375e-06,
      "loss": 0.3394,
      "num_input_tokens_seen": 22774360,
      "step": 34835
    },
    {
      "epoch": 18.259958071278827,
      "grad_norm": 0.1483643501996994,
      "learning_rate": 1.144730476115019e-06,
      "loss": 0.3831,
      "num_input_tokens_seen": 22778712,
      "step": 34840
    },
    {
      "epoch": 18.2625786163522,
      "grad_norm": 0.2076200395822525,
      "learning_rate": 1.1413125758237304e-06,
      "loss": 0.4273,
      "num_input_tokens_seen": 22781368,
      "step": 34845
    },
    {
      "epoch": 18.265199161425578,
      "grad_norm": 0.12045242637395859,
      "learning_rate": 1.1378996664976577e-06,
      "loss": 0.3631,
      "num_input_tokens_seen": 22784120,
      "step": 34850
    },
    {
      "epoch": 18.26781970649895,
      "grad_norm": 0.12750479578971863,
      "learning_rate": 1.134491748850741e-06,
      "loss": 0.3536,
      "num_input_tokens_seen": 22787160,
      "step": 34855
    },
    {
      "epoch": 18.270440251572328,
      "grad_norm": 0.12237785756587982,
      "learning_rate": 1.13108882359588e-06,
      "loss": 0.5013,
      "num_input_tokens_seen": 22790584,
      "step": 34860
    },
    {
      "epoch": 18.2730607966457,
      "grad_norm": 0.11513547599315643,
      "learning_rate": 1.127690891444924e-06,
      "loss": 0.4758,
      "num_input_tokens_seen": 22794168,
      "step": 34865
    },
    {
      "epoch": 18.27568134171908,
      "grad_norm": 0.22406640648841858,
      "learning_rate": 1.1242979531086829e-06,
      "loss": 0.5366,
      "num_input_tokens_seen": 22796600,
      "step": 34870
    },
    {
      "epoch": 18.278301886792452,
      "grad_norm": 0.1438007801771164,
      "learning_rate": 1.1209100092969244e-06,
      "loss": 0.5438,
      "num_input_tokens_seen": 22799896,
      "step": 34875
    },
    {
      "epoch": 18.28092243186583,
      "grad_norm": 0.16552379727363586,
      "learning_rate": 1.1175270607183625e-06,
      "loss": 0.5764,
      "num_input_tokens_seen": 22803416,
      "step": 34880
    },
    {
      "epoch": 18.283542976939202,
      "grad_norm": 0.13369764387607574,
      "learning_rate": 1.114149108080678e-06,
      "loss": 0.5688,
      "num_input_tokens_seen": 22806552,
      "step": 34885
    },
    {
      "epoch": 18.28616352201258,
      "grad_norm": 0.12356818467378616,
      "learning_rate": 1.1107761520904925e-06,
      "loss": 0.4559,
      "num_input_tokens_seen": 22809496,
      "step": 34890
    },
    {
      "epoch": 18.288784067085953,
      "grad_norm": 0.16181594133377075,
      "learning_rate": 1.1074081934533908e-06,
      "loss": 0.3862,
      "num_input_tokens_seen": 22812664,
      "step": 34895
    },
    {
      "epoch": 18.29140461215933,
      "grad_norm": 0.20048539340496063,
      "learning_rate": 1.1040452328739153e-06,
      "loss": 0.3734,
      "num_input_tokens_seen": 22815992,
      "step": 34900
    },
    {
      "epoch": 18.294025157232703,
      "grad_norm": 0.12144790589809418,
      "learning_rate": 1.1006872710555615e-06,
      "loss": 0.4606,
      "num_input_tokens_seen": 22818968,
      "step": 34905
    },
    {
      "epoch": 18.29664570230608,
      "grad_norm": 0.22795286774635315,
      "learning_rate": 1.0973343087007703e-06,
      "loss": 0.4235,
      "num_input_tokens_seen": 22821624,
      "step": 34910
    },
    {
      "epoch": 18.299266247379457,
      "grad_norm": 0.25547951459884644,
      "learning_rate": 1.0939863465109473e-06,
      "loss": 0.6249,
      "num_input_tokens_seen": 22824696,
      "step": 34915
    },
    {
      "epoch": 18.30188679245283,
      "grad_norm": 0.13021856546401978,
      "learning_rate": 1.0906433851864433e-06,
      "loss": 0.4522,
      "num_input_tokens_seen": 22827864,
      "step": 34920
    },
    {
      "epoch": 18.304507337526207,
      "grad_norm": 0.15464380383491516,
      "learning_rate": 1.0873054254265713e-06,
      "loss": 0.6838,
      "num_input_tokens_seen": 22831384,
      "step": 34925
    },
    {
      "epoch": 18.30712788259958,
      "grad_norm": 0.16598917543888092,
      "learning_rate": 1.0839724679295976e-06,
      "loss": 0.5716,
      "num_input_tokens_seen": 22835928,
      "step": 34930
    },
    {
      "epoch": 18.309748427672957,
      "grad_norm": 0.13366957008838654,
      "learning_rate": 1.0806445133927311e-06,
      "loss": 0.55,
      "num_input_tokens_seen": 22838776,
      "step": 34935
    },
    {
      "epoch": 18.31236897274633,
      "grad_norm": 0.15137527883052826,
      "learning_rate": 1.0773215625121457e-06,
      "loss": 0.4543,
      "num_input_tokens_seen": 22842008,
      "step": 34940
    },
    {
      "epoch": 18.314989517819708,
      "grad_norm": 0.16372928023338318,
      "learning_rate": 1.0740036159829715e-06,
      "loss": 0.5504,
      "num_input_tokens_seen": 22846040,
      "step": 34945
    },
    {
      "epoch": 18.31761006289308,
      "grad_norm": 0.17005380988121033,
      "learning_rate": 1.0706906744992724e-06,
      "loss": 0.3854,
      "num_input_tokens_seen": 22850328,
      "step": 34950
    },
    {
      "epoch": 18.320230607966458,
      "grad_norm": 0.3304808735847473,
      "learning_rate": 1.0673827387540857e-06,
      "loss": 0.5786,
      "num_input_tokens_seen": 22853624,
      "step": 34955
    },
    {
      "epoch": 18.32285115303983,
      "grad_norm": 0.3329010009765625,
      "learning_rate": 1.0640798094393916e-06,
      "loss": 0.6421,
      "num_input_tokens_seen": 22857240,
      "step": 34960
    },
    {
      "epoch": 18.32547169811321,
      "grad_norm": 0.15335321426391602,
      "learning_rate": 1.060781887246126e-06,
      "loss": 0.416,
      "num_input_tokens_seen": 22860216,
      "step": 34965
    },
    {
      "epoch": 18.328092243186582,
      "grad_norm": 0.1864629089832306,
      "learning_rate": 1.057488972864179e-06,
      "loss": 0.3453,
      "num_input_tokens_seen": 22864920,
      "step": 34970
    },
    {
      "epoch": 18.33071278825996,
      "grad_norm": 0.1840818077325821,
      "learning_rate": 1.0542010669823855e-06,
      "loss": 0.5024,
      "num_input_tokens_seen": 22867928,
      "step": 34975
    },
    {
      "epoch": 18.333333333333332,
      "grad_norm": 0.12909506261348724,
      "learning_rate": 1.0509181702885395e-06,
      "loss": 0.4994,
      "num_input_tokens_seen": 22871256,
      "step": 34980
    },
    {
      "epoch": 18.33595387840671,
      "grad_norm": 0.3747239112854004,
      "learning_rate": 1.047640283469392e-06,
      "loss": 0.5738,
      "num_input_tokens_seen": 22875544,
      "step": 34985
    },
    {
      "epoch": 18.338574423480082,
      "grad_norm": 0.19621814787387848,
      "learning_rate": 1.0443674072106308e-06,
      "loss": 0.5409,
      "num_input_tokens_seen": 22878456,
      "step": 34990
    },
    {
      "epoch": 18.34119496855346,
      "grad_norm": 0.14647622406482697,
      "learning_rate": 1.041099542196905e-06,
      "loss": 0.5867,
      "num_input_tokens_seen": 22882168,
      "step": 34995
    },
    {
      "epoch": 18.343815513626833,
      "grad_norm": 0.13716241717338562,
      "learning_rate": 1.0378366891118234e-06,
      "loss": 0.4293,
      "num_input_tokens_seen": 22884824,
      "step": 35000
    },
    {
      "epoch": 18.34643605870021,
      "grad_norm": 0.1249300166964531,
      "learning_rate": 1.0345788486379265e-06,
      "loss": 0.4807,
      "num_input_tokens_seen": 22887864,
      "step": 35005
    },
    {
      "epoch": 18.349056603773583,
      "grad_norm": 0.1493472456932068,
      "learning_rate": 1.0313260214567272e-06,
      "loss": 0.5737,
      "num_input_tokens_seen": 22892184,
      "step": 35010
    },
    {
      "epoch": 18.35167714884696,
      "grad_norm": 0.18071524798870087,
      "learning_rate": 1.0280782082486729e-06,
      "loss": 0.5277,
      "num_input_tokens_seen": 22895096,
      "step": 35015
    },
    {
      "epoch": 18.354297693920337,
      "grad_norm": 0.15772992372512817,
      "learning_rate": 1.02483540969317e-06,
      "loss": 0.4892,
      "num_input_tokens_seen": 22898168,
      "step": 35020
    },
    {
      "epoch": 18.35691823899371,
      "grad_norm": 0.10174944251775742,
      "learning_rate": 1.0215976264685763e-06,
      "loss": 0.4718,
      "num_input_tokens_seen": 22901368,
      "step": 35025
    },
    {
      "epoch": 18.359538784067087,
      "grad_norm": 0.09410137683153152,
      "learning_rate": 1.0183648592522021e-06,
      "loss": 0.5924,
      "num_input_tokens_seen": 22905048,
      "step": 35030
    },
    {
      "epoch": 18.36215932914046,
      "grad_norm": 0.13888996839523315,
      "learning_rate": 1.0151371087203015e-06,
      "loss": 0.4939,
      "num_input_tokens_seen": 22907800,
      "step": 35035
    },
    {
      "epoch": 18.364779874213838,
      "grad_norm": 0.18397746980190277,
      "learning_rate": 1.0119143755480815e-06,
      "loss": 0.5045,
      "num_input_tokens_seen": 22911480,
      "step": 35040
    },
    {
      "epoch": 18.36740041928721,
      "grad_norm": 0.15538714826107025,
      "learning_rate": 1.0086966604097026e-06,
      "loss": 0.4024,
      "num_input_tokens_seen": 22914936,
      "step": 35045
    },
    {
      "epoch": 18.370020964360588,
      "grad_norm": 0.18989454209804535,
      "learning_rate": 1.0054839639782766e-06,
      "loss": 0.7257,
      "num_input_tokens_seen": 22918424,
      "step": 35050
    },
    {
      "epoch": 18.37264150943396,
      "grad_norm": 0.16075845062732697,
      "learning_rate": 1.00227628692586e-06,
      "loss": 0.5147,
      "num_input_tokens_seen": 22921112,
      "step": 35055
    },
    {
      "epoch": 18.37526205450734,
      "grad_norm": 0.18721430003643036,
      "learning_rate": 9.990736299234692e-07,
      "loss": 0.4996,
      "num_input_tokens_seen": 22924024,
      "step": 35060
    },
    {
      "epoch": 18.377882599580712,
      "grad_norm": 0.1941836029291153,
      "learning_rate": 9.958759936410573e-07,
      "loss": 0.3625,
      "num_input_tokens_seen": 22927128,
      "step": 35065
    },
    {
      "epoch": 18.38050314465409,
      "grad_norm": 0.11393356323242188,
      "learning_rate": 9.926833787475304e-07,
      "loss": 0.33,
      "num_input_tokens_seen": 22929912,
      "step": 35070
    },
    {
      "epoch": 18.383123689727462,
      "grad_norm": 0.14666345715522766,
      "learning_rate": 9.89495785910749e-07,
      "loss": 0.4341,
      "num_input_tokens_seen": 22932504,
      "step": 35075
    },
    {
      "epoch": 18.38574423480084,
      "grad_norm": 0.1772172898054123,
      "learning_rate": 9.86313215797524e-07,
      "loss": 0.384,
      "num_input_tokens_seen": 22936600,
      "step": 35080
    },
    {
      "epoch": 18.388364779874212,
      "grad_norm": 0.210874542593956,
      "learning_rate": 9.831356690736149e-07,
      "loss": 0.4732,
      "num_input_tokens_seen": 22939864,
      "step": 35085
    },
    {
      "epoch": 18.39098532494759,
      "grad_norm": 0.1834060400724411,
      "learning_rate": 9.799631464037256e-07,
      "loss": 0.6091,
      "num_input_tokens_seen": 22943480,
      "step": 35090
    },
    {
      "epoch": 18.393605870020963,
      "grad_norm": 0.13647718727588654,
      "learning_rate": 9.76795648451509e-07,
      "loss": 0.5458,
      "num_input_tokens_seen": 22946712,
      "step": 35095
    },
    {
      "epoch": 18.39622641509434,
      "grad_norm": 0.21514543890953064,
      "learning_rate": 9.73633175879568e-07,
      "loss": 0.5107,
      "num_input_tokens_seen": 22950392,
      "step": 35100
    },
    {
      "epoch": 18.398846960167713,
      "grad_norm": 0.1247333362698555,
      "learning_rate": 9.704757293494627e-07,
      "loss": 0.5007,
      "num_input_tokens_seen": 22954264,
      "step": 35105
    },
    {
      "epoch": 18.40146750524109,
      "grad_norm": 0.14134326577186584,
      "learning_rate": 9.673233095216922e-07,
      "loss": 0.4014,
      "num_input_tokens_seen": 22958232,
      "step": 35110
    },
    {
      "epoch": 18.404088050314467,
      "grad_norm": 0.21597670018672943,
      "learning_rate": 9.641759170557068e-07,
      "loss": 0.67,
      "num_input_tokens_seen": 22961208,
      "step": 35115
    },
    {
      "epoch": 18.40670859538784,
      "grad_norm": 0.1827617734670639,
      "learning_rate": 9.610335526099024e-07,
      "loss": 0.3628,
      "num_input_tokens_seen": 22966296,
      "step": 35120
    },
    {
      "epoch": 18.409329140461217,
      "grad_norm": 0.19411641359329224,
      "learning_rate": 9.578962168416278e-07,
      "loss": 0.4562,
      "num_input_tokens_seen": 22968856,
      "step": 35125
    },
    {
      "epoch": 18.41194968553459,
      "grad_norm": 0.15964339673519135,
      "learning_rate": 9.547639104071775e-07,
      "loss": 0.3192,
      "num_input_tokens_seen": 22972056,
      "step": 35130
    },
    {
      "epoch": 18.414570230607968,
      "grad_norm": 0.2550467848777771,
      "learning_rate": 9.516366339617944e-07,
      "loss": 0.3711,
      "num_input_tokens_seen": 22974840,
      "step": 35135
    },
    {
      "epoch": 18.41719077568134,
      "grad_norm": 0.13930854201316833,
      "learning_rate": 9.485143881596714e-07,
      "loss": 0.5327,
      "num_input_tokens_seen": 22978520,
      "step": 35140
    },
    {
      "epoch": 18.419811320754718,
      "grad_norm": 0.18615728616714478,
      "learning_rate": 9.453971736539447e-07,
      "loss": 0.4524,
      "num_input_tokens_seen": 22982136,
      "step": 35145
    },
    {
      "epoch": 18.42243186582809,
      "grad_norm": 0.3079990744590759,
      "learning_rate": 9.42284991096698e-07,
      "loss": 0.4697,
      "num_input_tokens_seen": 22985304,
      "step": 35150
    },
    {
      "epoch": 18.42505241090147,
      "grad_norm": 0.1763480007648468,
      "learning_rate": 9.391778411389634e-07,
      "loss": 0.444,
      "num_input_tokens_seen": 22988856,
      "step": 35155
    },
    {
      "epoch": 18.427672955974842,
      "grad_norm": 0.11841606348752975,
      "learning_rate": 9.360757244307267e-07,
      "loss": 0.4057,
      "num_input_tokens_seen": 22991896,
      "step": 35160
    },
    {
      "epoch": 18.43029350104822,
      "grad_norm": 0.15007302165031433,
      "learning_rate": 9.329786416209157e-07,
      "loss": 0.4598,
      "num_input_tokens_seen": 22994808,
      "step": 35165
    },
    {
      "epoch": 18.432914046121592,
      "grad_norm": 0.170922189950943,
      "learning_rate": 9.298865933574014e-07,
      "loss": 0.4823,
      "num_input_tokens_seen": 22997560,
      "step": 35170
    },
    {
      "epoch": 18.43553459119497,
      "grad_norm": 0.14604505896568298,
      "learning_rate": 9.267995802870022e-07,
      "loss": 0.5996,
      "num_input_tokens_seen": 23000696,
      "step": 35175
    },
    {
      "epoch": 18.438155136268342,
      "grad_norm": 0.2527996897697449,
      "learning_rate": 9.237176030554878e-07,
      "loss": 0.4745,
      "num_input_tokens_seen": 23003320,
      "step": 35180
    },
    {
      "epoch": 18.44077568134172,
      "grad_norm": 0.33457791805267334,
      "learning_rate": 9.206406623075758e-07,
      "loss": 0.5135,
      "num_input_tokens_seen": 23006456,
      "step": 35185
    },
    {
      "epoch": 18.443396226415093,
      "grad_norm": 0.15089528262615204,
      "learning_rate": 9.175687586869292e-07,
      "loss": 0.4353,
      "num_input_tokens_seen": 23010712,
      "step": 35190
    },
    {
      "epoch": 18.44601677148847,
      "grad_norm": 0.22230233252048492,
      "learning_rate": 9.145018928361476e-07,
      "loss": 0.4854,
      "num_input_tokens_seen": 23014232,
      "step": 35195
    },
    {
      "epoch": 18.448637316561843,
      "grad_norm": 0.1338147521018982,
      "learning_rate": 9.114400653967903e-07,
      "loss": 0.4428,
      "num_input_tokens_seen": 23016952,
      "step": 35200
    },
    {
      "epoch": 18.45125786163522,
      "grad_norm": 0.16553336381912231,
      "learning_rate": 9.083832770093531e-07,
      "loss": 0.4645,
      "num_input_tokens_seen": 23019832,
      "step": 35205
    },
    {
      "epoch": 18.453878406708597,
      "grad_norm": 0.20341245830059052,
      "learning_rate": 9.053315283132802e-07,
      "loss": 0.5517,
      "num_input_tokens_seen": 23022264,
      "step": 35210
    },
    {
      "epoch": 18.45649895178197,
      "grad_norm": 0.15748031437397003,
      "learning_rate": 9.022848199469663e-07,
      "loss": 0.3976,
      "num_input_tokens_seen": 23025912,
      "step": 35215
    },
    {
      "epoch": 18.459119496855347,
      "grad_norm": 0.15398649871349335,
      "learning_rate": 8.992431525477518e-07,
      "loss": 0.3996,
      "num_input_tokens_seen": 23029560,
      "step": 35220
    },
    {
      "epoch": 18.46174004192872,
      "grad_norm": 0.14077453315258026,
      "learning_rate": 8.962065267519082e-07,
      "loss": 0.4938,
      "num_input_tokens_seen": 23033176,
      "step": 35225
    },
    {
      "epoch": 18.464360587002098,
      "grad_norm": 0.15485620498657227,
      "learning_rate": 8.931749431946718e-07,
      "loss": 0.4625,
      "num_input_tokens_seen": 23036568,
      "step": 35230
    },
    {
      "epoch": 18.46698113207547,
      "grad_norm": 0.35363808274269104,
      "learning_rate": 8.901484025102103e-07,
      "loss": 0.5316,
      "num_input_tokens_seen": 23039448,
      "step": 35235
    },
    {
      "epoch": 18.469601677148848,
      "grad_norm": 0.15658128261566162,
      "learning_rate": 8.871269053316422e-07,
      "loss": 0.4954,
      "num_input_tokens_seen": 23043384,
      "step": 35240
    },
    {
      "epoch": 18.47222222222222,
      "grad_norm": 0.22082079946994781,
      "learning_rate": 8.841104522910343e-07,
      "loss": 0.595,
      "num_input_tokens_seen": 23047128,
      "step": 35245
    },
    {
      "epoch": 18.4748427672956,
      "grad_norm": 0.14530499279499054,
      "learning_rate": 8.8109904401939e-07,
      "loss": 0.5177,
      "num_input_tokens_seen": 23050072,
      "step": 35250
    },
    {
      "epoch": 18.47746331236897,
      "grad_norm": 0.15128330886363983,
      "learning_rate": 8.780926811466639e-07,
      "loss": 0.486,
      "num_input_tokens_seen": 23053688,
      "step": 35255
    },
    {
      "epoch": 18.48008385744235,
      "grad_norm": 0.1623433381319046,
      "learning_rate": 8.7509136430175e-07,
      "loss": 0.4018,
      "num_input_tokens_seen": 23056440,
      "step": 35260
    },
    {
      "epoch": 18.482704402515722,
      "grad_norm": 0.13425423204898834,
      "learning_rate": 8.720950941124877e-07,
      "loss": 0.4344,
      "num_input_tokens_seen": 23060120,
      "step": 35265
    },
    {
      "epoch": 18.4853249475891,
      "grad_norm": 0.19407503306865692,
      "learning_rate": 8.691038712056732e-07,
      "loss": 0.5524,
      "num_input_tokens_seen": 23063544,
      "step": 35270
    },
    {
      "epoch": 18.487945492662472,
      "grad_norm": 0.1869829297065735,
      "learning_rate": 8.66117696207025e-07,
      "loss": 0.394,
      "num_input_tokens_seen": 23067096,
      "step": 35275
    },
    {
      "epoch": 18.49056603773585,
      "grad_norm": 0.11900676041841507,
      "learning_rate": 8.631365697412186e-07,
      "loss": 0.3864,
      "num_input_tokens_seen": 23070328,
      "step": 35280
    },
    {
      "epoch": 18.493186582809223,
      "grad_norm": 0.12126904726028442,
      "learning_rate": 8.601604924318802e-07,
      "loss": 0.5685,
      "num_input_tokens_seen": 23074648,
      "step": 35285
    },
    {
      "epoch": 18.4958071278826,
      "grad_norm": 0.1264030784368515,
      "learning_rate": 8.571894649015588e-07,
      "loss": 0.4489,
      "num_input_tokens_seen": 23077400,
      "step": 35290
    },
    {
      "epoch": 18.498427672955973,
      "grad_norm": 0.15772683918476105,
      "learning_rate": 8.542234877717686e-07,
      "loss": 0.5857,
      "num_input_tokens_seen": 23079992,
      "step": 35295
    },
    {
      "epoch": 18.50104821802935,
      "grad_norm": 0.21047505736351013,
      "learning_rate": 8.512625616629549e-07,
      "loss": 0.413,
      "num_input_tokens_seen": 23082936,
      "step": 35300
    },
    {
      "epoch": 18.503668763102727,
      "grad_norm": 0.2053804248571396,
      "learning_rate": 8.483066871945083e-07,
      "loss": 0.4488,
      "num_input_tokens_seen": 23085528,
      "step": 35305
    },
    {
      "epoch": 18.5062893081761,
      "grad_norm": 0.2744680643081665,
      "learning_rate": 8.453558649847676e-07,
      "loss": 0.4908,
      "num_input_tokens_seen": 23089016,
      "step": 35310
    },
    {
      "epoch": 18.508909853249477,
      "grad_norm": 0.16642320156097412,
      "learning_rate": 8.424100956510112e-07,
      "loss": 0.4617,
      "num_input_tokens_seen": 23091512,
      "step": 35315
    },
    {
      "epoch": 18.51153039832285,
      "grad_norm": 0.25256115198135376,
      "learning_rate": 8.394693798094544e-07,
      "loss": 0.6718,
      "num_input_tokens_seen": 23095096,
      "step": 35320
    },
    {
      "epoch": 18.514150943396228,
      "grad_norm": 0.09945955127477646,
      "learning_rate": 8.365337180752691e-07,
      "loss": 0.4831,
      "num_input_tokens_seen": 23098904,
      "step": 35325
    },
    {
      "epoch": 18.5167714884696,
      "grad_norm": 0.20956669747829437,
      "learning_rate": 8.336031110625558e-07,
      "loss": 0.3937,
      "num_input_tokens_seen": 23101944,
      "step": 35330
    },
    {
      "epoch": 18.519392033542978,
      "grad_norm": 0.15376578271389008,
      "learning_rate": 8.306775593843657e-07,
      "loss": 0.4363,
      "num_input_tokens_seen": 23105816,
      "step": 35335
    },
    {
      "epoch": 18.52201257861635,
      "grad_norm": 0.1323937624692917,
      "learning_rate": 8.277570636526955e-07,
      "loss": 0.4513,
      "num_input_tokens_seen": 23109304,
      "step": 35340
    },
    {
      "epoch": 18.52463312368973,
      "grad_norm": 0.12626752257347107,
      "learning_rate": 8.24841624478473e-07,
      "loss": 0.4345,
      "num_input_tokens_seen": 23112984,
      "step": 35345
    },
    {
      "epoch": 18.5272536687631,
      "grad_norm": 0.18798890709877014,
      "learning_rate": 8.219312424715825e-07,
      "loss": 0.4219,
      "num_input_tokens_seen": 23115832,
      "step": 35350
    },
    {
      "epoch": 18.52987421383648,
      "grad_norm": 0.13462521135807037,
      "learning_rate": 8.190259182408345e-07,
      "loss": 0.5102,
      "num_input_tokens_seen": 23118872,
      "step": 35355
    },
    {
      "epoch": 18.532494758909852,
      "grad_norm": 0.19713714718818665,
      "learning_rate": 8.161256523939925e-07,
      "loss": 0.6024,
      "num_input_tokens_seen": 23121880,
      "step": 35360
    },
    {
      "epoch": 18.53511530398323,
      "grad_norm": 0.13252700865268707,
      "learning_rate": 8.132304455377576e-07,
      "loss": 0.4903,
      "num_input_tokens_seen": 23125528,
      "step": 35365
    },
    {
      "epoch": 18.537735849056602,
      "grad_norm": 0.1974513679742813,
      "learning_rate": 8.103402982777813e-07,
      "loss": 0.5947,
      "num_input_tokens_seen": 23130360,
      "step": 35370
    },
    {
      "epoch": 18.54035639412998,
      "grad_norm": 0.18436993658542633,
      "learning_rate": 8.07455211218644e-07,
      "loss": 0.4047,
      "num_input_tokens_seen": 23133912,
      "step": 35375
    },
    {
      "epoch": 18.542976939203353,
      "grad_norm": 0.1256558895111084,
      "learning_rate": 8.045751849638738e-07,
      "loss": 0.5035,
      "num_input_tokens_seen": 23136440,
      "step": 35380
    },
    {
      "epoch": 18.54559748427673,
      "grad_norm": 0.1687905192375183,
      "learning_rate": 8.017002201159363e-07,
      "loss": 0.4814,
      "num_input_tokens_seen": 23139544,
      "step": 35385
    },
    {
      "epoch": 18.548218029350103,
      "grad_norm": 0.12194141745567322,
      "learning_rate": 7.988303172762418e-07,
      "loss": 0.4079,
      "num_input_tokens_seen": 23144408,
      "step": 35390
    },
    {
      "epoch": 18.55083857442348,
      "grad_norm": 0.22242777049541473,
      "learning_rate": 7.959654770451463e-07,
      "loss": 0.5361,
      "num_input_tokens_seen": 23147960,
      "step": 35395
    },
    {
      "epoch": 18.553459119496857,
      "grad_norm": 0.27511879801750183,
      "learning_rate": 7.9310570002194e-07,
      "loss": 0.4295,
      "num_input_tokens_seen": 23151096,
      "step": 35400
    },
    {
      "epoch": 18.55607966457023,
      "grad_norm": 0.16785204410552979,
      "learning_rate": 7.902509868048552e-07,
      "loss": 0.4769,
      "num_input_tokens_seen": 23154424,
      "step": 35405
    },
    {
      "epoch": 18.558700209643607,
      "grad_norm": 0.1484881043434143,
      "learning_rate": 7.874013379910616e-07,
      "loss": 0.3535,
      "num_input_tokens_seen": 23157048,
      "step": 35410
    },
    {
      "epoch": 18.56132075471698,
      "grad_norm": 0.1286952793598175,
      "learning_rate": 7.845567541766769e-07,
      "loss": 0.3748,
      "num_input_tokens_seen": 23160792,
      "step": 35415
    },
    {
      "epoch": 18.563941299790358,
      "grad_norm": 0.12560130655765533,
      "learning_rate": 7.817172359567526e-07,
      "loss": 0.4829,
      "num_input_tokens_seen": 23164248,
      "step": 35420
    },
    {
      "epoch": 18.56656184486373,
      "grad_norm": 0.2708837687969208,
      "learning_rate": 7.788827839252888e-07,
      "loss": 0.598,
      "num_input_tokens_seen": 23167832,
      "step": 35425
    },
    {
      "epoch": 18.569182389937108,
      "grad_norm": 0.1151028499007225,
      "learning_rate": 7.760533986752222e-07,
      "loss": 0.4055,
      "num_input_tokens_seen": 23172152,
      "step": 35430
    },
    {
      "epoch": 18.57180293501048,
      "grad_norm": 0.15131188929080963,
      "learning_rate": 7.732290807984155e-07,
      "loss": 0.4386,
      "num_input_tokens_seen": 23175000,
      "step": 35435
    },
    {
      "epoch": 18.57442348008386,
      "grad_norm": 0.20513547956943512,
      "learning_rate": 7.704098308856933e-07,
      "loss": 0.4927,
      "num_input_tokens_seen": 23177976,
      "step": 35440
    },
    {
      "epoch": 18.57704402515723,
      "grad_norm": 0.14322520792484283,
      "learning_rate": 7.675956495268088e-07,
      "loss": 0.4589,
      "num_input_tokens_seen": 23180792,
      "step": 35445
    },
    {
      "epoch": 18.57966457023061,
      "grad_norm": 0.13368795812129974,
      "learning_rate": 7.647865373104551e-07,
      "loss": 0.6853,
      "num_input_tokens_seen": 23184344,
      "step": 35450
    },
    {
      "epoch": 18.582285115303982,
      "grad_norm": 0.13263879716396332,
      "learning_rate": 7.619824948242732e-07,
      "loss": 0.4811,
      "num_input_tokens_seen": 23187640,
      "step": 35455
    },
    {
      "epoch": 18.58490566037736,
      "grad_norm": 0.20143456757068634,
      "learning_rate": 7.5918352265483e-07,
      "loss": 0.6901,
      "num_input_tokens_seen": 23191256,
      "step": 35460
    },
    {
      "epoch": 18.587526205450732,
      "grad_norm": 0.1657903641462326,
      "learning_rate": 7.563896213876348e-07,
      "loss": 0.5452,
      "num_input_tokens_seen": 23193944,
      "step": 35465
    },
    {
      "epoch": 18.59014675052411,
      "grad_norm": 0.17689833045005798,
      "learning_rate": 7.53600791607148e-07,
      "loss": 0.5506,
      "num_input_tokens_seen": 23197464,
      "step": 35470
    },
    {
      "epoch": 18.592767295597483,
      "grad_norm": 0.1417190283536911,
      "learning_rate": 7.508170338967585e-07,
      "loss": 0.4011,
      "num_input_tokens_seen": 23200024,
      "step": 35475
    },
    {
      "epoch": 18.59538784067086,
      "grad_norm": 0.15802739560604095,
      "learning_rate": 7.480383488387976e-07,
      "loss": 0.4914,
      "num_input_tokens_seen": 23202968,
      "step": 35480
    },
    {
      "epoch": 18.598008385744233,
      "grad_norm": 0.14603760838508606,
      "learning_rate": 7.452647370145338e-07,
      "loss": 0.4438,
      "num_input_tokens_seen": 23206296,
      "step": 35485
    },
    {
      "epoch": 18.60062893081761,
      "grad_norm": 0.14003023505210876,
      "learning_rate": 7.424961990041696e-07,
      "loss": 0.4851,
      "num_input_tokens_seen": 23209304,
      "step": 35490
    },
    {
      "epoch": 18.603249475890987,
      "grad_norm": 0.15862469375133514,
      "learning_rate": 7.397327353868583e-07,
      "loss": 0.38,
      "num_input_tokens_seen": 23213080,
      "step": 35495
    },
    {
      "epoch": 18.60587002096436,
      "grad_norm": 0.13443943858146667,
      "learning_rate": 7.369743467406848e-07,
      "loss": 0.4583,
      "num_input_tokens_seen": 23215928,
      "step": 35500
    },
    {
      "epoch": 18.608490566037737,
      "grad_norm": 0.13520710170269012,
      "learning_rate": 7.34221033642668e-07,
      "loss": 0.4974,
      "num_input_tokens_seen": 23219320,
      "step": 35505
    },
    {
      "epoch": 18.61111111111111,
      "grad_norm": 0.13332436978816986,
      "learning_rate": 7.314727966687751e-07,
      "loss": 0.4262,
      "num_input_tokens_seen": 23222296,
      "step": 35510
    },
    {
      "epoch": 18.613731656184488,
      "grad_norm": 0.1879418045282364,
      "learning_rate": 7.287296363939045e-07,
      "loss": 0.5021,
      "num_input_tokens_seen": 23224728,
      "step": 35515
    },
    {
      "epoch": 18.61635220125786,
      "grad_norm": 0.15984730422496796,
      "learning_rate": 7.259915533918888e-07,
      "loss": 0.4221,
      "num_input_tokens_seen": 23228248,
      "step": 35520
    },
    {
      "epoch": 18.618972746331238,
      "grad_norm": 0.21155349910259247,
      "learning_rate": 7.232585482355086e-07,
      "loss": 0.47,
      "num_input_tokens_seen": 23232600,
      "step": 35525
    },
    {
      "epoch": 18.62159329140461,
      "grad_norm": 0.2504297196865082,
      "learning_rate": 7.205306214964736e-07,
      "loss": 0.5163,
      "num_input_tokens_seen": 23235384,
      "step": 35530
    },
    {
      "epoch": 18.62421383647799,
      "grad_norm": 0.12772174179553986,
      "learning_rate": 7.178077737454436e-07,
      "loss": 0.4619,
      "num_input_tokens_seen": 23238744,
      "step": 35535
    },
    {
      "epoch": 18.62683438155136,
      "grad_norm": 0.14916515350341797,
      "learning_rate": 7.150900055519993e-07,
      "loss": 0.4641,
      "num_input_tokens_seen": 23241720,
      "step": 35540
    },
    {
      "epoch": 18.62945492662474,
      "grad_norm": 0.14865754544734955,
      "learning_rate": 7.123773174846693e-07,
      "loss": 0.5143,
      "num_input_tokens_seen": 23244696,
      "step": 35545
    },
    {
      "epoch": 18.632075471698112,
      "grad_norm": 0.09523794054985046,
      "learning_rate": 7.096697101109134e-07,
      "loss": 0.3664,
      "num_input_tokens_seen": 23248536,
      "step": 35550
    },
    {
      "epoch": 18.63469601677149,
      "grad_norm": 0.18950410187244415,
      "learning_rate": 7.069671839971371e-07,
      "loss": 0.4322,
      "num_input_tokens_seen": 23251352,
      "step": 35555
    },
    {
      "epoch": 18.637316561844862,
      "grad_norm": 0.14175696671009064,
      "learning_rate": 7.042697397086795e-07,
      "loss": 0.4098,
      "num_input_tokens_seen": 23254520,
      "step": 35560
    },
    {
      "epoch": 18.63993710691824,
      "grad_norm": 0.09578465670347214,
      "learning_rate": 7.015773778098089e-07,
      "loss": 0.5812,
      "num_input_tokens_seen": 23261752,
      "step": 35565
    },
    {
      "epoch": 18.642557651991613,
      "grad_norm": 0.17997489869594574,
      "learning_rate": 6.988900988637415e-07,
      "loss": 0.5625,
      "num_input_tokens_seen": 23265176,
      "step": 35570
    },
    {
      "epoch": 18.64517819706499,
      "grad_norm": 0.32629215717315674,
      "learning_rate": 6.962079034326218e-07,
      "loss": 0.4588,
      "num_input_tokens_seen": 23267800,
      "step": 35575
    },
    {
      "epoch": 18.647798742138363,
      "grad_norm": 0.2674938142299652,
      "learning_rate": 6.935307920775346e-07,
      "loss": 0.4087,
      "num_input_tokens_seen": 23270488,
      "step": 35580
    },
    {
      "epoch": 18.65041928721174,
      "grad_norm": 0.15083834528923035,
      "learning_rate": 6.90858765358507e-07,
      "loss": 0.5108,
      "num_input_tokens_seen": 23274008,
      "step": 35585
    },
    {
      "epoch": 18.653039832285117,
      "grad_norm": 0.24915438890457153,
      "learning_rate": 6.881918238344915e-07,
      "loss": 0.3167,
      "num_input_tokens_seen": 23277240,
      "step": 35590
    },
    {
      "epoch": 18.65566037735849,
      "grad_norm": 0.2583247125148773,
      "learning_rate": 6.855299680633808e-07,
      "loss": 0.5475,
      "num_input_tokens_seen": 23279768,
      "step": 35595
    },
    {
      "epoch": 18.658280922431867,
      "grad_norm": 0.2079475224018097,
      "learning_rate": 6.828731986020104e-07,
      "loss": 0.5637,
      "num_input_tokens_seen": 23283128,
      "step": 35600
    },
    {
      "epoch": 18.66090146750524,
      "grad_norm": 0.17474882304668427,
      "learning_rate": 6.802215160061381e-07,
      "loss": 0.4306,
      "num_input_tokens_seen": 23286424,
      "step": 35605
    },
    {
      "epoch": 18.663522012578618,
      "grad_norm": 0.14565733075141907,
      "learning_rate": 6.775749208304704e-07,
      "loss": 0.5598,
      "num_input_tokens_seen": 23289880,
      "step": 35610
    },
    {
      "epoch": 18.66614255765199,
      "grad_norm": 0.19898931682109833,
      "learning_rate": 6.749334136286478e-07,
      "loss": 0.4934,
      "num_input_tokens_seen": 23293528,
      "step": 35615
    },
    {
      "epoch": 18.668763102725368,
      "grad_norm": 0.22261762619018555,
      "learning_rate": 6.722969949532365e-07,
      "loss": 0.3559,
      "num_input_tokens_seen": 23296952,
      "step": 35620
    },
    {
      "epoch": 18.67138364779874,
      "grad_norm": 0.2140236347913742,
      "learning_rate": 6.69665665355751e-07,
      "loss": 0.4485,
      "num_input_tokens_seen": 23299384,
      "step": 35625
    },
    {
      "epoch": 18.67400419287212,
      "grad_norm": 0.16182366013526917,
      "learning_rate": 6.670394253866314e-07,
      "loss": 0.4796,
      "num_input_tokens_seen": 23306232,
      "step": 35630
    },
    {
      "epoch": 18.67662473794549,
      "grad_norm": 0.23496124148368835,
      "learning_rate": 6.644182755952605e-07,
      "loss": 0.5896,
      "num_input_tokens_seen": 23309144,
      "step": 35635
    },
    {
      "epoch": 18.67924528301887,
      "grad_norm": 0.1227615624666214,
      "learning_rate": 6.618022165299526e-07,
      "loss": 0.4343,
      "num_input_tokens_seen": 23312472,
      "step": 35640
    },
    {
      "epoch": 18.681865828092242,
      "grad_norm": 0.18853628635406494,
      "learning_rate": 6.591912487379531e-07,
      "loss": 0.3926,
      "num_input_tokens_seen": 23315320,
      "step": 35645
    },
    {
      "epoch": 18.68448637316562,
      "grad_norm": 0.24648351967334747,
      "learning_rate": 6.565853727654503e-07,
      "loss": 0.4866,
      "num_input_tokens_seen": 23318360,
      "step": 35650
    },
    {
      "epoch": 18.687106918238992,
      "grad_norm": 0.2653360664844513,
      "learning_rate": 6.539845891575636e-07,
      "loss": 0.4276,
      "num_input_tokens_seen": 23320824,
      "step": 35655
    },
    {
      "epoch": 18.68972746331237,
      "grad_norm": 0.1621062457561493,
      "learning_rate": 6.513888984583466e-07,
      "loss": 0.534,
      "num_input_tokens_seen": 23324216,
      "step": 35660
    },
    {
      "epoch": 18.692348008385743,
      "grad_norm": 0.30806422233581543,
      "learning_rate": 6.487983012107873e-07,
      "loss": 0.5073,
      "num_input_tokens_seen": 23326840,
      "step": 35665
    },
    {
      "epoch": 18.69496855345912,
      "grad_norm": 0.2599101662635803,
      "learning_rate": 6.462127979568133e-07,
      "loss": 0.5034,
      "num_input_tokens_seen": 23329464,
      "step": 35670
    },
    {
      "epoch": 18.697589098532493,
      "grad_norm": 0.2497348189353943,
      "learning_rate": 6.436323892372781e-07,
      "loss": 0.6048,
      "num_input_tokens_seen": 23332280,
      "step": 35675
    },
    {
      "epoch": 18.70020964360587,
      "grad_norm": 0.15242984890937805,
      "learning_rate": 6.410570755919721e-07,
      "loss": 0.4714,
      "num_input_tokens_seen": 23335384,
      "step": 35680
    },
    {
      "epoch": 18.702830188679247,
      "grad_norm": 0.1020333468914032,
      "learning_rate": 6.384868575596281e-07,
      "loss": 0.4276,
      "num_input_tokens_seen": 23338776,
      "step": 35685
    },
    {
      "epoch": 18.70545073375262,
      "grad_norm": 0.20430637896060944,
      "learning_rate": 6.359217356779024e-07,
      "loss": 0.5786,
      "num_input_tokens_seen": 23341624,
      "step": 35690
    },
    {
      "epoch": 18.708071278825997,
      "grad_norm": 0.12357322871685028,
      "learning_rate": 6.333617104833878e-07,
      "loss": 0.4833,
      "num_input_tokens_seen": 23345144,
      "step": 35695
    },
    {
      "epoch": 18.71069182389937,
      "grad_norm": 0.13309867680072784,
      "learning_rate": 6.308067825116142e-07,
      "loss": 0.3572,
      "num_input_tokens_seen": 23350328,
      "step": 35700
    },
    {
      "epoch": 18.713312368972748,
      "grad_norm": 0.2673215866088867,
      "learning_rate": 6.282569522970433e-07,
      "loss": 0.3921,
      "num_input_tokens_seen": 23353240,
      "step": 35705
    },
    {
      "epoch": 18.71593291404612,
      "grad_norm": 0.2446601539850235,
      "learning_rate": 6.257122203730731e-07,
      "loss": 0.5621,
      "num_input_tokens_seen": 23356120,
      "step": 35710
    },
    {
      "epoch": 18.718553459119498,
      "grad_norm": 0.211801677942276,
      "learning_rate": 6.231725872720251e-07,
      "loss": 0.605,
      "num_input_tokens_seen": 23359032,
      "step": 35715
    },
    {
      "epoch": 18.72117400419287,
      "grad_norm": 0.13057872653007507,
      "learning_rate": 6.206380535251716e-07,
      "loss": 0.4971,
      "num_input_tokens_seen": 23363064,
      "step": 35720
    },
    {
      "epoch": 18.72379454926625,
      "grad_norm": 0.20409640669822693,
      "learning_rate": 6.181086196626995e-07,
      "loss": 0.5284,
      "num_input_tokens_seen": 23365432,
      "step": 35725
    },
    {
      "epoch": 18.72641509433962,
      "grad_norm": 0.201237753033638,
      "learning_rate": 6.155842862137384e-07,
      "loss": 0.5327,
      "num_input_tokens_seen": 23368120,
      "step": 35730
    },
    {
      "epoch": 18.729035639413,
      "grad_norm": 0.1641024500131607,
      "learning_rate": 6.130650537063548e-07,
      "loss": 0.5206,
      "num_input_tokens_seen": 23371576,
      "step": 35735
    },
    {
      "epoch": 18.731656184486372,
      "grad_norm": 0.15386663377285004,
      "learning_rate": 6.105509226675438e-07,
      "loss": 0.4509,
      "num_input_tokens_seen": 23376216,
      "step": 35740
    },
    {
      "epoch": 18.73427672955975,
      "grad_norm": 0.17095868289470673,
      "learning_rate": 6.08041893623229e-07,
      "loss": 0.4069,
      "num_input_tokens_seen": 23379256,
      "step": 35745
    },
    {
      "epoch": 18.736897274633122,
      "grad_norm": 0.26678282022476196,
      "learning_rate": 6.055379670982713e-07,
      "loss": 0.4567,
      "num_input_tokens_seen": 23381912,
      "step": 35750
    },
    {
      "epoch": 18.7395178197065,
      "grad_norm": 0.18573465943336487,
      "learning_rate": 6.030391436164628e-07,
      "loss": 0.4599,
      "num_input_tokens_seen": 23384568,
      "step": 35755
    },
    {
      "epoch": 18.742138364779873,
      "grad_norm": 0.2663213312625885,
      "learning_rate": 6.005454237005298e-07,
      "loss": 0.4536,
      "num_input_tokens_seen": 23387320,
      "step": 35760
    },
    {
      "epoch": 18.74475890985325,
      "grad_norm": 0.2065044790506363,
      "learning_rate": 5.980568078721327e-07,
      "loss": 0.4297,
      "num_input_tokens_seen": 23390168,
      "step": 35765
    },
    {
      "epoch": 18.747379454926623,
      "grad_norm": 0.1605825126171112,
      "learning_rate": 5.955732966518606e-07,
      "loss": 0.5082,
      "num_input_tokens_seen": 23393080,
      "step": 35770
    },
    {
      "epoch": 18.75,
      "grad_norm": 0.1564423143863678,
      "learning_rate": 5.930948905592343e-07,
      "loss": 0.3648,
      "num_input_tokens_seen": 23395800,
      "step": 35775
    },
    {
      "epoch": 18.752620545073377,
      "grad_norm": 0.19706028699874878,
      "learning_rate": 5.906215901127054e-07,
      "loss": 0.4177,
      "num_input_tokens_seen": 23398424,
      "step": 35780
    },
    {
      "epoch": 18.75524109014675,
      "grad_norm": 0.19990426301956177,
      "learning_rate": 5.881533958296631e-07,
      "loss": 0.4712,
      "num_input_tokens_seen": 23400984,
      "step": 35785
    },
    {
      "epoch": 18.757861635220127,
      "grad_norm": 0.15784935653209686,
      "learning_rate": 5.856903082264248e-07,
      "loss": 0.4536,
      "num_input_tokens_seen": 23404952,
      "step": 35790
    },
    {
      "epoch": 18.7604821802935,
      "grad_norm": 0.1596129983663559,
      "learning_rate": 5.832323278182422e-07,
      "loss": 0.5044,
      "num_input_tokens_seen": 23409048,
      "step": 35795
    },
    {
      "epoch": 18.763102725366878,
      "grad_norm": 0.15215498208999634,
      "learning_rate": 5.807794551192986e-07,
      "loss": 0.5211,
      "num_input_tokens_seen": 23411864,
      "step": 35800
    },
    {
      "epoch": 18.76572327044025,
      "grad_norm": 0.1933017522096634,
      "learning_rate": 5.783316906427e-07,
      "loss": 0.4155,
      "num_input_tokens_seen": 23414936,
      "step": 35805
    },
    {
      "epoch": 18.768343815513628,
      "grad_norm": 0.21797218918800354,
      "learning_rate": 5.758890349004925e-07,
      "loss": 0.4854,
      "num_input_tokens_seen": 23417976,
      "step": 35810
    },
    {
      "epoch": 18.770964360587,
      "grad_norm": 0.1569690704345703,
      "learning_rate": 5.734514884036563e-07,
      "loss": 0.4018,
      "num_input_tokens_seen": 23420984,
      "step": 35815
    },
    {
      "epoch": 18.77358490566038,
      "grad_norm": 0.12824197113513947,
      "learning_rate": 5.710190516620945e-07,
      "loss": 0.4059,
      "num_input_tokens_seen": 23424120,
      "step": 35820
    },
    {
      "epoch": 18.77620545073375,
      "grad_norm": 0.174139603972435,
      "learning_rate": 5.685917251846501e-07,
      "loss": 0.5631,
      "num_input_tokens_seen": 23426584,
      "step": 35825
    },
    {
      "epoch": 18.77882599580713,
      "grad_norm": 0.2852259576320648,
      "learning_rate": 5.661695094790892e-07,
      "loss": 0.469,
      "num_input_tokens_seen": 23429880,
      "step": 35830
    },
    {
      "epoch": 18.781446540880502,
      "grad_norm": 0.1839556097984314,
      "learning_rate": 5.637524050521064e-07,
      "loss": 0.4246,
      "num_input_tokens_seen": 23432728,
      "step": 35835
    },
    {
      "epoch": 18.78406708595388,
      "grad_norm": 0.17641349136829376,
      "learning_rate": 5.613404124093391e-07,
      "loss": 0.5554,
      "num_input_tokens_seen": 23435704,
      "step": 35840
    },
    {
      "epoch": 18.786687631027252,
      "grad_norm": 0.29268595576286316,
      "learning_rate": 5.589335320553446e-07,
      "loss": 0.3991,
      "num_input_tokens_seen": 23439416,
      "step": 35845
    },
    {
      "epoch": 18.78930817610063,
      "grad_norm": 0.19756658375263214,
      "learning_rate": 5.565317644936202e-07,
      "loss": 0.4752,
      "num_input_tokens_seen": 23441880,
      "step": 35850
    },
    {
      "epoch": 18.791928721174003,
      "grad_norm": 0.18285080790519714,
      "learning_rate": 5.541351102265863e-07,
      "loss": 0.5058,
      "num_input_tokens_seen": 23445560,
      "step": 35855
    },
    {
      "epoch": 18.79454926624738,
      "grad_norm": 0.1700986921787262,
      "learning_rate": 5.51743569755589e-07,
      "loss": 0.5323,
      "num_input_tokens_seen": 23448760,
      "step": 35860
    },
    {
      "epoch": 18.797169811320753,
      "grad_norm": 0.18816134333610535,
      "learning_rate": 5.493571435809197e-07,
      "loss": 0.3166,
      "num_input_tokens_seen": 23451864,
      "step": 35865
    },
    {
      "epoch": 18.79979035639413,
      "grad_norm": 0.20279794931411743,
      "learning_rate": 5.469758322017849e-07,
      "loss": 0.4283,
      "num_input_tokens_seen": 23458552,
      "step": 35870
    },
    {
      "epoch": 18.802410901467507,
      "grad_norm": 0.1281176209449768,
      "learning_rate": 5.445996361163358e-07,
      "loss": 0.4125,
      "num_input_tokens_seen": 23461656,
      "step": 35875
    },
    {
      "epoch": 18.80503144654088,
      "grad_norm": 0.09314396977424622,
      "learning_rate": 5.422285558216389e-07,
      "loss": 0.517,
      "num_input_tokens_seen": 23464664,
      "step": 35880
    },
    {
      "epoch": 18.807651991614257,
      "grad_norm": 0.16662749648094177,
      "learning_rate": 5.398625918137001e-07,
      "loss": 0.5183,
      "num_input_tokens_seen": 23468216,
      "step": 35885
    },
    {
      "epoch": 18.81027253668763,
      "grad_norm": 0.1721613109111786,
      "learning_rate": 5.375017445874486e-07,
      "loss": 0.5345,
      "num_input_tokens_seen": 23471064,
      "step": 35890
    },
    {
      "epoch": 18.812893081761008,
      "grad_norm": 0.12494269013404846,
      "learning_rate": 5.351460146367504e-07,
      "loss": 0.3837,
      "num_input_tokens_seen": 23474264,
      "step": 35895
    },
    {
      "epoch": 18.81551362683438,
      "grad_norm": 0.3949890434741974,
      "learning_rate": 5.327954024543946e-07,
      "loss": 0.4074,
      "num_input_tokens_seen": 23477720,
      "step": 35900
    },
    {
      "epoch": 18.818134171907758,
      "grad_norm": 0.14564961194992065,
      "learning_rate": 5.304499085321074e-07,
      "loss": 0.3776,
      "num_input_tokens_seen": 23480344,
      "step": 35905
    },
    {
      "epoch": 18.82075471698113,
      "grad_norm": 0.17036713659763336,
      "learning_rate": 5.281095333605352e-07,
      "loss": 0.4835,
      "num_input_tokens_seen": 23483544,
      "step": 35910
    },
    {
      "epoch": 18.82337526205451,
      "grad_norm": 0.13715708255767822,
      "learning_rate": 5.257742774292556e-07,
      "loss": 0.5053,
      "num_input_tokens_seen": 23486392,
      "step": 35915
    },
    {
      "epoch": 18.82599580712788,
      "grad_norm": 0.1428823322057724,
      "learning_rate": 5.234441412267837e-07,
      "loss": 0.6596,
      "num_input_tokens_seen": 23490200,
      "step": 35920
    },
    {
      "epoch": 18.82861635220126,
      "grad_norm": 0.09972639381885529,
      "learning_rate": 5.211191252405517e-07,
      "loss": 0.4634,
      "num_input_tokens_seen": 23493048,
      "step": 35925
    },
    {
      "epoch": 18.831236897274632,
      "grad_norm": 0.15498675405979156,
      "learning_rate": 5.187992299569316e-07,
      "loss": 0.6183,
      "num_input_tokens_seen": 23496440,
      "step": 35930
    },
    {
      "epoch": 18.83385744234801,
      "grad_norm": 0.12853486835956573,
      "learning_rate": 5.164844558612131e-07,
      "loss": 0.4305,
      "num_input_tokens_seen": 23499832,
      "step": 35935
    },
    {
      "epoch": 18.836477987421382,
      "grad_norm": 0.16197335720062256,
      "learning_rate": 5.141748034376254e-07,
      "loss": 0.5607,
      "num_input_tokens_seen": 23503416,
      "step": 35940
    },
    {
      "epoch": 18.83909853249476,
      "grad_norm": 0.23181447386741638,
      "learning_rate": 5.118702731693209e-07,
      "loss": 0.377,
      "num_input_tokens_seen": 23506616,
      "step": 35945
    },
    {
      "epoch": 18.841719077568133,
      "grad_norm": 0.1737394481897354,
      "learning_rate": 5.095708655383751e-07,
      "loss": 0.5968,
      "num_input_tokens_seen": 23509912,
      "step": 35950
    },
    {
      "epoch": 18.84433962264151,
      "grad_norm": 0.12031903862953186,
      "learning_rate": 5.07276581025809e-07,
      "loss": 0.4354,
      "num_input_tokens_seen": 23513272,
      "step": 35955
    },
    {
      "epoch": 18.846960167714883,
      "grad_norm": 0.16716930270195007,
      "learning_rate": 5.049874201115495e-07,
      "loss": 0.4642,
      "num_input_tokens_seen": 23515992,
      "step": 35960
    },
    {
      "epoch": 18.84958071278826,
      "grad_norm": 0.18482309579849243,
      "learning_rate": 5.027033832744693e-07,
      "loss": 0.4395,
      "num_input_tokens_seen": 23519032,
      "step": 35965
    },
    {
      "epoch": 18.852201257861637,
      "grad_norm": 0.12916916608810425,
      "learning_rate": 5.004244709923639e-07,
      "loss": 0.4222,
      "num_input_tokens_seen": 23522392,
      "step": 35970
    },
    {
      "epoch": 18.85482180293501,
      "grad_norm": 0.16396461427211761,
      "learning_rate": 4.981506837419519e-07,
      "loss": 0.3732,
      "num_input_tokens_seen": 23525240,
      "step": 35975
    },
    {
      "epoch": 18.857442348008387,
      "grad_norm": 0.1761413961648941,
      "learning_rate": 4.958820219988836e-07,
      "loss": 0.537,
      "num_input_tokens_seen": 23528568,
      "step": 35980
    },
    {
      "epoch": 18.86006289308176,
      "grad_norm": 0.19826509058475494,
      "learning_rate": 4.936184862377402e-07,
      "loss": 0.4513,
      "num_input_tokens_seen": 23531288,
      "step": 35985
    },
    {
      "epoch": 18.862683438155138,
      "grad_norm": 0.13058944046497345,
      "learning_rate": 4.913600769320237e-07,
      "loss": 0.4601,
      "num_input_tokens_seen": 23534520,
      "step": 35990
    },
    {
      "epoch": 18.86530398322851,
      "grad_norm": 0.16971686482429504,
      "learning_rate": 4.891067945541728e-07,
      "loss": 0.537,
      "num_input_tokens_seen": 23537880,
      "step": 35995
    },
    {
      "epoch": 18.867924528301888,
      "grad_norm": 0.10607772320508957,
      "learning_rate": 4.868586395755437e-07,
      "loss": 0.4532,
      "num_input_tokens_seen": 23541464,
      "step": 36000
    },
    {
      "epoch": 18.87054507337526,
      "grad_norm": 0.2516566812992096,
      "learning_rate": 4.846156124664242e-07,
      "loss": 0.3149,
      "num_input_tokens_seen": 23544760,
      "step": 36005
    },
    {
      "epoch": 18.87316561844864,
      "grad_norm": 0.14547684788703918,
      "learning_rate": 4.823777136960362e-07,
      "loss": 0.4612,
      "num_input_tokens_seen": 23547832,
      "step": 36010
    },
    {
      "epoch": 18.87578616352201,
      "grad_norm": 0.1719508022069931,
      "learning_rate": 4.801449437325162e-07,
      "loss": 0.3964,
      "num_input_tokens_seen": 23550968,
      "step": 36015
    },
    {
      "epoch": 18.87840670859539,
      "grad_norm": 0.14328333735466003,
      "learning_rate": 4.779173030429379e-07,
      "loss": 0.3968,
      "num_input_tokens_seen": 23554680,
      "step": 36020
    },
    {
      "epoch": 18.881027253668762,
      "grad_norm": 0.10778255015611649,
      "learning_rate": 4.7569479209329804e-07,
      "loss": 0.452,
      "num_input_tokens_seen": 23558168,
      "step": 36025
    },
    {
      "epoch": 18.88364779874214,
      "grad_norm": 0.1767611801624298,
      "learning_rate": 4.734774113485163e-07,
      "loss": 0.4628,
      "num_input_tokens_seen": 23560952,
      "step": 36030
    },
    {
      "epoch": 18.886268343815512,
      "grad_norm": 0.13870686292648315,
      "learning_rate": 4.7126516127244945e-07,
      "loss": 0.431,
      "num_input_tokens_seen": 23563096,
      "step": 36035
    },
    {
      "epoch": 18.88888888888889,
      "grad_norm": 0.13720671832561493,
      "learning_rate": 4.690580423278718e-07,
      "loss": 0.578,
      "num_input_tokens_seen": 23565400,
      "step": 36040
    },
    {
      "epoch": 18.891509433962263,
      "grad_norm": 0.18145547807216644,
      "learning_rate": 4.668560549764861e-07,
      "loss": 0.4342,
      "num_input_tokens_seen": 23569016,
      "step": 36045
    },
    {
      "epoch": 18.89412997903564,
      "grad_norm": 0.1673235148191452,
      "learning_rate": 4.646591996789268e-07,
      "loss": 0.545,
      "num_input_tokens_seen": 23572216,
      "step": 36050
    },
    {
      "epoch": 18.896750524109013,
      "grad_norm": 0.17800317704677582,
      "learning_rate": 4.6246747689474847e-07,
      "loss": 0.5421,
      "num_input_tokens_seen": 23576056,
      "step": 36055
    },
    {
      "epoch": 18.89937106918239,
      "grad_norm": 0.08691395074129105,
      "learning_rate": 4.6028088708243434e-07,
      "loss": 0.607,
      "num_input_tokens_seen": 23579096,
      "step": 36060
    },
    {
      "epoch": 18.901991614255767,
      "grad_norm": 0.2512148916721344,
      "learning_rate": 4.580994306993991e-07,
      "loss": 0.4508,
      "num_input_tokens_seen": 23581720,
      "step": 36065
    },
    {
      "epoch": 18.90461215932914,
      "grad_norm": 0.2628519833087921,
      "learning_rate": 4.5592310820196936e-07,
      "loss": 0.7108,
      "num_input_tokens_seen": 23584824,
      "step": 36070
    },
    {
      "epoch": 18.907232704402517,
      "grad_norm": 0.2798810303211212,
      "learning_rate": 4.537519200454116e-07,
      "loss": 0.4651,
      "num_input_tokens_seen": 23587800,
      "step": 36075
    },
    {
      "epoch": 18.90985324947589,
      "grad_norm": 0.15980936586856842,
      "learning_rate": 4.515858666839151e-07,
      "loss": 0.6023,
      "num_input_tokens_seen": 23590584,
      "step": 36080
    },
    {
      "epoch": 18.912473794549268,
      "grad_norm": 0.1507088840007782,
      "learning_rate": 4.494249485705926e-07,
      "loss": 0.644,
      "num_input_tokens_seen": 23594392,
      "step": 36085
    },
    {
      "epoch": 18.91509433962264,
      "grad_norm": 0.1412121206521988,
      "learning_rate": 4.472691661574824e-07,
      "loss": 0.4494,
      "num_input_tokens_seen": 23597176,
      "step": 36090
    },
    {
      "epoch": 18.917714884696018,
      "grad_norm": 0.1510694921016693,
      "learning_rate": 4.4511851989555156e-07,
      "loss": 0.4742,
      "num_input_tokens_seen": 23600600,
      "step": 36095
    },
    {
      "epoch": 18.92033542976939,
      "grad_norm": 0.1619713306427002,
      "learning_rate": 4.429730102346874e-07,
      "loss": 0.5889,
      "num_input_tokens_seen": 23603480,
      "step": 36100
    },
    {
      "epoch": 18.92295597484277,
      "grad_norm": 0.1655864417552948,
      "learning_rate": 4.4083263762370873e-07,
      "loss": 0.3745,
      "num_input_tokens_seen": 23605784,
      "step": 36105
    },
    {
      "epoch": 18.92557651991614,
      "grad_norm": 0.14060933887958527,
      "learning_rate": 4.386974025103574e-07,
      "loss": 0.4637,
      "num_input_tokens_seen": 23609144,
      "step": 36110
    },
    {
      "epoch": 18.92819706498952,
      "grad_norm": 0.1949763149023056,
      "learning_rate": 4.365673053413011e-07,
      "loss": 0.5504,
      "num_input_tokens_seen": 23611960,
      "step": 36115
    },
    {
      "epoch": 18.930817610062892,
      "grad_norm": 0.16029268503189087,
      "learning_rate": 4.34442346562125e-07,
      "loss": 0.4779,
      "num_input_tokens_seen": 23614680,
      "step": 36120
    },
    {
      "epoch": 18.93343815513627,
      "grad_norm": 0.1955728828907013,
      "learning_rate": 4.323225266173542e-07,
      "loss": 0.462,
      "num_input_tokens_seen": 23617592,
      "step": 36125
    },
    {
      "epoch": 18.936058700209642,
      "grad_norm": 0.3572552502155304,
      "learning_rate": 4.302078459504283e-07,
      "loss": 0.4644,
      "num_input_tokens_seen": 23620440,
      "step": 36130
    },
    {
      "epoch": 18.93867924528302,
      "grad_norm": 0.12400087714195251,
      "learning_rate": 4.2809830500371294e-07,
      "loss": 0.36,
      "num_input_tokens_seen": 23623320,
      "step": 36135
    },
    {
      "epoch": 18.941299790356393,
      "grad_norm": 0.2684534788131714,
      "learning_rate": 4.2599390421850236e-07,
      "loss": 0.3725,
      "num_input_tokens_seen": 23626168,
      "step": 36140
    },
    {
      "epoch": 18.94392033542977,
      "grad_norm": 0.25590744614601135,
      "learning_rate": 4.2389464403501113e-07,
      "loss": 0.5861,
      "num_input_tokens_seen": 23629240,
      "step": 36145
    },
    {
      "epoch": 18.946540880503143,
      "grad_norm": 0.17150406539440155,
      "learning_rate": 4.218005248923823e-07,
      "loss": 0.4324,
      "num_input_tokens_seen": 23632472,
      "step": 36150
    },
    {
      "epoch": 18.94916142557652,
      "grad_norm": 0.1211383193731308,
      "learning_rate": 4.1971154722867946e-07,
      "loss": 0.5251,
      "num_input_tokens_seen": 23635864,
      "step": 36155
    },
    {
      "epoch": 18.951781970649897,
      "grad_norm": 0.1665397584438324,
      "learning_rate": 4.176277114808946e-07,
      "loss": 0.44,
      "num_input_tokens_seen": 23638712,
      "step": 36160
    },
    {
      "epoch": 18.95440251572327,
      "grad_norm": 0.20717324316501617,
      "learning_rate": 4.155490180849431e-07,
      "loss": 0.4706,
      "num_input_tokens_seen": 23641592,
      "step": 36165
    },
    {
      "epoch": 18.957023060796647,
      "grad_norm": 0.11202828586101532,
      "learning_rate": 4.1347546747566304e-07,
      "loss": 0.4337,
      "num_input_tokens_seen": 23645912,
      "step": 36170
    },
    {
      "epoch": 18.95964360587002,
      "grad_norm": 0.14074759185314178,
      "learning_rate": 4.114070600868131e-07,
      "loss": 0.4615,
      "num_input_tokens_seen": 23648824,
      "step": 36175
    },
    {
      "epoch": 18.962264150943398,
      "grad_norm": 0.23598936200141907,
      "learning_rate": 4.093437963510888e-07,
      "loss": 0.4698,
      "num_input_tokens_seen": 23651096,
      "step": 36180
    },
    {
      "epoch": 18.96488469601677,
      "grad_norm": 0.19923001527786255,
      "learning_rate": 4.072856767000949e-07,
      "loss": 0.4655,
      "num_input_tokens_seen": 23654616,
      "step": 36185
    },
    {
      "epoch": 18.967505241090148,
      "grad_norm": 0.13760891556739807,
      "learning_rate": 4.052327015643703e-07,
      "loss": 0.5283,
      "num_input_tokens_seen": 23658200,
      "step": 36190
    },
    {
      "epoch": 18.97012578616352,
      "grad_norm": 0.25116637349128723,
      "learning_rate": 4.031848713733744e-07,
      "loss": 0.5089,
      "num_input_tokens_seen": 23660856,
      "step": 36195
    },
    {
      "epoch": 18.9727463312369,
      "grad_norm": 0.11905413120985031,
      "learning_rate": 4.011421865554837e-07,
      "loss": 0.4746,
      "num_input_tokens_seen": 23664056,
      "step": 36200
    },
    {
      "epoch": 18.97536687631027,
      "grad_norm": 0.12274079769849777,
      "learning_rate": 3.9910464753801224e-07,
      "loss": 0.5262,
      "num_input_tokens_seen": 23667256,
      "step": 36205
    },
    {
      "epoch": 18.97798742138365,
      "grad_norm": 0.21254274249076843,
      "learning_rate": 3.9707225474718557e-07,
      "loss": 0.5139,
      "num_input_tokens_seen": 23670456,
      "step": 36210
    },
    {
      "epoch": 18.980607966457022,
      "grad_norm": 0.461382657289505,
      "learning_rate": 3.950450086081553e-07,
      "loss": 0.4958,
      "num_input_tokens_seen": 23673656,
      "step": 36215
    },
    {
      "epoch": 18.9832285115304,
      "grad_norm": 0.15080313384532928,
      "learning_rate": 3.930229095450072e-07,
      "loss": 0.5408,
      "num_input_tokens_seen": 23676824,
      "step": 36220
    },
    {
      "epoch": 18.985849056603772,
      "grad_norm": 0.17587876319885254,
      "learning_rate": 3.9100595798073345e-07,
      "loss": 0.7257,
      "num_input_tokens_seen": 23679832,
      "step": 36225
    },
    {
      "epoch": 18.98846960167715,
      "grad_norm": 0.32975149154663086,
      "learning_rate": 3.8899415433725493e-07,
      "loss": 0.4604,
      "num_input_tokens_seen": 23682808,
      "step": 36230
    },
    {
      "epoch": 18.991090146750523,
      "grad_norm": 0.17999133467674255,
      "learning_rate": 3.8698749903542666e-07,
      "loss": 0.584,
      "num_input_tokens_seen": 23685976,
      "step": 36235
    },
    {
      "epoch": 18.9937106918239,
      "grad_norm": 0.23421718180179596,
      "learning_rate": 3.8498599249501e-07,
      "loss": 0.5389,
      "num_input_tokens_seen": 23688952,
      "step": 36240
    },
    {
      "epoch": 18.996331236897273,
      "grad_norm": 0.36124294996261597,
      "learning_rate": 3.8298963513470345e-07,
      "loss": 0.4875,
      "num_input_tokens_seen": 23691832,
      "step": 36245
    },
    {
      "epoch": 18.99895178197065,
      "grad_norm": 0.11574102938175201,
      "learning_rate": 3.8099842737211734e-07,
      "loss": 0.4151,
      "num_input_tokens_seen": 23695544,
      "step": 36250
    },
    {
      "epoch": 19.0,
      "eval_loss": 0.4916197657585144,
      "eval_runtime": 10.1554,
      "eval_samples_per_second": 83.502,
      "eval_steps_per_second": 20.876,
      "num_input_tokens_seen": 23696296,
      "step": 36252
    },
    {
      "epoch": 19.001572327044027,
      "grad_norm": 0.14248766005039215,
      "learning_rate": 3.7901236962379627e-07,
      "loss": 0.5434,
      "num_input_tokens_seen": 23697768,
      "step": 36255
    },
    {
      "epoch": 19.0041928721174,
      "grad_norm": 0.17610225081443787,
      "learning_rate": 3.7703146230519126e-07,
      "loss": 0.4666,
      "num_input_tokens_seen": 23700072,
      "step": 36260
    },
    {
      "epoch": 19.006813417190777,
      "grad_norm": 0.17321327328681946,
      "learning_rate": 3.7505570583069306e-07,
      "loss": 0.5916,
      "num_input_tokens_seen": 23702856,
      "step": 36265
    },
    {
      "epoch": 19.00943396226415,
      "grad_norm": 0.23178760707378387,
      "learning_rate": 3.7308510061360435e-07,
      "loss": 0.5192,
      "num_input_tokens_seen": 23706184,
      "step": 36270
    },
    {
      "epoch": 19.012054507337528,
      "grad_norm": 0.18483880162239075,
      "learning_rate": 3.7111964706615374e-07,
      "loss": 0.5139,
      "num_input_tokens_seen": 23709128,
      "step": 36275
    },
    {
      "epoch": 19.0146750524109,
      "grad_norm": 0.20671696960926056,
      "learning_rate": 3.691593455994929e-07,
      "loss": 0.6513,
      "num_input_tokens_seen": 23712808,
      "step": 36280
    },
    {
      "epoch": 19.017295597484278,
      "grad_norm": 0.12258938699960709,
      "learning_rate": 3.672041966236911e-07,
      "loss": 0.4497,
      "num_input_tokens_seen": 23716168,
      "step": 36285
    },
    {
      "epoch": 19.01991614255765,
      "grad_norm": 0.1272817850112915,
      "learning_rate": 3.6525420054774884e-07,
      "loss": 0.4027,
      "num_input_tokens_seen": 23719688,
      "step": 36290
    },
    {
      "epoch": 19.02253668763103,
      "grad_norm": 0.10527720302343369,
      "learning_rate": 3.6330935777957597e-07,
      "loss": 0.4549,
      "num_input_tokens_seen": 23723752,
      "step": 36295
    },
    {
      "epoch": 19.0251572327044,
      "grad_norm": 0.15074430406093597,
      "learning_rate": 3.613696687260193e-07,
      "loss": 0.3835,
      "num_input_tokens_seen": 23727432,
      "step": 36300
    },
    {
      "epoch": 19.02777777777778,
      "grad_norm": 0.18723906576633453,
      "learning_rate": 3.59435133792832e-07,
      "loss": 0.4875,
      "num_input_tokens_seen": 23730312,
      "step": 36305
    },
    {
      "epoch": 19.030398322851152,
      "grad_norm": 0.21061773598194122,
      "learning_rate": 3.575057533847043e-07,
      "loss": 0.5221,
      "num_input_tokens_seen": 23733448,
      "step": 36310
    },
    {
      "epoch": 19.03301886792453,
      "grad_norm": 0.1435963660478592,
      "learning_rate": 3.555815279052327e-07,
      "loss": 0.479,
      "num_input_tokens_seen": 23737320,
      "step": 36315
    },
    {
      "epoch": 19.035639412997902,
      "grad_norm": 0.13524532318115234,
      "learning_rate": 3.5366245775694805e-07,
      "loss": 0.5568,
      "num_input_tokens_seen": 23739752,
      "step": 36320
    },
    {
      "epoch": 19.03825995807128,
      "grad_norm": 0.2060118019580841,
      "learning_rate": 3.517485433412987e-07,
      "loss": 0.3924,
      "num_input_tokens_seen": 23742152,
      "step": 36325
    },
    {
      "epoch": 19.040880503144653,
      "grad_norm": 0.12425532937049866,
      "learning_rate": 3.4983978505865044e-07,
      "loss": 0.4335,
      "num_input_tokens_seen": 23745448,
      "step": 36330
    },
    {
      "epoch": 19.04350104821803,
      "grad_norm": 0.17450478672981262,
      "learning_rate": 3.47936183308295e-07,
      "loss": 0.4357,
      "num_input_tokens_seen": 23749864,
      "step": 36335
    },
    {
      "epoch": 19.046121593291403,
      "grad_norm": 0.14789612591266632,
      "learning_rate": 3.460377384884472e-07,
      "loss": 0.511,
      "num_input_tokens_seen": 23752552,
      "step": 36340
    },
    {
      "epoch": 19.04874213836478,
      "grad_norm": 0.21620070934295654,
      "learning_rate": 3.4414445099623384e-07,
      "loss": 0.3934,
      "num_input_tokens_seen": 23755528,
      "step": 36345
    },
    {
      "epoch": 19.051362683438157,
      "grad_norm": 0.20135609805583954,
      "learning_rate": 3.42256321227713e-07,
      "loss": 0.3861,
      "num_input_tokens_seen": 23758600,
      "step": 36350
    },
    {
      "epoch": 19.05398322851153,
      "grad_norm": 0.28290873765945435,
      "learning_rate": 3.403733495778605e-07,
      "loss": 0.464,
      "num_input_tokens_seen": 23761480,
      "step": 36355
    },
    {
      "epoch": 19.056603773584907,
      "grad_norm": 0.2739580571651459,
      "learning_rate": 3.3849553644056955e-07,
      "loss": 0.44,
      "num_input_tokens_seen": 23764168,
      "step": 36360
    },
    {
      "epoch": 19.05922431865828,
      "grad_norm": 0.2807627022266388,
      "learning_rate": 3.36622882208662e-07,
      "loss": 0.5612,
      "num_input_tokens_seen": 23766920,
      "step": 36365
    },
    {
      "epoch": 19.061844863731658,
      "grad_norm": 0.2584194540977478,
      "learning_rate": 3.347553872738718e-07,
      "loss": 0.497,
      "num_input_tokens_seen": 23770056,
      "step": 36370
    },
    {
      "epoch": 19.06446540880503,
      "grad_norm": 0.1298241764307022,
      "learning_rate": 3.3289305202685583e-07,
      "loss": 0.3496,
      "num_input_tokens_seen": 23773096,
      "step": 36375
    },
    {
      "epoch": 19.067085953878408,
      "grad_norm": 0.289257675409317,
      "learning_rate": 3.310358768571997e-07,
      "loss": 0.4386,
      "num_input_tokens_seen": 23775496,
      "step": 36380
    },
    {
      "epoch": 19.06970649895178,
      "grad_norm": 0.22906085848808289,
      "learning_rate": 3.2918386215339827e-07,
      "loss": 0.4879,
      "num_input_tokens_seen": 23778600,
      "step": 36385
    },
    {
      "epoch": 19.072327044025158,
      "grad_norm": 0.04543339088559151,
      "learning_rate": 3.2733700830287205e-07,
      "loss": 0.4739,
      "num_input_tokens_seen": 23783688,
      "step": 36390
    },
    {
      "epoch": 19.07494758909853,
      "grad_norm": 0.19927982985973358,
      "learning_rate": 3.254953156919649e-07,
      "loss": 0.5519,
      "num_input_tokens_seen": 23786440,
      "step": 36395
    },
    {
      "epoch": 19.07756813417191,
      "grad_norm": 0.1812404841184616,
      "learning_rate": 3.2365878470593527e-07,
      "loss": 0.5366,
      "num_input_tokens_seen": 23789768,
      "step": 36400
    },
    {
      "epoch": 19.080188679245282,
      "grad_norm": 0.1889735609292984,
      "learning_rate": 3.2182741572896747e-07,
      "loss": 0.5162,
      "num_input_tokens_seen": 23792264,
      "step": 36405
    },
    {
      "epoch": 19.08280922431866,
      "grad_norm": 0.15834641456604004,
      "learning_rate": 3.2000120914415786e-07,
      "loss": 0.4485,
      "num_input_tokens_seen": 23795816,
      "step": 36410
    },
    {
      "epoch": 19.085429769392032,
      "grad_norm": 0.1542017012834549,
      "learning_rate": 3.181801653335315e-07,
      "loss": 0.5187,
      "num_input_tokens_seen": 23798984,
      "step": 36415
    },
    {
      "epoch": 19.08805031446541,
      "grad_norm": 0.11863952875137329,
      "learning_rate": 3.163642846780307e-07,
      "loss": 0.468,
      "num_input_tokens_seen": 23802056,
      "step": 36420
    },
    {
      "epoch": 19.090670859538783,
      "grad_norm": 0.1996084749698639,
      "learning_rate": 3.1455356755751565e-07,
      "loss": 0.4361,
      "num_input_tokens_seen": 23804424,
      "step": 36425
    },
    {
      "epoch": 19.09329140461216,
      "grad_norm": 0.13490177690982819,
      "learning_rate": 3.127480143507666e-07,
      "loss": 0.4118,
      "num_input_tokens_seen": 23807112,
      "step": 36430
    },
    {
      "epoch": 19.095911949685533,
      "grad_norm": 0.1482820063829422,
      "learning_rate": 3.1094762543548704e-07,
      "loss": 0.4212,
      "num_input_tokens_seen": 23810184,
      "step": 36435
    },
    {
      "epoch": 19.09853249475891,
      "grad_norm": 0.1278182715177536,
      "learning_rate": 3.091524011882951e-07,
      "loss": 0.4393,
      "num_input_tokens_seen": 23812776,
      "step": 36440
    },
    {
      "epoch": 19.101153039832287,
      "grad_norm": 0.15828105807304382,
      "learning_rate": 3.073623419847349e-07,
      "loss": 0.564,
      "num_input_tokens_seen": 23816296,
      "step": 36445
    },
    {
      "epoch": 19.10377358490566,
      "grad_norm": 0.14489337801933289,
      "learning_rate": 3.0557744819926235e-07,
      "loss": 0.5373,
      "num_input_tokens_seen": 23819432,
      "step": 36450
    },
    {
      "epoch": 19.106394129979037,
      "grad_norm": 0.2336847484111786,
      "learning_rate": 3.037977202052594e-07,
      "loss": 0.5488,
      "num_input_tokens_seen": 23822600,
      "step": 36455
    },
    {
      "epoch": 19.10901467505241,
      "grad_norm": 0.1496361941099167,
      "learning_rate": 3.020231583750255e-07,
      "loss": 0.4605,
      "num_input_tokens_seen": 23825512,
      "step": 36460
    },
    {
      "epoch": 19.111635220125788,
      "grad_norm": 0.13052037358283997,
      "learning_rate": 3.002537630797747e-07,
      "loss": 0.4901,
      "num_input_tokens_seen": 23828456,
      "step": 36465
    },
    {
      "epoch": 19.11425576519916,
      "grad_norm": 0.11558560281991959,
      "learning_rate": 2.984895346896499e-07,
      "loss": 0.3901,
      "num_input_tokens_seen": 23831656,
      "step": 36470
    },
    {
      "epoch": 19.116876310272538,
      "grad_norm": 0.24400383234024048,
      "learning_rate": 2.967304735737031e-07,
      "loss": 0.4377,
      "num_input_tokens_seen": 23834312,
      "step": 36475
    },
    {
      "epoch": 19.11949685534591,
      "grad_norm": 0.19291667640209198,
      "learning_rate": 2.9497658009991214e-07,
      "loss": 0.4874,
      "num_input_tokens_seen": 23836968,
      "step": 36480
    },
    {
      "epoch": 19.122117400419288,
      "grad_norm": 0.21200694143772125,
      "learning_rate": 2.932278546351724e-07,
      "loss": 0.4648,
      "num_input_tokens_seen": 23840616,
      "step": 36485
    },
    {
      "epoch": 19.12473794549266,
      "grad_norm": 0.19892965257167816,
      "learning_rate": 2.9148429754529137e-07,
      "loss": 0.444,
      "num_input_tokens_seen": 23844168,
      "step": 36490
    },
    {
      "epoch": 19.12735849056604,
      "grad_norm": 0.15533596277236938,
      "learning_rate": 2.897459091950078e-07,
      "loss": 0.4052,
      "num_input_tokens_seen": 23846888,
      "step": 36495
    },
    {
      "epoch": 19.129979035639412,
      "grad_norm": 0.11440496891736984,
      "learning_rate": 2.880126899479724e-07,
      "loss": 0.5239,
      "num_input_tokens_seen": 23850856,
      "step": 36500
    },
    {
      "epoch": 19.13259958071279,
      "grad_norm": 0.17100778222084045,
      "learning_rate": 2.862846401667507e-07,
      "loss": 0.4748,
      "num_input_tokens_seen": 23853992,
      "step": 36505
    },
    {
      "epoch": 19.135220125786162,
      "grad_norm": 0.20359471440315247,
      "learning_rate": 2.8456176021283697e-07,
      "loss": 0.4695,
      "num_input_tokens_seen": 23857864,
      "step": 36510
    },
    {
      "epoch": 19.13784067085954,
      "grad_norm": 0.16065964102745056,
      "learning_rate": 2.8284405044662897e-07,
      "loss": 0.4857,
      "num_input_tokens_seen": 23861480,
      "step": 36515
    },
    {
      "epoch": 19.140461215932913,
      "grad_norm": 0.18495014309883118,
      "learning_rate": 2.8113151122745874e-07,
      "loss": 0.4484,
      "num_input_tokens_seen": 23864424,
      "step": 36520
    },
    {
      "epoch": 19.14308176100629,
      "grad_norm": 0.23466187715530396,
      "learning_rate": 2.7942414291357023e-07,
      "loss": 0.5878,
      "num_input_tokens_seen": 23867272,
      "step": 36525
    },
    {
      "epoch": 19.145702306079663,
      "grad_norm": 0.19996339082717896,
      "learning_rate": 2.777219458621222e-07,
      "loss": 0.409,
      "num_input_tokens_seen": 23871272,
      "step": 36530
    },
    {
      "epoch": 19.14832285115304,
      "grad_norm": 0.2846128046512604,
      "learning_rate": 2.7602492042919923e-07,
      "loss": 0.4425,
      "num_input_tokens_seen": 23874376,
      "step": 36535
    },
    {
      "epoch": 19.150943396226417,
      "grad_norm": 0.1590155065059662,
      "learning_rate": 2.7433306696979523e-07,
      "loss": 0.402,
      "num_input_tokens_seen": 23877096,
      "step": 36540
    },
    {
      "epoch": 19.15356394129979,
      "grad_norm": 0.21327245235443115,
      "learning_rate": 2.7264638583782973e-07,
      "loss": 0.4004,
      "num_input_tokens_seen": 23880776,
      "step": 36545
    },
    {
      "epoch": 19.156184486373167,
      "grad_norm": 0.09799119085073471,
      "learning_rate": 2.7096487738613177e-07,
      "loss": 0.341,
      "num_input_tokens_seen": 23885000,
      "step": 36550
    },
    {
      "epoch": 19.15880503144654,
      "grad_norm": 0.14439363777637482,
      "learning_rate": 2.6928854196646147e-07,
      "loss": 0.3942,
      "num_input_tokens_seen": 23887720,
      "step": 36555
    },
    {
      "epoch": 19.161425576519918,
      "grad_norm": 0.13623474538326263,
      "learning_rate": 2.6761737992948565e-07,
      "loss": 0.4316,
      "num_input_tokens_seen": 23890760,
      "step": 36560
    },
    {
      "epoch": 19.16404612159329,
      "grad_norm": 0.11927511543035507,
      "learning_rate": 2.6595139162479134e-07,
      "loss": 0.4309,
      "num_input_tokens_seen": 23893640,
      "step": 36565
    },
    {
      "epoch": 19.166666666666668,
      "grad_norm": 0.12075139582157135,
      "learning_rate": 2.6429057740088303e-07,
      "loss": 0.4322,
      "num_input_tokens_seen": 23896584,
      "step": 36570
    },
    {
      "epoch": 19.16928721174004,
      "grad_norm": 0.12288840115070343,
      "learning_rate": 2.6263493760518564e-07,
      "loss": 0.5423,
      "num_input_tokens_seen": 23901480,
      "step": 36575
    },
    {
      "epoch": 19.171907756813418,
      "grad_norm": 0.15259939432144165,
      "learning_rate": 2.6098447258404157e-07,
      "loss": 0.5042,
      "num_input_tokens_seen": 23904680,
      "step": 36580
    },
    {
      "epoch": 19.17452830188679,
      "grad_norm": 0.2804325222969055,
      "learning_rate": 2.593391826827107e-07,
      "loss": 0.3411,
      "num_input_tokens_seen": 23907144,
      "step": 36585
    },
    {
      "epoch": 19.17714884696017,
      "grad_norm": 0.15077383816242218,
      "learning_rate": 2.5769906824536225e-07,
      "loss": 0.4682,
      "num_input_tokens_seen": 23910888,
      "step": 36590
    },
    {
      "epoch": 19.179769392033542,
      "grad_norm": 0.22189244627952576,
      "learning_rate": 2.560641296150967e-07,
      "loss": 0.6199,
      "num_input_tokens_seen": 23914184,
      "step": 36595
    },
    {
      "epoch": 19.18238993710692,
      "grad_norm": 0.11906933784484863,
      "learning_rate": 2.544343671339211e-07,
      "loss": 0.4237,
      "num_input_tokens_seen": 23917608,
      "step": 36600
    },
    {
      "epoch": 19.185010482180292,
      "grad_norm": 0.1820777803659439,
      "learning_rate": 2.5280978114276264e-07,
      "loss": 0.4956,
      "num_input_tokens_seen": 23922312,
      "step": 36605
    },
    {
      "epoch": 19.18763102725367,
      "grad_norm": 0.1848500818014145,
      "learning_rate": 2.511903719814662e-07,
      "loss": 0.532,
      "num_input_tokens_seen": 23925096,
      "step": 36610
    },
    {
      "epoch": 19.190251572327043,
      "grad_norm": 0.1827874481678009,
      "learning_rate": 2.4957613998879695e-07,
      "loss": 0.4857,
      "num_input_tokens_seen": 23927976,
      "step": 36615
    },
    {
      "epoch": 19.19287211740042,
      "grad_norm": 0.19666388630867004,
      "learning_rate": 2.479670855024291e-07,
      "loss": 0.4551,
      "num_input_tokens_seen": 23934760,
      "step": 36620
    },
    {
      "epoch": 19.195492662473793,
      "grad_norm": 0.12105768173933029,
      "learning_rate": 2.463632088589629e-07,
      "loss": 0.5063,
      "num_input_tokens_seen": 23937960,
      "step": 36625
    },
    {
      "epoch": 19.19811320754717,
      "grad_norm": 0.15254251658916473,
      "learning_rate": 2.447645103939078e-07,
      "loss": 0.394,
      "num_input_tokens_seen": 23940904,
      "step": 36630
    },
    {
      "epoch": 19.200733752620547,
      "grad_norm": 0.15718185901641846,
      "learning_rate": 2.431709904416962e-07,
      "loss": 0.4336,
      "num_input_tokens_seen": 23943848,
      "step": 36635
    },
    {
      "epoch": 19.20335429769392,
      "grad_norm": 0.17095372080802917,
      "learning_rate": 2.4158264933567264e-07,
      "loss": 0.4739,
      "num_input_tokens_seen": 23947112,
      "step": 36640
    },
    {
      "epoch": 19.205974842767297,
      "grad_norm": 0.17918090522289276,
      "learning_rate": 2.399994874080963e-07,
      "loss": 0.3708,
      "num_input_tokens_seen": 23950056,
      "step": 36645
    },
    {
      "epoch": 19.20859538784067,
      "grad_norm": 0.2078704535961151,
      "learning_rate": 2.3842150499015504e-07,
      "loss": 0.5266,
      "num_input_tokens_seen": 23953320,
      "step": 36650
    },
    {
      "epoch": 19.211215932914047,
      "grad_norm": 0.19367091357707977,
      "learning_rate": 2.368487024119348e-07,
      "loss": 0.4856,
      "num_input_tokens_seen": 23956168,
      "step": 36655
    },
    {
      "epoch": 19.21383647798742,
      "grad_norm": 0.15044939517974854,
      "learning_rate": 2.3528108000245573e-07,
      "loss": 0.4502,
      "num_input_tokens_seen": 23959528,
      "step": 36660
    },
    {
      "epoch": 19.216457023060798,
      "grad_norm": 0.12515658140182495,
      "learning_rate": 2.3371863808964445e-07,
      "loss": 0.4917,
      "num_input_tokens_seen": 23962504,
      "step": 36665
    },
    {
      "epoch": 19.21907756813417,
      "grad_norm": 0.1256079524755478,
      "learning_rate": 2.32161377000345e-07,
      "loss": 0.4239,
      "num_input_tokens_seen": 23965736,
      "step": 36670
    },
    {
      "epoch": 19.221698113207548,
      "grad_norm": 0.16206932067871094,
      "learning_rate": 2.306092970603163e-07,
      "loss": 0.3992,
      "num_input_tokens_seen": 23968328,
      "step": 36675
    },
    {
      "epoch": 19.22431865828092,
      "grad_norm": 0.27076154947280884,
      "learning_rate": 2.2906239859424306e-07,
      "loss": 0.4192,
      "num_input_tokens_seen": 23972360,
      "step": 36680
    },
    {
      "epoch": 19.2269392033543,
      "grad_norm": 0.16972507536411285,
      "learning_rate": 2.2752068192571084e-07,
      "loss": 0.4793,
      "num_input_tokens_seen": 23975688,
      "step": 36685
    },
    {
      "epoch": 19.229559748427672,
      "grad_norm": 0.21051505208015442,
      "learning_rate": 2.2598414737723396e-07,
      "loss": 0.5279,
      "num_input_tokens_seen": 23979656,
      "step": 36690
    },
    {
      "epoch": 19.23218029350105,
      "grad_norm": 0.16888292133808136,
      "learning_rate": 2.2445279527023578e-07,
      "loss": 0.5054,
      "num_input_tokens_seen": 23982568,
      "step": 36695
    },
    {
      "epoch": 19.234800838574422,
      "grad_norm": 0.15352296829223633,
      "learning_rate": 2.2292662592506007e-07,
      "loss": 0.5355,
      "num_input_tokens_seen": 23985640,
      "step": 36700
    },
    {
      "epoch": 19.2374213836478,
      "grad_norm": 0.17411652207374573,
      "learning_rate": 2.2140563966095985e-07,
      "loss": 0.5151,
      "num_input_tokens_seen": 23989000,
      "step": 36705
    },
    {
      "epoch": 19.240041928721173,
      "grad_norm": 0.12041397392749786,
      "learning_rate": 2.1988983679611386e-07,
      "loss": 0.4803,
      "num_input_tokens_seen": 23991784,
      "step": 36710
    },
    {
      "epoch": 19.24266247379455,
      "grad_norm": 0.23983430862426758,
      "learning_rate": 2.1837921764760462e-07,
      "loss": 0.4539,
      "num_input_tokens_seen": 23996328,
      "step": 36715
    },
    {
      "epoch": 19.245283018867923,
      "grad_norm": 0.19940213859081268,
      "learning_rate": 2.1687378253144043e-07,
      "loss": 0.5597,
      "num_input_tokens_seen": 23999816,
      "step": 36720
    },
    {
      "epoch": 19.2479035639413,
      "grad_norm": 0.12053532898426056,
      "learning_rate": 2.153735317625416e-07,
      "loss": 0.5244,
      "num_input_tokens_seen": 24002472,
      "step": 36725
    },
    {
      "epoch": 19.250524109014677,
      "grad_norm": 0.1419299840927124,
      "learning_rate": 2.1387846565474045e-07,
      "loss": 0.333,
      "num_input_tokens_seen": 24005256,
      "step": 36730
    },
    {
      "epoch": 19.25314465408805,
      "grad_norm": 0.15421532094478607,
      "learning_rate": 2.123885845207896e-07,
      "loss": 0.4489,
      "num_input_tokens_seen": 24008776,
      "step": 36735
    },
    {
      "epoch": 19.255765199161427,
      "grad_norm": 0.13994421064853668,
      "learning_rate": 2.1090388867235356e-07,
      "loss": 0.358,
      "num_input_tokens_seen": 24012904,
      "step": 36740
    },
    {
      "epoch": 19.2583857442348,
      "grad_norm": 0.26077553629875183,
      "learning_rate": 2.0942437842001728e-07,
      "loss": 0.4307,
      "num_input_tokens_seen": 24016040,
      "step": 36745
    },
    {
      "epoch": 19.261006289308177,
      "grad_norm": 0.1564512401819229,
      "learning_rate": 2.0795005407327483e-07,
      "loss": 0.4825,
      "num_input_tokens_seen": 24018760,
      "step": 36750
    },
    {
      "epoch": 19.26362683438155,
      "grad_norm": 0.11045882105827332,
      "learning_rate": 2.0648091594053508e-07,
      "loss": 0.5221,
      "num_input_tokens_seen": 24022088,
      "step": 36755
    },
    {
      "epoch": 19.266247379454928,
      "grad_norm": 0.2697448134422302,
      "learning_rate": 2.0501696432913275e-07,
      "loss": 0.4623,
      "num_input_tokens_seen": 24025064,
      "step": 36760
    },
    {
      "epoch": 19.2688679245283,
      "grad_norm": 0.14740586280822754,
      "learning_rate": 2.035581995453034e-07,
      "loss": 0.4537,
      "num_input_tokens_seen": 24028264,
      "step": 36765
    },
    {
      "epoch": 19.271488469601678,
      "grad_norm": 0.12919996678829193,
      "learning_rate": 2.0210462189420577e-07,
      "loss": 0.5202,
      "num_input_tokens_seen": 24031560,
      "step": 36770
    },
    {
      "epoch": 19.27410901467505,
      "grad_norm": 0.09164828807115555,
      "learning_rate": 2.006562316799132e-07,
      "loss": 0.5218,
      "num_input_tokens_seen": 24035944,
      "step": 36775
    },
    {
      "epoch": 19.27672955974843,
      "grad_norm": 0.18859843909740448,
      "learning_rate": 1.992130292054084e-07,
      "loss": 0.6558,
      "num_input_tokens_seen": 24039912,
      "step": 36780
    },
    {
      "epoch": 19.279350104821802,
      "grad_norm": 0.14751626551151276,
      "learning_rate": 1.9777501477259708e-07,
      "loss": 0.4042,
      "num_input_tokens_seen": 24042792,
      "step": 36785
    },
    {
      "epoch": 19.28197064989518,
      "grad_norm": 0.1614667773246765,
      "learning_rate": 1.9634218868229414e-07,
      "loss": 0.472,
      "num_input_tokens_seen": 24046184,
      "step": 36790
    },
    {
      "epoch": 19.284591194968552,
      "grad_norm": 0.1699284464120865,
      "learning_rate": 1.9491455123423207e-07,
      "loss": 0.4979,
      "num_input_tokens_seen": 24049576,
      "step": 36795
    },
    {
      "epoch": 19.28721174004193,
      "grad_norm": 0.20575812458992004,
      "learning_rate": 1.9349210272705253e-07,
      "loss": 0.438,
      "num_input_tokens_seen": 24051944,
      "step": 36800
    },
    {
      "epoch": 19.289832285115303,
      "grad_norm": 0.12138701230287552,
      "learning_rate": 1.920748434583175e-07,
      "loss": 0.5201,
      "num_input_tokens_seen": 24055112,
      "step": 36805
    },
    {
      "epoch": 19.29245283018868,
      "grad_norm": 0.16437731683254242,
      "learning_rate": 1.906627737245009e-07,
      "loss": 0.4726,
      "num_input_tokens_seen": 24058888,
      "step": 36810
    },
    {
      "epoch": 19.295073375262053,
      "grad_norm": 0.06021406501531601,
      "learning_rate": 1.8925589382099152e-07,
      "loss": 0.4197,
      "num_input_tokens_seen": 24062728,
      "step": 36815
    },
    {
      "epoch": 19.29769392033543,
      "grad_norm": 0.11115076392889023,
      "learning_rate": 1.8785420404209553e-07,
      "loss": 0.6581,
      "num_input_tokens_seen": 24066504,
      "step": 36820
    },
    {
      "epoch": 19.300314465408803,
      "grad_norm": 0.14121806621551514,
      "learning_rate": 1.864577046810284e-07,
      "loss": 0.4405,
      "num_input_tokens_seen": 24069256,
      "step": 36825
    },
    {
      "epoch": 19.30293501048218,
      "grad_norm": 0.14867502450942993,
      "learning_rate": 1.8506639602992038e-07,
      "loss": 0.4318,
      "num_input_tokens_seen": 24071720,
      "step": 36830
    },
    {
      "epoch": 19.305555555555557,
      "grad_norm": 0.0969482809305191,
      "learning_rate": 1.836802783798164e-07,
      "loss": 0.3409,
      "num_input_tokens_seen": 24075720,
      "step": 36835
    },
    {
      "epoch": 19.30817610062893,
      "grad_norm": 0.1672462671995163,
      "learning_rate": 1.822993520206817e-07,
      "loss": 0.3962,
      "num_input_tokens_seen": 24078920,
      "step": 36840
    },
    {
      "epoch": 19.310796645702307,
      "grad_norm": 0.14037956297397614,
      "learning_rate": 1.8092361724138528e-07,
      "loss": 0.4492,
      "num_input_tokens_seen": 24082376,
      "step": 36845
    },
    {
      "epoch": 19.31341719077568,
      "grad_norm": 0.10262007266283035,
      "learning_rate": 1.7955307432971914e-07,
      "loss": 0.4164,
      "num_input_tokens_seen": 24085928,
      "step": 36850
    },
    {
      "epoch": 19.316037735849058,
      "grad_norm": 0.21036340296268463,
      "learning_rate": 1.7818772357238445e-07,
      "loss": 0.4818,
      "num_input_tokens_seen": 24088648,
      "step": 36855
    },
    {
      "epoch": 19.31865828092243,
      "grad_norm": 0.21378669142723083,
      "learning_rate": 1.7682756525499166e-07,
      "loss": 0.4024,
      "num_input_tokens_seen": 24092136,
      "step": 36860
    },
    {
      "epoch": 19.321278825995808,
      "grad_norm": 0.2074742615222931,
      "learning_rate": 1.7547259966207708e-07,
      "loss": 0.529,
      "num_input_tokens_seen": 24094632,
      "step": 36865
    },
    {
      "epoch": 19.32389937106918,
      "grad_norm": 0.1752775013446808,
      "learning_rate": 1.7412282707708062e-07,
      "loss": 0.506,
      "num_input_tokens_seen": 24098248,
      "step": 36870
    },
    {
      "epoch": 19.32651991614256,
      "grad_norm": 0.25242576003074646,
      "learning_rate": 1.7277824778235974e-07,
      "loss": 0.5026,
      "num_input_tokens_seen": 24100872,
      "step": 36875
    },
    {
      "epoch": 19.329140461215932,
      "grad_norm": 0.22977299988269806,
      "learning_rate": 1.714388620591867e-07,
      "loss": 0.5472,
      "num_input_tokens_seen": 24104520,
      "step": 36880
    },
    {
      "epoch": 19.33176100628931,
      "grad_norm": 0.034697260707616806,
      "learning_rate": 1.701046701877429e-07,
      "loss": 0.4802,
      "num_input_tokens_seen": 24111048,
      "step": 36885
    },
    {
      "epoch": 19.334381551362682,
      "grad_norm": 0.2147148847579956,
      "learning_rate": 1.6877567244713011e-07,
      "loss": 0.4772,
      "num_input_tokens_seen": 24113448,
      "step": 36890
    },
    {
      "epoch": 19.33700209643606,
      "grad_norm": 0.16595937311649323,
      "learning_rate": 1.6745186911535372e-07,
      "loss": 0.4438,
      "num_input_tokens_seen": 24116264,
      "step": 36895
    },
    {
      "epoch": 19.339622641509433,
      "grad_norm": 0.17138724029064178,
      "learning_rate": 1.6613326046934497e-07,
      "loss": 0.4407,
      "num_input_tokens_seen": 24118888,
      "step": 36900
    },
    {
      "epoch": 19.34224318658281,
      "grad_norm": 0.10653537511825562,
      "learning_rate": 1.64819846784936e-07,
      "loss": 0.51,
      "num_input_tokens_seen": 24122056,
      "step": 36905
    },
    {
      "epoch": 19.344863731656183,
      "grad_norm": 0.1503615379333496,
      "learning_rate": 1.6351162833688204e-07,
      "loss": 0.5598,
      "num_input_tokens_seen": 24125288,
      "step": 36910
    },
    {
      "epoch": 19.34748427672956,
      "grad_norm": 0.12147736549377441,
      "learning_rate": 1.6220860539884475e-07,
      "loss": 0.5469,
      "num_input_tokens_seen": 24128904,
      "step": 36915
    },
    {
      "epoch": 19.350104821802937,
      "grad_norm": 0.20049268007278442,
      "learning_rate": 1.6091077824340327e-07,
      "loss": 0.578,
      "num_input_tokens_seen": 24131880,
      "step": 36920
    },
    {
      "epoch": 19.35272536687631,
      "grad_norm": 0.21147814393043518,
      "learning_rate": 1.5961814714204605e-07,
      "loss": 0.4573,
      "num_input_tokens_seen": 24134600,
      "step": 36925
    },
    {
      "epoch": 19.355345911949687,
      "grad_norm": 0.08658681809902191,
      "learning_rate": 1.5833071236518172e-07,
      "loss": 0.4402,
      "num_input_tokens_seen": 24137448,
      "step": 36930
    },
    {
      "epoch": 19.35796645702306,
      "grad_norm": 0.1480594128370285,
      "learning_rate": 1.5704847418212266e-07,
      "loss": 0.6452,
      "num_input_tokens_seen": 24141064,
      "step": 36935
    },
    {
      "epoch": 19.360587002096437,
      "grad_norm": 0.14671917259693146,
      "learning_rate": 1.5577143286109875e-07,
      "loss": 0.3971,
      "num_input_tokens_seen": 24143688,
      "step": 36940
    },
    {
      "epoch": 19.36320754716981,
      "grad_norm": 0.14881066977977753,
      "learning_rate": 1.5449958866925185e-07,
      "loss": 0.3445,
      "num_input_tokens_seen": 24146984,
      "step": 36945
    },
    {
      "epoch": 19.365828092243188,
      "grad_norm": 0.11315201967954636,
      "learning_rate": 1.532329418726386e-07,
      "loss": 0.5744,
      "num_input_tokens_seen": 24151784,
      "step": 36950
    },
    {
      "epoch": 19.36844863731656,
      "grad_norm": 0.11555378139019012,
      "learning_rate": 1.5197149273623036e-07,
      "loss": 0.4942,
      "num_input_tokens_seen": 24154792,
      "step": 36955
    },
    {
      "epoch": 19.371069182389938,
      "grad_norm": 0.4364275634288788,
      "learning_rate": 1.5071524152389936e-07,
      "loss": 0.5099,
      "num_input_tokens_seen": 24157512,
      "step": 36960
    },
    {
      "epoch": 19.37368972746331,
      "grad_norm": 0.12588165700435638,
      "learning_rate": 1.4946418849844934e-07,
      "loss": 0.5915,
      "num_input_tokens_seen": 24160392,
      "step": 36965
    },
    {
      "epoch": 19.37631027253669,
      "grad_norm": 0.24326357245445251,
      "learning_rate": 1.4821833392157923e-07,
      "loss": 0.4437,
      "num_input_tokens_seen": 24163656,
      "step": 36970
    },
    {
      "epoch": 19.378930817610062,
      "grad_norm": 0.1334674060344696,
      "learning_rate": 1.4697767805390838e-07,
      "loss": 0.4914,
      "num_input_tokens_seen": 24166632,
      "step": 36975
    },
    {
      "epoch": 19.38155136268344,
      "grad_norm": 0.22578978538513184,
      "learning_rate": 1.4574222115497084e-07,
      "loss": 0.5025,
      "num_input_tokens_seen": 24169320,
      "step": 36980
    },
    {
      "epoch": 19.384171907756812,
      "grad_norm": 0.2226237803697586,
      "learning_rate": 1.4451196348320706e-07,
      "loss": 0.5889,
      "num_input_tokens_seen": 24172520,
      "step": 36985
    },
    {
      "epoch": 19.38679245283019,
      "grad_norm": 0.20476548373699188,
      "learning_rate": 1.4328690529597232e-07,
      "loss": 0.3603,
      "num_input_tokens_seen": 24176232,
      "step": 36990
    },
    {
      "epoch": 19.389412997903563,
      "grad_norm": 0.21670621633529663,
      "learning_rate": 1.4206704684953943e-07,
      "loss": 0.5083,
      "num_input_tokens_seen": 24179400,
      "step": 36995
    },
    {
      "epoch": 19.39203354297694,
      "grad_norm": 0.25262004137039185,
      "learning_rate": 1.4085238839908476e-07,
      "loss": 0.5193,
      "num_input_tokens_seen": 24182568,
      "step": 37000
    },
    {
      "epoch": 19.394654088050313,
      "grad_norm": 0.15859027206897736,
      "learning_rate": 1.3964293019870235e-07,
      "loss": 0.4855,
      "num_input_tokens_seen": 24185640,
      "step": 37005
    },
    {
      "epoch": 19.39727463312369,
      "grad_norm": 0.21127179265022278,
      "learning_rate": 1.3843867250139807e-07,
      "loss": 0.5545,
      "num_input_tokens_seen": 24189256,
      "step": 37010
    },
    {
      "epoch": 19.399895178197063,
      "grad_norm": 0.11557449400424957,
      "learning_rate": 1.3723961555908715e-07,
      "loss": 0.4527,
      "num_input_tokens_seen": 24192744,
      "step": 37015
    },
    {
      "epoch": 19.40251572327044,
      "grad_norm": 0.15103912353515625,
      "learning_rate": 1.3604575962259947e-07,
      "loss": 0.5243,
      "num_input_tokens_seen": 24197128,
      "step": 37020
    },
    {
      "epoch": 19.405136268343817,
      "grad_norm": 0.12068962305784225,
      "learning_rate": 1.3485710494167692e-07,
      "loss": 0.3801,
      "num_input_tokens_seen": 24200552,
      "step": 37025
    },
    {
      "epoch": 19.40775681341719,
      "grad_norm": 0.1559356451034546,
      "learning_rate": 1.3367365176496783e-07,
      "loss": 0.4573,
      "num_input_tokens_seen": 24203880,
      "step": 37030
    },
    {
      "epoch": 19.410377358490567,
      "grad_norm": 0.31045371294021606,
      "learning_rate": 1.324954003400436e-07,
      "loss": 0.5071,
      "num_input_tokens_seen": 24207752,
      "step": 37035
    },
    {
      "epoch": 19.41299790356394,
      "grad_norm": 0.3114710748195648,
      "learning_rate": 1.313223509133793e-07,
      "loss": 0.5586,
      "num_input_tokens_seen": 24211752,
      "step": 37040
    },
    {
      "epoch": 19.415618448637318,
      "grad_norm": 0.2886182963848114,
      "learning_rate": 1.301545037303592e-07,
      "loss": 0.4605,
      "num_input_tokens_seen": 24213992,
      "step": 37045
    },
    {
      "epoch": 19.41823899371069,
      "grad_norm": 0.17657199501991272,
      "learning_rate": 1.2899185903529066e-07,
      "loss": 0.5253,
      "num_input_tokens_seen": 24216584,
      "step": 37050
    },
    {
      "epoch": 19.420859538784068,
      "grad_norm": 0.22978997230529785,
      "learning_rate": 1.278344170713791e-07,
      "loss": 0.451,
      "num_input_tokens_seen": 24219336,
      "step": 37055
    },
    {
      "epoch": 19.42348008385744,
      "grad_norm": 0.2982656955718994,
      "learning_rate": 1.2668217808075033e-07,
      "loss": 0.4501,
      "num_input_tokens_seen": 24222024,
      "step": 37060
    },
    {
      "epoch": 19.42610062893082,
      "grad_norm": 0.18304219841957092,
      "learning_rate": 1.255351423044393e-07,
      "loss": 0.4921,
      "num_input_tokens_seen": 24225384,
      "step": 37065
    },
    {
      "epoch": 19.428721174004192,
      "grad_norm": 0.20131084322929382,
      "learning_rate": 1.2439330998239572e-07,
      "loss": 0.4334,
      "num_input_tokens_seen": 24228232,
      "step": 37070
    },
    {
      "epoch": 19.43134171907757,
      "grad_norm": 0.2612343728542328,
      "learning_rate": 1.2325668135347578e-07,
      "loss": 0.5516,
      "num_input_tokens_seen": 24231432,
      "step": 37075
    },
    {
      "epoch": 19.433962264150942,
      "grad_norm": 0.13631510734558105,
      "learning_rate": 1.221252566554476e-07,
      "loss": 0.4119,
      "num_input_tokens_seen": 24234952,
      "step": 37080
    },
    {
      "epoch": 19.43658280922432,
      "grad_norm": 0.1314769983291626,
      "learning_rate": 1.2099903612499685e-07,
      "loss": 0.4257,
      "num_input_tokens_seen": 24237768,
      "step": 37085
    },
    {
      "epoch": 19.439203354297693,
      "grad_norm": 0.099318727850914,
      "learning_rate": 1.198780199977101e-07,
      "loss": 0.5322,
      "num_input_tokens_seen": 24241032,
      "step": 37090
    },
    {
      "epoch": 19.44182389937107,
      "grad_norm": 0.10979068279266357,
      "learning_rate": 1.1876220850809416e-07,
      "loss": 0.5228,
      "num_input_tokens_seen": 24245416,
      "step": 37095
    },
    {
      "epoch": 19.444444444444443,
      "grad_norm": 0.34850311279296875,
      "learning_rate": 1.1765160188956514e-07,
      "loss": 0.5401,
      "num_input_tokens_seen": 24249128,
      "step": 37100
    },
    {
      "epoch": 19.44706498951782,
      "grad_norm": 0.10290640592575073,
      "learning_rate": 1.1654620037445107e-07,
      "loss": 0.4143,
      "num_input_tokens_seen": 24252200,
      "step": 37105
    },
    {
      "epoch": 19.449685534591197,
      "grad_norm": 0.12795044481754303,
      "learning_rate": 1.1544600419398088e-07,
      "loss": 0.4581,
      "num_input_tokens_seen": 24255112,
      "step": 37110
    },
    {
      "epoch": 19.45230607966457,
      "grad_norm": 0.16593891382217407,
      "learning_rate": 1.1435101357831213e-07,
      "loss": 0.6241,
      "num_input_tokens_seen": 24258504,
      "step": 37115
    },
    {
      "epoch": 19.454926624737947,
      "grad_norm": 0.23162516951560974,
      "learning_rate": 1.1326122875650047e-07,
      "loss": 0.5455,
      "num_input_tokens_seen": 24260776,
      "step": 37120
    },
    {
      "epoch": 19.45754716981132,
      "grad_norm": 0.26229721307754517,
      "learning_rate": 1.1217664995651633e-07,
      "loss": 0.4215,
      "num_input_tokens_seen": 24263976,
      "step": 37125
    },
    {
      "epoch": 19.460167714884697,
      "grad_norm": 0.18814820051193237,
      "learning_rate": 1.1109727740523935e-07,
      "loss": 0.4662,
      "num_input_tokens_seen": 24267144,
      "step": 37130
    },
    {
      "epoch": 19.46278825995807,
      "grad_norm": 0.10682903230190277,
      "learning_rate": 1.1002311132846944e-07,
      "loss": 0.4762,
      "num_input_tokens_seen": 24270312,
      "step": 37135
    },
    {
      "epoch": 19.465408805031448,
      "grad_norm": 0.281489759683609,
      "learning_rate": 1.0895415195090186e-07,
      "loss": 0.4759,
      "num_input_tokens_seen": 24274632,
      "step": 37140
    },
    {
      "epoch": 19.46802935010482,
      "grad_norm": 0.256018728017807,
      "learning_rate": 1.0789039949615221e-07,
      "loss": 0.6213,
      "num_input_tokens_seen": 24279144,
      "step": 37145
    },
    {
      "epoch": 19.470649895178198,
      "grad_norm": 0.23404787480831146,
      "learning_rate": 1.0683185418674801e-07,
      "loss": 0.5443,
      "num_input_tokens_seen": 24282024,
      "step": 37150
    },
    {
      "epoch": 19.47327044025157,
      "grad_norm": 0.41540247201919556,
      "learning_rate": 1.0577851624412604e-07,
      "loss": 0.5483,
      "num_input_tokens_seen": 24285288,
      "step": 37155
    },
    {
      "epoch": 19.47589098532495,
      "grad_norm": 0.11846696585416794,
      "learning_rate": 1.0473038588862672e-07,
      "loss": 0.387,
      "num_input_tokens_seen": 24289384,
      "step": 37160
    },
    {
      "epoch": 19.478511530398322,
      "grad_norm": 0.3220769762992859,
      "learning_rate": 1.0368746333951351e-07,
      "loss": 0.6228,
      "num_input_tokens_seen": 24291848,
      "step": 37165
    },
    {
      "epoch": 19.4811320754717,
      "grad_norm": 0.1768198013305664,
      "learning_rate": 1.0264974881494805e-07,
      "loss": 0.3749,
      "num_input_tokens_seen": 24296296,
      "step": 37170
    },
    {
      "epoch": 19.483752620545072,
      "grad_norm": 0.15081505477428436,
      "learning_rate": 1.0161724253201222e-07,
      "loss": 0.7242,
      "num_input_tokens_seen": 24299176,
      "step": 37175
    },
    {
      "epoch": 19.48637316561845,
      "grad_norm": 0.13985468447208405,
      "learning_rate": 1.0058994470669436e-07,
      "loss": 0.4024,
      "num_input_tokens_seen": 24302216,
      "step": 37180
    },
    {
      "epoch": 19.488993710691823,
      "grad_norm": 0.19731876254081726,
      "learning_rate": 9.956785555388926e-08,
      "loss": 0.5562,
      "num_input_tokens_seen": 24304648,
      "step": 37185
    },
    {
      "epoch": 19.4916142557652,
      "grad_norm": 0.3207213878631592,
      "learning_rate": 9.855097528741475e-08,
      "loss": 0.4581,
      "num_input_tokens_seen": 24308712,
      "step": 37190
    },
    {
      "epoch": 19.494234800838573,
      "grad_norm": 0.17399099469184875,
      "learning_rate": 9.753930411998402e-08,
      "loss": 0.4429,
      "num_input_tokens_seen": 24312072,
      "step": 37195
    },
    {
      "epoch": 19.49685534591195,
      "grad_norm": 0.09184210002422333,
      "learning_rate": 9.653284226322501e-08,
      "loss": 0.3952,
      "num_input_tokens_seen": 24318312,
      "step": 37200
    },
    {
      "epoch": 19.499475890985323,
      "grad_norm": 0.17076046764850616,
      "learning_rate": 9.553158992768318e-08,
      "loss": 0.5248,
      "num_input_tokens_seen": 24321128,
      "step": 37205
    },
    {
      "epoch": 19.5020964360587,
      "grad_norm": 0.12580044567584991,
      "learning_rate": 9.453554732281044e-08,
      "loss": 0.4233,
      "num_input_tokens_seen": 24324392,
      "step": 37210
    },
    {
      "epoch": 19.504716981132077,
      "grad_norm": 0.956707239151001,
      "learning_rate": 9.354471465696236e-08,
      "loss": 0.6943,
      "num_input_tokens_seen": 24327560,
      "step": 37215
    },
    {
      "epoch": 19.50733752620545,
      "grad_norm": 0.14964285492897034,
      "learning_rate": 9.255909213741199e-08,
      "loss": 0.5719,
      "num_input_tokens_seen": 24330504,
      "step": 37220
    },
    {
      "epoch": 19.509958071278827,
      "grad_norm": 0.20070257782936096,
      "learning_rate": 9.157867997034164e-08,
      "loss": 0.4015,
      "num_input_tokens_seen": 24333800,
      "step": 37225
    },
    {
      "epoch": 19.5125786163522,
      "grad_norm": 0.14650817215442657,
      "learning_rate": 9.060347836084004e-08,
      "loss": 0.4572,
      "num_input_tokens_seen": 24336264,
      "step": 37230
    },
    {
      "epoch": 19.515199161425578,
      "grad_norm": 0.24171744287014008,
      "learning_rate": 8.96334875129079e-08,
      "loss": 0.5549,
      "num_input_tokens_seen": 24338696,
      "step": 37235
    },
    {
      "epoch": 19.51781970649895,
      "grad_norm": 0.3085666000843048,
      "learning_rate": 8.866870762946067e-08,
      "loss": 0.4526,
      "num_input_tokens_seen": 24342056,
      "step": 37240
    },
    {
      "epoch": 19.520440251572328,
      "grad_norm": 0.14373600482940674,
      "learning_rate": 8.77091389123147e-08,
      "loss": 0.4906,
      "num_input_tokens_seen": 24345480,
      "step": 37245
    },
    {
      "epoch": 19.5230607966457,
      "grad_norm": 0.11734937876462936,
      "learning_rate": 8.675478156220385e-08,
      "loss": 0.493,
      "num_input_tokens_seen": 24348680,
      "step": 37250
    },
    {
      "epoch": 19.52568134171908,
      "grad_norm": 0.09425126016139984,
      "learning_rate": 8.580563577876566e-08,
      "loss": 0.4179,
      "num_input_tokens_seen": 24351624,
      "step": 37255
    },
    {
      "epoch": 19.528301886792452,
      "grad_norm": 0.11387787759304047,
      "learning_rate": 8.486170176055242e-08,
      "loss": 0.514,
      "num_input_tokens_seen": 24356552,
      "step": 37260
    },
    {
      "epoch": 19.53092243186583,
      "grad_norm": 0.15142332017421722,
      "learning_rate": 8.392297970502282e-08,
      "loss": 0.4417,
      "num_input_tokens_seen": 24359688,
      "step": 37265
    },
    {
      "epoch": 19.533542976939202,
      "grad_norm": 0.132102832198143,
      "learning_rate": 8.298946980855315e-08,
      "loss": 0.473,
      "num_input_tokens_seen": 24364328,
      "step": 37270
    },
    {
      "epoch": 19.53616352201258,
      "grad_norm": 0.1520739644765854,
      "learning_rate": 8.206117226641497e-08,
      "loss": 0.3277,
      "num_input_tokens_seen": 24367752,
      "step": 37275
    },
    {
      "epoch": 19.538784067085953,
      "grad_norm": 0.14022913575172424,
      "learning_rate": 8.113808727280015e-08,
      "loss": 0.501,
      "num_input_tokens_seen": 24371208,
      "step": 37280
    },
    {
      "epoch": 19.54140461215933,
      "grad_norm": 0.20970706641674042,
      "learning_rate": 8.022021502080979e-08,
      "loss": 0.3794,
      "num_input_tokens_seen": 24374824,
      "step": 37285
    },
    {
      "epoch": 19.544025157232703,
      "grad_norm": 0.12580150365829468,
      "learning_rate": 7.930755570244863e-08,
      "loss": 0.3896,
      "num_input_tokens_seen": 24378408,
      "step": 37290
    },
    {
      "epoch": 19.54664570230608,
      "grad_norm": 0.26175057888031006,
      "learning_rate": 7.840010950863619e-08,
      "loss": 0.6056,
      "num_input_tokens_seen": 24381160,
      "step": 37295
    },
    {
      "epoch": 19.549266247379457,
      "grad_norm": 0.15000960230827332,
      "learning_rate": 7.749787662920116e-08,
      "loss": 0.618,
      "num_input_tokens_seen": 24384072,
      "step": 37300
    },
    {
      "epoch": 19.55188679245283,
      "grad_norm": 0.22161222994327545,
      "learning_rate": 7.660085725288146e-08,
      "loss": 0.3911,
      "num_input_tokens_seen": 24386952,
      "step": 37305
    },
    {
      "epoch": 19.554507337526207,
      "grad_norm": 0.11263281852006912,
      "learning_rate": 7.570905156731867e-08,
      "loss": 0.4154,
      "num_input_tokens_seen": 24390600,
      "step": 37310
    },
    {
      "epoch": 19.55712788259958,
      "grad_norm": 0.13541141152381897,
      "learning_rate": 7.482245975907188e-08,
      "loss": 0.4988,
      "num_input_tokens_seen": 24393416,
      "step": 37315
    },
    {
      "epoch": 19.559748427672957,
      "grad_norm": 0.23910684883594513,
      "learning_rate": 7.394108201360383e-08,
      "loss": 0.4438,
      "num_input_tokens_seen": 24396328,
      "step": 37320
    },
    {
      "epoch": 19.56236897274633,
      "grad_norm": 0.09970199316740036,
      "learning_rate": 7.306491851529206e-08,
      "loss": 0.3869,
      "num_input_tokens_seen": 24404040,
      "step": 37325
    },
    {
      "epoch": 19.564989517819708,
      "grad_norm": 0.1551792472600937,
      "learning_rate": 7.219396944741774e-08,
      "loss": 0.5111,
      "num_input_tokens_seen": 24407304,
      "step": 37330
    },
    {
      "epoch": 19.56761006289308,
      "grad_norm": 0.11970236897468567,
      "learning_rate": 7.132823499217123e-08,
      "loss": 0.5796,
      "num_input_tokens_seen": 24411496,
      "step": 37335
    },
    {
      "epoch": 19.570230607966458,
      "grad_norm": 0.22819989919662476,
      "learning_rate": 7.046771533065766e-08,
      "loss": 0.4482,
      "num_input_tokens_seen": 24414280,
      "step": 37340
    },
    {
      "epoch": 19.57285115303983,
      "grad_norm": 0.15776516497135162,
      "learning_rate": 6.961241064288859e-08,
      "loss": 0.3811,
      "num_input_tokens_seen": 24417352,
      "step": 37345
    },
    {
      "epoch": 19.57547169811321,
      "grad_norm": 0.20650920271873474,
      "learning_rate": 6.876232110778202e-08,
      "loss": 0.5579,
      "num_input_tokens_seen": 24421192,
      "step": 37350
    },
    {
      "epoch": 19.578092243186582,
      "grad_norm": 0.15685507655143738,
      "learning_rate": 6.791744690316515e-08,
      "loss": 0.507,
      "num_input_tokens_seen": 24424744,
      "step": 37355
    },
    {
      "epoch": 19.58071278825996,
      "grad_norm": 0.1276089996099472,
      "learning_rate": 6.707778820577715e-08,
      "loss": 0.4731,
      "num_input_tokens_seen": 24427592,
      "step": 37360
    },
    {
      "epoch": 19.583333333333332,
      "grad_norm": 0.20101770758628845,
      "learning_rate": 6.624334519126919e-08,
      "loss": 0.5735,
      "num_input_tokens_seen": 24430600,
      "step": 37365
    },
    {
      "epoch": 19.58595387840671,
      "grad_norm": 0.1434648483991623,
      "learning_rate": 6.541411803419329e-08,
      "loss": 0.4694,
      "num_input_tokens_seen": 24433160,
      "step": 37370
    },
    {
      "epoch": 19.588574423480082,
      "grad_norm": 0.22521834075450897,
      "learning_rate": 6.459010690801626e-08,
      "loss": 0.6294,
      "num_input_tokens_seen": 24435656,
      "step": 37375
    },
    {
      "epoch": 19.59119496855346,
      "grad_norm": 0.1954823136329651,
      "learning_rate": 6.377131198510855e-08,
      "loss": 0.5482,
      "num_input_tokens_seen": 24438440,
      "step": 37380
    },
    {
      "epoch": 19.593815513626833,
      "grad_norm": 0.1583237200975418,
      "learning_rate": 6.295773343675537e-08,
      "loss": 0.4501,
      "num_input_tokens_seen": 24441416,
      "step": 37385
    },
    {
      "epoch": 19.59643605870021,
      "grad_norm": 0.16795805096626282,
      "learning_rate": 6.214937143314837e-08,
      "loss": 0.3757,
      "num_input_tokens_seen": 24444328,
      "step": 37390
    },
    {
      "epoch": 19.599056603773583,
      "grad_norm": 0.2873033881187439,
      "learning_rate": 6.134622614338559e-08,
      "loss": 0.4472,
      "num_input_tokens_seen": 24446472,
      "step": 37395
    },
    {
      "epoch": 19.60167714884696,
      "grad_norm": 0.2456432580947876,
      "learning_rate": 6.05482977354771e-08,
      "loss": 0.5215,
      "num_input_tokens_seen": 24449544,
      "step": 37400
    },
    {
      "epoch": 19.604297693920337,
      "grad_norm": 0.165083110332489,
      "learning_rate": 5.975558637634215e-08,
      "loss": 0.495,
      "num_input_tokens_seen": 24452776,
      "step": 37405
    },
    {
      "epoch": 19.60691823899371,
      "grad_norm": 0.12834955751895905,
      "learning_rate": 5.896809223180366e-08,
      "loss": 0.4846,
      "num_input_tokens_seen": 24455560,
      "step": 37410
    },
    {
      "epoch": 19.609538784067087,
      "grad_norm": 0.2982673943042755,
      "learning_rate": 5.8185815466596516e-08,
      "loss": 0.5049,
      "num_input_tokens_seen": 24458408,
      "step": 37415
    },
    {
      "epoch": 19.61215932914046,
      "grad_norm": 0.17480556666851044,
      "learning_rate": 5.7408756244370365e-08,
      "loss": 0.4152,
      "num_input_tokens_seen": 24461576,
      "step": 37420
    },
    {
      "epoch": 19.614779874213838,
      "grad_norm": 0.13569064438343048,
      "learning_rate": 5.663691472766741e-08,
      "loss": 0.4674,
      "num_input_tokens_seen": 24464840,
      "step": 37425
    },
    {
      "epoch": 19.61740041928721,
      "grad_norm": 0.20568878948688507,
      "learning_rate": 5.587029107795572e-08,
      "loss": 0.4485,
      "num_input_tokens_seen": 24468200,
      "step": 37430
    },
    {
      "epoch": 19.620020964360588,
      "grad_norm": 0.1430869847536087,
      "learning_rate": 5.510888545559867e-08,
      "loss": 0.5351,
      "num_input_tokens_seen": 24471176,
      "step": 37435
    },
    {
      "epoch": 19.62264150943396,
      "grad_norm": 0.11490800976753235,
      "learning_rate": 5.435269801987997e-08,
      "loss": 0.3566,
      "num_input_tokens_seen": 24473960,
      "step": 37440
    },
    {
      "epoch": 19.62526205450734,
      "grad_norm": 0.1782723069190979,
      "learning_rate": 5.360172892897863e-08,
      "loss": 0.3775,
      "num_input_tokens_seen": 24476776,
      "step": 37445
    },
    {
      "epoch": 19.627882599580712,
      "grad_norm": 0.09594607353210449,
      "learning_rate": 5.285597833999401e-08,
      "loss": 0.5195,
      "num_input_tokens_seen": 24480456,
      "step": 37450
    },
    {
      "epoch": 19.63050314465409,
      "grad_norm": 0.2421015501022339,
      "learning_rate": 5.2115446408926295e-08,
      "loss": 0.3481,
      "num_input_tokens_seen": 24483400,
      "step": 37455
    },
    {
      "epoch": 19.633123689727462,
      "grad_norm": 0.16476324200630188,
      "learning_rate": 5.138013329068769e-08,
      "loss": 0.6452,
      "num_input_tokens_seen": 24486536,
      "step": 37460
    },
    {
      "epoch": 19.63574423480084,
      "grad_norm": 0.12639924883842468,
      "learning_rate": 5.065003913909405e-08,
      "loss": 0.3776,
      "num_input_tokens_seen": 24490056,
      "step": 37465
    },
    {
      "epoch": 19.638364779874212,
      "grad_norm": 0.16546881198883057,
      "learning_rate": 4.9925164106875975e-08,
      "loss": 0.4936,
      "num_input_tokens_seen": 24493160,
      "step": 37470
    },
    {
      "epoch": 19.64098532494759,
      "grad_norm": 0.25046399235725403,
      "learning_rate": 4.9205508345670506e-08,
      "loss": 0.5683,
      "num_input_tokens_seen": 24496200,
      "step": 37475
    },
    {
      "epoch": 19.643605870020963,
      "grad_norm": 0.15977948904037476,
      "learning_rate": 4.8491072006015546e-08,
      "loss": 0.3779,
      "num_input_tokens_seen": 24500680,
      "step": 37480
    },
    {
      "epoch": 19.64622641509434,
      "grad_norm": 0.18073497712612152,
      "learning_rate": 4.778185523736933e-08,
      "loss": 0.6918,
      "num_input_tokens_seen": 24503720,
      "step": 37485
    },
    {
      "epoch": 19.648846960167717,
      "grad_norm": 0.1806933432817459,
      "learning_rate": 4.7077858188085414e-08,
      "loss": 0.488,
      "num_input_tokens_seen": 24506440,
      "step": 37490
    },
    {
      "epoch": 19.65146750524109,
      "grad_norm": 0.1700839102268219,
      "learning_rate": 4.6379081005437644e-08,
      "loss": 0.5317,
      "num_input_tokens_seen": 24509384,
      "step": 37495
    },
    {
      "epoch": 19.654088050314467,
      "grad_norm": 0.1312578171491623,
      "learning_rate": 4.568552383560076e-08,
      "loss": 0.4502,
      "num_input_tokens_seen": 24511720,
      "step": 37500
    },
    {
      "epoch": 19.65670859538784,
      "grad_norm": 0.16730955243110657,
      "learning_rate": 4.4997186823655945e-08,
      "loss": 0.52,
      "num_input_tokens_seen": 24514472,
      "step": 37505
    },
    {
      "epoch": 19.659329140461217,
      "grad_norm": 0.10073717683553696,
      "learning_rate": 4.431407011360189e-08,
      "loss": 0.4621,
      "num_input_tokens_seen": 24517224,
      "step": 37510
    },
    {
      "epoch": 19.66194968553459,
      "grad_norm": 0.14911331236362457,
      "learning_rate": 4.3636173848332627e-08,
      "loss": 0.4709,
      "num_input_tokens_seen": 24519944,
      "step": 37515
    },
    {
      "epoch": 19.664570230607968,
      "grad_norm": 0.21599207818508148,
      "learning_rate": 4.2963498169656946e-08,
      "loss": 0.5195,
      "num_input_tokens_seen": 24524840,
      "step": 37520
    },
    {
      "epoch": 19.66719077568134,
      "grad_norm": 0.12451428174972534,
      "learning_rate": 4.229604321829561e-08,
      "loss": 0.4899,
      "num_input_tokens_seen": 24527912,
      "step": 37525
    },
    {
      "epoch": 19.669811320754718,
      "grad_norm": 0.16178491711616516,
      "learning_rate": 4.16338091338675e-08,
      "loss": 0.3948,
      "num_input_tokens_seen": 24530696,
      "step": 37530
    },
    {
      "epoch": 19.67243186582809,
      "grad_norm": 0.15931926667690277,
      "learning_rate": 4.097679605490623e-08,
      "loss": 0.3182,
      "num_input_tokens_seen": 24533256,
      "step": 37535
    },
    {
      "epoch": 19.67505241090147,
      "grad_norm": 0.16348816454410553,
      "learning_rate": 4.032500411885465e-08,
      "loss": 0.5096,
      "num_input_tokens_seen": 24536552,
      "step": 37540
    },
    {
      "epoch": 19.677672955974842,
      "grad_norm": 0.12927381694316864,
      "learning_rate": 3.967843346205369e-08,
      "loss": 0.4749,
      "num_input_tokens_seen": 24539464,
      "step": 37545
    },
    {
      "epoch": 19.68029350104822,
      "grad_norm": 0.1630571484565735,
      "learning_rate": 3.903708421976182e-08,
      "loss": 0.4108,
      "num_input_tokens_seen": 24542312,
      "step": 37550
    },
    {
      "epoch": 19.682914046121592,
      "grad_norm": 0.21273362636566162,
      "learning_rate": 3.840095652614395e-08,
      "loss": 0.4486,
      "num_input_tokens_seen": 24546056,
      "step": 37555
    },
    {
      "epoch": 19.68553459119497,
      "grad_norm": 0.1239013746380806,
      "learning_rate": 3.777005051426863e-08,
      "loss": 0.4677,
      "num_input_tokens_seen": 24549512,
      "step": 37560
    },
    {
      "epoch": 19.688155136268342,
      "grad_norm": 0.2786845266819,
      "learning_rate": 3.71443663161164e-08,
      "loss": 0.4516,
      "num_input_tokens_seen": 24553000,
      "step": 37565
    },
    {
      "epoch": 19.69077568134172,
      "grad_norm": 0.12332155555486679,
      "learning_rate": 3.652390406256867e-08,
      "loss": 0.686,
      "num_input_tokens_seen": 24555944,
      "step": 37570
    },
    {
      "epoch": 19.693396226415093,
      "grad_norm": 0.254210889339447,
      "learning_rate": 3.5908663883421624e-08,
      "loss": 0.3794,
      "num_input_tokens_seen": 24558600,
      "step": 37575
    },
    {
      "epoch": 19.69601677148847,
      "grad_norm": 0.19072946906089783,
      "learning_rate": 3.529864590737786e-08,
      "loss": 0.4597,
      "num_input_tokens_seen": 24562440,
      "step": 37580
    },
    {
      "epoch": 19.698637316561843,
      "grad_norm": 0.10869759321212769,
      "learning_rate": 3.4693850262046415e-08,
      "loss": 0.4708,
      "num_input_tokens_seen": 24565736,
      "step": 37585
    },
    {
      "epoch": 19.70125786163522,
      "grad_norm": 0.12728217244148254,
      "learning_rate": 3.409427707394275e-08,
      "loss": 0.4642,
      "num_input_tokens_seen": 24569064,
      "step": 37590
    },
    {
      "epoch": 19.703878406708597,
      "grad_norm": 0.1686820685863495,
      "learning_rate": 3.349992646848599e-08,
      "loss": 0.4747,
      "num_input_tokens_seen": 24572584,
      "step": 37595
    },
    {
      "epoch": 19.70649895178197,
      "grad_norm": 0.2030370533466339,
      "learning_rate": 3.291079857001556e-08,
      "loss": 0.6414,
      "num_input_tokens_seen": 24575240,
      "step": 37600
    },
    {
      "epoch": 19.709119496855347,
      "grad_norm": 0.17926740646362305,
      "learning_rate": 3.2326893501763436e-08,
      "loss": 0.6442,
      "num_input_tokens_seen": 24578408,
      "step": 37605
    },
    {
      "epoch": 19.71174004192872,
      "grad_norm": 0.3540093004703522,
      "learning_rate": 3.174821138588191e-08,
      "loss": 0.4688,
      "num_input_tokens_seen": 24581064,
      "step": 37610
    },
    {
      "epoch": 19.714360587002098,
      "grad_norm": 0.1600237339735031,
      "learning_rate": 3.11747523434186e-08,
      "loss": 0.436,
      "num_input_tokens_seen": 24584200,
      "step": 37615
    },
    {
      "epoch": 19.71698113207547,
      "grad_norm": 0.17090918123722076,
      "learning_rate": 3.060651649434143e-08,
      "loss": 0.5772,
      "num_input_tokens_seen": 24587400,
      "step": 37620
    },
    {
      "epoch": 19.719601677148848,
      "grad_norm": 0.16196022927761078,
      "learning_rate": 3.004350395751088e-08,
      "loss": 0.3854,
      "num_input_tokens_seen": 24590088,
      "step": 37625
    },
    {
      "epoch": 19.72222222222222,
      "grad_norm": 0.2799566090106964,
      "learning_rate": 2.94857148507105e-08,
      "loss": 0.6069,
      "num_input_tokens_seen": 24593064,
      "step": 37630
    },
    {
      "epoch": 19.7248427672956,
      "grad_norm": 0.09714753180742264,
      "learning_rate": 2.893314929061641e-08,
      "loss": 0.4933,
      "num_input_tokens_seen": 24595816,
      "step": 37635
    },
    {
      "epoch": 19.72746331236897,
      "grad_norm": 0.14056318998336792,
      "learning_rate": 2.8385807392825037e-08,
      "loss": 0.3819,
      "num_input_tokens_seen": 24598824,
      "step": 37640
    },
    {
      "epoch": 19.73008385744235,
      "grad_norm": 0.17467047274112701,
      "learning_rate": 2.7843689271830897e-08,
      "loss": 0.4786,
      "num_input_tokens_seen": 24601480,
      "step": 37645
    },
    {
      "epoch": 19.732704402515722,
      "grad_norm": 0.18234819173812866,
      "learning_rate": 2.7306795041040502e-08,
      "loss": 0.3932,
      "num_input_tokens_seen": 24604296,
      "step": 37650
    },
    {
      "epoch": 19.7353249475891,
      "grad_norm": 0.09949353337287903,
      "learning_rate": 2.677512481276401e-08,
      "loss": 0.4229,
      "num_input_tokens_seen": 24607336,
      "step": 37655
    },
    {
      "epoch": 19.737945492662472,
      "grad_norm": 0.10179416835308075,
      "learning_rate": 2.6248678698220785e-08,
      "loss": 0.5668,
      "num_input_tokens_seen": 24610984,
      "step": 37660
    },
    {
      "epoch": 19.74056603773585,
      "grad_norm": 0.1155388355255127,
      "learning_rate": 2.5727456807542183e-08,
      "loss": 0.4235,
      "num_input_tokens_seen": 24613800,
      "step": 37665
    },
    {
      "epoch": 19.743186582809223,
      "grad_norm": 0.11007735133171082,
      "learning_rate": 2.5211459249754875e-08,
      "loss": 0.3817,
      "num_input_tokens_seen": 24617064,
      "step": 37670
    },
    {
      "epoch": 19.7458071278826,
      "grad_norm": 0.1575925052165985,
      "learning_rate": 2.4700686132803076e-08,
      "loss": 0.4449,
      "num_input_tokens_seen": 24621416,
      "step": 37675
    },
    {
      "epoch": 19.748427672955973,
      "grad_norm": 0.13053466379642487,
      "learning_rate": 2.4195137563534643e-08,
      "loss": 0.4867,
      "num_input_tokens_seen": 24625160,
      "step": 37680
    },
    {
      "epoch": 19.75104821802935,
      "grad_norm": 0.13805224001407623,
      "learning_rate": 2.369481364770387e-08,
      "loss": 0.4384,
      "num_input_tokens_seen": 24628520,
      "step": 37685
    },
    {
      "epoch": 19.753668763102727,
      "grad_norm": 0.14687877893447876,
      "learning_rate": 2.3199714489974266e-08,
      "loss": 0.571,
      "num_input_tokens_seen": 24631528,
      "step": 37690
    },
    {
      "epoch": 19.7562893081761,
      "grad_norm": 0.13426779210567474,
      "learning_rate": 2.2709840193912978e-08,
      "loss": 0.5918,
      "num_input_tokens_seen": 24634440,
      "step": 37695
    },
    {
      "epoch": 19.758909853249477,
      "grad_norm": 0.11101856082677841,
      "learning_rate": 2.2225190861999146e-08,
      "loss": 0.4859,
      "num_input_tokens_seen": 24637704,
      "step": 37700
    },
    {
      "epoch": 19.76153039832285,
      "grad_norm": 0.24711523950099945,
      "learning_rate": 2.1745766595612783e-08,
      "loss": 0.62,
      "num_input_tokens_seen": 24640776,
      "step": 37705
    },
    {
      "epoch": 19.764150943396228,
      "grad_norm": 0.18578092753887177,
      "learning_rate": 2.1271567495045886e-08,
      "loss": 0.3635,
      "num_input_tokens_seen": 24643624,
      "step": 37710
    },
    {
      "epoch": 19.7667714884696,
      "grad_norm": 0.1422283798456192,
      "learning_rate": 2.0802593659491333e-08,
      "loss": 0.5363,
      "num_input_tokens_seen": 24647496,
      "step": 37715
    },
    {
      "epoch": 19.769392033542978,
      "grad_norm": 0.13568788766860962,
      "learning_rate": 2.0338845187059528e-08,
      "loss": 0.6095,
      "num_input_tokens_seen": 24651048,
      "step": 37720
    },
    {
      "epoch": 19.77201257861635,
      "grad_norm": 0.16370555758476257,
      "learning_rate": 1.988032217475899e-08,
      "loss": 0.5234,
      "num_input_tokens_seen": 24653736,
      "step": 37725
    },
    {
      "epoch": 19.77463312368973,
      "grad_norm": 0.15755027532577515,
      "learning_rate": 1.942702471850466e-08,
      "loss": 0.4638,
      "num_input_tokens_seen": 24657032,
      "step": 37730
    },
    {
      "epoch": 19.7772536687631,
      "grad_norm": 0.15648628771305084,
      "learning_rate": 1.897895291312346e-08,
      "loss": 0.4674,
      "num_input_tokens_seen": 24660136,
      "step": 37735
    },
    {
      "epoch": 19.77987421383648,
      "grad_norm": 0.13391363620758057,
      "learning_rate": 1.8536106852348746e-08,
      "loss": 0.6448,
      "num_input_tokens_seen": 24663048,
      "step": 37740
    },
    {
      "epoch": 19.782494758909852,
      "grad_norm": 0.14405228197574615,
      "learning_rate": 1.8098486628814748e-08,
      "loss": 0.4548,
      "num_input_tokens_seen": 24667016,
      "step": 37745
    },
    {
      "epoch": 19.78511530398323,
      "grad_norm": 0.1439061164855957,
      "learning_rate": 1.7666092334067684e-08,
      "loss": 0.5049,
      "num_input_tokens_seen": 24670568,
      "step": 37750
    },
    {
      "epoch": 19.787735849056602,
      "grad_norm": 0.22043722867965698,
      "learning_rate": 1.7238924058562977e-08,
      "loss": 0.5701,
      "num_input_tokens_seen": 24674024,
      "step": 37755
    },
    {
      "epoch": 19.79035639412998,
      "grad_norm": 0.21461568772792816,
      "learning_rate": 1.6816981891656924e-08,
      "loss": 0.5677,
      "num_input_tokens_seen": 24677320,
      "step": 37760
    },
    {
      "epoch": 19.792976939203353,
      "grad_norm": 0.12357223033905029,
      "learning_rate": 1.6400265921615032e-08,
      "loss": 0.4354,
      "num_input_tokens_seen": 24681096,
      "step": 37765
    },
    {
      "epoch": 19.79559748427673,
      "grad_norm": 0.14756448566913605,
      "learning_rate": 1.598877623560924e-08,
      "loss": 0.508,
      "num_input_tokens_seen": 24684168,
      "step": 37770
    },
    {
      "epoch": 19.798218029350103,
      "grad_norm": 0.15344731509685516,
      "learning_rate": 1.5582512919717908e-08,
      "loss": 0.4714,
      "num_input_tokens_seen": 24687464,
      "step": 37775
    },
    {
      "epoch": 19.80083857442348,
      "grad_norm": 0.21900491416454315,
      "learning_rate": 1.5181476058928613e-08,
      "loss": 0.54,
      "num_input_tokens_seen": 24690920,
      "step": 37780
    },
    {
      "epoch": 19.803459119496857,
      "grad_norm": 0.18005743622779846,
      "learning_rate": 1.4785665737129805e-08,
      "loss": 0.3423,
      "num_input_tokens_seen": 24693768,
      "step": 37785
    },
    {
      "epoch": 19.80607966457023,
      "grad_norm": 0.22985053062438965,
      "learning_rate": 1.4395082037127471e-08,
      "loss": 0.5147,
      "num_input_tokens_seen": 24696552,
      "step": 37790
    },
    {
      "epoch": 19.808700209643607,
      "grad_norm": 0.24189789593219757,
      "learning_rate": 1.4009725040620148e-08,
      "loss": 0.4315,
      "num_input_tokens_seen": 24699656,
      "step": 37795
    },
    {
      "epoch": 19.81132075471698,
      "grad_norm": 0.15280477702617645,
      "learning_rate": 1.3629594828223902e-08,
      "loss": 0.5753,
      "num_input_tokens_seen": 24703496,
      "step": 37800
    },
    {
      "epoch": 19.813941299790358,
      "grad_norm": 0.1336512267589569,
      "learning_rate": 1.3254691479458458e-08,
      "loss": 0.567,
      "num_input_tokens_seen": 24706760,
      "step": 37805
    },
    {
      "epoch": 19.81656184486373,
      "grad_norm": 0.17471672594547272,
      "learning_rate": 1.2885015072747197e-08,
      "loss": 0.5292,
      "num_input_tokens_seen": 24710952,
      "step": 37810
    },
    {
      "epoch": 19.819182389937108,
      "grad_norm": 0.16411495208740234,
      "learning_rate": 1.2520565685419927e-08,
      "loss": 0.3524,
      "num_input_tokens_seen": 24714088,
      "step": 37815
    },
    {
      "epoch": 19.82180293501048,
      "grad_norm": 0.37867245078086853,
      "learning_rate": 1.2161343393721214e-08,
      "loss": 0.4907,
      "num_input_tokens_seen": 24717064,
      "step": 37820
    },
    {
      "epoch": 19.82442348008386,
      "grad_norm": 0.5201301574707031,
      "learning_rate": 1.1807348272790952e-08,
      "loss": 0.5641,
      "num_input_tokens_seen": 24720712,
      "step": 37825
    },
    {
      "epoch": 19.82704402515723,
      "grad_norm": 0.21186700463294983,
      "learning_rate": 1.1458580396683794e-08,
      "loss": 0.6732,
      "num_input_tokens_seen": 24723560,
      "step": 37830
    },
    {
      "epoch": 19.82966457023061,
      "grad_norm": 0.14974066615104675,
      "learning_rate": 1.1115039838358044e-08,
      "loss": 0.6196,
      "num_input_tokens_seen": 24731656,
      "step": 37835
    },
    {
      "epoch": 19.832285115303982,
      "grad_norm": 0.2821713984012604,
      "learning_rate": 1.0776726669678439e-08,
      "loss": 0.692,
      "num_input_tokens_seen": 24734280,
      "step": 37840
    },
    {
      "epoch": 19.83490566037736,
      "grad_norm": 0.41050824522972107,
      "learning_rate": 1.044364096141337e-08,
      "loss": 0.6299,
      "num_input_tokens_seen": 24736840,
      "step": 37845
    },
    {
      "epoch": 19.837526205450732,
      "grad_norm": 0.09713126718997955,
      "learning_rate": 1.0115782783243211e-08,
      "loss": 0.4602,
      "num_input_tokens_seen": 24740200,
      "step": 37850
    },
    {
      "epoch": 19.84014675052411,
      "grad_norm": 0.1381097435951233,
      "learning_rate": 9.793152203751988e-09,
      "loss": 0.4397,
      "num_input_tokens_seen": 24743592,
      "step": 37855
    },
    {
      "epoch": 19.842767295597483,
      "grad_norm": 0.2976767420768738,
      "learning_rate": 9.475749290430158e-09,
      "loss": 0.4826,
      "num_input_tokens_seen": 24746344,
      "step": 37860
    },
    {
      "epoch": 19.84538784067086,
      "grad_norm": 0.19266337156295776,
      "learning_rate": 9.16357410967461e-09,
      "loss": 0.3905,
      "num_input_tokens_seen": 24748712,
      "step": 37865
    },
    {
      "epoch": 19.848008385744233,
      "grad_norm": 0.16422967612743378,
      "learning_rate": 8.856626726788664e-09,
      "loss": 0.4549,
      "num_input_tokens_seen": 24752264,
      "step": 37870
    },
    {
      "epoch": 19.85062893081761,
      "grad_norm": 0.1472795158624649,
      "learning_rate": 8.554907205982066e-09,
      "loss": 0.3785,
      "num_input_tokens_seen": 24755816,
      "step": 37875
    },
    {
      "epoch": 19.853249475890987,
      "grad_norm": 0.16859357059001923,
      "learning_rate": 8.258415610373772e-09,
      "loss": 0.4963,
      "num_input_tokens_seen": 24758792,
      "step": 37880
    },
    {
      "epoch": 19.85587002096436,
      "grad_norm": 0.12334322184324265,
      "learning_rate": 7.967152001980838e-09,
      "loss": 0.4467,
      "num_input_tokens_seen": 24762504,
      "step": 37885
    },
    {
      "epoch": 19.858490566037737,
      "grad_norm": 0.17467941343784332,
      "learning_rate": 7.681116441737856e-09,
      "loss": 0.5767,
      "num_input_tokens_seen": 24766760,
      "step": 37890
    },
    {
      "epoch": 19.86111111111111,
      "grad_norm": 0.20479431748390198,
      "learning_rate": 7.4003089894747464e-09,
      "loss": 0.5279,
      "num_input_tokens_seen": 24770536,
      "step": 37895
    },
    {
      "epoch": 19.863731656184488,
      "grad_norm": 0.17506444454193115,
      "learning_rate": 7.12472970393896e-09,
      "loss": 0.4323,
      "num_input_tokens_seen": 24773960,
      "step": 37900
    },
    {
      "epoch": 19.86635220125786,
      "grad_norm": 0.1679806262254715,
      "learning_rate": 6.854378642776049e-09,
      "loss": 0.4313,
      "num_input_tokens_seen": 24776744,
      "step": 37905
    },
    {
      "epoch": 19.868972746331238,
      "grad_norm": 0.13666872680187225,
      "learning_rate": 6.589255862538002e-09,
      "loss": 0.5174,
      "num_input_tokens_seen": 24781128,
      "step": 37910
    },
    {
      "epoch": 19.87159329140461,
      "grad_norm": 0.14641599357128143,
      "learning_rate": 6.32936141869156e-09,
      "loss": 0.4006,
      "num_input_tokens_seen": 24784072,
      "step": 37915
    },
    {
      "epoch": 19.87421383647799,
      "grad_norm": 0.209395170211792,
      "learning_rate": 6.074695365596017e-09,
      "loss": 0.5293,
      "num_input_tokens_seen": 24787304,
      "step": 37920
    },
    {
      "epoch": 19.87683438155136,
      "grad_norm": 0.2769449055194855,
      "learning_rate": 5.8252577565309775e-09,
      "loss": 0.4305,
      "num_input_tokens_seen": 24790216,
      "step": 37925
    },
    {
      "epoch": 19.87945492662474,
      "grad_norm": 0.1486600786447525,
      "learning_rate": 5.581048643671371e-09,
      "loss": 0.5034,
      "num_input_tokens_seen": 24794024,
      "step": 37930
    },
    {
      "epoch": 19.882075471698112,
      "grad_norm": 0.27782589197158813,
      "learning_rate": 5.342068078106888e-09,
      "loss": 0.473,
      "num_input_tokens_seen": 24797032,
      "step": 37935
    },
    {
      "epoch": 19.88469601677149,
      "grad_norm": 0.20168325304985046,
      "learning_rate": 5.1083161098253175e-09,
      "loss": 0.44,
      "num_input_tokens_seen": 24799592,
      "step": 37940
    },
    {
      "epoch": 19.887316561844862,
      "grad_norm": 0.3207286596298218,
      "learning_rate": 4.87979278772921e-09,
      "loss": 0.4911,
      "num_input_tokens_seen": 24803944,
      "step": 37945
    },
    {
      "epoch": 19.88993710691824,
      "grad_norm": 0.16677545011043549,
      "learning_rate": 4.656498159619216e-09,
      "loss": 0.5348,
      "num_input_tokens_seen": 24807816,
      "step": 37950
    },
    {
      "epoch": 19.892557651991613,
      "grad_norm": 0.21440015733242035,
      "learning_rate": 4.438432272210746e-09,
      "loss": 0.4277,
      "num_input_tokens_seen": 24811272,
      "step": 37955
    },
    {
      "epoch": 19.89517819706499,
      "grad_norm": 0.1312991827726364,
      "learning_rate": 4.225595171114538e-09,
      "loss": 0.4595,
      "num_input_tokens_seen": 24815048,
      "step": 37960
    },
    {
      "epoch": 19.897798742138363,
      "grad_norm": 0.1342458873987198,
      "learning_rate": 4.017986900858861e-09,
      "loss": 0.5223,
      "num_input_tokens_seen": 24818120,
      "step": 37965
    },
    {
      "epoch": 19.90041928721174,
      "grad_norm": 0.2281801849603653,
      "learning_rate": 3.815607504870089e-09,
      "loss": 0.4592,
      "num_input_tokens_seen": 24822280,
      "step": 37970
    },
    {
      "epoch": 19.903039832285117,
      "grad_norm": 0.109359510242939,
      "learning_rate": 3.618457025486577e-09,
      "loss": 0.4709,
      "num_input_tokens_seen": 24826568,
      "step": 37975
    },
    {
      "epoch": 19.90566037735849,
      "grad_norm": 0.11079663038253784,
      "learning_rate": 3.426535503947559e-09,
      "loss": 0.4862,
      "num_input_tokens_seen": 24829768,
      "step": 37980
    },
    {
      "epoch": 19.908280922431867,
      "grad_norm": 0.5207259058952332,
      "learning_rate": 3.2398429803986997e-09,
      "loss": 0.4457,
      "num_input_tokens_seen": 24832392,
      "step": 37985
    },
    {
      "epoch": 19.91090146750524,
      "grad_norm": 0.2809367775917053,
      "learning_rate": 3.058379493900421e-09,
      "loss": 0.6043,
      "num_input_tokens_seen": 24834856,
      "step": 37990
    },
    {
      "epoch": 19.913522012578618,
      "grad_norm": 0.12115887552499771,
      "learning_rate": 2.882145082405696e-09,
      "loss": 0.653,
      "num_input_tokens_seen": 24838248,
      "step": 37995
    },
    {
      "epoch": 19.91614255765199,
      "grad_norm": 0.2670835256576538,
      "learning_rate": 2.711139782785033e-09,
      "loss": 0.4212,
      "num_input_tokens_seen": 24841864,
      "step": 38000
    },
    {
      "epoch": 19.918763102725368,
      "grad_norm": 0.21714776754379272,
      "learning_rate": 2.5453636308098163e-09,
      "loss": 0.4719,
      "num_input_tokens_seen": 24844776,
      "step": 38005
    },
    {
      "epoch": 19.92138364779874,
      "grad_norm": 0.17086894810199738,
      "learning_rate": 2.3848166611606383e-09,
      "loss": 0.502,
      "num_input_tokens_seen": 24848360,
      "step": 38010
    },
    {
      "epoch": 19.92400419287212,
      "grad_norm": 0.20069195330142975,
      "learning_rate": 2.2294989074161943e-09,
      "loss": 0.5348,
      "num_input_tokens_seen": 24851496,
      "step": 38015
    },
    {
      "epoch": 19.92662473794549,
      "grad_norm": 0.14896352589130402,
      "learning_rate": 2.0794104020727125e-09,
      "loss": 0.4106,
      "num_input_tokens_seen": 24855144,
      "step": 38020
    },
    {
      "epoch": 19.92924528301887,
      "grad_norm": 0.1534729301929474,
      "learning_rate": 1.9345511765272993e-09,
      "loss": 0.4534,
      "num_input_tokens_seen": 24858120,
      "step": 38025
    },
    {
      "epoch": 19.931865828092242,
      "grad_norm": 0.1799229234457016,
      "learning_rate": 1.794921261077942e-09,
      "loss": 0.5211,
      "num_input_tokens_seen": 24862248,
      "step": 38030
    },
    {
      "epoch": 19.93448637316562,
      "grad_norm": 0.09142064303159714,
      "learning_rate": 1.6605206849373833e-09,
      "loss": 0.5108,
      "num_input_tokens_seen": 24864776,
      "step": 38035
    },
    {
      "epoch": 19.937106918238992,
      "grad_norm": 0.1622813642024994,
      "learning_rate": 1.531349476219246e-09,
      "loss": 0.4933,
      "num_input_tokens_seen": 24868200,
      "step": 38040
    },
    {
      "epoch": 19.93972746331237,
      "grad_norm": 0.15348295867443085,
      "learning_rate": 1.4074076619463584e-09,
      "loss": 0.4651,
      "num_input_tokens_seen": 24870824,
      "step": 38045
    },
    {
      "epoch": 19.942348008385743,
      "grad_norm": 0.14982031285762787,
      "learning_rate": 1.2886952680452036e-09,
      "loss": 0.4877,
      "num_input_tokens_seen": 24874056,
      "step": 38050
    },
    {
      "epoch": 19.94496855345912,
      "grad_norm": 0.2632051706314087,
      "learning_rate": 1.1752123193459197e-09,
      "loss": 0.405,
      "num_input_tokens_seen": 24877096,
      "step": 38055
    },
    {
      "epoch": 19.947589098532493,
      "grad_norm": 0.1754467934370041,
      "learning_rate": 1.0669588395934017e-09,
      "loss": 0.4559,
      "num_input_tokens_seen": 24880104,
      "step": 38060
    },
    {
      "epoch": 19.95020964360587,
      "grad_norm": 0.15379731357097626,
      "learning_rate": 9.639348514278722e-10,
      "loss": 0.551,
      "num_input_tokens_seen": 24883112,
      "step": 38065
    },
    {
      "epoch": 19.952830188679247,
      "grad_norm": 0.15631844103336334,
      "learning_rate": 8.661403764043119e-10,
      "loss": 0.4899,
      "num_input_tokens_seen": 24886824,
      "step": 38070
    },
    {
      "epoch": 19.95545073375262,
      "grad_norm": 0.26567378640174866,
      "learning_rate": 7.735754349785795e-10,
      "loss": 0.4706,
      "num_input_tokens_seen": 24889096,
      "step": 38075
    },
    {
      "epoch": 19.958071278825997,
      "grad_norm": 0.16255298256874084,
      "learning_rate": 6.862400465157403e-10,
      "loss": 0.4507,
      "num_input_tokens_seen": 24893640,
      "step": 38080
    },
    {
      "epoch": 19.96069182389937,
      "grad_norm": 0.19397079944610596,
      "learning_rate": 6.041342292817387e-10,
      "loss": 0.5175,
      "num_input_tokens_seen": 24896264,
      "step": 38085
    },
    {
      "epoch": 19.963312368972748,
      "grad_norm": 0.14237205684185028,
      "learning_rate": 5.272580004572758e-10,
      "loss": 0.4459,
      "num_input_tokens_seen": 24899080,
      "step": 38090
    },
    {
      "epoch": 19.96593291404612,
      "grad_norm": 0.19796358048915863,
      "learning_rate": 4.556113761211567e-10,
      "loss": 0.4577,
      "num_input_tokens_seen": 24902024,
      "step": 38095
    },
    {
      "epoch": 19.968553459119498,
      "grad_norm": 0.16886959969997406,
      "learning_rate": 3.8919437125861657e-10,
      "loss": 0.42,
      "num_input_tokens_seen": 24906088,
      "step": 38100
    },
    {
      "epoch": 19.97117400419287,
      "grad_norm": 0.3357365131378174,
      "learning_rate": 3.280069997696478e-10,
      "loss": 0.5363,
      "num_input_tokens_seen": 24908968,
      "step": 38105
    },
    {
      "epoch": 19.97379454926625,
      "grad_norm": 0.16267715394496918,
      "learning_rate": 2.720492744467951e-10,
      "loss": 0.5019,
      "num_input_tokens_seen": 24912104,
      "step": 38110
    },
    {
      "epoch": 19.97641509433962,
      "grad_norm": 0.2149432897567749,
      "learning_rate": 2.2132120700291136e-10,
      "loss": 0.5302,
      "num_input_tokens_seen": 24915464,
      "step": 38115
    },
    {
      "epoch": 19.979035639413,
      "grad_norm": 0.188021719455719,
      "learning_rate": 1.7582280804340213e-10,
      "loss": 0.4026,
      "num_input_tokens_seen": 24918632,
      "step": 38120
    },
    {
      "epoch": 19.981656184486372,
      "grad_norm": 0.19746297597885132,
      "learning_rate": 1.3555408708842977e-10,
      "loss": 0.5699,
      "num_input_tokens_seen": 24921032,
      "step": 38125
    },
    {
      "epoch": 19.98427672955975,
      "grad_norm": 0.16712085902690887,
      "learning_rate": 1.0051505256458704e-10,
      "loss": 0.4789,
      "num_input_tokens_seen": 24923912,
      "step": 38130
    },
    {
      "epoch": 19.986897274633122,
      "grad_norm": 0.15796568989753723,
      "learning_rate": 7.070571179657037e-11,
      "loss": 0.518,
      "num_input_tokens_seen": 24926952,
      "step": 38135
    },
    {
      "epoch": 19.9895178197065,
      "grad_norm": 0.0933530181646347,
      "learning_rate": 4.612607102383315e-11,
      "loss": 0.5527,
      "num_input_tokens_seen": 24930984,
      "step": 38140
    },
    {
      "epoch": 19.992138364779873,
      "grad_norm": 0.15408851206302643,
      "learning_rate": 2.677613538670798e-11,
      "loss": 0.6305,
      "num_input_tokens_seen": 24933832,
      "step": 38145
    },
    {
      "epoch": 19.99475890985325,
      "grad_norm": 0.13851699233055115,
      "learning_rate": 1.2655908931957782e-11,
      "loss": 0.4765,
      "num_input_tokens_seen": 24936584,
      "step": 38150
    },
    {
      "epoch": 19.997379454926623,
      "grad_norm": 0.43802228569984436,
      "learning_rate": 3.765394615551366e-12,
      "loss": 0.4144,
      "num_input_tokens_seen": 24939432,
      "step": 38155
    },
    {
      "epoch": 20.0,
      "grad_norm": 0.3498309850692749,
      "learning_rate": 1.0459429711229262e-13,
      "loss": 0.6589,
      "num_input_tokens_seen": 24941912,
      "step": 38160
    },
    {
      "epoch": 20.0,
      "eval_loss": 0.49166518449783325,
      "eval_runtime": 10.1418,
      "eval_samples_per_second": 83.615,
      "eval_steps_per_second": 20.904,
      "num_input_tokens_seen": 24941912,
      "step": 38160
    },
    {
      "epoch": 20.0,
      "num_input_tokens_seen": 24941912,
      "step": 38160,
      "total_flos": 1.1231524470905242e+18,
      "train_loss": 0.53017604626887,
      "train_runtime": 6987.3529,
      "train_samples_per_second": 21.837,
      "train_steps_per_second": 5.461
    }
  ],
  "logging_steps": 5,
  "max_steps": 38160,
  "num_input_tokens_seen": 24941912,
  "num_train_epochs": 20,
  "save_steps": 1908,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.1231524470905242e+18,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}