{
  "best_metric": 0.22251690924167633,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.21180831347630394,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0010590415673815197,
      "grad_norm": 6.466629981994629,
      "learning_rate": 1e-05,
      "loss": 1.2913,
      "step": 1
    },
    {
      "epoch": 0.0010590415673815197,
      "eval_loss": 1.8021806478500366,
      "eval_runtime": 67.1063,
      "eval_samples_per_second": 23.694,
      "eval_steps_per_second": 5.931,
      "step": 1
    },
    {
      "epoch": 0.0021180831347630395,
      "grad_norm": 3.6866455078125,
      "learning_rate": 2e-05,
      "loss": 1.1358,
      "step": 2
    },
    {
      "epoch": 0.003177124702144559,
      "grad_norm": 2.5750315189361572,
      "learning_rate": 3e-05,
      "loss": 1.0576,
      "step": 3
    },
    {
      "epoch": 0.004236166269526079,
      "grad_norm": 2.527364730834961,
      "learning_rate": 4e-05,
      "loss": 1.0166,
      "step": 4
    },
    {
      "epoch": 0.0052952078369075985,
      "grad_norm": 2.1124846935272217,
      "learning_rate": 5e-05,
      "loss": 0.8114,
      "step": 5
    },
    {
      "epoch": 0.006354249404289118,
      "grad_norm": 1.7748686075210571,
      "learning_rate": 6e-05,
      "loss": 0.748,
      "step": 6
    },
    {
      "epoch": 0.007413290971670638,
      "grad_norm": 1.861733317375183,
      "learning_rate": 7e-05,
      "loss": 0.6555,
      "step": 7
    },
    {
      "epoch": 0.008472332539052158,
      "grad_norm": 1.3713303804397583,
      "learning_rate": 8e-05,
      "loss": 0.5035,
      "step": 8
    },
    {
      "epoch": 0.009531374106433678,
      "grad_norm": 1.7268099784851074,
      "learning_rate": 9e-05,
      "loss": 0.7231,
      "step": 9
    },
    {
      "epoch": 0.010590415673815197,
      "grad_norm": 1.8979824781417847,
      "learning_rate": 0.0001,
      "loss": 0.5476,
      "step": 10
    },
    {
      "epoch": 0.011649457241196717,
      "grad_norm": 1.6777318716049194,
      "learning_rate": 9.999316524962345e-05,
      "loss": 0.5349,
      "step": 11
    },
    {
      "epoch": 0.012708498808578236,
      "grad_norm": 1.9419457912445068,
      "learning_rate": 9.997266286704631e-05,
      "loss": 0.6929,
      "step": 12
    },
    {
      "epoch": 0.013767540375959756,
      "grad_norm": 3.9112260341644287,
      "learning_rate": 9.993849845741524e-05,
      "loss": 0.8575,
      "step": 13
    },
    {
      "epoch": 0.014826581943341277,
      "grad_norm": 2.743795156478882,
      "learning_rate": 9.989068136093873e-05,
      "loss": 0.6012,
      "step": 14
    },
    {
      "epoch": 0.015885623510722795,
      "grad_norm": 2.5246307849884033,
      "learning_rate": 9.98292246503335e-05,
      "loss": 0.6209,
      "step": 15
    },
    {
      "epoch": 0.016944665078104316,
      "grad_norm": 1.7863917350769043,
      "learning_rate": 9.975414512725057e-05,
      "loss": 0.6629,
      "step": 16
    },
    {
      "epoch": 0.018003706645485836,
      "grad_norm": 1.5928558111190796,
      "learning_rate": 9.966546331768191e-05,
      "loss": 0.6872,
      "step": 17
    },
    {
      "epoch": 0.019062748212867357,
      "grad_norm": 5.618917465209961,
      "learning_rate": 9.956320346634876e-05,
      "loss": 0.5394,
      "step": 18
    },
    {
      "epoch": 0.020121789780248873,
      "grad_norm": 1.4306669235229492,
      "learning_rate": 9.944739353007344e-05,
      "loss": 0.2793,
      "step": 19
    },
    {
      "epoch": 0.021180831347630394,
      "grad_norm": 1.084677815437317,
      "learning_rate": 9.931806517013612e-05,
      "loss": 0.207,
      "step": 20
    },
    {
      "epoch": 0.022239872915011914,
      "grad_norm": 6.180703639984131,
      "learning_rate": 9.917525374361912e-05,
      "loss": 0.7382,
      "step": 21
    },
    {
      "epoch": 0.023298914482393435,
      "grad_norm": 3.444171667098999,
      "learning_rate": 9.901899829374047e-05,
      "loss": 0.5246,
      "step": 22
    },
    {
      "epoch": 0.024357956049774955,
      "grad_norm": 1.930855393409729,
      "learning_rate": 9.884934153917997e-05,
      "loss": 0.7433,
      "step": 23
    },
    {
      "epoch": 0.025416997617156472,
      "grad_norm": 1.6847729682922363,
      "learning_rate": 9.86663298624003e-05,
      "loss": 0.3954,
      "step": 24
    },
    {
      "epoch": 0.026476039184537992,
      "grad_norm": 1.975999116897583,
      "learning_rate": 9.847001329696653e-05,
      "loss": 0.5326,
      "step": 25
    },
    {
      "epoch": 0.027535080751919513,
      "grad_norm": 1.9394807815551758,
      "learning_rate": 9.826044551386744e-05,
      "loss": 0.559,
      "step": 26
    },
    {
      "epoch": 0.028594122319301033,
      "grad_norm": 1.390851616859436,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.6024,
      "step": 27
    },
    {
      "epoch": 0.029653163886682554,
      "grad_norm": 1.3196767568588257,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.4952,
      "step": 28
    },
    {
      "epoch": 0.03071220545406407,
      "grad_norm": 2.9127204418182373,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.6642,
      "step": 29
    },
    {
      "epoch": 0.03177124702144559,
      "grad_norm": 1.320184588432312,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.4532,
      "step": 30
    },
    {
      "epoch": 0.03283028858882711,
      "grad_norm": 1.2366173267364502,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.6775,
      "step": 31
    },
    {
      "epoch": 0.03388933015620863,
      "grad_norm": 1.341973900794983,
      "learning_rate": 9.672822322997305e-05,
      "loss": 0.4613,
      "step": 32
    },
    {
      "epoch": 0.03494837172359015,
      "grad_norm": 24.46187400817871,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.557,
      "step": 33
    },
    {
      "epoch": 0.03600741329097167,
      "grad_norm": 6.664669036865234,
      "learning_rate": 9.611448774886924e-05,
      "loss": 0.5654,
      "step": 34
    },
    {
      "epoch": 0.03706645485835319,
      "grad_norm": 5.941922187805176,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.7246,
      "step": 35
    },
    {
      "epoch": 0.03812549642573471,
      "grad_norm": 12.943635940551758,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.0159,
      "step": 36
    },
    {
      "epoch": 0.03918453799311623,
      "grad_norm": 72.04322052001953,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.3638,
      "step": 37
    },
    {
      "epoch": 0.04024357956049775,
      "grad_norm": 8.62022876739502,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.1737,
      "step": 38
    },
    {
      "epoch": 0.04130262112787927,
      "grad_norm": 5.7977213859558105,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.6908,
      "step": 39
    },
    {
      "epoch": 0.04236166269526079,
      "grad_norm": 4.385583877563477,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.7221,
      "step": 40
    },
    {
      "epoch": 0.04342070426264231,
      "grad_norm": 26.598480224609375,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.8431,
      "step": 41
    },
    {
      "epoch": 0.04447974583002383,
      "grad_norm": 4.537588119506836,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.7111,
      "step": 42
    },
    {
      "epoch": 0.04553878739740535,
      "grad_norm": 2.166567325592041,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.5642,
      "step": 43
    },
    {
      "epoch": 0.04659782896478687,
      "grad_norm": 3.8406102657318115,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.4837,
      "step": 44
    },
    {
      "epoch": 0.04765687053216839,
      "grad_norm": 1.1608014106750488,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.4859,
      "step": 45
    },
    {
      "epoch": 0.04871591209954991,
      "grad_norm": 3.252889633178711,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.6159,
      "step": 46
    },
    {
      "epoch": 0.04977495366693143,
      "grad_norm": 5.995134353637695,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.4822,
      "step": 47
    },
    {
      "epoch": 0.050833995234312944,
      "grad_norm": 2.3585846424102783,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.5257,
      "step": 48
    },
    {
      "epoch": 0.051893036801694464,
      "grad_norm": 18.709585189819336,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.5885,
      "step": 49
    },
    {
      "epoch": 0.052952078369075985,
      "grad_norm": 6.8019843101501465,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.622,
      "step": 50
    },
    {
      "epoch": 0.052952078369075985,
      "eval_loss": 0.8600424528121948,
      "eval_runtime": 68.3138,
      "eval_samples_per_second": 23.275,
      "eval_steps_per_second": 5.826,
      "step": 50
    },
    {
      "epoch": 0.054011119936457505,
      "grad_norm": 13.320000648498535,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.2674,
      "step": 51
    },
    {
      "epoch": 0.055070161503839025,
      "grad_norm": 8.385751724243164,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.0851,
      "step": 52
    },
    {
      "epoch": 0.056129203071220546,
      "grad_norm": 9.528212547302246,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.8958,
      "step": 53
    },
    {
      "epoch": 0.057188244638602066,
      "grad_norm": 9.573410987854004,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.5851,
      "step": 54
    },
    {
      "epoch": 0.05824728620598359,
      "grad_norm": 2.2730178833007812,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.4714,
      "step": 55
    },
    {
      "epoch": 0.05930632777336511,
      "grad_norm": 1.7303773164749146,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.3051,
      "step": 56
    },
    {
      "epoch": 0.06036536934074663,
      "grad_norm": 1.8023905754089355,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.3993,
      "step": 57
    },
    {
      "epoch": 0.06142441090812814,
      "grad_norm": 1.3482708930969238,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.3187,
      "step": 58
    },
    {
      "epoch": 0.06248345247550966,
      "grad_norm": 1.1417664289474487,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.3241,
      "step": 59
    },
    {
      "epoch": 0.06354249404289118,
      "grad_norm": 1.1761040687561035,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.3104,
      "step": 60
    },
    {
      "epoch": 0.06460153561027271,
      "grad_norm": 0.889615535736084,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.2541,
      "step": 61
    },
    {
      "epoch": 0.06566057717765422,
      "grad_norm": 0.9282159209251404,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.225,
      "step": 62
    },
    {
      "epoch": 0.06671961874503574,
      "grad_norm": 0.9758453965187073,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.3365,
      "step": 63
    },
    {
      "epoch": 0.06777866031241726,
      "grad_norm": 1.6036334037780762,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.451,
      "step": 64
    },
    {
      "epoch": 0.06883770187979878,
      "grad_norm": 0.8630813360214233,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.2243,
      "step": 65
    },
    {
      "epoch": 0.0698967434471803,
      "grad_norm": 0.7337073683738708,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.1872,
      "step": 66
    },
    {
      "epoch": 0.07095578501456182,
      "grad_norm": 0.8603715896606445,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.2916,
      "step": 67
    },
    {
      "epoch": 0.07201482658194334,
      "grad_norm": 0.6783121824264526,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.1535,
      "step": 68
    },
    {
      "epoch": 0.07307386814932486,
      "grad_norm": 1.063816785812378,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.1144,
      "step": 69
    },
    {
      "epoch": 0.07413290971670639,
      "grad_norm": 0.6977053284645081,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.1028,
      "step": 70
    },
    {
      "epoch": 0.0751919512840879,
      "grad_norm": 0.37291035056114197,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.08,
      "step": 71
    },
    {
      "epoch": 0.07625099285146943,
      "grad_norm": 0.7426699995994568,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.1739,
      "step": 72
    },
    {
      "epoch": 0.07731003441885094,
      "grad_norm": 1.473480224609375,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.4169,
      "step": 73
    },
    {
      "epoch": 0.07836907598623245,
      "grad_norm": 1.481460690498352,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.4291,
      "step": 74
    },
    {
      "epoch": 0.07942811755361398,
      "grad_norm": 1.4350910186767578,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.3616,
      "step": 75
    },
    {
      "epoch": 0.0804871591209955,
      "grad_norm": 1.3191903829574585,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.4038,
      "step": 76
    },
    {
      "epoch": 0.08154620068837702,
      "grad_norm": 0.9689040184020996,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.3365,
      "step": 77
    },
    {
      "epoch": 0.08260524225575853,
      "grad_norm": 1.0821114778518677,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.4324,
      "step": 78
    },
    {
      "epoch": 0.08366428382314006,
      "grad_norm": 0.9169743657112122,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.4332,
      "step": 79
    },
    {
      "epoch": 0.08472332539052158,
      "grad_norm": 1.2360223531723022,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.4482,
      "step": 80
    },
    {
      "epoch": 0.0857823669579031,
      "grad_norm": 1.0113508701324463,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.4029,
      "step": 81
    },
    {
      "epoch": 0.08684140852528462,
      "grad_norm": 0.8452385067939758,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.4975,
      "step": 82
    },
    {
      "epoch": 0.08790045009266614,
      "grad_norm": 1.172534704208374,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.4505,
      "step": 83
    },
    {
      "epoch": 0.08895949166004766,
      "grad_norm": 0.8616721034049988,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.2889,
      "step": 84
    },
    {
      "epoch": 0.09001853322742917,
      "grad_norm": 0.7209553122520447,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.3609,
      "step": 85
    },
    {
      "epoch": 0.0910775747948107,
      "grad_norm": 0.6880238056182861,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.3271,
      "step": 86
    },
    {
      "epoch": 0.09213661636219221,
      "grad_norm": 0.47881606221199036,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.257,
      "step": 87
    },
    {
      "epoch": 0.09319565792957374,
      "grad_norm": 0.7327137589454651,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.2573,
      "step": 88
    },
    {
      "epoch": 0.09425469949695525,
      "grad_norm": 0.5549834966659546,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.2742,
      "step": 89
    },
    {
      "epoch": 0.09531374106433678,
      "grad_norm": 0.6294392943382263,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.2555,
      "step": 90
    },
    {
      "epoch": 0.09637278263171829,
      "grad_norm": 0.6991058588027954,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.2769,
      "step": 91
    },
    {
      "epoch": 0.09743182419909982,
      "grad_norm": 0.5928325057029724,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.2361,
      "step": 92
    },
    {
      "epoch": 0.09849086576648133,
      "grad_norm": 0.6267572045326233,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.2361,
      "step": 93
    },
    {
      "epoch": 0.09954990733386286,
      "grad_norm": 0.7301824688911438,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.1672,
      "step": 94
    },
    {
      "epoch": 0.10060894890124437,
      "grad_norm": 0.7125401496887207,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.1885,
      "step": 95
    },
    {
      "epoch": 0.10166799046862589,
      "grad_norm": 1.2743104696273804,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.4712,
      "step": 96
    },
    {
      "epoch": 0.10272703203600742,
      "grad_norm": 1.5936224460601807,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.3883,
      "step": 97
    },
    {
      "epoch": 0.10378607360338893,
      "grad_norm": 1.1847281455993652,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.2563,
      "step": 98
    },
    {
      "epoch": 0.10484511517077046,
      "grad_norm": 0.7801738381385803,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.1774,
      "step": 99
    },
    {
      "epoch": 0.10590415673815197,
      "grad_norm": 1.6040635108947754,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.2692,
      "step": 100
    },
    {
      "epoch": 0.10590415673815197,
      "eval_loss": 0.5275586247444153,
      "eval_runtime": 68.1194,
      "eval_samples_per_second": 23.341,
      "eval_steps_per_second": 5.843,
      "step": 100
    },
    {
      "epoch": 0.1069631983055335,
      "grad_norm": 5.80922794342041,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.3767,
      "step": 101
    },
    {
      "epoch": 0.10802223987291501,
      "grad_norm": 3.241302967071533,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.6015,
      "step": 102
    },
    {
      "epoch": 0.10908128144029654,
      "grad_norm": 4.702502727508545,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.8657,
      "step": 103
    },
    {
      "epoch": 0.11014032300767805,
      "grad_norm": 5.423489570617676,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.6292,
      "step": 104
    },
    {
      "epoch": 0.11119936457505956,
      "grad_norm": 3.159705877304077,
      "learning_rate": 5e-05,
      "loss": 0.3756,
      "step": 105
    },
    {
      "epoch": 0.11225840614244109,
      "grad_norm": 0.8795123100280762,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.2365,
      "step": 106
    },
    {
      "epoch": 0.1133174477098226,
      "grad_norm": 3.298581600189209,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.3797,
      "step": 107
    },
    {
      "epoch": 0.11437648927720413,
      "grad_norm": 4.502202033996582,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.3446,
      "step": 108
    },
    {
      "epoch": 0.11543553084458565,
      "grad_norm": 1.6163389682769775,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.3028,
      "step": 109
    },
    {
      "epoch": 0.11649457241196717,
      "grad_norm": 1.044812798500061,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.2733,
      "step": 110
    },
    {
      "epoch": 0.11755361397934869,
      "grad_norm": 0.9134991765022278,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.2026,
      "step": 111
    },
    {
      "epoch": 0.11861265554673021,
      "grad_norm": 1.4023849964141846,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.2324,
      "step": 112
    },
    {
      "epoch": 0.11967169711411173,
      "grad_norm": 1.173660159111023,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.263,
      "step": 113
    },
    {
      "epoch": 0.12073073868149325,
      "grad_norm": 1.1302968263626099,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.2441,
      "step": 114
    },
    {
      "epoch": 0.12178978024887477,
      "grad_norm": 1.1289983987808228,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.1165,
      "step": 115
    },
    {
      "epoch": 0.12284882181625628,
      "grad_norm": 0.43796804547309875,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.1161,
      "step": 116
    },
    {
      "epoch": 0.12390786338363781,
      "grad_norm": 0.8190528154373169,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.1993,
      "step": 117
    },
    {
      "epoch": 0.12496690495101932,
      "grad_norm": 0.8702072501182556,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.19,
      "step": 118
    },
    {
      "epoch": 0.12602594651840085,
      "grad_norm": 0.7176076769828796,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.1215,
      "step": 119
    },
    {
      "epoch": 0.12708498808578236,
      "grad_norm": 2.6524486541748047,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.1813,
      "step": 120
    },
    {
      "epoch": 0.12814402965316388,
      "grad_norm": 2.9871771335601807,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.2169,
      "step": 121
    },
    {
      "epoch": 0.12920307122054542,
      "grad_norm": 3.62376070022583,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.2176,
      "step": 122
    },
    {
      "epoch": 0.13026211278792693,
      "grad_norm": 0.8743485808372498,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.3019,
      "step": 123
    },
    {
      "epoch": 0.13132115435530844,
      "grad_norm": 0.6141861081123352,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.2031,
      "step": 124
    },
    {
      "epoch": 0.13238019592268996,
      "grad_norm": 0.8120310306549072,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.2883,
      "step": 125
    },
    {
      "epoch": 0.13343923749007147,
      "grad_norm": 1.0505905151367188,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.3317,
      "step": 126
    },
    {
      "epoch": 0.134498279057453,
      "grad_norm": 1.2326226234436035,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.4347,
      "step": 127
    },
    {
      "epoch": 0.13555732062483453,
      "grad_norm": 0.6433725357055664,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.3062,
      "step": 128
    },
    {
      "epoch": 0.13661636219221604,
      "grad_norm": 0.7737178206443787,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.3204,
      "step": 129
    },
    {
      "epoch": 0.13767540375959755,
      "grad_norm": 0.9729801416397095,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.3586,
      "step": 130
    },
    {
      "epoch": 0.1387344453269791,
      "grad_norm": 0.7147522568702698,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.2758,
      "step": 131
    },
    {
      "epoch": 0.1397934868943606,
      "grad_norm": 0.5694999694824219,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.2216,
      "step": 132
    },
    {
      "epoch": 0.14085252846174212,
      "grad_norm": 0.6949028968811035,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.2413,
      "step": 133
    },
    {
      "epoch": 0.14191157002912363,
      "grad_norm": 0.8832529187202454,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.3362,
      "step": 134
    },
    {
      "epoch": 0.14297061159650518,
      "grad_norm": 0.7539413571357727,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.3256,
      "step": 135
    },
    {
      "epoch": 0.1440296531638867,
      "grad_norm": 0.8469157218933105,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.2638,
      "step": 136
    },
    {
      "epoch": 0.1450886947312682,
      "grad_norm": 0.5857799053192139,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.1745,
      "step": 137
    },
    {
      "epoch": 0.14614773629864972,
      "grad_norm": 0.6602780222892761,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.1931,
      "step": 138
    },
    {
      "epoch": 0.14720677786603123,
      "grad_norm": 0.6062917113304138,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.1439,
      "step": 139
    },
    {
      "epoch": 0.14826581943341277,
      "grad_norm": 0.6118988394737244,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.2357,
      "step": 140
    },
    {
      "epoch": 0.14932486100079428,
      "grad_norm": 0.5995131731033325,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.1657,
      "step": 141
    },
    {
      "epoch": 0.1503839025681758,
      "grad_norm": 0.5409533977508545,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.1981,
      "step": 142
    },
    {
      "epoch": 0.1514429441355573,
      "grad_norm": 0.8835098147392273,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.2167,
      "step": 143
    },
    {
      "epoch": 0.15250198570293885,
      "grad_norm": 0.6345929503440857,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.1838,
      "step": 144
    },
    {
      "epoch": 0.15356102727032037,
      "grad_norm": 0.6283819675445557,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.2039,
      "step": 145
    },
    {
      "epoch": 0.15462006883770188,
      "grad_norm": 0.8313515782356262,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.1767,
      "step": 146
    },
    {
      "epoch": 0.1556791104050834,
      "grad_norm": 0.9614133238792419,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.1956,
      "step": 147
    },
    {
      "epoch": 0.1567381519724649,
      "grad_norm": 1.100778341293335,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.212,
      "step": 148
    },
    {
      "epoch": 0.15779719353984645,
      "grad_norm": 0.8009694814682007,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.1721,
      "step": 149
    },
    {
      "epoch": 0.15885623510722796,
      "grad_norm": 0.7843473553657532,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.1372,
      "step": 150
    },
    {
      "epoch": 0.15885623510722796,
      "eval_loss": 0.36114931106567383,
      "eval_runtime": 68.5499,
      "eval_samples_per_second": 23.195,
      "eval_steps_per_second": 5.806,
      "step": 150
    },
    {
      "epoch": 0.15991527667460947,
      "grad_norm": 3.7045722007751465,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.1776,
      "step": 151
    },
    {
      "epoch": 0.160974318241991,
      "grad_norm": 2.4158377647399902,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.2958,
      "step": 152
    },
    {
      "epoch": 0.16203335980937253,
      "grad_norm": 4.951345443725586,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.6655,
      "step": 153
    },
    {
      "epoch": 0.16309240137675404,
      "grad_norm": 4.614811897277832,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.5613,
      "step": 154
    },
    {
      "epoch": 0.16415144294413556,
      "grad_norm": 5.582204341888428,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.5106,
      "step": 155
    },
    {
      "epoch": 0.16521048451151707,
      "grad_norm": 3.2766034603118896,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.3556,
      "step": 156
    },
    {
      "epoch": 0.16626952607889858,
      "grad_norm": 2.1364059448242188,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.3699,
      "step": 157
    },
    {
      "epoch": 0.16732856764628012,
      "grad_norm": 1.3831149339675903,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.3027,
      "step": 158
    },
    {
      "epoch": 0.16838760921366164,
      "grad_norm": 0.609874963760376,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.1605,
      "step": 159
    },
    {
      "epoch": 0.16944665078104315,
      "grad_norm": 0.7327921986579895,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.1352,
      "step": 160
    },
    {
      "epoch": 0.17050569234842466,
      "grad_norm": 0.646138072013855,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.1644,
      "step": 161
    },
    {
      "epoch": 0.1715647339158062,
      "grad_norm": 1.8058477640151978,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.1754,
      "step": 162
    },
    {
      "epoch": 0.17262377548318772,
      "grad_norm": 0.5481339693069458,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.1563,
      "step": 163
    },
    {
      "epoch": 0.17368281705056923,
      "grad_norm": 0.8295053243637085,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.132,
      "step": 164
    },
    {
      "epoch": 0.17474185861795075,
      "grad_norm": 0.9174979329109192,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.1945,
      "step": 165
    },
    {
      "epoch": 0.1758009001853323,
      "grad_norm": 0.8813605904579163,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.1583,
      "step": 166
    },
    {
      "epoch": 0.1768599417527138,
      "grad_norm": 1.1465108394622803,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.229,
      "step": 167
    },
    {
      "epoch": 0.17791898332009531,
      "grad_norm": 0.694993793964386,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.1748,
      "step": 168
    },
    {
      "epoch": 0.17897802488747683,
      "grad_norm": 1.2563068866729736,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.2038,
      "step": 169
    },
    {
      "epoch": 0.18003706645485834,
      "grad_norm": 0.905998170375824,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.1639,
      "step": 170
    },
    {
      "epoch": 0.18109610802223988,
      "grad_norm": 1.100280523300171,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.3124,
      "step": 171
    },
    {
      "epoch": 0.1821551495896214,
      "grad_norm": 1.5016824007034302,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.3384,
      "step": 172
    },
    {
      "epoch": 0.1832141911570029,
      "grad_norm": 0.9917207956314087,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.2804,
      "step": 173
    },
    {
      "epoch": 0.18427323272438442,
      "grad_norm": 0.8501930236816406,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.2914,
      "step": 174
    },
    {
      "epoch": 0.18533227429176596,
      "grad_norm": 0.8428164124488831,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.2615,
      "step": 175
    },
    {
      "epoch": 0.18639131585914748,
      "grad_norm": 0.663905918598175,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.2427,
      "step": 176
    },
    {
      "epoch": 0.187450357426529,
      "grad_norm": 0.7672511339187622,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.3332,
      "step": 177
    },
    {
      "epoch": 0.1885093989939105,
      "grad_norm": 0.7166119813919067,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.2407,
      "step": 178
    },
    {
      "epoch": 0.18956844056129202,
      "grad_norm": 0.7371949553489685,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.3839,
      "step": 179
    },
    {
      "epoch": 0.19062748212867356,
      "grad_norm": 0.9255362749099731,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.2698,
      "step": 180
    },
    {
      "epoch": 0.19168652369605507,
      "grad_norm": 0.7725550532341003,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.2499,
      "step": 181
    },
    {
      "epoch": 0.19274556526343659,
      "grad_norm": 1.1561079025268555,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.3418,
      "step": 182
    },
    {
      "epoch": 0.1938046068308181,
      "grad_norm": 0.8105995655059814,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.283,
      "step": 183
    },
    {
      "epoch": 0.19486364839819964,
      "grad_norm": 0.717684268951416,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.2568,
      "step": 184
    },
    {
      "epoch": 0.19592268996558115,
      "grad_norm": 0.582086980342865,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.1517,
      "step": 185
    },
    {
      "epoch": 0.19698173153296267,
      "grad_norm": 0.6231020092964172,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.2368,
      "step": 186
    },
    {
      "epoch": 0.19804077310034418,
      "grad_norm": 0.8702476024627686,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.3019,
      "step": 187
    },
    {
      "epoch": 0.19909981466772572,
      "grad_norm": 0.9144231081008911,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.2559,
      "step": 188
    },
    {
      "epoch": 0.20015885623510724,
      "grad_norm": 0.6421347856521606,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.2007,
      "step": 189
    },
    {
      "epoch": 0.20121789780248875,
      "grad_norm": 0.6394674777984619,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.2335,
      "step": 190
    },
    {
      "epoch": 0.20227693936987026,
      "grad_norm": 1.008832335472107,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.3647,
      "step": 191
    },
    {
      "epoch": 0.20333598093725178,
      "grad_norm": 0.9754836559295654,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.315,
      "step": 192
    },
    {
      "epoch": 0.20439502250463332,
      "grad_norm": 0.7258129715919495,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.2364,
      "step": 193
    },
    {
      "epoch": 0.20545406407201483,
      "grad_norm": 0.5731558799743652,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.1663,
      "step": 194
    },
    {
      "epoch": 0.20651310563939634,
      "grad_norm": 0.6555661559104919,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.2063,
      "step": 195
    },
    {
      "epoch": 0.20757214720677786,
      "grad_norm": 0.4508231282234192,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.1162,
      "step": 196
    },
    {
      "epoch": 0.2086311887741594,
      "grad_norm": 0.4226987063884735,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.0976,
      "step": 197
    },
    {
      "epoch": 0.2096902303415409,
      "grad_norm": 0.6174224615097046,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.139,
      "step": 198
    },
    {
      "epoch": 0.21074927190892243,
      "grad_norm": 0.8022536635398865,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.1875,
      "step": 199
    },
    {
      "epoch": 0.21180831347630394,
      "grad_norm": 0.6075817346572876,
      "learning_rate": 0.0,
      "loss": 0.1515,
      "step": 200
    },
    {
      "epoch": 0.21180831347630394,
      "eval_loss": 0.22251690924167633,
      "eval_runtime": 68.2146,
      "eval_samples_per_second": 23.309,
      "eval_steps_per_second": 5.835,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.34707043893248e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}