{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.410958904109589,
  "eval_steps": 500,
  "global_step": 3000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0013698630136986301,
      "grad_norm": 1.4028738737106323,
      "learning_rate": 3e-06,
      "loss": 2.2484,
      "step": 10
    },
    {
      "epoch": 0.0027397260273972603,
      "grad_norm": 1.4300401210784912,
      "learning_rate": 6.333333333333334e-06,
      "loss": 2.2741,
      "step": 20
    },
    {
      "epoch": 0.00410958904109589,
      "grad_norm": 4.389048099517822,
      "learning_rate": 9.666666666666667e-06,
      "loss": 2.3968,
      "step": 30
    },
    {
      "epoch": 0.005479452054794521,
      "grad_norm": 1.3157492876052856,
      "learning_rate": 1.3000000000000001e-05,
      "loss": 2.347,
      "step": 40
    },
    {
      "epoch": 0.00684931506849315,
      "grad_norm": 1.290838360786438,
      "learning_rate": 1.6333333333333335e-05,
      "loss": 2.0741,
      "step": 50
    },
    {
      "epoch": 0.00821917808219178,
      "grad_norm": 1.7722171545028687,
      "learning_rate": 1.9666666666666666e-05,
      "loss": 1.7538,
      "step": 60
    },
    {
      "epoch": 0.009589041095890411,
      "grad_norm": 1.3361868858337402,
      "learning_rate": 2.3000000000000003e-05,
      "loss": 1.8606,
      "step": 70
    },
    {
      "epoch": 0.010958904109589041,
      "grad_norm": 1.5917797088623047,
      "learning_rate": 2.633333333333333e-05,
      "loss": 2.0042,
      "step": 80
    },
    {
      "epoch": 0.012328767123287671,
      "grad_norm": 1.829978346824646,
      "learning_rate": 2.9666666666666672e-05,
      "loss": 1.848,
      "step": 90
    },
    {
      "epoch": 0.0136986301369863,
      "grad_norm": 1.3388415575027466,
      "learning_rate": 3.3e-05,
      "loss": 1.8417,
      "step": 100
    },
    {
      "epoch": 0.015068493150684932,
      "grad_norm": 1.4123870134353638,
      "learning_rate": 3.633333333333333e-05,
      "loss": 2.3187,
      "step": 110
    },
    {
      "epoch": 0.01643835616438356,
      "grad_norm": 1.5437504053115845,
      "learning_rate": 3.966666666666667e-05,
      "loss": 1.8643,
      "step": 120
    },
    {
      "epoch": 0.01780821917808219,
      "grad_norm": 1.496188998222351,
      "learning_rate": 4.3e-05,
      "loss": 1.6831,
      "step": 130
    },
    {
      "epoch": 0.019178082191780823,
      "grad_norm": 2.099055290222168,
      "learning_rate": 4.633333333333333e-05,
      "loss": 1.693,
      "step": 140
    },
    {
      "epoch": 0.02054794520547945,
      "grad_norm": 3.8160312175750732,
      "learning_rate": 4.966666666666667e-05,
      "loss": 1.7158,
      "step": 150
    },
    {
      "epoch": 0.021917808219178082,
      "grad_norm": 1.1084319353103638,
      "learning_rate": 5.300000000000001e-05,
      "loss": 1.78,
      "step": 160
    },
    {
      "epoch": 0.023287671232876714,
      "grad_norm": 1.714733600616455,
      "learning_rate": 5.633333333333334e-05,
      "loss": 1.8776,
      "step": 170
    },
    {
      "epoch": 0.024657534246575342,
      "grad_norm": 1.0134049654006958,
      "learning_rate": 5.966666666666667e-05,
      "loss": 1.9483,
      "step": 180
    },
    {
      "epoch": 0.026027397260273973,
      "grad_norm": 0.7298793792724609,
      "learning_rate": 6.3e-05,
      "loss": 1.7299,
      "step": 190
    },
    {
      "epoch": 0.0273972602739726,
      "grad_norm": 1.709596872329712,
      "learning_rate": 6.633333333333334e-05,
      "loss": 2.1109,
      "step": 200
    },
    {
      "epoch": 0.028767123287671233,
      "grad_norm": 1.8335596323013306,
      "learning_rate": 6.966666666666668e-05,
      "loss": 1.7318,
      "step": 210
    },
    {
      "epoch": 0.030136986301369864,
      "grad_norm": 2.340693235397339,
      "learning_rate": 7.3e-05,
      "loss": 1.9166,
      "step": 220
    },
    {
      "epoch": 0.031506849315068496,
      "grad_norm": 2.241060972213745,
      "learning_rate": 7.633333333333334e-05,
      "loss": 1.8311,
      "step": 230
    },
    {
      "epoch": 0.03287671232876712,
      "grad_norm": 0.9591296911239624,
      "learning_rate": 7.966666666666666e-05,
      "loss": 1.8488,
      "step": 240
    },
    {
      "epoch": 0.03424657534246575,
      "grad_norm": 1.4175859689712524,
      "learning_rate": 8.3e-05,
      "loss": 1.7258,
      "step": 250
    },
    {
      "epoch": 0.03561643835616438,
      "grad_norm": 2.6265478134155273,
      "learning_rate": 8.633333333333334e-05,
      "loss": 1.7014,
      "step": 260
    },
    {
      "epoch": 0.036986301369863014,
      "grad_norm": 1.3878326416015625,
      "learning_rate": 8.966666666666666e-05,
      "loss": 1.8881,
      "step": 270
    },
    {
      "epoch": 0.038356164383561646,
      "grad_norm": 1.3768271207809448,
      "learning_rate": 9.300000000000001e-05,
      "loss": 2.0887,
      "step": 280
    },
    {
      "epoch": 0.03972602739726028,
      "grad_norm": 1.9178138971328735,
      "learning_rate": 9.633333333333335e-05,
      "loss": 1.5802,
      "step": 290
    },
    {
      "epoch": 0.0410958904109589,
      "grad_norm": 2.208899736404419,
      "learning_rate": 9.966666666666667e-05,
      "loss": 2.2524,
      "step": 300
    },
    {
      "epoch": 0.04246575342465753,
      "grad_norm": 1.5130765438079834,
      "learning_rate": 9.999725846827562e-05,
      "loss": 1.9414,
      "step": 310
    },
    {
      "epoch": 0.043835616438356165,
      "grad_norm": 1.3225808143615723,
      "learning_rate": 9.998778195446311e-05,
      "loss": 1.8956,
      "step": 320
    },
    {
      "epoch": 0.045205479452054796,
      "grad_norm": 2.186032772064209,
      "learning_rate": 9.997153789515461e-05,
      "loss": 1.9144,
      "step": 330
    },
    {
      "epoch": 0.04657534246575343,
      "grad_norm": 1.4805994033813477,
      "learning_rate": 9.994852848953574e-05,
      "loss": 1.8424,
      "step": 340
    },
    {
      "epoch": 0.04794520547945205,
      "grad_norm": 1.9020469188690186,
      "learning_rate": 9.991875685271168e-05,
      "loss": 1.9322,
      "step": 350
    },
    {
      "epoch": 0.049315068493150684,
      "grad_norm": 1.9721382856369019,
      "learning_rate": 9.988222701528547e-05,
      "loss": 1.7567,
      "step": 360
    },
    {
      "epoch": 0.050684931506849315,
      "grad_norm": 2.7525320053100586,
      "learning_rate": 9.983894392281237e-05,
      "loss": 1.7599,
      "step": 370
    },
    {
      "epoch": 0.052054794520547946,
      "grad_norm": 1.7699042558670044,
      "learning_rate": 9.978891343513023e-05,
      "loss": 1.8447,
      "step": 380
    },
    {
      "epoch": 0.05342465753424658,
      "grad_norm": 1.6476136445999146,
      "learning_rate": 9.973214232556622e-05,
      "loss": 1.8282,
      "step": 390
    },
    {
      "epoch": 0.0547945205479452,
      "grad_norm": 1.5267293453216553,
      "learning_rate": 9.966863828001982e-05,
      "loss": 1.823,
      "step": 400
    },
    {
      "epoch": 0.056164383561643834,
      "grad_norm": 3.080139636993408,
      "learning_rate": 9.959840989592226e-05,
      "loss": 1.8589,
      "step": 410
    },
    {
      "epoch": 0.057534246575342465,
      "grad_norm": 1.95379638671875,
      "learning_rate": 9.952146668107254e-05,
      "loss": 1.7409,
      "step": 420
    },
    {
      "epoch": 0.0589041095890411,
      "grad_norm": 1.5778534412384033,
      "learning_rate": 9.94378190523503e-05,
      "loss": 1.7094,
      "step": 430
    },
    {
      "epoch": 0.06027397260273973,
      "grad_norm": 1.2501163482666016,
      "learning_rate": 9.934747833430547e-05,
      "loss": 1.7797,
      "step": 440
    },
    {
      "epoch": 0.06164383561643835,
      "grad_norm": 2.6678783893585205,
      "learning_rate": 9.925045675762514e-05,
      "loss": 1.7278,
      "step": 450
    },
    {
      "epoch": 0.06301369863013699,
      "grad_norm": 2.3310539722442627,
      "learning_rate": 9.914676745747772e-05,
      "loss": 1.9377,
      "step": 460
    },
    {
      "epoch": 0.06438356164383562,
      "grad_norm": 2.122758150100708,
      "learning_rate": 9.903642447173465e-05,
      "loss": 1.6575,
      "step": 470
    },
    {
      "epoch": 0.06575342465753424,
      "grad_norm": 1.6059435606002808,
      "learning_rate": 9.891944273906986e-05,
      "loss": 1.6817,
      "step": 480
    },
    {
      "epoch": 0.06712328767123288,
      "grad_norm": 2.2808139324188232,
      "learning_rate": 9.879583809693738e-05,
      "loss": 1.6041,
      "step": 490
    },
    {
      "epoch": 0.0684931506849315,
      "grad_norm": 3.285200595855713,
      "learning_rate": 9.866562727942714e-05,
      "loss": 2.1497,
      "step": 500
    },
    {
      "epoch": 0.06986301369863014,
      "grad_norm": 2.0148391723632812,
      "learning_rate": 9.85288279149995e-05,
      "loss": 1.9,
      "step": 510
    },
    {
      "epoch": 0.07123287671232877,
      "grad_norm": 3.0683975219726562,
      "learning_rate": 9.838545852409857e-05,
      "loss": 1.9064,
      "step": 520
    },
    {
      "epoch": 0.07260273972602739,
      "grad_norm": 1.8953275680541992,
      "learning_rate": 9.823553851664489e-05,
      "loss": 2.0165,
      "step": 530
    },
    {
      "epoch": 0.07397260273972603,
      "grad_norm": 1.1835041046142578,
      "learning_rate": 9.807908818940761e-05,
      "loss": 1.7635,
      "step": 540
    },
    {
      "epoch": 0.07534246575342465,
      "grad_norm": 2.961548089981079,
      "learning_rate": 9.791612872325667e-05,
      "loss": 1.976,
      "step": 550
    },
    {
      "epoch": 0.07671232876712329,
      "grad_norm": 1.2752394676208496,
      "learning_rate": 9.77466821802952e-05,
      "loss": 1.5748,
      "step": 560
    },
    {
      "epoch": 0.07808219178082192,
      "grad_norm": 1.5959957838058472,
      "learning_rate": 9.75707715008727e-05,
      "loss": 1.5759,
      "step": 570
    },
    {
      "epoch": 0.07945205479452055,
      "grad_norm": 1.7095568180084229,
      "learning_rate": 9.73884205004793e-05,
      "loss": 1.6938,
      "step": 580
    },
    {
      "epoch": 0.08082191780821918,
      "grad_norm": 2.9951586723327637,
      "learning_rate": 9.719965386652141e-05,
      "loss": 1.7049,
      "step": 590
    },
    {
      "epoch": 0.0821917808219178,
      "grad_norm": 1.3047794103622437,
      "learning_rate": 9.700449715497961e-05,
      "loss": 1.7304,
      "step": 600
    },
    {
      "epoch": 0.08356164383561644,
      "grad_norm": 2.5035972595214844,
      "learning_rate": 9.680297678694867e-05,
      "loss": 1.6454,
      "step": 610
    },
    {
      "epoch": 0.08493150684931507,
      "grad_norm": 1.5420308113098145,
      "learning_rate": 9.659512004506057e-05,
      "loss": 1.8747,
      "step": 620
    },
    {
      "epoch": 0.0863013698630137,
      "grad_norm": 2.6465156078338623,
      "learning_rate": 9.63809550697909e-05,
      "loss": 1.2745,
      "step": 630
    },
    {
      "epoch": 0.08767123287671233,
      "grad_norm": 2.176868200302124,
      "learning_rate": 9.616051085564906e-05,
      "loss": 1.8517,
      "step": 640
    },
    {
      "epoch": 0.08904109589041095,
      "grad_norm": 2.929405689239502,
      "learning_rate": 9.593381724725285e-05,
      "loss": 1.7283,
      "step": 650
    },
    {
      "epoch": 0.09041095890410959,
      "grad_norm": 1.6962724924087524,
      "learning_rate": 9.570090493528809e-05,
      "loss": 1.7032,
      "step": 660
    },
    {
      "epoch": 0.09178082191780822,
      "grad_norm": 1.4863412380218506,
      "learning_rate": 9.546180545235344e-05,
      "loss": 1.7415,
      "step": 670
    },
    {
      "epoch": 0.09315068493150686,
      "grad_norm": 1.5905004739761353,
      "learning_rate": 9.52165511686915e-05,
      "loss": 1.9648,
      "step": 680
    },
    {
      "epoch": 0.09452054794520548,
      "grad_norm": 1.0756253004074097,
      "learning_rate": 9.496517528780637e-05,
      "loss": 1.7415,
      "step": 690
    },
    {
      "epoch": 0.0958904109589041,
      "grad_norm": 2.223599910736084,
      "learning_rate": 9.47077118419684e-05,
      "loss": 1.6522,
      "step": 700
    },
    {
      "epoch": 0.09726027397260274,
      "grad_norm": 3.08367657661438,
      "learning_rate": 9.444419568760684e-05,
      "loss": 1.7242,
      "step": 710
    },
    {
      "epoch": 0.09863013698630137,
      "grad_norm": 2.0376291275024414,
      "learning_rate": 9.417466250059073e-05,
      "loss": 1.8525,
      "step": 720
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.322006940841675,
      "learning_rate": 9.389914877139903e-05,
      "loss": 1.8933,
      "step": 730
    },
    {
      "epoch": 0.10136986301369863,
      "grad_norm": 2.1793291568756104,
      "learning_rate": 9.361769180018038e-05,
      "loss": 1.4555,
      "step": 740
    },
    {
      "epoch": 0.10273972602739725,
      "grad_norm": 1.3473883867263794,
      "learning_rate": 9.333032969170326e-05,
      "loss": 1.6996,
      "step": 750
    },
    {
      "epoch": 0.10410958904109589,
      "grad_norm": 1.4925732612609863,
      "learning_rate": 9.30371013501972e-05,
      "loss": 2.0061,
      "step": 760
    },
    {
      "epoch": 0.10547945205479452,
      "grad_norm": 2.43156099319458,
      "learning_rate": 9.273804647408575e-05,
      "loss": 2.0925,
      "step": 770
    },
    {
      "epoch": 0.10684931506849316,
      "grad_norm": 4.304801940917969,
      "learning_rate": 9.243320555061205e-05,
      "loss": 1.4975,
      "step": 780
    },
    {
      "epoch": 0.10821917808219178,
      "grad_norm": 1.838025689125061,
      "learning_rate": 9.212261985035739e-05,
      "loss": 1.8231,
      "step": 790
    },
    {
      "epoch": 0.1095890410958904,
      "grad_norm": 3.2455596923828125,
      "learning_rate": 9.180633142165384e-05,
      "loss": 1.5903,
      "step": 800
    },
    {
      "epoch": 0.11095890410958904,
      "grad_norm": 1.955100178718567,
      "learning_rate": 9.148438308489168e-05,
      "loss": 1.7339,
      "step": 810
    },
    {
      "epoch": 0.11232876712328767,
      "grad_norm": 1.5595628023147583,
      "learning_rate": 9.11568184267221e-05,
      "loss": 1.4822,
      "step": 820
    },
    {
      "epoch": 0.1136986301369863,
      "grad_norm": 2.0874927043914795,
      "learning_rate": 9.082368179415632e-05,
      "loss": 1.6121,
      "step": 830
    },
    {
      "epoch": 0.11506849315068493,
      "grad_norm": 1.7389984130859375,
      "learning_rate": 9.04850182885617e-05,
      "loss": 1.6468,
      "step": 840
    },
    {
      "epoch": 0.11643835616438356,
      "grad_norm": 2.019792318344116,
      "learning_rate": 9.014087375955573e-05,
      "loss": 1.4844,
      "step": 850
    },
    {
      "epoch": 0.1178082191780822,
      "grad_norm": 1.5820057392120361,
      "learning_rate": 8.979129479879873e-05,
      "loss": 1.6963,
      "step": 860
    },
    {
      "epoch": 0.11917808219178082,
      "grad_norm": 2.9591360092163086,
      "learning_rate": 8.943632873368611e-05,
      "loss": 1.6382,
      "step": 870
    },
    {
      "epoch": 0.12054794520547946,
      "grad_norm": 1.5505706071853638,
      "learning_rate": 8.907602362094094e-05,
      "loss": 1.8158,
      "step": 880
    },
    {
      "epoch": 0.12191780821917808,
      "grad_norm": 1.7320066690444946,
      "learning_rate": 8.871042824010791e-05,
      "loss": 1.3332,
      "step": 890
    },
    {
      "epoch": 0.1232876712328767,
      "grad_norm": 3.1696863174438477,
      "learning_rate": 8.833959208694929e-05,
      "loss": 1.6221,
      "step": 900
    },
    {
      "epoch": 0.12465753424657534,
      "grad_norm": 2.0459401607513428,
      "learning_rate": 8.796356536674403e-05,
      "loss": 1.7272,
      "step": 910
    },
    {
      "epoch": 0.12602739726027398,
      "grad_norm": 4.202193737030029,
      "learning_rate": 8.758239898749085e-05,
      "loss": 1.7802,
      "step": 920
    },
    {
      "epoch": 0.1273972602739726,
      "grad_norm": 2.30110239982605,
      "learning_rate": 8.719614455301593e-05,
      "loss": 1.6778,
      "step": 930
    },
    {
      "epoch": 0.12876712328767123,
      "grad_norm": 3.40289044380188,
      "learning_rate": 8.680485435598673e-05,
      "loss": 1.4947,
      "step": 940
    },
    {
      "epoch": 0.13013698630136986,
      "grad_norm": 2.5461068153381348,
      "learning_rate": 8.640858137083232e-05,
      "loss": 1.4905,
      "step": 950
    },
    {
      "epoch": 0.13150684931506848,
      "grad_norm": 1.995659351348877,
      "learning_rate": 8.600737924657156e-05,
      "loss": 1.4755,
      "step": 960
    },
    {
      "epoch": 0.13287671232876713,
      "grad_norm": 2.8345696926116943,
      "learning_rate": 8.560130229954984e-05,
      "loss": 1.762,
      "step": 970
    },
    {
      "epoch": 0.13424657534246576,
      "grad_norm": 2.7922985553741455,
      "learning_rate": 8.519040550608546e-05,
      "loss": 1.6869,
      "step": 980
    },
    {
      "epoch": 0.13561643835616438,
      "grad_norm": 1.7801475524902344,
      "learning_rate": 8.477474449502682e-05,
      "loss": 1.8112,
      "step": 990
    },
    {
      "epoch": 0.136986301369863,
      "grad_norm": 1.9390945434570312,
      "learning_rate": 8.435437554022115e-05,
      "loss": 1.6323,
      "step": 1000
    },
    {
      "epoch": 0.13835616438356163,
      "grad_norm": 2.17783522605896,
      "learning_rate": 8.392935555289584e-05,
      "loss": 1.778,
      "step": 1010
    },
    {
      "epoch": 0.13972602739726028,
      "grad_norm": 2.8870582580566406,
      "learning_rate": 8.349974207395366e-05,
      "loss": 1.7704,
      "step": 1020
    },
    {
      "epoch": 0.1410958904109589,
      "grad_norm": 2.317627191543579,
      "learning_rate": 8.306559326618259e-05,
      "loss": 1.8989,
      "step": 1030
    },
    {
      "epoch": 0.14246575342465753,
      "grad_norm": 1.2422226667404175,
      "learning_rate": 8.26269679063816e-05,
      "loss": 1.4631,
      "step": 1040
    },
    {
      "epoch": 0.14383561643835616,
      "grad_norm": 1.404496192932129,
      "learning_rate": 8.218392537740305e-05,
      "loss": 1.8917,
      "step": 1050
    },
    {
      "epoch": 0.14520547945205478,
      "grad_norm": 2.3392229080200195,
      "learning_rate": 8.173652566011338e-05,
      "loss": 1.4527,
      "step": 1060
    },
    {
      "epoch": 0.14657534246575343,
      "grad_norm": 2.2701287269592285,
      "learning_rate": 8.128482932527255e-05,
      "loss": 1.4515,
      "step": 1070
    },
    {
      "epoch": 0.14794520547945206,
      "grad_norm": 1.7164267301559448,
      "learning_rate": 8.082889752533375e-05,
      "loss": 1.976,
      "step": 1080
    },
    {
      "epoch": 0.14931506849315068,
      "grad_norm": 1.9462798833847046,
      "learning_rate": 8.036879198616434e-05,
      "loss": 1.492,
      "step": 1090
    },
    {
      "epoch": 0.1506849315068493,
      "grad_norm": 1.8207604885101318,
      "learning_rate": 7.990457499868919e-05,
      "loss": 1.719,
      "step": 1100
    },
    {
      "epoch": 0.15205479452054796,
      "grad_norm": 2.4192392826080322,
      "learning_rate": 7.943630941045744e-05,
      "loss": 1.7396,
      "step": 1110
    },
    {
      "epoch": 0.15342465753424658,
      "grad_norm": 1.4872899055480957,
      "learning_rate": 7.896405861713394e-05,
      "loss": 1.6986,
      "step": 1120
    },
    {
      "epoch": 0.1547945205479452,
      "grad_norm": 1.2965902090072632,
      "learning_rate": 7.848788655391658e-05,
      "loss": 1.7037,
      "step": 1130
    },
    {
      "epoch": 0.15616438356164383,
      "grad_norm": 2.5743491649627686,
      "learning_rate": 7.800785768688035e-05,
      "loss": 1.7557,
      "step": 1140
    },
    {
      "epoch": 0.15753424657534246,
      "grad_norm": 1.7150269746780396,
      "learning_rate": 7.752403700424979e-05,
      "loss": 1.5882,
      "step": 1150
    },
    {
      "epoch": 0.1589041095890411,
      "grad_norm": 1.951343297958374,
      "learning_rate": 7.703649000760053e-05,
      "loss": 1.6334,
      "step": 1160
    },
    {
      "epoch": 0.16027397260273973,
      "grad_norm": 1.4758542776107788,
      "learning_rate": 7.654528270299154e-05,
      "loss": 1.4411,
      "step": 1170
    },
    {
      "epoch": 0.16164383561643836,
      "grad_norm": 2.057770013809204,
      "learning_rate": 7.605048159202883e-05,
      "loss": 1.4802,
      "step": 1180
    },
    {
      "epoch": 0.16301369863013698,
      "grad_norm": 1.9429600238800049,
      "learning_rate": 7.555215366286227e-05,
      "loss": 1.8208,
      "step": 1190
    },
    {
      "epoch": 0.1643835616438356,
      "grad_norm": 2.5410385131835938,
      "learning_rate": 7.505036638111648e-05,
      "loss": 1.6631,
      "step": 1200
    },
    {
      "epoch": 0.16575342465753426,
      "grad_norm": 2.681525468826294,
      "learning_rate": 7.454518768075704e-05,
      "loss": 1.7817,
      "step": 1210
    },
    {
      "epoch": 0.16712328767123288,
      "grad_norm": 1.9880177974700928,
      "learning_rate": 7.403668595489333e-05,
      "loss": 1.5418,
      "step": 1220
    },
    {
      "epoch": 0.1684931506849315,
      "grad_norm": 1.3528841733932495,
      "learning_rate": 7.352493004651916e-05,
      "loss": 1.4801,
      "step": 1230
    },
    {
      "epoch": 0.16986301369863013,
      "grad_norm": 1.925644040107727,
      "learning_rate": 7.300998923919259e-05,
      "loss": 1.6479,
      "step": 1240
    },
    {
      "epoch": 0.17123287671232876,
      "grad_norm": 1.71919846534729,
      "learning_rate": 7.249193324765599e-05,
      "loss": 1.6604,
      "step": 1250
    },
    {
      "epoch": 0.1726027397260274,
      "grad_norm": 3.4710135459899902,
      "learning_rate": 7.197083220839785e-05,
      "loss": 1.5677,
      "step": 1260
    },
    {
      "epoch": 0.17397260273972603,
      "grad_norm": 2.525416374206543,
      "learning_rate": 7.14467566701573e-05,
      "loss": 1.2976,
      "step": 1270
    },
    {
      "epoch": 0.17534246575342466,
      "grad_norm": 1.8520058393478394,
      "learning_rate": 7.091977758437311e-05,
      "loss": 1.8077,
      "step": 1280
    },
    {
      "epoch": 0.17671232876712328,
      "grad_norm": 2.294551372528076,
      "learning_rate": 7.038996629557783e-05,
      "loss": 1.3442,
      "step": 1290
    },
    {
      "epoch": 0.1780821917808219,
      "grad_norm": 3.1857283115386963,
      "learning_rate": 6.985739453173903e-05,
      "loss": 1.6914,
      "step": 1300
    },
    {
      "epoch": 0.17945205479452056,
      "grad_norm": 3.152991771697998,
      "learning_rate": 6.932213439454837e-05,
      "loss": 1.6722,
      "step": 1310
    },
    {
      "epoch": 0.18082191780821918,
      "grad_norm": 2.886530876159668,
      "learning_rate": 6.87842583496602e-05,
      "loss": 1.6481,
      "step": 1320
    },
    {
      "epoch": 0.1821917808219178,
      "grad_norm": 1.9449256658554077,
      "learning_rate": 6.824383921688098e-05,
      "loss": 1.7951,
      "step": 1330
    },
    {
      "epoch": 0.18356164383561643,
      "grad_norm": 2.6453981399536133,
      "learning_rate": 6.77009501603105e-05,
      "loss": 1.4426,
      "step": 1340
    },
    {
      "epoch": 0.18493150684931506,
      "grad_norm": 1.520572543144226,
      "learning_rate": 6.71556646784367e-05,
      "loss": 1.5972,
      "step": 1350
    },
    {
      "epoch": 0.1863013698630137,
      "grad_norm": 1.8179829120635986,
      "learning_rate": 6.660805659418516e-05,
      "loss": 1.539,
      "step": 1360
    },
    {
      "epoch": 0.18767123287671234,
      "grad_norm": 1.310356616973877,
      "learning_rate": 6.605820004492467e-05,
      "loss": 1.6673,
      "step": 1370
    },
    {
      "epoch": 0.18904109589041096,
      "grad_norm": 1.5523558855056763,
      "learning_rate": 6.550616947243009e-05,
      "loss": 1.7355,
      "step": 1380
    },
    {
      "epoch": 0.19041095890410958,
      "grad_norm": 2.022332191467285,
      "learning_rate": 6.495203961280434e-05,
      "loss": 2.0947,
      "step": 1390
    },
    {
      "epoch": 0.1917808219178082,
      "grad_norm": 1.9009687900543213,
      "learning_rate": 6.439588548636016e-05,
      "loss": 1.7585,
      "step": 1400
    },
    {
      "epoch": 0.19315068493150686,
      "grad_norm": 1.7422246932983398,
      "learning_rate": 6.38377823874636e-05,
      "loss": 1.8785,
      "step": 1410
    },
    {
      "epoch": 0.19452054794520549,
      "grad_norm": 2.1474485397338867,
      "learning_rate": 6.327780587434044e-05,
      "loss": 1.9686,
      "step": 1420
    },
    {
      "epoch": 0.1958904109589041,
      "grad_norm": 2.906446933746338,
      "learning_rate": 6.27160317588467e-05,
      "loss": 1.8049,
      "step": 1430
    },
    {
      "epoch": 0.19726027397260273,
      "grad_norm": 1.4890384674072266,
      "learning_rate": 6.215253609620498e-05,
      "loss": 1.5907,
      "step": 1440
    },
    {
      "epoch": 0.19863013698630136,
      "grad_norm": 2.201432943344116,
      "learning_rate": 6.158739517470786e-05,
      "loss": 1.4835,
      "step": 1450
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.544870138168335,
      "learning_rate": 6.102068550538962e-05,
      "loss": 1.498,
      "step": 1460
    },
    {
      "epoch": 0.20136986301369864,
      "grad_norm": 1.5199240446090698,
      "learning_rate": 6.045248381166783e-05,
      "loss": 1.5222,
      "step": 1470
    },
    {
      "epoch": 0.20273972602739726,
      "grad_norm": 2.430156707763672,
      "learning_rate": 5.988286701895631e-05,
      "loss": 1.7141,
      "step": 1480
    },
    {
      "epoch": 0.20410958904109588,
      "grad_norm": 3.335169792175293,
      "learning_rate": 5.9311912244250675e-05,
      "loss": 1.4424,
      "step": 1490
    },
    {
      "epoch": 0.2054794520547945,
      "grad_norm": 2.818110227584839,
      "learning_rate": 5.873969678568784e-05,
      "loss": 1.3691,
      "step": 1500
    },
    {
      "epoch": 0.20684931506849316,
      "grad_norm": 2.5193567276000977,
      "learning_rate": 5.816629811208112e-05,
      "loss": 1.9398,
      "step": 1510
    },
    {
      "epoch": 0.20821917808219179,
      "grad_norm": 1.702822208404541,
      "learning_rate": 5.759179385243224e-05,
      "loss": 1.5451,
      "step": 1520
    },
    {
      "epoch": 0.2095890410958904,
      "grad_norm": 2.1284539699554443,
      "learning_rate": 5.701626178542158e-05,
      "loss": 1.8442,
      "step": 1530
    },
    {
      "epoch": 0.21095890410958903,
      "grad_norm": 2.4425137042999268,
      "learning_rate": 5.643977982887815e-05,
      "loss": 1.5843,
      "step": 1540
    },
    {
      "epoch": 0.21232876712328766,
      "grad_norm": 2.1926443576812744,
      "learning_rate": 5.586242602923081e-05,
      "loss": 1.7854,
      "step": 1550
    },
    {
      "epoch": 0.2136986301369863,
      "grad_norm": 2.7360684871673584,
      "learning_rate": 5.528427855094206e-05,
      "loss": 1.6639,
      "step": 1560
    },
    {
      "epoch": 0.21506849315068494,
      "grad_norm": 1.9223748445510864,
      "learning_rate": 5.470541566592573e-05,
      "loss": 1.4212,
      "step": 1570
    },
    {
      "epoch": 0.21643835616438356,
      "grad_norm": 2.355661630630493,
      "learning_rate": 5.4125915742950275e-05,
      "loss": 1.6376,
      "step": 1580
    },
    {
      "epoch": 0.21780821917808219,
      "grad_norm": 3.028079032897949,
      "learning_rate": 5.354585723702893e-05,
      "loss": 1.4013,
      "step": 1590
    },
    {
      "epoch": 0.2191780821917808,
      "grad_norm": 1.155507206916809,
      "learning_rate": 5.296531867879809e-05,
      "loss": 1.3336,
      "step": 1600
    },
    {
      "epoch": 0.22054794520547946,
      "grad_norm": 1.4151854515075684,
      "learning_rate": 5.2384378663885545e-05,
      "loss": 1.4375,
      "step": 1610
    },
    {
      "epoch": 0.2219178082191781,
      "grad_norm": 1.5798234939575195,
      "learning_rate": 5.180311584226991e-05,
      "loss": 1.8073,
      "step": 1620
    },
    {
      "epoch": 0.2232876712328767,
      "grad_norm": 3.0987579822540283,
      "learning_rate": 5.1221608907632665e-05,
      "loss": 1.5686,
      "step": 1630
    },
    {
      "epoch": 0.22465753424657534,
      "grad_norm": 3.5521326065063477,
      "learning_rate": 5.063993658670425e-05,
      "loss": 1.6035,
      "step": 1640
    },
    {
      "epoch": 0.22602739726027396,
      "grad_norm": 2.7165257930755615,
      "learning_rate": 5.0058177628605795e-05,
      "loss": 1.7542,
      "step": 1650
    },
    {
      "epoch": 0.2273972602739726,
      "grad_norm": 2.1085574626922607,
      "learning_rate": 4.947641079418773e-05,
      "loss": 1.6202,
      "step": 1660
    },
    {
      "epoch": 0.22876712328767124,
      "grad_norm": 1.6660608053207397,
      "learning_rate": 4.889471484536672e-05,
      "loss": 1.3809,
      "step": 1670
    },
    {
      "epoch": 0.23013698630136986,
      "grad_norm": 1.4744638204574585,
      "learning_rate": 4.83131685344628e-05,
      "loss": 1.7246,
      "step": 1680
    },
    {
      "epoch": 0.23150684931506849,
      "grad_norm": 1.5128684043884277,
      "learning_rate": 4.773185059353732e-05,
      "loss": 1.5884,
      "step": 1690
    },
    {
      "epoch": 0.2328767123287671,
      "grad_norm": 2.4789443016052246,
      "learning_rate": 4.715083972373401e-05,
      "loss": 1.3332,
      "step": 1700
    },
    {
      "epoch": 0.23424657534246576,
      "grad_norm": 2.1762492656707764,
      "learning_rate": 4.657021458462409e-05,
      "loss": 1.3395,
      "step": 1710
    },
    {
      "epoch": 0.2356164383561644,
      "grad_norm": 2.474536180496216,
      "learning_rate": 4.599005378355706e-05,
      "loss": 1.5043,
      "step": 1720
    },
    {
      "epoch": 0.236986301369863,
      "grad_norm": 1.550284743309021,
      "learning_rate": 4.541043586501842e-05,
      "loss": 1.6574,
      "step": 1730
    },
    {
      "epoch": 0.23835616438356164,
      "grad_norm": 3.5782651901245117,
      "learning_rate": 4.4831439299996084e-05,
      "loss": 1.3442,
      "step": 1740
    },
    {
      "epoch": 0.23972602739726026,
      "grad_norm": 3.107822895050049,
      "learning_rate": 4.425314247535668e-05,
      "loss": 1.5739,
      "step": 1750
    },
    {
      "epoch": 0.2410958904109589,
      "grad_norm": 2.3097240924835205,
      "learning_rate": 4.3675623683233135e-05,
      "loss": 1.639,
      "step": 1760
    },
    {
      "epoch": 0.24246575342465754,
      "grad_norm": 3.2882018089294434,
      "learning_rate": 4.309896111042529e-05,
      "loss": 1.5524,
      "step": 1770
    },
    {
      "epoch": 0.24383561643835616,
      "grad_norm": 2.627248764038086,
      "learning_rate": 4.252323282781453e-05,
      "loss": 1.8598,
      "step": 1780
    },
    {
      "epoch": 0.2452054794520548,
      "grad_norm": 2.3391318321228027,
      "learning_rate": 4.1948516779794364e-05,
      "loss": 1.4296,
      "step": 1790
    },
    {
      "epoch": 0.2465753424657534,
      "grad_norm": 1.7528724670410156,
      "learning_rate": 4.137489077371787e-05,
      "loss": 1.7706,
      "step": 1800
    },
    {
      "epoch": 0.24794520547945206,
      "grad_norm": 2.4802026748657227,
      "learning_rate": 4.080243246936399e-05,
      "loss": 1.7099,
      "step": 1810
    },
    {
      "epoch": 0.2493150684931507,
      "grad_norm": 1.308081865310669,
      "learning_rate": 4.0231219368423466e-05,
      "loss": 1.5738,
      "step": 1820
    },
    {
      "epoch": 0.25068493150684934,
      "grad_norm": 3.3826613426208496,
      "learning_rate": 3.9661328804006475e-05,
      "loss": 1.6931,
      "step": 1830
    },
    {
      "epoch": 0.25205479452054796,
      "grad_norm": 2.954171895980835,
      "learning_rate": 3.9092837930172884e-05,
      "loss": 1.7415,
      "step": 1840
    },
    {
      "epoch": 0.2534246575342466,
      "grad_norm": 2.82603120803833,
      "learning_rate": 3.852582371148687e-05,
      "loss": 1.6954,
      "step": 1850
    },
    {
      "epoch": 0.2547945205479452,
      "grad_norm": 4.872611999511719,
      "learning_rate": 3.796036291259718e-05,
      "loss": 1.6146,
      "step": 1860
    },
    {
      "epoch": 0.25616438356164384,
      "grad_norm": 2.3178672790527344,
      "learning_rate": 3.739653208784432e-05,
      "loss": 1.5219,
      "step": 1870
    },
    {
      "epoch": 0.25753424657534246,
      "grad_norm": 1.4849178791046143,
      "learning_rate": 3.683440757089646e-05,
      "loss": 1.6899,
      "step": 1880
    },
    {
      "epoch": 0.2589041095890411,
      "grad_norm": 2.602975368499756,
      "learning_rate": 3.627406546441494e-05,
      "loss": 1.7254,
      "step": 1890
    },
    {
      "epoch": 0.2602739726027397,
      "grad_norm": 2.0159435272216797,
      "learning_rate": 3.5715581629751326e-05,
      "loss": 1.8248,
      "step": 1900
    },
    {
      "epoch": 0.26164383561643834,
      "grad_norm": 1.9132189750671387,
      "learning_rate": 3.515903167667686e-05,
      "loss": 1.6893,
      "step": 1910
    },
    {
      "epoch": 0.26301369863013696,
      "grad_norm": 1.4500598907470703,
      "learning_rate": 3.460449095314621e-05,
      "loss": 1.6949,
      "step": 1920
    },
    {
      "epoch": 0.26438356164383564,
      "grad_norm": 2.5748395919799805,
      "learning_rate": 3.40520345350965e-05,
      "loss": 1.7592,
      "step": 1930
    },
    {
      "epoch": 0.26575342465753427,
      "grad_norm": 3.0504002571105957,
      "learning_rate": 3.35017372162833e-05,
      "loss": 1.2447,
      "step": 1940
    },
    {
      "epoch": 0.2671232876712329,
      "grad_norm": 2.7209043502807617,
      "learning_rate": 3.295367349815469e-05,
      "loss": 1.5326,
      "step": 1950
    },
    {
      "epoch": 0.2684931506849315,
      "grad_norm": 2.3009421825408936,
      "learning_rate": 3.240791757976491e-05,
      "loss": 1.7541,
      "step": 1960
    },
    {
      "epoch": 0.26986301369863014,
      "grad_norm": 2.9373281002044678,
      "learning_rate": 3.186454334772916e-05,
      "loss": 1.5251,
      "step": 1970
    },
    {
      "epoch": 0.27123287671232876,
      "grad_norm": 1.99565851688385,
      "learning_rate": 3.132362436622035e-05,
      "loss": 1.8553,
      "step": 1980
    },
    {
      "epoch": 0.2726027397260274,
      "grad_norm": 2.116396427154541,
      "learning_rate": 3.078523386700982e-05,
      "loss": 1.7764,
      "step": 1990
    },
    {
      "epoch": 0.273972602739726,
      "grad_norm": 1.9144208431243896,
      "learning_rate": 3.0249444739552844e-05,
      "loss": 1.5227,
      "step": 2000
    },
    {
      "epoch": 0.27534246575342464,
      "grad_norm": 1.4267282485961914,
      "learning_rate": 2.971632952112066e-05,
      "loss": 1.5648,
      "step": 2010
    },
    {
      "epoch": 0.27671232876712326,
      "grad_norm": 1.9111319780349731,
      "learning_rate": 2.918596038697995e-05,
      "loss": 1.6373,
      "step": 2020
    },
    {
      "epoch": 0.27808219178082194,
      "grad_norm": 1.5937117338180542,
      "learning_rate": 2.86584091406216e-05,
      "loss": 1.5564,
      "step": 2030
    },
    {
      "epoch": 0.27945205479452057,
      "grad_norm": 3.2327396869659424,
      "learning_rate": 2.8133747204039574e-05,
      "loss": 1.4473,
      "step": 2040
    },
    {
      "epoch": 0.2808219178082192,
      "grad_norm": 2.090632677078247,
      "learning_rate": 2.761204560806152e-05,
      "loss": 1.4269,
      "step": 2050
    },
    {
      "epoch": 0.2821917808219178,
      "grad_norm": 1.5311341285705566,
      "learning_rate": 2.709337498273243e-05,
      "loss": 1.6679,
      "step": 2060
    },
    {
      "epoch": 0.28356164383561644,
      "grad_norm": 2.3477697372436523,
      "learning_rate": 2.65778055477523e-05,
      "loss": 1.4417,
      "step": 2070
    },
    {
      "epoch": 0.28493150684931506,
      "grad_norm": 2.742414712905884,
      "learning_rate": 2.6065407102969664e-05,
      "loss": 1.6031,
      "step": 2080
    },
    {
      "epoch": 0.2863013698630137,
      "grad_norm": 1.9928070306777954,
      "learning_rate": 2.555624901893171e-05,
      "loss": 1.7178,
      "step": 2090
    },
    {
      "epoch": 0.2876712328767123,
      "grad_norm": 2.1661367416381836,
      "learning_rate": 2.505040022749265e-05,
      "loss": 1.6949,
      "step": 2100
    },
    {
      "epoch": 0.28904109589041094,
      "grad_norm": 2.8512043952941895,
      "learning_rate": 2.4547929212481435e-05,
      "loss": 1.6156,
      "step": 2110
    },
    {
      "epoch": 0.29041095890410956,
      "grad_norm": 2.101837635040283,
      "learning_rate": 2.404890400043023e-05,
      "loss": 1.5359,
      "step": 2120
    },
    {
      "epoch": 0.29178082191780824,
      "grad_norm": 3.111027479171753,
      "learning_rate": 2.3553392151364536e-05,
      "loss": 1.5937,
      "step": 2130
    },
    {
      "epoch": 0.29315068493150687,
      "grad_norm": 1.679614782333374,
      "learning_rate": 2.3061460749656844e-05,
      "loss": 1.2465,
      "step": 2140
    },
    {
      "epoch": 0.2945205479452055,
      "grad_norm": 1.1103602647781372,
      "learning_rate": 2.2573176394944328e-05,
      "loss": 1.6416,
      "step": 2150
    },
    {
      "epoch": 0.2958904109589041,
      "grad_norm": 2.2801201343536377,
      "learning_rate": 2.2088605193112383e-05,
      "loss": 1.6485,
      "step": 2160
    },
    {
      "epoch": 0.29726027397260274,
      "grad_norm": 3.090172052383423,
      "learning_rate": 2.160781274734495e-05,
      "loss": 1.3705,
      "step": 2170
    },
    {
      "epoch": 0.29863013698630136,
      "grad_norm": 1.51850426197052,
      "learning_rate": 2.1130864149242878e-05,
      "loss": 1.4144,
      "step": 2180
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7803024053573608,
      "learning_rate": 2.0657823970011618e-05,
      "loss": 1.4928,
      "step": 2190
    },
    {
      "epoch": 0.3013698630136986,
      "grad_norm": 1.7049388885498047,
      "learning_rate": 2.0188756251719203e-05,
      "loss": 1.7443,
      "step": 2200
    },
    {
      "epoch": 0.30273972602739724,
      "grad_norm": 1.9908382892608643,
      "learning_rate": 1.9723724498626105e-05,
      "loss": 1.6456,
      "step": 2210
    },
    {
      "epoch": 0.3041095890410959,
      "grad_norm": 3.0804522037506104,
      "learning_rate": 1.9262791668587676e-05,
      "loss": 1.4869,
      "step": 2220
    },
    {
      "epoch": 0.30547945205479454,
      "grad_norm": 1.7918072938919067,
      "learning_rate": 1.8806020164530702e-05,
      "loss": 1.5418,
      "step": 2230
    },
    {
      "epoch": 0.30684931506849317,
      "grad_norm": 2.762115240097046,
      "learning_rate": 1.8353471826005036e-05,
      "loss": 1.5662,
      "step": 2240
    },
    {
      "epoch": 0.3082191780821918,
      "grad_norm": 2.011409044265747,
      "learning_rate": 1.7905207920811572e-05,
      "loss": 1.6886,
      "step": 2250
    },
    {
      "epoch": 0.3095890410958904,
      "grad_norm": 2.0907769203186035,
      "learning_rate": 1.746128913670746e-05,
      "loss": 1.8852,
      "step": 2260
    },
    {
      "epoch": 0.31095890410958904,
      "grad_norm": 1.4657833576202393,
      "learning_rate": 1.7021775573190013e-05,
      "loss": 1.636,
      "step": 2270
    },
    {
      "epoch": 0.31232876712328766,
      "grad_norm": 2.9756340980529785,
      "learning_rate": 1.6586726733360237e-05,
      "loss": 1.3435,
      "step": 2280
    },
    {
      "epoch": 0.3136986301369863,
      "grad_norm": 2.4570364952087402,
      "learning_rate": 1.615620151586697e-05,
      "loss": 1.885,
      "step": 2290
    },
    {
      "epoch": 0.3150684931506849,
      "grad_norm": 2.5543649196624756,
      "learning_rate": 1.5730258206933025e-05,
      "loss": 1.5222,
      "step": 2300
    },
    {
      "epoch": 0.31643835616438354,
      "grad_norm": 1.7562202215194702,
      "learning_rate": 1.530895447246411e-05,
      "loss": 1.6443,
      "step": 2310
    },
    {
      "epoch": 0.3178082191780822,
      "grad_norm": 2.0639760494232178,
      "learning_rate": 1.4892347350241881e-05,
      "loss": 1.5368,
      "step": 2320
    },
    {
      "epoch": 0.31917808219178084,
      "grad_norm": 1.7844699621200562,
      "learning_rate": 1.448049324220181e-05,
      "loss": 1.5212,
      "step": 2330
    },
    {
      "epoch": 0.32054794520547947,
      "grad_norm": 2.314725637435913,
      "learning_rate": 1.4073447906797376e-05,
      "loss": 1.5756,
      "step": 2340
    },
    {
      "epoch": 0.3219178082191781,
      "grad_norm": 3.013345956802368,
      "learning_rate": 1.367126645145121e-05,
      "loss": 1.4461,
      "step": 2350
    },
    {
      "epoch": 0.3232876712328767,
      "grad_norm": 2.2153244018554688,
      "learning_rate": 1.327400332509442e-05,
      "loss": 1.6265,
      "step": 2360
    },
    {
      "epoch": 0.32465753424657534,
      "grad_norm": 2.4383397102355957,
      "learning_rate": 1.2881712310795118e-05,
      "loss": 1.5656,
      "step": 2370
    },
    {
      "epoch": 0.32602739726027397,
      "grad_norm": 1.7660318613052368,
      "learning_rate": 1.2494446518477022e-05,
      "loss": 1.4731,
      "step": 2380
    },
    {
      "epoch": 0.3273972602739726,
      "grad_norm": 2.3250346183776855,
      "learning_rate": 1.2112258377729274e-05,
      "loss": 1.7248,
      "step": 2390
    },
    {
      "epoch": 0.3287671232876712,
      "grad_norm": 1.7260797023773193,
      "learning_rate": 1.1735199630708222e-05,
      "loss": 1.5978,
      "step": 2400
    },
    {
      "epoch": 0.33013698630136984,
      "grad_norm": 2.4176435470581055,
      "learning_rate": 1.1363321325132447e-05,
      "loss": 1.454,
      "step": 2410
    },
    {
      "epoch": 0.3315068493150685,
      "grad_norm": 2.250056028366089,
      "learning_rate": 1.0996673807371677e-05,
      "loss": 1.2518,
      "step": 2420
    },
    {
      "epoch": 0.33287671232876714,
      "grad_norm": 2.250523328781128,
      "learning_rate": 1.0635306715630682e-05,
      "loss": 1.4483,
      "step": 2430
    },
    {
      "epoch": 0.33424657534246577,
      "grad_norm": 1.8544931411743164,
      "learning_rate": 1.0279268973229089e-05,
      "loss": 1.3334,
      "step": 2440
    },
    {
      "epoch": 0.3356164383561644,
      "grad_norm": 2.062802791595459,
      "learning_rate": 9.928608781977966e-06,
      "loss": 1.3373,
      "step": 2450
    },
    {
      "epoch": 0.336986301369863,
      "grad_norm": 2.6046059131622314,
      "learning_rate": 9.583373615653978e-06,
      "loss": 1.556,
      "step": 2460
    },
    {
      "epoch": 0.33835616438356164,
      "grad_norm": 1.7562941312789917,
      "learning_rate": 9.243610213572285e-06,
      "loss": 1.3916,
      "step": 2470
    },
    {
      "epoch": 0.33972602739726027,
      "grad_norm": 0.9646738767623901,
      "learning_rate": 8.909364574258793e-06,
      "loss": 1.4581,
      "step": 2480
    },
    {
      "epoch": 0.3410958904109589,
      "grad_norm": 1.8336948156356812,
      "learning_rate": 8.580681949222568e-06,
      "loss": 1.5921,
      "step": 2490
    },
    {
      "epoch": 0.3424657534246575,
      "grad_norm": 2.8797600269317627,
      "learning_rate": 8.257606836829678e-06,
      "loss": 1.5602,
      "step": 2500
    },
    {
      "epoch": 0.34383561643835614,
      "grad_norm": 1.4857509136199951,
      "learning_rate": 7.940182976278692e-06,
      "loss": 1.4826,
      "step": 2510
    },
    {
      "epoch": 0.3452054794520548,
      "grad_norm": 2.2163922786712646,
      "learning_rate": 7.6284533416791814e-06,
      "loss": 1.7773,
      "step": 2520
    },
    {
      "epoch": 0.34657534246575344,
      "grad_norm": 1.7907894849777222,
      "learning_rate": 7.322460136233622e-06,
      "loss": 1.4813,
      "step": 2530
    },
    {
      "epoch": 0.34794520547945207,
      "grad_norm": 2.7916197776794434,
      "learning_rate": 7.02224478652388e-06,
      "loss": 1.2967,
      "step": 2540
    },
    {
      "epoch": 0.3493150684931507,
      "grad_norm": 1.6418206691741943,
      "learning_rate": 6.727847936902543e-06,
      "loss": 1.5247,
      "step": 2550
    },
    {
      "epoch": 0.3506849315068493,
      "grad_norm": 2.005744457244873,
      "learning_rate": 6.439309443990532e-06,
      "loss": 1.332,
      "step": 2560
    },
    {
      "epoch": 0.35205479452054794,
      "grad_norm": 2.3452343940734863,
      "learning_rate": 6.1566683712809824e-06,
      "loss": 1.623,
      "step": 2570
    },
    {
      "epoch": 0.35342465753424657,
      "grad_norm": 2.700227975845337,
      "learning_rate": 5.879962983850745e-06,
      "loss": 1.7463,
      "step": 2580
    },
    {
      "epoch": 0.3547945205479452,
      "grad_norm": 1.9577980041503906,
      "learning_rate": 5.6092307431799384e-06,
      "loss": 1.4809,
      "step": 2590
    },
    {
      "epoch": 0.3561643835616438,
      "grad_norm": 2.223386287689209,
      "learning_rate": 5.34450830208017e-06,
      "loss": 1.5018,
      "step": 2600
    },
    {
      "epoch": 0.35753424657534244,
      "grad_norm": 3.3687524795532227,
      "learning_rate": 5.08583149973243e-06,
      "loss": 1.6345,
      "step": 2610
    },
    {
      "epoch": 0.3589041095890411,
      "grad_norm": 2.8207719326019287,
      "learning_rate": 4.833235356834959e-06,
      "loss": 1.7605,
      "step": 2620
    },
    {
      "epoch": 0.36027397260273974,
      "grad_norm": 1.8670095205307007,
      "learning_rate": 4.586754070862099e-06,
      "loss": 1.6041,
      "step": 2630
    },
    {
      "epoch": 0.36164383561643837,
      "grad_norm": 1.9502205848693848,
      "learning_rate": 4.346421011434382e-06,
      "loss": 1.4593,
      "step": 2640
    },
    {
      "epoch": 0.363013698630137,
      "grad_norm": 1.7307766675949097,
      "learning_rate": 4.112268715800943e-06,
      "loss": 1.6873,
      "step": 2650
    },
    {
      "epoch": 0.3643835616438356,
      "grad_norm": 1.3438829183578491,
      "learning_rate": 3.884328884434402e-06,
      "loss": 1.4745,
      "step": 2660
    },
    {
      "epoch": 0.36575342465753424,
      "grad_norm": 1.6506098508834839,
      "learning_rate": 3.6626323767391777e-06,
      "loss": 1.4283,
      "step": 2670
    },
    {
      "epoch": 0.36712328767123287,
      "grad_norm": 1.3745683431625366,
      "learning_rate": 3.4472092068735916e-06,
      "loss": 1.2547,
      "step": 2680
    },
    {
      "epoch": 0.3684931506849315,
      "grad_norm": 2.131854295730591,
      "learning_rate": 3.238088539686451e-06,
      "loss": 1.7309,
      "step": 2690
    },
    {
      "epoch": 0.3698630136986301,
      "grad_norm": 2.2386934757232666,
      "learning_rate": 3.0352986867686007e-06,
      "loss": 1.6385,
      "step": 2700
    },
    {
      "epoch": 0.37123287671232874,
      "grad_norm": 1.1820231676101685,
      "learning_rate": 2.8388671026199522e-06,
      "loss": 1.5724,
      "step": 2710
    },
    {
      "epoch": 0.3726027397260274,
      "grad_norm": 1.3294404745101929,
      "learning_rate": 2.6488203809326207e-06,
      "loss": 1.5353,
      "step": 2720
    },
    {
      "epoch": 0.37397260273972605,
      "grad_norm": 1.6606618165969849,
      "learning_rate": 2.4651842509905487e-06,
      "loss": 1.6569,
      "step": 2730
    },
    {
      "epoch": 0.37534246575342467,
      "grad_norm": 1.345475196838379,
      "learning_rate": 2.2879835741861586e-06,
      "loss": 1.4381,
      "step": 2740
    },
    {
      "epoch": 0.3767123287671233,
      "grad_norm": 2.877779483795166,
      "learning_rate": 2.1172423406545516e-06,
      "loss": 1.5604,
      "step": 2750
    },
    {
      "epoch": 0.3780821917808219,
      "grad_norm": 2.1802237033843994,
      "learning_rate": 1.9529836660256096e-06,
      "loss": 1.5397,
      "step": 2760
    },
    {
      "epoch": 0.37945205479452054,
      "grad_norm": 1.4911959171295166,
      "learning_rate": 1.7952297882945003e-06,
      "loss": 1.7633,
      "step": 2770
    },
    {
      "epoch": 0.38082191780821917,
      "grad_norm": 1.5174319744110107,
      "learning_rate": 1.6440020648110067e-06,
      "loss": 1.6316,
      "step": 2780
    },
    {
      "epoch": 0.3821917808219178,
      "grad_norm": 1.5157442092895508,
      "learning_rate": 1.4993209693881183e-06,
      "loss": 1.2263,
      "step": 2790
    },
    {
      "epoch": 0.3835616438356164,
      "grad_norm": 2.8943581581115723,
      "learning_rate": 1.3612060895301759e-06,
      "loss": 1.5867,
      "step": 2800
    },
    {
      "epoch": 0.38493150684931504,
      "grad_norm": 2.1494810581207275,
      "learning_rate": 1.2296761237810207e-06,
      "loss": 1.3249,
      "step": 2810
    },
    {
      "epoch": 0.3863013698630137,
      "grad_norm": 2.02693247795105,
      "learning_rate": 1.104748879192552e-06,
      "loss": 1.4641,
      "step": 2820
    },
    {
      "epoch": 0.38767123287671235,
      "grad_norm": 2.7323265075683594,
      "learning_rate": 9.864412689139123e-07,
      "loss": 1.4395,
      "step": 2830
    },
    {
      "epoch": 0.38904109589041097,
      "grad_norm": 1.8553956747055054,
      "learning_rate": 8.747693099017129e-07,
      "loss": 1.4012,
      "step": 2840
    },
    {
      "epoch": 0.3904109589041096,
      "grad_norm": 1.5457431077957153,
      "learning_rate": 7.697481207516289e-07,
      "loss": 1.8086,
      "step": 2850
    },
    {
      "epoch": 0.3917808219178082,
      "grad_norm": 1.3996024131774902,
      "learning_rate": 6.713919196515317e-07,
      "loss": 1.7806,
      "step": 2860
    },
    {
      "epoch": 0.39315068493150684,
      "grad_norm": 2.355863332748413,
      "learning_rate": 5.797140224566122e-07,
      "loss": 1.7083,
      "step": 2870
    },
    {
      "epoch": 0.39452054794520547,
      "grad_norm": 2.2099246978759766,
      "learning_rate": 4.947268408866113e-07,
      "loss": 1.3791,
      "step": 2880
    },
    {
      "epoch": 0.3958904109589041,
      "grad_norm": 1.2900251150131226,
      "learning_rate": 4.1644188084548063e-07,
      "loss": 1.4308,
      "step": 2890
    },
    {
      "epoch": 0.3972602739726027,
      "grad_norm": 2.1631548404693604,
      "learning_rate": 3.4486974086366253e-07,
      "loss": 1.3797,
      "step": 2900
    },
    {
      "epoch": 0.39863013698630134,
      "grad_norm": 1.1319299936294556,
      "learning_rate": 2.800201106632205e-07,
      "loss": 1.5695,
      "step": 2910
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7180445194244385,
      "learning_rate": 2.219017698460002e-07,
      "loss": 1.5461,
      "step": 2920
    },
    {
      "epoch": 0.40136986301369865,
      "grad_norm": 1.7271549701690674,
      "learning_rate": 1.7052258670501308e-07,
      "loss": 1.5524,
      "step": 2930
    },
    {
      "epoch": 0.40273972602739727,
      "grad_norm": 1.233688235282898,
      "learning_rate": 1.2588951715921116e-07,
      "loss": 1.4612,
      "step": 2940
    },
    {
      "epoch": 0.4041095890410959,
      "grad_norm": 2.4479854106903076,
      "learning_rate": 8.800860381173448e-08,
      "loss": 1.4658,
      "step": 2950
    },
    {
      "epoch": 0.4054794520547945,
      "grad_norm": 1.8242405652999878,
      "learning_rate": 5.688497513188229e-08,
      "loss": 1.5335,
      "step": 2960
    },
    {
      "epoch": 0.40684931506849314,
      "grad_norm": 1.8931853771209717,
      "learning_rate": 3.2522844760762836e-08,
      "loss": 1.266,
      "step": 2970
    },
    {
      "epoch": 0.40821917808219177,
      "grad_norm": 2.2524218559265137,
      "learning_rate": 1.4925510940844156e-08,
      "loss": 1.6866,
      "step": 2980
    },
    {
      "epoch": 0.4095890410958904,
      "grad_norm": 1.7584060430526733,
      "learning_rate": 4.095356069439005e-09,
      "loss": 1.4982,
      "step": 2990
    },
    {
      "epoch": 0.410958904109589,
      "grad_norm": 1.325352430343628,
      "learning_rate": 3.384637615733155e-11,
      "loss": 1.487,
      "step": 3000
    }
  ],
  "logging_steps": 10,
  "max_steps": 3000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 300,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.0814314512384e+17,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}