{
  "best_metric": 1.4909777641296387,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.1718213058419244,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.000859106529209622,
      "grad_norm": 5.928728103637695,
      "learning_rate": 1.001e-05,
      "loss": 2.3539,
      "step": 1
    },
    {
      "epoch": 0.000859106529209622,
      "eval_loss": 2.5483806133270264,
      "eval_runtime": 26.3739,
      "eval_samples_per_second": 18.579,
      "eval_steps_per_second": 4.664,
      "step": 1
    },
    {
      "epoch": 0.001718213058419244,
      "grad_norm": 4.801278114318848,
      "learning_rate": 2.002e-05,
      "loss": 2.0539,
      "step": 2
    },
    {
      "epoch": 0.002577319587628866,
      "grad_norm": 4.651398181915283,
      "learning_rate": 3.0029999999999995e-05,
      "loss": 2.2714,
      "step": 3
    },
    {
      "epoch": 0.003436426116838488,
      "grad_norm": 4.354385852813721,
      "learning_rate": 4.004e-05,
      "loss": 2.1376,
      "step": 4
    },
    {
      "epoch": 0.00429553264604811,
      "grad_norm": 3.075941324234009,
      "learning_rate": 5.005e-05,
      "loss": 2.0429,
      "step": 5
    },
    {
      "epoch": 0.005154639175257732,
      "grad_norm": 1.7578625679016113,
      "learning_rate": 6.005999999999999e-05,
      "loss": 1.9386,
      "step": 6
    },
    {
      "epoch": 0.006013745704467354,
      "grad_norm": 1.0474644899368286,
      "learning_rate": 7.006999999999998e-05,
      "loss": 1.7412,
      "step": 7
    },
    {
      "epoch": 0.006872852233676976,
      "grad_norm": 0.8305192589759827,
      "learning_rate": 8.008e-05,
      "loss": 1.8618,
      "step": 8
    },
    {
      "epoch": 0.007731958762886598,
      "grad_norm": 0.6909606456756592,
      "learning_rate": 9.009e-05,
      "loss": 1.5999,
      "step": 9
    },
    {
      "epoch": 0.00859106529209622,
      "grad_norm": 0.7085854411125183,
      "learning_rate": 0.0001001,
      "loss": 1.6194,
      "step": 10
    },
    {
      "epoch": 0.009450171821305841,
      "grad_norm": 1.100886344909668,
      "learning_rate": 9.957315789473684e-05,
      "loss": 1.932,
      "step": 11
    },
    {
      "epoch": 0.010309278350515464,
      "grad_norm": 0.9694746732711792,
      "learning_rate": 9.904631578947367e-05,
      "loss": 1.5086,
      "step": 12
    },
    {
      "epoch": 0.011168384879725086,
      "grad_norm": 0.8996667861938477,
      "learning_rate": 9.851947368421052e-05,
      "loss": 1.7042,
      "step": 13
    },
    {
      "epoch": 0.012027491408934709,
      "grad_norm": 0.7877088785171509,
      "learning_rate": 9.799263157894736e-05,
      "loss": 1.6645,
      "step": 14
    },
    {
      "epoch": 0.01288659793814433,
      "grad_norm": 0.8245639801025391,
      "learning_rate": 9.746578947368421e-05,
      "loss": 2.0033,
      "step": 15
    },
    {
      "epoch": 0.013745704467353952,
      "grad_norm": 0.6540593504905701,
      "learning_rate": 9.693894736842104e-05,
      "loss": 1.6979,
      "step": 16
    },
    {
      "epoch": 0.014604810996563574,
      "grad_norm": 0.5481818318367004,
      "learning_rate": 9.641210526315789e-05,
      "loss": 1.4844,
      "step": 17
    },
    {
      "epoch": 0.015463917525773196,
      "grad_norm": 0.5299813151359558,
      "learning_rate": 9.588526315789473e-05,
      "loss": 1.6551,
      "step": 18
    },
    {
      "epoch": 0.01632302405498282,
      "grad_norm": 0.5546098947525024,
      "learning_rate": 9.535842105263157e-05,
      "loss": 1.7145,
      "step": 19
    },
    {
      "epoch": 0.01718213058419244,
      "grad_norm": 0.5181926488876343,
      "learning_rate": 9.483157894736841e-05,
      "loss": 1.5542,
      "step": 20
    },
    {
      "epoch": 0.01804123711340206,
      "grad_norm": 0.464667409658432,
      "learning_rate": 9.430473684210526e-05,
      "loss": 1.4155,
      "step": 21
    },
    {
      "epoch": 0.018900343642611683,
      "grad_norm": 0.5198472738265991,
      "learning_rate": 9.37778947368421e-05,
      "loss": 1.5655,
      "step": 22
    },
    {
      "epoch": 0.019759450171821305,
      "grad_norm": 0.5255956053733826,
      "learning_rate": 9.325105263157894e-05,
      "loss": 1.5633,
      "step": 23
    },
    {
      "epoch": 0.020618556701030927,
      "grad_norm": 0.44441619515419006,
      "learning_rate": 9.272421052631578e-05,
      "loss": 1.5421,
      "step": 24
    },
    {
      "epoch": 0.02147766323024055,
      "grad_norm": 0.4783642888069153,
      "learning_rate": 9.219736842105263e-05,
      "loss": 1.4316,
      "step": 25
    },
    {
      "epoch": 0.022336769759450172,
      "grad_norm": 0.4691917896270752,
      "learning_rate": 9.167052631578946e-05,
      "loss": 1.5855,
      "step": 26
    },
    {
      "epoch": 0.023195876288659795,
      "grad_norm": 0.4796159863471985,
      "learning_rate": 9.114368421052632e-05,
      "loss": 1.5567,
      "step": 27
    },
    {
      "epoch": 0.024054982817869417,
      "grad_norm": 0.5819175243377686,
      "learning_rate": 9.061684210526315e-05,
      "loss": 1.7546,
      "step": 28
    },
    {
      "epoch": 0.02491408934707904,
      "grad_norm": 0.4925194978713989,
      "learning_rate": 9.009e-05,
      "loss": 1.7332,
      "step": 29
    },
    {
      "epoch": 0.02577319587628866,
      "grad_norm": 0.500887393951416,
      "learning_rate": 8.956315789473683e-05,
      "loss": 1.5155,
      "step": 30
    },
    {
      "epoch": 0.02663230240549828,
      "grad_norm": 0.48832571506500244,
      "learning_rate": 8.903631578947368e-05,
      "loss": 1.4386,
      "step": 31
    },
    {
      "epoch": 0.027491408934707903,
      "grad_norm": 0.45185863971710205,
      "learning_rate": 8.850947368421052e-05,
      "loss": 1.5028,
      "step": 32
    },
    {
      "epoch": 0.028350515463917526,
      "grad_norm": 0.49141839146614075,
      "learning_rate": 8.798263157894736e-05,
      "loss": 1.642,
      "step": 33
    },
    {
      "epoch": 0.029209621993127148,
      "grad_norm": 0.44430410861968994,
      "learning_rate": 8.745578947368422e-05,
      "loss": 1.3989,
      "step": 34
    },
    {
      "epoch": 0.03006872852233677,
      "grad_norm": 0.43278783559799194,
      "learning_rate": 8.692894736842105e-05,
      "loss": 1.3636,
      "step": 35
    },
    {
      "epoch": 0.030927835051546393,
      "grad_norm": 0.46601980924606323,
      "learning_rate": 8.64021052631579e-05,
      "loss": 1.6403,
      "step": 36
    },
    {
      "epoch": 0.03178694158075601,
      "grad_norm": 0.46934980154037476,
      "learning_rate": 8.587526315789473e-05,
      "loss": 1.6551,
      "step": 37
    },
    {
      "epoch": 0.03264604810996564,
      "grad_norm": 0.44036510586738586,
      "learning_rate": 8.534842105263157e-05,
      "loss": 1.5409,
      "step": 38
    },
    {
      "epoch": 0.03350515463917526,
      "grad_norm": 0.43330177664756775,
      "learning_rate": 8.482157894736842e-05,
      "loss": 1.4959,
      "step": 39
    },
    {
      "epoch": 0.03436426116838488,
      "grad_norm": 0.6163046956062317,
      "learning_rate": 8.429473684210525e-05,
      "loss": 1.6538,
      "step": 40
    },
    {
      "epoch": 0.0352233676975945,
      "grad_norm": 0.4682713747024536,
      "learning_rate": 8.376789473684211e-05,
      "loss": 1.3941,
      "step": 41
    },
    {
      "epoch": 0.03608247422680412,
      "grad_norm": 0.5424863696098328,
      "learning_rate": 8.324105263157894e-05,
      "loss": 1.541,
      "step": 42
    },
    {
      "epoch": 0.036941580756013746,
      "grad_norm": 0.5334774851799011,
      "learning_rate": 8.271421052631579e-05,
      "loss": 1.5979,
      "step": 43
    },
    {
      "epoch": 0.037800687285223365,
      "grad_norm": 0.6095847487449646,
      "learning_rate": 8.218736842105262e-05,
      "loss": 1.562,
      "step": 44
    },
    {
      "epoch": 0.03865979381443299,
      "grad_norm": 0.6060227155685425,
      "learning_rate": 8.166052631578947e-05,
      "loss": 1.4832,
      "step": 45
    },
    {
      "epoch": 0.03951890034364261,
      "grad_norm": 0.5880366563796997,
      "learning_rate": 8.113368421052631e-05,
      "loss": 1.7103,
      "step": 46
    },
    {
      "epoch": 0.040378006872852236,
      "grad_norm": 0.5826221108436584,
      "learning_rate": 8.060684210526315e-05,
      "loss": 1.6033,
      "step": 47
    },
    {
      "epoch": 0.041237113402061855,
      "grad_norm": 0.6990689635276794,
      "learning_rate": 8.008e-05,
      "loss": 1.7349,
      "step": 48
    },
    {
      "epoch": 0.04209621993127148,
      "grad_norm": 0.80436772108078,
      "learning_rate": 7.955315789473684e-05,
      "loss": 1.7074,
      "step": 49
    },
    {
      "epoch": 0.0429553264604811,
      "grad_norm": 0.8325445652008057,
      "learning_rate": 7.902631578947368e-05,
      "loss": 1.891,
      "step": 50
    },
    {
      "epoch": 0.0429553264604811,
      "eval_loss": 1.5498371124267578,
      "eval_runtime": 24.9498,
      "eval_samples_per_second": 19.639,
      "eval_steps_per_second": 4.93,
      "step": 50
    },
    {
      "epoch": 0.04381443298969072,
      "grad_norm": 0.44505757093429565,
      "learning_rate": 7.849947368421052e-05,
      "loss": 1.4996,
      "step": 51
    },
    {
      "epoch": 0.044673539518900345,
      "grad_norm": 0.4838855266571045,
      "learning_rate": 7.797263157894736e-05,
      "loss": 1.5216,
      "step": 52
    },
    {
      "epoch": 0.04553264604810996,
      "grad_norm": 0.4800682067871094,
      "learning_rate": 7.744578947368421e-05,
      "loss": 1.5385,
      "step": 53
    },
    {
      "epoch": 0.04639175257731959,
      "grad_norm": 0.44465377926826477,
      "learning_rate": 7.691894736842104e-05,
      "loss": 1.4122,
      "step": 54
    },
    {
      "epoch": 0.04725085910652921,
      "grad_norm": 0.4586678743362427,
      "learning_rate": 7.63921052631579e-05,
      "loss": 1.6508,
      "step": 55
    },
    {
      "epoch": 0.048109965635738834,
      "grad_norm": 0.42472296953201294,
      "learning_rate": 7.586526315789473e-05,
      "loss": 1.4994,
      "step": 56
    },
    {
      "epoch": 0.04896907216494845,
      "grad_norm": 0.4265405237674713,
      "learning_rate": 7.533842105263158e-05,
      "loss": 1.3472,
      "step": 57
    },
    {
      "epoch": 0.04982817869415808,
      "grad_norm": 0.4571130573749542,
      "learning_rate": 7.481157894736841e-05,
      "loss": 1.4216,
      "step": 58
    },
    {
      "epoch": 0.0506872852233677,
      "grad_norm": 0.43544867634773254,
      "learning_rate": 7.428473684210526e-05,
      "loss": 1.5216,
      "step": 59
    },
    {
      "epoch": 0.05154639175257732,
      "grad_norm": 0.4616435766220093,
      "learning_rate": 7.375789473684209e-05,
      "loss": 1.5369,
      "step": 60
    },
    {
      "epoch": 0.05240549828178694,
      "grad_norm": 0.44542810320854187,
      "learning_rate": 7.323105263157895e-05,
      "loss": 1.4106,
      "step": 61
    },
    {
      "epoch": 0.05326460481099656,
      "grad_norm": 0.48202329874038696,
      "learning_rate": 7.270421052631578e-05,
      "loss": 1.548,
      "step": 62
    },
    {
      "epoch": 0.05412371134020619,
      "grad_norm": 0.4340774416923523,
      "learning_rate": 7.217736842105263e-05,
      "loss": 1.4627,
      "step": 63
    },
    {
      "epoch": 0.054982817869415807,
      "grad_norm": 0.47386494278907776,
      "learning_rate": 7.165052631578947e-05,
      "loss": 1.5337,
      "step": 64
    },
    {
      "epoch": 0.05584192439862543,
      "grad_norm": 0.5032212138175964,
      "learning_rate": 7.11236842105263e-05,
      "loss": 1.9408,
      "step": 65
    },
    {
      "epoch": 0.05670103092783505,
      "grad_norm": 0.4906267821788788,
      "learning_rate": 7.059684210526315e-05,
      "loss": 1.7887,
      "step": 66
    },
    {
      "epoch": 0.05756013745704467,
      "grad_norm": 0.4755704998970032,
      "learning_rate": 7.006999999999998e-05,
      "loss": 1.6393,
      "step": 67
    },
    {
      "epoch": 0.058419243986254296,
      "grad_norm": 0.42872437834739685,
      "learning_rate": 6.954315789473684e-05,
      "loss": 1.3954,
      "step": 68
    },
    {
      "epoch": 0.059278350515463915,
      "grad_norm": 0.42157912254333496,
      "learning_rate": 6.901631578947368e-05,
      "loss": 1.4267,
      "step": 69
    },
    {
      "epoch": 0.06013745704467354,
      "grad_norm": 0.48660025000572205,
      "learning_rate": 6.848947368421052e-05,
      "loss": 1.8621,
      "step": 70
    },
    {
      "epoch": 0.06099656357388316,
      "grad_norm": 0.4611688256263733,
      "learning_rate": 6.796263157894737e-05,
      "loss": 1.5259,
      "step": 71
    },
    {
      "epoch": 0.061855670103092786,
      "grad_norm": 0.4612456262111664,
      "learning_rate": 6.74357894736842e-05,
      "loss": 1.6302,
      "step": 72
    },
    {
      "epoch": 0.0627147766323024,
      "grad_norm": 0.4128530025482178,
      "learning_rate": 6.690894736842105e-05,
      "loss": 1.3599,
      "step": 73
    },
    {
      "epoch": 0.06357388316151202,
      "grad_norm": 0.46616268157958984,
      "learning_rate": 6.638210526315788e-05,
      "loss": 1.8427,
      "step": 74
    },
    {
      "epoch": 0.06443298969072164,
      "grad_norm": 0.4137905240058899,
      "learning_rate": 6.585526315789474e-05,
      "loss": 1.3743,
      "step": 75
    },
    {
      "epoch": 0.06529209621993128,
      "grad_norm": 0.41269126534461975,
      "learning_rate": 6.532842105263157e-05,
      "loss": 1.4494,
      "step": 76
    },
    {
      "epoch": 0.0661512027491409,
      "grad_norm": 0.4382306933403015,
      "learning_rate": 6.480157894736842e-05,
      "loss": 1.5754,
      "step": 77
    },
    {
      "epoch": 0.06701030927835051,
      "grad_norm": 0.43855199217796326,
      "learning_rate": 6.427473684210526e-05,
      "loss": 1.5385,
      "step": 78
    },
    {
      "epoch": 0.06786941580756013,
      "grad_norm": 0.43043795228004456,
      "learning_rate": 6.37478947368421e-05,
      "loss": 1.4418,
      "step": 79
    },
    {
      "epoch": 0.06872852233676977,
      "grad_norm": 0.41860294342041016,
      "learning_rate": 6.322105263157894e-05,
      "loss": 1.4541,
      "step": 80
    },
    {
      "epoch": 0.06958762886597938,
      "grad_norm": 0.4255295395851135,
      "learning_rate": 6.269421052631577e-05,
      "loss": 1.5295,
      "step": 81
    },
    {
      "epoch": 0.070446735395189,
      "grad_norm": 0.41935423016548157,
      "learning_rate": 6.216736842105263e-05,
      "loss": 1.4867,
      "step": 82
    },
    {
      "epoch": 0.07130584192439862,
      "grad_norm": 0.4140067398548126,
      "learning_rate": 6.164052631578947e-05,
      "loss": 1.5371,
      "step": 83
    },
    {
      "epoch": 0.07216494845360824,
      "grad_norm": 0.42022502422332764,
      "learning_rate": 6.111368421052631e-05,
      "loss": 1.3979,
      "step": 84
    },
    {
      "epoch": 0.07302405498281787,
      "grad_norm": 0.4209689795970917,
      "learning_rate": 6.058684210526315e-05,
      "loss": 1.5414,
      "step": 85
    },
    {
      "epoch": 0.07388316151202749,
      "grad_norm": 0.4335344135761261,
      "learning_rate": 6.005999999999999e-05,
      "loss": 1.3725,
      "step": 86
    },
    {
      "epoch": 0.07474226804123711,
      "grad_norm": 0.4103253483772278,
      "learning_rate": 5.953315789473684e-05,
      "loss": 1.3898,
      "step": 87
    },
    {
      "epoch": 0.07560137457044673,
      "grad_norm": 0.4868164360523224,
      "learning_rate": 5.9006315789473676e-05,
      "loss": 1.5759,
      "step": 88
    },
    {
      "epoch": 0.07646048109965636,
      "grad_norm": 0.4783855080604553,
      "learning_rate": 5.847947368421053e-05,
      "loss": 1.4645,
      "step": 89
    },
    {
      "epoch": 0.07731958762886598,
      "grad_norm": 0.5467380881309509,
      "learning_rate": 5.795263157894737e-05,
      "loss": 1.642,
      "step": 90
    },
    {
      "epoch": 0.0781786941580756,
      "grad_norm": 0.5185117721557617,
      "learning_rate": 5.742578947368421e-05,
      "loss": 1.6723,
      "step": 91
    },
    {
      "epoch": 0.07903780068728522,
      "grad_norm": 0.535789966583252,
      "learning_rate": 5.6898947368421046e-05,
      "loss": 1.619,
      "step": 92
    },
    {
      "epoch": 0.07989690721649484,
      "grad_norm": 0.5576356649398804,
      "learning_rate": 5.6372105263157886e-05,
      "loss": 1.6906,
      "step": 93
    },
    {
      "epoch": 0.08075601374570447,
      "grad_norm": 0.5344556570053101,
      "learning_rate": 5.584526315789473e-05,
      "loss": 1.6378,
      "step": 94
    },
    {
      "epoch": 0.08161512027491409,
      "grad_norm": 0.5366571545600891,
      "learning_rate": 5.531842105263158e-05,
      "loss": 1.3028,
      "step": 95
    },
    {
      "epoch": 0.08247422680412371,
      "grad_norm": 0.5787448883056641,
      "learning_rate": 5.4791578947368424e-05,
      "loss": 1.6638,
      "step": 96
    },
    {
      "epoch": 0.08333333333333333,
      "grad_norm": 0.5679211616516113,
      "learning_rate": 5.426473684210526e-05,
      "loss": 1.4092,
      "step": 97
    },
    {
      "epoch": 0.08419243986254296,
      "grad_norm": 0.6531189680099487,
      "learning_rate": 5.37378947368421e-05,
      "loss": 1.4579,
      "step": 98
    },
    {
      "epoch": 0.08505154639175258,
      "grad_norm": 0.8260589838027954,
      "learning_rate": 5.321105263157894e-05,
      "loss": 1.5204,
      "step": 99
    },
    {
      "epoch": 0.0859106529209622,
      "grad_norm": 0.7999553084373474,
      "learning_rate": 5.268421052631578e-05,
      "loss": 1.6669,
      "step": 100
    },
    {
      "epoch": 0.0859106529209622,
      "eval_loss": 1.5160658359527588,
      "eval_runtime": 24.1171,
      "eval_samples_per_second": 20.318,
      "eval_steps_per_second": 5.1,
      "step": 100
    },
    {
      "epoch": 0.08676975945017182,
      "grad_norm": 0.48756492137908936,
      "learning_rate": 5.2157368421052626e-05,
      "loss": 1.4015,
      "step": 101
    },
    {
      "epoch": 0.08762886597938144,
      "grad_norm": 0.42037105560302734,
      "learning_rate": 5.163052631578947e-05,
      "loss": 1.4429,
      "step": 102
    },
    {
      "epoch": 0.08848797250859107,
      "grad_norm": 0.4923235774040222,
      "learning_rate": 5.110368421052632e-05,
      "loss": 1.4581,
      "step": 103
    },
    {
      "epoch": 0.08934707903780069,
      "grad_norm": 0.41760286688804626,
      "learning_rate": 5.057684210526316e-05,
      "loss": 1.1997,
      "step": 104
    },
    {
      "epoch": 0.09020618556701031,
      "grad_norm": 0.4583994746208191,
      "learning_rate": 5.005e-05,
      "loss": 1.4086,
      "step": 105
    },
    {
      "epoch": 0.09106529209621993,
      "grad_norm": 0.458413302898407,
      "learning_rate": 4.9523157894736836e-05,
      "loss": 1.5256,
      "step": 106
    },
    {
      "epoch": 0.09192439862542956,
      "grad_norm": 0.4747663140296936,
      "learning_rate": 4.899631578947368e-05,
      "loss": 1.3945,
      "step": 107
    },
    {
      "epoch": 0.09278350515463918,
      "grad_norm": 0.4569973647594452,
      "learning_rate": 4.846947368421052e-05,
      "loss": 1.4371,
      "step": 108
    },
    {
      "epoch": 0.0936426116838488,
      "grad_norm": 0.4393562972545624,
      "learning_rate": 4.794263157894737e-05,
      "loss": 1.4525,
      "step": 109
    },
    {
      "epoch": 0.09450171821305842,
      "grad_norm": 0.4615718424320221,
      "learning_rate": 4.7415789473684206e-05,
      "loss": 1.578,
      "step": 110
    },
    {
      "epoch": 0.09536082474226804,
      "grad_norm": 0.44367459416389465,
      "learning_rate": 4.688894736842105e-05,
      "loss": 1.4314,
      "step": 111
    },
    {
      "epoch": 0.09621993127147767,
      "grad_norm": 0.4110322594642639,
      "learning_rate": 4.636210526315789e-05,
      "loss": 1.4254,
      "step": 112
    },
    {
      "epoch": 0.09707903780068729,
      "grad_norm": 0.4521294832229614,
      "learning_rate": 4.583526315789473e-05,
      "loss": 1.5936,
      "step": 113
    },
    {
      "epoch": 0.0979381443298969,
      "grad_norm": 0.4364003837108612,
      "learning_rate": 4.530842105263158e-05,
      "loss": 1.5987,
      "step": 114
    },
    {
      "epoch": 0.09879725085910653,
      "grad_norm": 0.4320337772369385,
      "learning_rate": 4.4781578947368416e-05,
      "loss": 1.5928,
      "step": 115
    },
    {
      "epoch": 0.09965635738831616,
      "grad_norm": 0.47091159224510193,
      "learning_rate": 4.425473684210526e-05,
      "loss": 1.5915,
      "step": 116
    },
    {
      "epoch": 0.10051546391752578,
      "grad_norm": 0.4349591135978699,
      "learning_rate": 4.372789473684211e-05,
      "loss": 1.4902,
      "step": 117
    },
    {
      "epoch": 0.1013745704467354,
      "grad_norm": 0.4385061264038086,
      "learning_rate": 4.320105263157895e-05,
      "loss": 1.6705,
      "step": 118
    },
    {
      "epoch": 0.10223367697594501,
      "grad_norm": 0.4168137311935425,
      "learning_rate": 4.2674210526315786e-05,
      "loss": 1.5591,
      "step": 119
    },
    {
      "epoch": 0.10309278350515463,
      "grad_norm": 0.42430955171585083,
      "learning_rate": 4.2147368421052626e-05,
      "loss": 1.4457,
      "step": 120
    },
    {
      "epoch": 0.10395189003436427,
      "grad_norm": 0.4482705295085907,
      "learning_rate": 4.162052631578947e-05,
      "loss": 1.6385,
      "step": 121
    },
    {
      "epoch": 0.10481099656357389,
      "grad_norm": 0.4236373007297516,
      "learning_rate": 4.109368421052631e-05,
      "loss": 1.5653,
      "step": 122
    },
    {
      "epoch": 0.1056701030927835,
      "grad_norm": 0.4247753620147705,
      "learning_rate": 4.056684210526316e-05,
      "loss": 1.4705,
      "step": 123
    },
    {
      "epoch": 0.10652920962199312,
      "grad_norm": 0.4102475345134735,
      "learning_rate": 4.004e-05,
      "loss": 1.387,
      "step": 124
    },
    {
      "epoch": 0.10738831615120274,
      "grad_norm": 0.4251536428928375,
      "learning_rate": 3.951315789473684e-05,
      "loss": 1.5281,
      "step": 125
    },
    {
      "epoch": 0.10824742268041238,
      "grad_norm": 0.41173189878463745,
      "learning_rate": 3.898631578947368e-05,
      "loss": 1.5701,
      "step": 126
    },
    {
      "epoch": 0.109106529209622,
      "grad_norm": 0.4324643015861511,
      "learning_rate": 3.845947368421052e-05,
      "loss": 1.4983,
      "step": 127
    },
    {
      "epoch": 0.10996563573883161,
      "grad_norm": 0.3974796235561371,
      "learning_rate": 3.7932631578947367e-05,
      "loss": 1.4384,
      "step": 128
    },
    {
      "epoch": 0.11082474226804123,
      "grad_norm": 0.43069207668304443,
      "learning_rate": 3.7405789473684206e-05,
      "loss": 1.5198,
      "step": 129
    },
    {
      "epoch": 0.11168384879725086,
      "grad_norm": 0.4067608416080475,
      "learning_rate": 3.6878947368421045e-05,
      "loss": 1.4449,
      "step": 130
    },
    {
      "epoch": 0.11254295532646048,
      "grad_norm": 0.38891565799713135,
      "learning_rate": 3.635210526315789e-05,
      "loss": 1.3167,
      "step": 131
    },
    {
      "epoch": 0.1134020618556701,
      "grad_norm": 0.41567280888557434,
      "learning_rate": 3.582526315789474e-05,
      "loss": 1.4248,
      "step": 132
    },
    {
      "epoch": 0.11426116838487972,
      "grad_norm": 0.44299566745758057,
      "learning_rate": 3.5298421052631576e-05,
      "loss": 1.5865,
      "step": 133
    },
    {
      "epoch": 0.11512027491408934,
      "grad_norm": 0.3968340754508972,
      "learning_rate": 3.477157894736842e-05,
      "loss": 1.5147,
      "step": 134
    },
    {
      "epoch": 0.11597938144329897,
      "grad_norm": 0.40769246220588684,
      "learning_rate": 3.424473684210526e-05,
      "loss": 1.5635,
      "step": 135
    },
    {
      "epoch": 0.11683848797250859,
      "grad_norm": 0.4221096336841583,
      "learning_rate": 3.37178947368421e-05,
      "loss": 1.447,
      "step": 136
    },
    {
      "epoch": 0.11769759450171821,
      "grad_norm": 0.38482198119163513,
      "learning_rate": 3.319105263157894e-05,
      "loss": 1.3877,
      "step": 137
    },
    {
      "epoch": 0.11855670103092783,
      "grad_norm": 0.4175458550453186,
      "learning_rate": 3.2664210526315786e-05,
      "loss": 1.5137,
      "step": 138
    },
    {
      "epoch": 0.11941580756013746,
      "grad_norm": 0.4658557176589966,
      "learning_rate": 3.213736842105263e-05,
      "loss": 1.366,
      "step": 139
    },
    {
      "epoch": 0.12027491408934708,
      "grad_norm": 0.4771043658256531,
      "learning_rate": 3.161052631578947e-05,
      "loss": 1.4614,
      "step": 140
    },
    {
      "epoch": 0.1211340206185567,
      "grad_norm": 0.42562994360923767,
      "learning_rate": 3.108368421052632e-05,
      "loss": 1.1766,
      "step": 141
    },
    {
      "epoch": 0.12199312714776632,
      "grad_norm": 0.45292311906814575,
      "learning_rate": 3.0556842105263156e-05,
      "loss": 1.3532,
      "step": 142
    },
    {
      "epoch": 0.12285223367697594,
      "grad_norm": 0.5051841735839844,
      "learning_rate": 3.0029999999999995e-05,
      "loss": 1.501,
      "step": 143
    },
    {
      "epoch": 0.12371134020618557,
      "grad_norm": 0.525606095790863,
      "learning_rate": 2.9503157894736838e-05,
      "loss": 1.4933,
      "step": 144
    },
    {
      "epoch": 0.12457044673539519,
      "grad_norm": 0.500237762928009,
      "learning_rate": 2.8976315789473684e-05,
      "loss": 1.2868,
      "step": 145
    },
    {
      "epoch": 0.1254295532646048,
      "grad_norm": 0.5657179355621338,
      "learning_rate": 2.8449473684210523e-05,
      "loss": 1.5876,
      "step": 146
    },
    {
      "epoch": 0.12628865979381443,
      "grad_norm": 0.5619215369224548,
      "learning_rate": 2.7922631578947366e-05,
      "loss": 1.4058,
      "step": 147
    },
    {
      "epoch": 0.12714776632302405,
      "grad_norm": 0.5765167474746704,
      "learning_rate": 2.7395789473684212e-05,
      "loss": 1.6314,
      "step": 148
    },
    {
      "epoch": 0.12800687285223367,
      "grad_norm": 0.77131587266922,
      "learning_rate": 2.686894736842105e-05,
      "loss": 1.9087,
      "step": 149
    },
    {
      "epoch": 0.12886597938144329,
      "grad_norm": 0.7527775168418884,
      "learning_rate": 2.634210526315789e-05,
      "loss": 1.4388,
      "step": 150
    },
    {
      "epoch": 0.12886597938144329,
      "eval_loss": 1.4959875345230103,
      "eval_runtime": 24.189,
      "eval_samples_per_second": 20.257,
      "eval_steps_per_second": 5.085,
      "step": 150
    },
    {
      "epoch": 0.12972508591065293,
      "grad_norm": 0.4515070617198944,
      "learning_rate": 2.5815263157894736e-05,
      "loss": 1.4754,
      "step": 151
    },
    {
      "epoch": 0.13058419243986255,
      "grad_norm": 0.3967503011226654,
      "learning_rate": 2.528842105263158e-05,
      "loss": 1.163,
      "step": 152
    },
    {
      "epoch": 0.13144329896907217,
      "grad_norm": 0.4591720998287201,
      "learning_rate": 2.4761578947368418e-05,
      "loss": 1.3497,
      "step": 153
    },
    {
      "epoch": 0.1323024054982818,
      "grad_norm": 0.47611069679260254,
      "learning_rate": 2.423473684210526e-05,
      "loss": 1.2411,
      "step": 154
    },
    {
      "epoch": 0.1331615120274914,
      "grad_norm": 0.445859432220459,
      "learning_rate": 2.3707894736842103e-05,
      "loss": 1.3745,
      "step": 155
    },
    {
      "epoch": 0.13402061855670103,
      "grad_norm": 0.4517122507095337,
      "learning_rate": 2.3181052631578946e-05,
      "loss": 1.4243,
      "step": 156
    },
    {
      "epoch": 0.13487972508591065,
      "grad_norm": 0.48769956827163696,
      "learning_rate": 2.265421052631579e-05,
      "loss": 1.4979,
      "step": 157
    },
    {
      "epoch": 0.13573883161512026,
      "grad_norm": 0.3892180323600769,
      "learning_rate": 2.212736842105263e-05,
      "loss": 1.368,
      "step": 158
    },
    {
      "epoch": 0.13659793814432988,
      "grad_norm": 0.4427521228790283,
      "learning_rate": 2.1600526315789474e-05,
      "loss": 1.6416,
      "step": 159
    },
    {
      "epoch": 0.13745704467353953,
      "grad_norm": 0.41149961948394775,
      "learning_rate": 2.1073684210526313e-05,
      "loss": 1.3097,
      "step": 160
    },
    {
      "epoch": 0.13831615120274915,
      "grad_norm": 0.44827204942703247,
      "learning_rate": 2.0546842105263155e-05,
      "loss": 1.488,
      "step": 161
    },
    {
      "epoch": 0.13917525773195877,
      "grad_norm": 0.4299844801425934,
      "learning_rate": 2.002e-05,
      "loss": 1.4034,
      "step": 162
    },
    {
      "epoch": 0.1400343642611684,
      "grad_norm": 0.4039885997772217,
      "learning_rate": 1.949315789473684e-05,
      "loss": 1.3447,
      "step": 163
    },
    {
      "epoch": 0.140893470790378,
      "grad_norm": 0.43283891677856445,
      "learning_rate": 1.8966315789473683e-05,
      "loss": 1.3755,
      "step": 164
    },
    {
      "epoch": 0.14175257731958762,
      "grad_norm": 0.4442145526409149,
      "learning_rate": 1.8439473684210522e-05,
      "loss": 1.6471,
      "step": 165
    },
    {
      "epoch": 0.14261168384879724,
      "grad_norm": 0.42544904351234436,
      "learning_rate": 1.791263157894737e-05,
      "loss": 1.5591,
      "step": 166
    },
    {
      "epoch": 0.14347079037800686,
      "grad_norm": 0.4591159522533417,
      "learning_rate": 1.738578947368421e-05,
      "loss": 1.4695,
      "step": 167
    },
    {
      "epoch": 0.14432989690721648,
      "grad_norm": 0.4475899636745453,
      "learning_rate": 1.685894736842105e-05,
      "loss": 1.7011,
      "step": 168
    },
    {
      "epoch": 0.14518900343642613,
      "grad_norm": 0.3987146019935608,
      "learning_rate": 1.6332105263157893e-05,
      "loss": 1.3711,
      "step": 169
    },
    {
      "epoch": 0.14604810996563575,
      "grad_norm": 0.4292147159576416,
      "learning_rate": 1.5805263157894735e-05,
      "loss": 1.5364,
      "step": 170
    },
    {
      "epoch": 0.14690721649484537,
      "grad_norm": 0.4621577262878418,
      "learning_rate": 1.5278421052631578e-05,
      "loss": 1.5781,
      "step": 171
    },
    {
      "epoch": 0.14776632302405499,
      "grad_norm": 0.4481081962585449,
      "learning_rate": 1.4751578947368419e-05,
      "loss": 1.5406,
      "step": 172
    },
    {
      "epoch": 0.1486254295532646,
      "grad_norm": 0.4390103220939636,
      "learning_rate": 1.4224736842105262e-05,
      "loss": 1.549,
      "step": 173
    },
    {
      "epoch": 0.14948453608247422,
      "grad_norm": 0.41087478399276733,
      "learning_rate": 1.3697894736842106e-05,
      "loss": 1.4282,
      "step": 174
    },
    {
      "epoch": 0.15034364261168384,
      "grad_norm": 0.4323715269565582,
      "learning_rate": 1.3171052631578945e-05,
      "loss": 1.421,
      "step": 175
    },
    {
      "epoch": 0.15120274914089346,
      "grad_norm": 0.4265320897102356,
      "learning_rate": 1.264421052631579e-05,
      "loss": 1.4488,
      "step": 176
    },
    {
      "epoch": 0.15206185567010308,
      "grad_norm": 0.4511622190475464,
      "learning_rate": 1.211736842105263e-05,
      "loss": 1.5736,
      "step": 177
    },
    {
      "epoch": 0.15292096219931273,
      "grad_norm": 0.417791485786438,
      "learning_rate": 1.1590526315789473e-05,
      "loss": 1.4141,
      "step": 178
    },
    {
      "epoch": 0.15378006872852235,
      "grad_norm": 0.4172487258911133,
      "learning_rate": 1.1063684210526316e-05,
      "loss": 1.3984,
      "step": 179
    },
    {
      "epoch": 0.15463917525773196,
      "grad_norm": 0.42654547095298767,
      "learning_rate": 1.0536842105263156e-05,
      "loss": 1.4239,
      "step": 180
    },
    {
      "epoch": 0.15549828178694158,
      "grad_norm": 0.4155035614967346,
      "learning_rate": 1.001e-05,
      "loss": 1.51,
      "step": 181
    },
    {
      "epoch": 0.1563573883161512,
      "grad_norm": 0.39767947793006897,
      "learning_rate": 9.483157894736842e-06,
      "loss": 1.4256,
      "step": 182
    },
    {
      "epoch": 0.15721649484536082,
      "grad_norm": 0.4175090789794922,
      "learning_rate": 8.956315789473684e-06,
      "loss": 1.3716,
      "step": 183
    },
    {
      "epoch": 0.15807560137457044,
      "grad_norm": 0.44680315256118774,
      "learning_rate": 8.429473684210525e-06,
      "loss": 1.5732,
      "step": 184
    },
    {
      "epoch": 0.15893470790378006,
      "grad_norm": 0.4153236150741577,
      "learning_rate": 7.902631578947368e-06,
      "loss": 1.5334,
      "step": 185
    },
    {
      "epoch": 0.15979381443298968,
      "grad_norm": 0.4097844660282135,
      "learning_rate": 7.3757894736842095e-06,
      "loss": 1.4617,
      "step": 186
    },
    {
      "epoch": 0.16065292096219932,
      "grad_norm": 0.44263848662376404,
      "learning_rate": 6.848947368421053e-06,
      "loss": 1.5138,
      "step": 187
    },
    {
      "epoch": 0.16151202749140894,
      "grad_norm": 0.3853647708892822,
      "learning_rate": 6.322105263157895e-06,
      "loss": 1.3983,
      "step": 188
    },
    {
      "epoch": 0.16237113402061856,
      "grad_norm": 0.4370990991592407,
      "learning_rate": 5.7952631578947365e-06,
      "loss": 1.4268,
      "step": 189
    },
    {
      "epoch": 0.16323024054982818,
      "grad_norm": 0.437597393989563,
      "learning_rate": 5.268421052631578e-06,
      "loss": 1.3218,
      "step": 190
    },
    {
      "epoch": 0.1640893470790378,
      "grad_norm": 0.49196720123291016,
      "learning_rate": 4.741578947368421e-06,
      "loss": 1.5127,
      "step": 191
    },
    {
      "epoch": 0.16494845360824742,
      "grad_norm": 0.450184166431427,
      "learning_rate": 4.2147368421052626e-06,
      "loss": 1.3564,
      "step": 192
    },
    {
      "epoch": 0.16580756013745704,
      "grad_norm": 0.5065258145332336,
      "learning_rate": 3.6878947368421047e-06,
      "loss": 1.6407,
      "step": 193
    },
    {
      "epoch": 0.16666666666666666,
      "grad_norm": 0.5404831767082214,
      "learning_rate": 3.1610526315789474e-06,
      "loss": 1.6031,
      "step": 194
    },
    {
      "epoch": 0.16752577319587628,
      "grad_norm": 0.5111208558082581,
      "learning_rate": 2.634210526315789e-06,
      "loss": 1.4117,
      "step": 195
    },
    {
      "epoch": 0.16838487972508592,
      "grad_norm": 0.628701388835907,
      "learning_rate": 2.1073684210526313e-06,
      "loss": 1.563,
      "step": 196
    },
    {
      "epoch": 0.16924398625429554,
      "grad_norm": 0.591088056564331,
      "learning_rate": 1.5805263157894737e-06,
      "loss": 1.4572,
      "step": 197
    },
    {
      "epoch": 0.17010309278350516,
      "grad_norm": 0.6137686371803284,
      "learning_rate": 1.0536842105263156e-06,
      "loss": 1.473,
      "step": 198
    },
    {
      "epoch": 0.17096219931271478,
      "grad_norm": 0.787673830986023,
      "learning_rate": 5.268421052631578e-07,
      "loss": 1.5638,
      "step": 199
    },
    {
      "epoch": 0.1718213058419244,
      "grad_norm": 0.8384268283843994,
      "learning_rate": 0.0,
      "loss": 1.466,
      "step": 200
    },
    {
      "epoch": 0.1718213058419244,
      "eval_loss": 1.4909777641296387,
      "eval_runtime": 24.2168,
      "eval_samples_per_second": 20.234,
      "eval_steps_per_second": 5.079,
      "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": 5.9886658387968e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}