{
  "best_metric": 0.22676756978034973,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.29027576197387517,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.001451378809869376,
      "grad_norm": 0.11311399191617966,
      "learning_rate": 1.013e-05,
      "loss": 0.2249,
      "step": 1
    },
    {
      "epoch": 0.001451378809869376,
      "eval_loss": 0.27678582072257996,
      "eval_runtime": 142.3159,
      "eval_samples_per_second": 2.045,
      "eval_steps_per_second": 0.513,
      "step": 1
    },
    {
      "epoch": 0.002902757619738752,
      "grad_norm": 0.11226661503314972,
      "learning_rate": 2.026e-05,
      "loss": 0.4195,
      "step": 2
    },
    {
      "epoch": 0.0043541364296081275,
      "grad_norm": 0.1760484129190445,
      "learning_rate": 3.039e-05,
      "loss": 0.3752,
      "step": 3
    },
    {
      "epoch": 0.005805515239477504,
      "grad_norm": 0.13543817400932312,
      "learning_rate": 4.052e-05,
      "loss": 0.3455,
      "step": 4
    },
    {
      "epoch": 0.00725689404934688,
      "grad_norm": 0.11544519662857056,
      "learning_rate": 5.065e-05,
      "loss": 0.1971,
      "step": 5
    },
    {
      "epoch": 0.008708272859216255,
      "grad_norm": 0.13015075027942657,
      "learning_rate": 6.078e-05,
      "loss": 0.2833,
      "step": 6
    },
    {
      "epoch": 0.010159651669085631,
      "grad_norm": 0.14773070812225342,
      "learning_rate": 7.091e-05,
      "loss": 0.4397,
      "step": 7
    },
    {
      "epoch": 0.011611030478955007,
      "grad_norm": 0.15062645077705383,
      "learning_rate": 8.104e-05,
      "loss": 0.4152,
      "step": 8
    },
    {
      "epoch": 0.013062409288824383,
      "grad_norm": 0.12542325258255005,
      "learning_rate": 9.117e-05,
      "loss": 0.3613,
      "step": 9
    },
    {
      "epoch": 0.01451378809869376,
      "grad_norm": 0.13777385652065277,
      "learning_rate": 0.0001013,
      "loss": 0.7267,
      "step": 10
    },
    {
      "epoch": 0.015965166908563134,
      "grad_norm": 0.16426308453083038,
      "learning_rate": 0.00010076684210526316,
      "loss": 0.5661,
      "step": 11
    },
    {
      "epoch": 0.01741654571843251,
      "grad_norm": 0.12953603267669678,
      "learning_rate": 0.0001002336842105263,
      "loss": 0.3289,
      "step": 12
    },
    {
      "epoch": 0.018867924528301886,
      "grad_norm": 0.1950962245464325,
      "learning_rate": 9.970052631578946e-05,
      "loss": 0.6131,
      "step": 13
    },
    {
      "epoch": 0.020319303338171262,
      "grad_norm": 0.17205172777175903,
      "learning_rate": 9.916736842105263e-05,
      "loss": 0.4953,
      "step": 14
    },
    {
      "epoch": 0.02177068214804064,
      "grad_norm": 0.13275088369846344,
      "learning_rate": 9.863421052631579e-05,
      "loss": 0.2192,
      "step": 15
    },
    {
      "epoch": 0.023222060957910014,
      "grad_norm": 0.13935284316539764,
      "learning_rate": 9.810105263157895e-05,
      "loss": 0.2338,
      "step": 16
    },
    {
      "epoch": 0.02467343976777939,
      "grad_norm": 0.1427544802427292,
      "learning_rate": 9.756789473684211e-05,
      "loss": 0.3142,
      "step": 17
    },
    {
      "epoch": 0.026124818577648767,
      "grad_norm": 0.13596214354038239,
      "learning_rate": 9.703473684210525e-05,
      "loss": 0.255,
      "step": 18
    },
    {
      "epoch": 0.027576197387518143,
      "grad_norm": 0.16519062221050262,
      "learning_rate": 9.650157894736842e-05,
      "loss": 0.5583,
      "step": 19
    },
    {
      "epoch": 0.02902757619738752,
      "grad_norm": 0.11557265371084213,
      "learning_rate": 9.596842105263158e-05,
      "loss": 0.1951,
      "step": 20
    },
    {
      "epoch": 0.030478955007256895,
      "grad_norm": 0.09901455044746399,
      "learning_rate": 9.543526315789474e-05,
      "loss": 0.1046,
      "step": 21
    },
    {
      "epoch": 0.03193033381712627,
      "grad_norm": 0.15276901423931122,
      "learning_rate": 9.49021052631579e-05,
      "loss": 0.445,
      "step": 22
    },
    {
      "epoch": 0.033381712626995644,
      "grad_norm": 0.15448720753192902,
      "learning_rate": 9.436894736842105e-05,
      "loss": 0.4333,
      "step": 23
    },
    {
      "epoch": 0.03483309143686502,
      "grad_norm": 0.18187403678894043,
      "learning_rate": 9.38357894736842e-05,
      "loss": 0.3444,
      "step": 24
    },
    {
      "epoch": 0.036284470246734396,
      "grad_norm": 0.13476215302944183,
      "learning_rate": 9.330263157894737e-05,
      "loss": 0.3449,
      "step": 25
    },
    {
      "epoch": 0.03773584905660377,
      "grad_norm": 0.15487666428089142,
      "learning_rate": 9.276947368421051e-05,
      "loss": 0.4944,
      "step": 26
    },
    {
      "epoch": 0.03918722786647315,
      "grad_norm": 0.17659035325050354,
      "learning_rate": 9.223631578947369e-05,
      "loss": 0.5834,
      "step": 27
    },
    {
      "epoch": 0.040638606676342524,
      "grad_norm": 0.1136946901679039,
      "learning_rate": 9.170315789473684e-05,
      "loss": 0.2565,
      "step": 28
    },
    {
      "epoch": 0.0420899854862119,
      "grad_norm": 0.17775152623653412,
      "learning_rate": 9.117e-05,
      "loss": 0.4925,
      "step": 29
    },
    {
      "epoch": 0.04354136429608128,
      "grad_norm": 0.2578783333301544,
      "learning_rate": 9.063684210526316e-05,
      "loss": 0.3918,
      "step": 30
    },
    {
      "epoch": 0.04499274310595065,
      "grad_norm": 0.1302911639213562,
      "learning_rate": 9.010368421052632e-05,
      "loss": 0.3904,
      "step": 31
    },
    {
      "epoch": 0.04644412191582003,
      "grad_norm": 0.12905248999595642,
      "learning_rate": 8.957052631578946e-05,
      "loss": 0.28,
      "step": 32
    },
    {
      "epoch": 0.047895500725689405,
      "grad_norm": 0.16814574599266052,
      "learning_rate": 8.903736842105263e-05,
      "loss": 0.4776,
      "step": 33
    },
    {
      "epoch": 0.04934687953555878,
      "grad_norm": 0.1260037124156952,
      "learning_rate": 8.850421052631579e-05,
      "loss": 0.2609,
      "step": 34
    },
    {
      "epoch": 0.05079825834542816,
      "grad_norm": 0.10647979378700256,
      "learning_rate": 8.797105263157895e-05,
      "loss": 0.2253,
      "step": 35
    },
    {
      "epoch": 0.05224963715529753,
      "grad_norm": 0.10167553275823593,
      "learning_rate": 8.743789473684211e-05,
      "loss": 0.2064,
      "step": 36
    },
    {
      "epoch": 0.05370101596516691,
      "grad_norm": 0.13301363587379456,
      "learning_rate": 8.690473684210526e-05,
      "loss": 0.2473,
      "step": 37
    },
    {
      "epoch": 0.055152394775036286,
      "grad_norm": 0.11595452576875687,
      "learning_rate": 8.637157894736842e-05,
      "loss": 0.2187,
      "step": 38
    },
    {
      "epoch": 0.05660377358490566,
      "grad_norm": 0.15230633318424225,
      "learning_rate": 8.583842105263158e-05,
      "loss": 0.3451,
      "step": 39
    },
    {
      "epoch": 0.05805515239477504,
      "grad_norm": 0.15629370510578156,
      "learning_rate": 8.530526315789472e-05,
      "loss": 0.6737,
      "step": 40
    },
    {
      "epoch": 0.059506531204644414,
      "grad_norm": 0.14271894097328186,
      "learning_rate": 8.47721052631579e-05,
      "loss": 0.216,
      "step": 41
    },
    {
      "epoch": 0.06095791001451379,
      "grad_norm": 0.23410384356975555,
      "learning_rate": 8.423894736842105e-05,
      "loss": 0.3089,
      "step": 42
    },
    {
      "epoch": 0.062409288824383166,
      "grad_norm": 0.13687372207641602,
      "learning_rate": 8.37057894736842e-05,
      "loss": 0.212,
      "step": 43
    },
    {
      "epoch": 0.06386066763425254,
      "grad_norm": 0.1781831830739975,
      "learning_rate": 8.317263157894737e-05,
      "loss": 0.6556,
      "step": 44
    },
    {
      "epoch": 0.06531204644412192,
      "grad_norm": 0.25682398676872253,
      "learning_rate": 8.263947368421053e-05,
      "loss": 0.5198,
      "step": 45
    },
    {
      "epoch": 0.06676342525399129,
      "grad_norm": 0.2759973406791687,
      "learning_rate": 8.210631578947368e-05,
      "loss": 0.9177,
      "step": 46
    },
    {
      "epoch": 0.06821480406386067,
      "grad_norm": 0.2668778598308563,
      "learning_rate": 8.157315789473684e-05,
      "loss": 0.8107,
      "step": 47
    },
    {
      "epoch": 0.06966618287373004,
      "grad_norm": 0.2767494022846222,
      "learning_rate": 8.104e-05,
      "loss": 0.606,
      "step": 48
    },
    {
      "epoch": 0.07111756168359942,
      "grad_norm": 0.5263920426368713,
      "learning_rate": 8.050684210526316e-05,
      "loss": 0.9517,
      "step": 49
    },
    {
      "epoch": 0.07256894049346879,
      "grad_norm": 0.4693450927734375,
      "learning_rate": 7.997368421052632e-05,
      "loss": 1.1133,
      "step": 50
    },
    {
      "epoch": 0.07256894049346879,
      "eval_loss": 0.2412186712026596,
      "eval_runtime": 145.5936,
      "eval_samples_per_second": 1.999,
      "eval_steps_per_second": 0.501,
      "step": 50
    },
    {
      "epoch": 0.07402031930333818,
      "grad_norm": 0.1490558385848999,
      "learning_rate": 7.944052631578947e-05,
      "loss": 0.5739,
      "step": 51
    },
    {
      "epoch": 0.07547169811320754,
      "grad_norm": 0.19806860387325287,
      "learning_rate": 7.890736842105263e-05,
      "loss": 0.3819,
      "step": 52
    },
    {
      "epoch": 0.07692307692307693,
      "grad_norm": 0.12554095685482025,
      "learning_rate": 7.837421052631579e-05,
      "loss": 0.3486,
      "step": 53
    },
    {
      "epoch": 0.0783744557329463,
      "grad_norm": 0.17904381453990936,
      "learning_rate": 7.784105263157893e-05,
      "loss": 0.6337,
      "step": 54
    },
    {
      "epoch": 0.07982583454281568,
      "grad_norm": 0.16063204407691956,
      "learning_rate": 7.730789473684211e-05,
      "loss": 0.4335,
      "step": 55
    },
    {
      "epoch": 0.08127721335268505,
      "grad_norm": 0.17032606899738312,
      "learning_rate": 7.677473684210526e-05,
      "loss": 0.5176,
      "step": 56
    },
    {
      "epoch": 0.08272859216255443,
      "grad_norm": 0.36131739616394043,
      "learning_rate": 7.624157894736842e-05,
      "loss": 0.3518,
      "step": 57
    },
    {
      "epoch": 0.0841799709724238,
      "grad_norm": 0.19793149828910828,
      "learning_rate": 7.570842105263158e-05,
      "loss": 0.8414,
      "step": 58
    },
    {
      "epoch": 0.08563134978229318,
      "grad_norm": 0.13127806782722473,
      "learning_rate": 7.517526315789474e-05,
      "loss": 0.3519,
      "step": 59
    },
    {
      "epoch": 0.08708272859216255,
      "grad_norm": 0.1507265418767929,
      "learning_rate": 7.464210526315789e-05,
      "loss": 0.8202,
      "step": 60
    },
    {
      "epoch": 0.08853410740203194,
      "grad_norm": 0.1529066562652588,
      "learning_rate": 7.410894736842106e-05,
      "loss": 0.5119,
      "step": 61
    },
    {
      "epoch": 0.0899854862119013,
      "grad_norm": 0.12405847012996674,
      "learning_rate": 7.35757894736842e-05,
      "loss": 0.5436,
      "step": 62
    },
    {
      "epoch": 0.09143686502177069,
      "grad_norm": 0.16444605588912964,
      "learning_rate": 7.304263157894737e-05,
      "loss": 0.4945,
      "step": 63
    },
    {
      "epoch": 0.09288824383164006,
      "grad_norm": 0.10274633765220642,
      "learning_rate": 7.250947368421053e-05,
      "loss": 0.2285,
      "step": 64
    },
    {
      "epoch": 0.09433962264150944,
      "grad_norm": 0.13879486918449402,
      "learning_rate": 7.197631578947368e-05,
      "loss": 0.3028,
      "step": 65
    },
    {
      "epoch": 0.09579100145137881,
      "grad_norm": 0.1328522115945816,
      "learning_rate": 7.144315789473684e-05,
      "loss": 0.4795,
      "step": 66
    },
    {
      "epoch": 0.09724238026124818,
      "grad_norm": 0.15546078979969025,
      "learning_rate": 7.091e-05,
      "loss": 0.5576,
      "step": 67
    },
    {
      "epoch": 0.09869375907111756,
      "grad_norm": 0.16497625410556793,
      "learning_rate": 7.037684210526316e-05,
      "loss": 0.271,
      "step": 68
    },
    {
      "epoch": 0.10014513788098693,
      "grad_norm": 0.1409892439842224,
      "learning_rate": 6.984368421052632e-05,
      "loss": 0.4059,
      "step": 69
    },
    {
      "epoch": 0.10159651669085631,
      "grad_norm": 0.173180490732193,
      "learning_rate": 6.931052631578947e-05,
      "loss": 0.5719,
      "step": 70
    },
    {
      "epoch": 0.10304789550072568,
      "grad_norm": 0.1634824275970459,
      "learning_rate": 6.877736842105263e-05,
      "loss": 0.3316,
      "step": 71
    },
    {
      "epoch": 0.10449927431059507,
      "grad_norm": 0.1520518958568573,
      "learning_rate": 6.824421052631579e-05,
      "loss": 0.749,
      "step": 72
    },
    {
      "epoch": 0.10595065312046444,
      "grad_norm": 0.11503203213214874,
      "learning_rate": 6.771105263157895e-05,
      "loss": 0.3243,
      "step": 73
    },
    {
      "epoch": 0.10740203193033382,
      "grad_norm": 0.09616807848215103,
      "learning_rate": 6.71778947368421e-05,
      "loss": 0.2127,
      "step": 74
    },
    {
      "epoch": 0.10885341074020319,
      "grad_norm": 0.12218757718801498,
      "learning_rate": 6.664473684210527e-05,
      "loss": 0.3151,
      "step": 75
    },
    {
      "epoch": 0.11030478955007257,
      "grad_norm": 0.11861466616392136,
      "learning_rate": 6.611157894736842e-05,
      "loss": 0.2481,
      "step": 76
    },
    {
      "epoch": 0.11175616835994194,
      "grad_norm": 0.08120733499526978,
      "learning_rate": 6.557842105263158e-05,
      "loss": 0.1281,
      "step": 77
    },
    {
      "epoch": 0.11320754716981132,
      "grad_norm": 0.11339303106069565,
      "learning_rate": 6.504526315789474e-05,
      "loss": 0.3048,
      "step": 78
    },
    {
      "epoch": 0.11465892597968069,
      "grad_norm": 0.12353406846523285,
      "learning_rate": 6.451210526315789e-05,
      "loss": 0.522,
      "step": 79
    },
    {
      "epoch": 0.11611030478955008,
      "grad_norm": 0.18549102544784546,
      "learning_rate": 6.397894736842105e-05,
      "loss": 1.1374,
      "step": 80
    },
    {
      "epoch": 0.11756168359941944,
      "grad_norm": 0.10565419495105743,
      "learning_rate": 6.344578947368421e-05,
      "loss": 0.2235,
      "step": 81
    },
    {
      "epoch": 0.11901306240928883,
      "grad_norm": 0.13751162588596344,
      "learning_rate": 6.291263157894737e-05,
      "loss": 0.4815,
      "step": 82
    },
    {
      "epoch": 0.1204644412191582,
      "grad_norm": 0.12583044171333313,
      "learning_rate": 6.237947368421053e-05,
      "loss": 0.4072,
      "step": 83
    },
    {
      "epoch": 0.12191582002902758,
      "grad_norm": 0.11933013796806335,
      "learning_rate": 6.184631578947368e-05,
      "loss": 0.4836,
      "step": 84
    },
    {
      "epoch": 0.12336719883889695,
      "grad_norm": 0.17227371037006378,
      "learning_rate": 6.131315789473684e-05,
      "loss": 0.6044,
      "step": 85
    },
    {
      "epoch": 0.12481857764876633,
      "grad_norm": 0.1414662003517151,
      "learning_rate": 6.078e-05,
      "loss": 0.5979,
      "step": 86
    },
    {
      "epoch": 0.1262699564586357,
      "grad_norm": 0.08421526849269867,
      "learning_rate": 6.024684210526315e-05,
      "loss": 0.1035,
      "step": 87
    },
    {
      "epoch": 0.12772133526850507,
      "grad_norm": 0.12226138263940811,
      "learning_rate": 5.9713684210526305e-05,
      "loss": 0.1514,
      "step": 88
    },
    {
      "epoch": 0.12917271407837447,
      "grad_norm": 0.08483120799064636,
      "learning_rate": 5.918052631578947e-05,
      "loss": 0.1681,
      "step": 89
    },
    {
      "epoch": 0.13062409288824384,
      "grad_norm": 0.16209205985069275,
      "learning_rate": 5.8647368421052634e-05,
      "loss": 1.0777,
      "step": 90
    },
    {
      "epoch": 0.1320754716981132,
      "grad_norm": 0.16335219144821167,
      "learning_rate": 5.811421052631579e-05,
      "loss": 0.8582,
      "step": 91
    },
    {
      "epoch": 0.13352685050798258,
      "grad_norm": 0.14614659547805786,
      "learning_rate": 5.758105263157894e-05,
      "loss": 0.8367,
      "step": 92
    },
    {
      "epoch": 0.13497822931785197,
      "grad_norm": 0.12466667592525482,
      "learning_rate": 5.70478947368421e-05,
      "loss": 0.2996,
      "step": 93
    },
    {
      "epoch": 0.13642960812772134,
      "grad_norm": 0.16494490206241608,
      "learning_rate": 5.6514736842105256e-05,
      "loss": 0.7263,
      "step": 94
    },
    {
      "epoch": 0.1378809869375907,
      "grad_norm": 0.1597548872232437,
      "learning_rate": 5.5981578947368424e-05,
      "loss": 0.5913,
      "step": 95
    },
    {
      "epoch": 0.13933236574746008,
      "grad_norm": 0.246367946267128,
      "learning_rate": 5.544842105263158e-05,
      "loss": 1.299,
      "step": 96
    },
    {
      "epoch": 0.14078374455732948,
      "grad_norm": 0.2906947731971741,
      "learning_rate": 5.491526315789474e-05,
      "loss": 1.0694,
      "step": 97
    },
    {
      "epoch": 0.14223512336719885,
      "grad_norm": 0.32492387294769287,
      "learning_rate": 5.438210526315789e-05,
      "loss": 0.6249,
      "step": 98
    },
    {
      "epoch": 0.14368650217706821,
      "grad_norm": 0.3696464002132416,
      "learning_rate": 5.384894736842105e-05,
      "loss": 1.3239,
      "step": 99
    },
    {
      "epoch": 0.14513788098693758,
      "grad_norm": 0.4203333854675293,
      "learning_rate": 5.331578947368421e-05,
      "loss": 0.976,
      "step": 100
    },
    {
      "epoch": 0.14513788098693758,
      "eval_loss": 0.2287154644727707,
      "eval_runtime": 144.369,
      "eval_samples_per_second": 2.016,
      "eval_steps_per_second": 0.506,
      "step": 100
    },
    {
      "epoch": 0.14658925979680695,
      "grad_norm": 0.14465273916721344,
      "learning_rate": 5.278263157894736e-05,
      "loss": 0.7933,
      "step": 101
    },
    {
      "epoch": 0.14804063860667635,
      "grad_norm": 0.1529458463191986,
      "learning_rate": 5.224947368421053e-05,
      "loss": 0.4874,
      "step": 102
    },
    {
      "epoch": 0.14949201741654572,
      "grad_norm": 0.09371291100978851,
      "learning_rate": 5.171631578947368e-05,
      "loss": 0.1754,
      "step": 103
    },
    {
      "epoch": 0.1509433962264151,
      "grad_norm": 0.11294244974851608,
      "learning_rate": 5.1183157894736844e-05,
      "loss": 0.4367,
      "step": 104
    },
    {
      "epoch": 0.15239477503628446,
      "grad_norm": 0.11110277473926544,
      "learning_rate": 5.065e-05,
      "loss": 0.3404,
      "step": 105
    },
    {
      "epoch": 0.15384615384615385,
      "grad_norm": 0.09716267883777618,
      "learning_rate": 5.011684210526315e-05,
      "loss": 0.3296,
      "step": 106
    },
    {
      "epoch": 0.15529753265602322,
      "grad_norm": 0.15396001935005188,
      "learning_rate": 4.958368421052631e-05,
      "loss": 0.772,
      "step": 107
    },
    {
      "epoch": 0.1567489114658926,
      "grad_norm": 0.17330443859100342,
      "learning_rate": 4.9050526315789473e-05,
      "loss": 0.4806,
      "step": 108
    },
    {
      "epoch": 0.15820029027576196,
      "grad_norm": 0.15418781340122223,
      "learning_rate": 4.851736842105263e-05,
      "loss": 0.6729,
      "step": 109
    },
    {
      "epoch": 0.15965166908563136,
      "grad_norm": 0.13719098269939423,
      "learning_rate": 4.798421052631579e-05,
      "loss": 0.2485,
      "step": 110
    },
    {
      "epoch": 0.16110304789550073,
      "grad_norm": 0.1265023797750473,
      "learning_rate": 4.745105263157895e-05,
      "loss": 0.3845,
      "step": 111
    },
    {
      "epoch": 0.1625544267053701,
      "grad_norm": 0.15158045291900635,
      "learning_rate": 4.69178947368421e-05,
      "loss": 0.4527,
      "step": 112
    },
    {
      "epoch": 0.16400580551523947,
      "grad_norm": 0.12712687253952026,
      "learning_rate": 4.638473684210526e-05,
      "loss": 0.3207,
      "step": 113
    },
    {
      "epoch": 0.16545718432510886,
      "grad_norm": 0.10735659301280975,
      "learning_rate": 4.585157894736842e-05,
      "loss": 0.2524,
      "step": 114
    },
    {
      "epoch": 0.16690856313497823,
      "grad_norm": 0.14663982391357422,
      "learning_rate": 4.531842105263158e-05,
      "loss": 0.7199,
      "step": 115
    },
    {
      "epoch": 0.1683599419448476,
      "grad_norm": 0.12890632450580597,
      "learning_rate": 4.478526315789473e-05,
      "loss": 0.5994,
      "step": 116
    },
    {
      "epoch": 0.16981132075471697,
      "grad_norm": 0.16796326637268066,
      "learning_rate": 4.425210526315789e-05,
      "loss": 0.6609,
      "step": 117
    },
    {
      "epoch": 0.17126269956458637,
      "grad_norm": 0.21263499557971954,
      "learning_rate": 4.3718947368421054e-05,
      "loss": 0.6373,
      "step": 118
    },
    {
      "epoch": 0.17271407837445574,
      "grad_norm": 0.15164527297019958,
      "learning_rate": 4.318578947368421e-05,
      "loss": 0.3913,
      "step": 119
    },
    {
      "epoch": 0.1741654571843251,
      "grad_norm": 0.09173654019832611,
      "learning_rate": 4.265263157894736e-05,
      "loss": 0.22,
      "step": 120
    },
    {
      "epoch": 0.17561683599419448,
      "grad_norm": 0.10727760940790176,
      "learning_rate": 4.211947368421052e-05,
      "loss": 0.2657,
      "step": 121
    },
    {
      "epoch": 0.17706821480406387,
      "grad_norm": 0.1093788743019104,
      "learning_rate": 4.1586315789473684e-05,
      "loss": 0.1771,
      "step": 122
    },
    {
      "epoch": 0.17851959361393324,
      "grad_norm": 0.11123545467853546,
      "learning_rate": 4.105315789473684e-05,
      "loss": 0.3519,
      "step": 123
    },
    {
      "epoch": 0.1799709724238026,
      "grad_norm": 0.11694060266017914,
      "learning_rate": 4.052e-05,
      "loss": 0.2824,
      "step": 124
    },
    {
      "epoch": 0.18142235123367198,
      "grad_norm": 0.1088772639632225,
      "learning_rate": 3.998684210526316e-05,
      "loss": 0.3736,
      "step": 125
    },
    {
      "epoch": 0.18287373004354138,
      "grad_norm": 0.07843092083930969,
      "learning_rate": 3.945368421052631e-05,
      "loss": 0.1476,
      "step": 126
    },
    {
      "epoch": 0.18432510885341075,
      "grad_norm": 0.12299855053424835,
      "learning_rate": 3.892052631578947e-05,
      "loss": 0.3185,
      "step": 127
    },
    {
      "epoch": 0.18577648766328012,
      "grad_norm": 0.13632050156593323,
      "learning_rate": 3.838736842105263e-05,
      "loss": 0.3271,
      "step": 128
    },
    {
      "epoch": 0.18722786647314948,
      "grad_norm": 0.11568715423345566,
      "learning_rate": 3.785421052631579e-05,
      "loss": 0.2214,
      "step": 129
    },
    {
      "epoch": 0.18867924528301888,
      "grad_norm": 0.11555133759975433,
      "learning_rate": 3.732105263157894e-05,
      "loss": 0.2503,
      "step": 130
    },
    {
      "epoch": 0.19013062409288825,
      "grad_norm": 0.11684191972017288,
      "learning_rate": 3.67878947368421e-05,
      "loss": 0.1773,
      "step": 131
    },
    {
      "epoch": 0.19158200290275762,
      "grad_norm": 0.09825399518013,
      "learning_rate": 3.6254736842105264e-05,
      "loss": 0.2053,
      "step": 132
    },
    {
      "epoch": 0.193033381712627,
      "grad_norm": 0.13328059017658234,
      "learning_rate": 3.572157894736842e-05,
      "loss": 0.1957,
      "step": 133
    },
    {
      "epoch": 0.19448476052249636,
      "grad_norm": 0.1092911884188652,
      "learning_rate": 3.518842105263158e-05,
      "loss": 0.2233,
      "step": 134
    },
    {
      "epoch": 0.19593613933236576,
      "grad_norm": 0.10300347954034805,
      "learning_rate": 3.465526315789473e-05,
      "loss": 0.3399,
      "step": 135
    },
    {
      "epoch": 0.19738751814223512,
      "grad_norm": 0.1662655472755432,
      "learning_rate": 3.4122105263157894e-05,
      "loss": 0.2793,
      "step": 136
    },
    {
      "epoch": 0.1988388969521045,
      "grad_norm": 0.08663914352655411,
      "learning_rate": 3.358894736842105e-05,
      "loss": 0.1391,
      "step": 137
    },
    {
      "epoch": 0.20029027576197386,
      "grad_norm": 0.15232573449611664,
      "learning_rate": 3.305578947368421e-05,
      "loss": 0.3642,
      "step": 138
    },
    {
      "epoch": 0.20174165457184326,
      "grad_norm": 0.10668061673641205,
      "learning_rate": 3.252263157894737e-05,
      "loss": 0.1986,
      "step": 139
    },
    {
      "epoch": 0.20319303338171263,
      "grad_norm": 0.09178131073713303,
      "learning_rate": 3.198947368421052e-05,
      "loss": 0.2896,
      "step": 140
    },
    {
      "epoch": 0.204644412191582,
      "grad_norm": 0.1355268359184265,
      "learning_rate": 3.1456315789473684e-05,
      "loss": 0.8356,
      "step": 141
    },
    {
      "epoch": 0.20609579100145137,
      "grad_norm": 0.13758248090744019,
      "learning_rate": 3.092315789473684e-05,
      "loss": 0.6022,
      "step": 142
    },
    {
      "epoch": 0.20754716981132076,
      "grad_norm": 0.12314596027135849,
      "learning_rate": 3.039e-05,
      "loss": 0.4143,
      "step": 143
    },
    {
      "epoch": 0.20899854862119013,
      "grad_norm": 0.11046963185071945,
      "learning_rate": 2.9856842105263153e-05,
      "loss": 0.2393,
      "step": 144
    },
    {
      "epoch": 0.2104499274310595,
      "grad_norm": 0.16623206436634064,
      "learning_rate": 2.9323684210526317e-05,
      "loss": 0.3518,
      "step": 145
    },
    {
      "epoch": 0.21190130624092887,
      "grad_norm": 0.19511793553829193,
      "learning_rate": 2.879052631578947e-05,
      "loss": 0.5,
      "step": 146
    },
    {
      "epoch": 0.21335268505079827,
      "grad_norm": 0.2845495939254761,
      "learning_rate": 2.8257368421052628e-05,
      "loss": 0.7584,
      "step": 147
    },
    {
      "epoch": 0.21480406386066764,
      "grad_norm": 0.27343127131462097,
      "learning_rate": 2.772421052631579e-05,
      "loss": 0.87,
      "step": 148
    },
    {
      "epoch": 0.216255442670537,
      "grad_norm": 0.41281190514564514,
      "learning_rate": 2.7191052631578946e-05,
      "loss": 0.8707,
      "step": 149
    },
    {
      "epoch": 0.21770682148040638,
      "grad_norm": 0.3892386257648468,
      "learning_rate": 2.6657894736842104e-05,
      "loss": 0.5039,
      "step": 150
    },
    {
      "epoch": 0.21770682148040638,
      "eval_loss": 0.22993537783622742,
      "eval_runtime": 142.9262,
      "eval_samples_per_second": 2.036,
      "eval_steps_per_second": 0.511,
      "step": 150
    },
    {
      "epoch": 0.21915820029027577,
      "grad_norm": 0.15316051244735718,
      "learning_rate": 2.6124736842105265e-05,
      "loss": 1.0054,
      "step": 151
    },
    {
      "epoch": 0.22060957910014514,
      "grad_norm": 0.18049275875091553,
      "learning_rate": 2.5591578947368422e-05,
      "loss": 0.8339,
      "step": 152
    },
    {
      "epoch": 0.2220609579100145,
      "grad_norm": 0.1573140174150467,
      "learning_rate": 2.5058421052631576e-05,
      "loss": 0.5323,
      "step": 153
    },
    {
      "epoch": 0.22351233671988388,
      "grad_norm": 0.1617121547460556,
      "learning_rate": 2.4525263157894737e-05,
      "loss": 1.0705,
      "step": 154
    },
    {
      "epoch": 0.22496371552975328,
      "grad_norm": 0.12206801027059555,
      "learning_rate": 2.3992105263157894e-05,
      "loss": 0.5119,
      "step": 155
    },
    {
      "epoch": 0.22641509433962265,
      "grad_norm": 0.9858077764511108,
      "learning_rate": 2.345894736842105e-05,
      "loss": 1.0604,
      "step": 156
    },
    {
      "epoch": 0.22786647314949202,
      "grad_norm": 0.15416570007801056,
      "learning_rate": 2.292578947368421e-05,
      "loss": 0.2413,
      "step": 157
    },
    {
      "epoch": 0.22931785195936139,
      "grad_norm": 0.18875420093536377,
      "learning_rate": 2.2392631578947366e-05,
      "loss": 0.7481,
      "step": 158
    },
    {
      "epoch": 0.23076923076923078,
      "grad_norm": 0.12112567573785782,
      "learning_rate": 2.1859473684210527e-05,
      "loss": 0.32,
      "step": 159
    },
    {
      "epoch": 0.23222060957910015,
      "grad_norm": 0.14630571007728577,
      "learning_rate": 2.132631578947368e-05,
      "loss": 0.3848,
      "step": 160
    },
    {
      "epoch": 0.23367198838896952,
      "grad_norm": 0.10878642648458481,
      "learning_rate": 2.0793157894736842e-05,
      "loss": 0.4093,
      "step": 161
    },
    {
      "epoch": 0.2351233671988389,
      "grad_norm": 0.14565850794315338,
      "learning_rate": 2.026e-05,
      "loss": 0.4859,
      "step": 162
    },
    {
      "epoch": 0.2365747460087083,
      "grad_norm": 0.14800970256328583,
      "learning_rate": 1.9726842105263157e-05,
      "loss": 0.3829,
      "step": 163
    },
    {
      "epoch": 0.23802612481857766,
      "grad_norm": 0.14318357408046722,
      "learning_rate": 1.9193684210526314e-05,
      "loss": 0.8232,
      "step": 164
    },
    {
      "epoch": 0.23947750362844702,
      "grad_norm": 0.11647699773311615,
      "learning_rate": 1.866052631578947e-05,
      "loss": 0.3973,
      "step": 165
    },
    {
      "epoch": 0.2409288824383164,
      "grad_norm": 0.17588454484939575,
      "learning_rate": 1.8127368421052632e-05,
      "loss": 1.0372,
      "step": 166
    },
    {
      "epoch": 0.24238026124818576,
      "grad_norm": 0.22197209298610687,
      "learning_rate": 1.759421052631579e-05,
      "loss": 0.6101,
      "step": 167
    },
    {
      "epoch": 0.24383164005805516,
      "grad_norm": 0.11662589013576508,
      "learning_rate": 1.7061052631578947e-05,
      "loss": 0.3538,
      "step": 168
    },
    {
      "epoch": 0.24528301886792453,
      "grad_norm": 0.12414685636758804,
      "learning_rate": 1.6527894736842104e-05,
      "loss": 0.5084,
      "step": 169
    },
    {
      "epoch": 0.2467343976777939,
      "grad_norm": 0.10851942747831345,
      "learning_rate": 1.599473684210526e-05,
      "loss": 0.3311,
      "step": 170
    },
    {
      "epoch": 0.24818577648766327,
      "grad_norm": 0.1293657124042511,
      "learning_rate": 1.546157894736842e-05,
      "loss": 0.3142,
      "step": 171
    },
    {
      "epoch": 0.24963715529753266,
      "grad_norm": 0.19298462569713593,
      "learning_rate": 1.4928421052631576e-05,
      "loss": 0.4494,
      "step": 172
    },
    {
      "epoch": 0.251088534107402,
      "grad_norm": 0.12975549697875977,
      "learning_rate": 1.4395263157894735e-05,
      "loss": 0.2316,
      "step": 173
    },
    {
      "epoch": 0.2525399129172714,
      "grad_norm": 0.09491987526416779,
      "learning_rate": 1.3862105263157895e-05,
      "loss": 0.3277,
      "step": 174
    },
    {
      "epoch": 0.2539912917271408,
      "grad_norm": 0.21053697168827057,
      "learning_rate": 1.3328947368421052e-05,
      "loss": 0.3236,
      "step": 175
    },
    {
      "epoch": 0.25544267053701014,
      "grad_norm": 0.12965311110019684,
      "learning_rate": 1.2795789473684211e-05,
      "loss": 0.9922,
      "step": 176
    },
    {
      "epoch": 0.25689404934687954,
      "grad_norm": 0.10223641246557236,
      "learning_rate": 1.2262631578947368e-05,
      "loss": 0.3105,
      "step": 177
    },
    {
      "epoch": 0.25834542815674894,
      "grad_norm": 0.15014851093292236,
      "learning_rate": 1.1729473684210526e-05,
      "loss": 0.5113,
      "step": 178
    },
    {
      "epoch": 0.2597968069666183,
      "grad_norm": 0.1316860169172287,
      "learning_rate": 1.1196315789473683e-05,
      "loss": 0.5018,
      "step": 179
    },
    {
      "epoch": 0.2612481857764877,
      "grad_norm": 0.11262688785791397,
      "learning_rate": 1.066315789473684e-05,
      "loss": 0.2846,
      "step": 180
    },
    {
      "epoch": 0.262699564586357,
      "grad_norm": 0.14847137033939362,
      "learning_rate": 1.013e-05,
      "loss": 0.3721,
      "step": 181
    },
    {
      "epoch": 0.2641509433962264,
      "grad_norm": 0.11905442178249359,
      "learning_rate": 9.596842105263157e-06,
      "loss": 0.5352,
      "step": 182
    },
    {
      "epoch": 0.2656023222060958,
      "grad_norm": 0.12718041241168976,
      "learning_rate": 9.063684210526316e-06,
      "loss": 0.298,
      "step": 183
    },
    {
      "epoch": 0.26705370101596515,
      "grad_norm": 0.09555162489414215,
      "learning_rate": 8.530526315789473e-06,
      "loss": 0.2167,
      "step": 184
    },
    {
      "epoch": 0.26850507982583455,
      "grad_norm": 0.09585528820753098,
      "learning_rate": 7.99736842105263e-06,
      "loss": 0.6557,
      "step": 185
    },
    {
      "epoch": 0.26995645863570394,
      "grad_norm": 0.09632135927677155,
      "learning_rate": 7.464210526315788e-06,
      "loss": 0.2777,
      "step": 186
    },
    {
      "epoch": 0.2714078374455733,
      "grad_norm": 0.10153202712535858,
      "learning_rate": 6.931052631578947e-06,
      "loss": 0.237,
      "step": 187
    },
    {
      "epoch": 0.2728592162554427,
      "grad_norm": 0.1199401244521141,
      "learning_rate": 6.3978947368421055e-06,
      "loss": 0.2484,
      "step": 188
    },
    {
      "epoch": 0.274310595065312,
      "grad_norm": 0.15911225974559784,
      "learning_rate": 5.864736842105263e-06,
      "loss": 0.2827,
      "step": 189
    },
    {
      "epoch": 0.2757619738751814,
      "grad_norm": 0.12618254125118256,
      "learning_rate": 5.33157894736842e-06,
      "loss": 0.2016,
      "step": 190
    },
    {
      "epoch": 0.2772133526850508,
      "grad_norm": 0.1026952788233757,
      "learning_rate": 4.7984210526315785e-06,
      "loss": 0.282,
      "step": 191
    },
    {
      "epoch": 0.27866473149492016,
      "grad_norm": 0.10733366012573242,
      "learning_rate": 4.265263157894737e-06,
      "loss": 0.1684,
      "step": 192
    },
    {
      "epoch": 0.28011611030478956,
      "grad_norm": 0.12453214824199677,
      "learning_rate": 3.732105263157894e-06,
      "loss": 0.4221,
      "step": 193
    },
    {
      "epoch": 0.28156748911465895,
      "grad_norm": 0.19456051290035248,
      "learning_rate": 3.1989473684210527e-06,
      "loss": 0.4257,
      "step": 194
    },
    {
      "epoch": 0.2830188679245283,
      "grad_norm": 0.18270336091518402,
      "learning_rate": 2.66578947368421e-06,
      "loss": 0.8292,
      "step": 195
    },
    {
      "epoch": 0.2844702467343977,
      "grad_norm": 0.1802217960357666,
      "learning_rate": 2.1326315789473684e-06,
      "loss": 0.8952,
      "step": 196
    },
    {
      "epoch": 0.28592162554426703,
      "grad_norm": 0.23863941431045532,
      "learning_rate": 1.5994736842105264e-06,
      "loss": 0.6211,
      "step": 197
    },
    {
      "epoch": 0.28737300435413643,
      "grad_norm": 0.282530277967453,
      "learning_rate": 1.0663157894736842e-06,
      "loss": 0.6815,
      "step": 198
    },
    {
      "epoch": 0.2888243831640058,
      "grad_norm": 0.46267229318618774,
      "learning_rate": 5.331578947368421e-07,
      "loss": 0.6762,
      "step": 199
    },
    {
      "epoch": 0.29027576197387517,
      "grad_norm": 0.4375733733177185,
      "learning_rate": 0.0,
      "loss": 1.0879,
      "step": 200
    },
    {
      "epoch": 0.29027576197387517,
      "eval_loss": 0.22676756978034973,
      "eval_runtime": 144.3205,
      "eval_samples_per_second": 2.016,
      "eval_steps_per_second": 0.506,
      "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": 2.4053840999350272e+17,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}