{
  "best_metric": 3.096618890762329,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.9673518742442564,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0048367593712212815,
      "grad_norm": 1.1617188453674316,
      "learning_rate": 1e-05,
      "loss": 4.7596,
      "step": 1
    },
    {
      "epoch": 0.0048367593712212815,
      "eval_loss": 6.157076835632324,
      "eval_runtime": 24.5444,
      "eval_samples_per_second": 14.219,
      "eval_steps_per_second": 3.585,
      "step": 1
    },
    {
      "epoch": 0.009673518742442563,
      "grad_norm": 0.9026299715042114,
      "learning_rate": 2e-05,
      "loss": 4.5941,
      "step": 2
    },
    {
      "epoch": 0.014510278113663845,
      "grad_norm": 0.93766850233078,
      "learning_rate": 3e-05,
      "loss": 4.5263,
      "step": 3
    },
    {
      "epoch": 0.019347037484885126,
      "grad_norm": 0.9980384111404419,
      "learning_rate": 4e-05,
      "loss": 4.896,
      "step": 4
    },
    {
      "epoch": 0.02418379685610641,
      "grad_norm": 1.0029842853546143,
      "learning_rate": 5e-05,
      "loss": 4.8572,
      "step": 5
    },
    {
      "epoch": 0.02902055622732769,
      "grad_norm": 0.9169318079948425,
      "learning_rate": 6e-05,
      "loss": 4.7819,
      "step": 6
    },
    {
      "epoch": 0.03385731559854897,
      "grad_norm": 0.8879703879356384,
      "learning_rate": 7e-05,
      "loss": 5.1249,
      "step": 7
    },
    {
      "epoch": 0.03869407496977025,
      "grad_norm": 0.7526925802230835,
      "learning_rate": 8e-05,
      "loss": 4.9538,
      "step": 8
    },
    {
      "epoch": 0.04353083434099154,
      "grad_norm": 0.8418651223182678,
      "learning_rate": 9e-05,
      "loss": 4.9324,
      "step": 9
    },
    {
      "epoch": 0.04836759371221282,
      "grad_norm": 0.9776986241340637,
      "learning_rate": 0.0001,
      "loss": 5.0089,
      "step": 10
    },
    {
      "epoch": 0.053204353083434096,
      "grad_norm": 1.1972429752349854,
      "learning_rate": 9.999316524962345e-05,
      "loss": 4.796,
      "step": 11
    },
    {
      "epoch": 0.05804111245465538,
      "grad_norm": 1.2151463031768799,
      "learning_rate": 9.997266286704631e-05,
      "loss": 4.8224,
      "step": 12
    },
    {
      "epoch": 0.06287787182587666,
      "grad_norm": 0.9506840109825134,
      "learning_rate": 9.993849845741524e-05,
      "loss": 4.8054,
      "step": 13
    },
    {
      "epoch": 0.06771463119709795,
      "grad_norm": 0.8992571830749512,
      "learning_rate": 9.989068136093873e-05,
      "loss": 4.7549,
      "step": 14
    },
    {
      "epoch": 0.07255139056831923,
      "grad_norm": 1.0656702518463135,
      "learning_rate": 9.98292246503335e-05,
      "loss": 4.5751,
      "step": 15
    },
    {
      "epoch": 0.0773881499395405,
      "grad_norm": 1.4029414653778076,
      "learning_rate": 9.975414512725057e-05,
      "loss": 4.5562,
      "step": 16
    },
    {
      "epoch": 0.08222490931076179,
      "grad_norm": 1.8448899984359741,
      "learning_rate": 9.966546331768191e-05,
      "loss": 4.4634,
      "step": 17
    },
    {
      "epoch": 0.08706166868198308,
      "grad_norm": 1.419966697692871,
      "learning_rate": 9.956320346634876e-05,
      "loss": 4.3018,
      "step": 18
    },
    {
      "epoch": 0.09189842805320435,
      "grad_norm": 1.185367465019226,
      "learning_rate": 9.944739353007344e-05,
      "loss": 4.3949,
      "step": 19
    },
    {
      "epoch": 0.09673518742442563,
      "grad_norm": 1.2497384548187256,
      "learning_rate": 9.931806517013612e-05,
      "loss": 4.0369,
      "step": 20
    },
    {
      "epoch": 0.10157194679564692,
      "grad_norm": 1.7013061046600342,
      "learning_rate": 9.917525374361912e-05,
      "loss": 4.256,
      "step": 21
    },
    {
      "epoch": 0.10640870616686819,
      "grad_norm": 1.7007272243499756,
      "learning_rate": 9.901899829374047e-05,
      "loss": 4.2902,
      "step": 22
    },
    {
      "epoch": 0.11124546553808948,
      "grad_norm": 1.3456192016601562,
      "learning_rate": 9.884934153917997e-05,
      "loss": 4.2021,
      "step": 23
    },
    {
      "epoch": 0.11608222490931076,
      "grad_norm": 1.1488994359970093,
      "learning_rate": 9.86663298624003e-05,
      "loss": 3.9547,
      "step": 24
    },
    {
      "epoch": 0.12091898428053205,
      "grad_norm": 1.1721539497375488,
      "learning_rate": 9.847001329696653e-05,
      "loss": 3.9169,
      "step": 25
    },
    {
      "epoch": 0.12575574365175332,
      "grad_norm": 1.5152736902236938,
      "learning_rate": 9.826044551386744e-05,
      "loss": 3.6748,
      "step": 26
    },
    {
      "epoch": 0.13059250302297462,
      "grad_norm": 1.2665867805480957,
      "learning_rate": 9.803768380684242e-05,
      "loss": 3.9372,
      "step": 27
    },
    {
      "epoch": 0.1354292623941959,
      "grad_norm": 1.237270474433899,
      "learning_rate": 9.780178907671789e-05,
      "loss": 4.1551,
      "step": 28
    },
    {
      "epoch": 0.14026602176541716,
      "grad_norm": 1.3872307538986206,
      "learning_rate": 9.755282581475769e-05,
      "loss": 3.839,
      "step": 29
    },
    {
      "epoch": 0.14510278113663846,
      "grad_norm": 1.3766710758209229,
      "learning_rate": 9.729086208503174e-05,
      "loss": 3.7746,
      "step": 30
    },
    {
      "epoch": 0.14993954050785974,
      "grad_norm": 1.529934048652649,
      "learning_rate": 9.701596950580806e-05,
      "loss": 3.6619,
      "step": 31
    },
    {
      "epoch": 0.154776299879081,
      "grad_norm": 1.3987491130828857,
      "learning_rate": 9.672822322997305e-05,
      "loss": 3.8394,
      "step": 32
    },
    {
      "epoch": 0.1596130592503023,
      "grad_norm": 1.4551591873168945,
      "learning_rate": 9.642770192448536e-05,
      "loss": 3.6552,
      "step": 33
    },
    {
      "epoch": 0.16444981862152358,
      "grad_norm": 2.5900394916534424,
      "learning_rate": 9.611448774886924e-05,
      "loss": 3.4262,
      "step": 34
    },
    {
      "epoch": 0.16928657799274485,
      "grad_norm": 1.5955469608306885,
      "learning_rate": 9.578866633275288e-05,
      "loss": 3.5815,
      "step": 35
    },
    {
      "epoch": 0.17412333736396615,
      "grad_norm": 1.6739338636398315,
      "learning_rate": 9.545032675245813e-05,
      "loss": 3.2588,
      "step": 36
    },
    {
      "epoch": 0.17896009673518742,
      "grad_norm": 1.7447373867034912,
      "learning_rate": 9.509956150664796e-05,
      "loss": 3.4028,
      "step": 37
    },
    {
      "epoch": 0.1837968561064087,
      "grad_norm": 1.8848189115524292,
      "learning_rate": 9.473646649103818e-05,
      "loss": 3.717,
      "step": 38
    },
    {
      "epoch": 0.18863361547763,
      "grad_norm": 1.897143006324768,
      "learning_rate": 9.43611409721806e-05,
      "loss": 3.9314,
      "step": 39
    },
    {
      "epoch": 0.19347037484885127,
      "grad_norm": 1.9058037996292114,
      "learning_rate": 9.397368756032445e-05,
      "loss": 3.6918,
      "step": 40
    },
    {
      "epoch": 0.19830713422007254,
      "grad_norm": 2.0588226318359375,
      "learning_rate": 9.357421218136386e-05,
      "loss": 3.6354,
      "step": 41
    },
    {
      "epoch": 0.20314389359129384,
      "grad_norm": 2.1984565258026123,
      "learning_rate": 9.316282404787871e-05,
      "loss": 3.7167,
      "step": 42
    },
    {
      "epoch": 0.2079806529625151,
      "grad_norm": 2.3980116844177246,
      "learning_rate": 9.273963562927695e-05,
      "loss": 3.7643,
      "step": 43
    },
    {
      "epoch": 0.21281741233373638,
      "grad_norm": 2.733450174331665,
      "learning_rate": 9.230476262104677e-05,
      "loss": 3.7228,
      "step": 44
    },
    {
      "epoch": 0.21765417170495768,
      "grad_norm": 2.4327785968780518,
      "learning_rate": 9.185832391312644e-05,
      "loss": 3.5608,
      "step": 45
    },
    {
      "epoch": 0.22249093107617895,
      "grad_norm": 3.0570452213287354,
      "learning_rate": 9.140044155740101e-05,
      "loss": 3.5869,
      "step": 46
    },
    {
      "epoch": 0.22732769044740025,
      "grad_norm": 2.8573720455169678,
      "learning_rate": 9.093124073433463e-05,
      "loss": 3.4583,
      "step": 47
    },
    {
      "epoch": 0.23216444981862153,
      "grad_norm": 3.0484259128570557,
      "learning_rate": 9.045084971874738e-05,
      "loss": 3.1126,
      "step": 48
    },
    {
      "epoch": 0.2370012091898428,
      "grad_norm": 3.359527587890625,
      "learning_rate": 8.995939984474624e-05,
      "loss": 3.2686,
      "step": 49
    },
    {
      "epoch": 0.2418379685610641,
      "grad_norm": 6.187241554260254,
      "learning_rate": 8.945702546981969e-05,
      "loss": 2.9971,
      "step": 50
    },
    {
      "epoch": 0.2418379685610641,
      "eval_loss": 3.9119009971618652,
      "eval_runtime": 25.0984,
      "eval_samples_per_second": 13.905,
      "eval_steps_per_second": 3.506,
      "step": 50
    },
    {
      "epoch": 0.24667472793228537,
      "grad_norm": 3.537184715270996,
      "learning_rate": 8.894386393810563e-05,
      "loss": 3.9745,
      "step": 51
    },
    {
      "epoch": 0.25151148730350664,
      "grad_norm": 3.478048324584961,
      "learning_rate": 8.842005554284296e-05,
      "loss": 3.7278,
      "step": 52
    },
    {
      "epoch": 0.25634824667472794,
      "grad_norm": 2.7068095207214355,
      "learning_rate": 8.788574348801675e-05,
      "loss": 3.6057,
      "step": 53
    },
    {
      "epoch": 0.26118500604594924,
      "grad_norm": 2.021810531616211,
      "learning_rate": 8.73410738492077e-05,
      "loss": 3.75,
      "step": 54
    },
    {
      "epoch": 0.2660217654171705,
      "grad_norm": 2.003779888153076,
      "learning_rate": 8.678619553365659e-05,
      "loss": 3.6097,
      "step": 55
    },
    {
      "epoch": 0.2708585247883918,
      "grad_norm": 2.335697650909424,
      "learning_rate": 8.622126023955446e-05,
      "loss": 3.8061,
      "step": 56
    },
    {
      "epoch": 0.2756952841596131,
      "grad_norm": 1.3209770917892456,
      "learning_rate": 8.564642241456986e-05,
      "loss": 3.4939,
      "step": 57
    },
    {
      "epoch": 0.28053204353083433,
      "grad_norm": 0.9901565313339233,
      "learning_rate": 8.506183921362443e-05,
      "loss": 3.6202,
      "step": 58
    },
    {
      "epoch": 0.28536880290205563,
      "grad_norm": 1.325579285621643,
      "learning_rate": 8.44676704559283e-05,
      "loss": 3.7492,
      "step": 59
    },
    {
      "epoch": 0.29020556227327693,
      "grad_norm": 1.4497206211090088,
      "learning_rate": 8.386407858128706e-05,
      "loss": 3.5262,
      "step": 60
    },
    {
      "epoch": 0.2950423216444982,
      "grad_norm": 1.1344389915466309,
      "learning_rate": 8.32512286056924e-05,
      "loss": 3.6636,
      "step": 61
    },
    {
      "epoch": 0.2998790810157195,
      "grad_norm": 1.0906139612197876,
      "learning_rate": 8.262928807620843e-05,
      "loss": 3.4333,
      "step": 62
    },
    {
      "epoch": 0.3047158403869408,
      "grad_norm": 1.3604044914245605,
      "learning_rate": 8.199842702516583e-05,
      "loss": 3.5378,
      "step": 63
    },
    {
      "epoch": 0.309552599758162,
      "grad_norm": 1.2874724864959717,
      "learning_rate": 8.135881792367686e-05,
      "loss": 3.809,
      "step": 64
    },
    {
      "epoch": 0.3143893591293833,
      "grad_norm": 1.2226186990737915,
      "learning_rate": 8.07106356344834e-05,
      "loss": 3.2982,
      "step": 65
    },
    {
      "epoch": 0.3192261185006046,
      "grad_norm": 1.129316806793213,
      "learning_rate": 8.005405736415126e-05,
      "loss": 3.5369,
      "step": 66
    },
    {
      "epoch": 0.32406287787182586,
      "grad_norm": 1.2191989421844482,
      "learning_rate": 7.938926261462366e-05,
      "loss": 3.6421,
      "step": 67
    },
    {
      "epoch": 0.32889963724304716,
      "grad_norm": 1.3816190958023071,
      "learning_rate": 7.871643313414718e-05,
      "loss": 3.4114,
      "step": 68
    },
    {
      "epoch": 0.33373639661426846,
      "grad_norm": 1.2303191423416138,
      "learning_rate": 7.803575286758364e-05,
      "loss": 3.2829,
      "step": 69
    },
    {
      "epoch": 0.3385731559854897,
      "grad_norm": 1.3407537937164307,
      "learning_rate": 7.734740790612136e-05,
      "loss": 3.5007,
      "step": 70
    },
    {
      "epoch": 0.343409915356711,
      "grad_norm": 1.4049428701400757,
      "learning_rate": 7.66515864363997e-05,
      "loss": 3.5486,
      "step": 71
    },
    {
      "epoch": 0.3482466747279323,
      "grad_norm": 1.3637009859085083,
      "learning_rate": 7.594847868906076e-05,
      "loss": 3.3421,
      "step": 72
    },
    {
      "epoch": 0.35308343409915355,
      "grad_norm": 1.2243515253067017,
      "learning_rate": 7.52382768867422e-05,
      "loss": 3.3868,
      "step": 73
    },
    {
      "epoch": 0.35792019347037485,
      "grad_norm": 1.277376651763916,
      "learning_rate": 7.452117519152542e-05,
      "loss": 3.475,
      "step": 74
    },
    {
      "epoch": 0.36275695284159615,
      "grad_norm": 1.3276245594024658,
      "learning_rate": 7.379736965185368e-05,
      "loss": 3.3594,
      "step": 75
    },
    {
      "epoch": 0.3675937122128174,
      "grad_norm": 1.8373863697052002,
      "learning_rate": 7.30670581489344e-05,
      "loss": 3.2602,
      "step": 76
    },
    {
      "epoch": 0.3724304715840387,
      "grad_norm": 1.611742615699768,
      "learning_rate": 7.233044034264034e-05,
      "loss": 3.6582,
      "step": 77
    },
    {
      "epoch": 0.37726723095526,
      "grad_norm": 1.431017518043518,
      "learning_rate": 7.158771761692464e-05,
      "loss": 3.3325,
      "step": 78
    },
    {
      "epoch": 0.38210399032648124,
      "grad_norm": 1.6275678873062134,
      "learning_rate": 7.083909302476453e-05,
      "loss": 3.372,
      "step": 79
    },
    {
      "epoch": 0.38694074969770254,
      "grad_norm": 1.5885802507400513,
      "learning_rate": 7.008477123264848e-05,
      "loss": 3.2323,
      "step": 80
    },
    {
      "epoch": 0.39177750906892383,
      "grad_norm": 1.5743188858032227,
      "learning_rate": 6.932495846462261e-05,
      "loss": 3.178,
      "step": 81
    },
    {
      "epoch": 0.3966142684401451,
      "grad_norm": 1.6304560899734497,
      "learning_rate": 6.855986244591104e-05,
      "loss": 3.2423,
      "step": 82
    },
    {
      "epoch": 0.4014510278113664,
      "grad_norm": 1.6588176488876343,
      "learning_rate": 6.778969234612584e-05,
      "loss": 3.2678,
      "step": 83
    },
    {
      "epoch": 0.4062877871825877,
      "grad_norm": 1.89093816280365,
      "learning_rate": 6.701465872208216e-05,
      "loss": 3.3034,
      "step": 84
    },
    {
      "epoch": 0.4111245465538089,
      "grad_norm": 1.8580236434936523,
      "learning_rate": 6.623497346023418e-05,
      "loss": 3.1634,
      "step": 85
    },
    {
      "epoch": 0.4159613059250302,
      "grad_norm": 1.809619426727295,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.8499,
      "step": 86
    },
    {
      "epoch": 0.4207980652962515,
      "grad_norm": 1.9497642517089844,
      "learning_rate": 6.466250186922325e-05,
      "loss": 2.8149,
      "step": 87
    },
    {
      "epoch": 0.42563482466747277,
      "grad_norm": 2.023190498352051,
      "learning_rate": 6.387014543809223e-05,
      "loss": 3.3851,
      "step": 88
    },
    {
      "epoch": 0.43047158403869407,
      "grad_norm": 2.2798399925231934,
      "learning_rate": 6.307399704769099e-05,
      "loss": 3.3802,
      "step": 89
    },
    {
      "epoch": 0.43530834340991537,
      "grad_norm": 2.407221794128418,
      "learning_rate": 6.227427435703997e-05,
      "loss": 3.1791,
      "step": 90
    },
    {
      "epoch": 0.44014510278113667,
      "grad_norm": 2.260432004928589,
      "learning_rate": 6.147119600233758e-05,
      "loss": 3.4068,
      "step": 91
    },
    {
      "epoch": 0.4449818621523579,
      "grad_norm": 2.4161081314086914,
      "learning_rate": 6.066498153718735e-05,
      "loss": 2.9127,
      "step": 92
    },
    {
      "epoch": 0.4498186215235792,
      "grad_norm": 2.496824026107788,
      "learning_rate": 5.985585137257401e-05,
      "loss": 3.4243,
      "step": 93
    },
    {
      "epoch": 0.4546553808948005,
      "grad_norm": 2.5099995136260986,
      "learning_rate": 5.90440267166055e-05,
      "loss": 3.2605,
      "step": 94
    },
    {
      "epoch": 0.45949214026602175,
      "grad_norm": 2.7672200202941895,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 3.0465,
      "step": 95
    },
    {
      "epoch": 0.46432889963724305,
      "grad_norm": 2.9241063594818115,
      "learning_rate": 5.74131823855921e-05,
      "loss": 3.4914,
      "step": 96
    },
    {
      "epoch": 0.46916565900846435,
      "grad_norm": 2.9847824573516846,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 2.8743,
      "step": 97
    },
    {
      "epoch": 0.4740024183796856,
      "grad_norm": 3.8728387355804443,
      "learning_rate": 5.577423184847932e-05,
      "loss": 2.9304,
      "step": 98
    },
    {
      "epoch": 0.4788391777509069,
      "grad_norm": 3.5250191688537598,
      "learning_rate": 5.495227651252315e-05,
      "loss": 2.9263,
      "step": 99
    },
    {
      "epoch": 0.4836759371221282,
      "grad_norm": 6.544610500335693,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 3.2858,
      "step": 100
    },
    {
      "epoch": 0.4836759371221282,
      "eval_loss": 3.3022475242614746,
      "eval_runtime": 25.0406,
      "eval_samples_per_second": 13.937,
      "eval_steps_per_second": 3.514,
      "step": 100
    },
    {
      "epoch": 0.48851269649334944,
      "grad_norm": 1.4022338390350342,
      "learning_rate": 5.330452921628497e-05,
      "loss": 3.225,
      "step": 101
    },
    {
      "epoch": 0.49334945586457074,
      "grad_norm": 1.4815634489059448,
      "learning_rate": 5.247918773366112e-05,
      "loss": 3.2691,
      "step": 102
    },
    {
      "epoch": 0.49818621523579204,
      "grad_norm": 1.7118796110153198,
      "learning_rate": 5.165316846586541e-05,
      "loss": 3.4775,
      "step": 103
    },
    {
      "epoch": 0.5030229746070133,
      "grad_norm": 1.5497167110443115,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 3.3042,
      "step": 104
    },
    {
      "epoch": 0.5078597339782346,
      "grad_norm": 1.317162036895752,
      "learning_rate": 5e-05,
      "loss": 3.2489,
      "step": 105
    },
    {
      "epoch": 0.5126964933494559,
      "grad_norm": 1.2906768321990967,
      "learning_rate": 4.917330276168208e-05,
      "loss": 3.285,
      "step": 106
    },
    {
      "epoch": 0.5175332527206772,
      "grad_norm": 1.1710935831069946,
      "learning_rate": 4.834683153413459e-05,
      "loss": 3.4524,
      "step": 107
    },
    {
      "epoch": 0.5223700120918985,
      "grad_norm": 1.1658213138580322,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 3.3257,
      "step": 108
    },
    {
      "epoch": 0.5272067714631197,
      "grad_norm": 1.058742642402649,
      "learning_rate": 4.669547078371504e-05,
      "loss": 3.2696,
      "step": 109
    },
    {
      "epoch": 0.532043530834341,
      "grad_norm": 1.1191959381103516,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 3.4153,
      "step": 110
    },
    {
      "epoch": 0.5368802902055623,
      "grad_norm": 1.112247347831726,
      "learning_rate": 4.504772348747687e-05,
      "loss": 3.2513,
      "step": 111
    },
    {
      "epoch": 0.5417170495767836,
      "grad_norm": 1.2679721117019653,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 3.3369,
      "step": 112
    },
    {
      "epoch": 0.5465538089480049,
      "grad_norm": 1.2211815118789673,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 3.2837,
      "step": 113
    },
    {
      "epoch": 0.5513905683192262,
      "grad_norm": 1.259421467781067,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 3.2335,
      "step": 114
    },
    {
      "epoch": 0.5562273276904474,
      "grad_norm": 1.4141755104064941,
      "learning_rate": 4.17702704859633e-05,
      "loss": 3.3882,
      "step": 115
    },
    {
      "epoch": 0.5610640870616687,
      "grad_norm": 1.3068320751190186,
      "learning_rate": 4.095597328339452e-05,
      "loss": 3.4796,
      "step": 116
    },
    {
      "epoch": 0.56590084643289,
      "grad_norm": 1.200391411781311,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 3.1921,
      "step": 117
    },
    {
      "epoch": 0.5707376058041113,
      "grad_norm": 1.2435513734817505,
      "learning_rate": 3.933501846281267e-05,
      "loss": 3.2918,
      "step": 118
    },
    {
      "epoch": 0.5755743651753326,
      "grad_norm": 1.3446563482284546,
      "learning_rate": 3.852880399766243e-05,
      "loss": 3.2688,
      "step": 119
    },
    {
      "epoch": 0.5804111245465539,
      "grad_norm": 1.393993854522705,
      "learning_rate": 3.772572564296005e-05,
      "loss": 3.1923,
      "step": 120
    },
    {
      "epoch": 0.585247883917775,
      "grad_norm": 1.396091103553772,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 3.3134,
      "step": 121
    },
    {
      "epoch": 0.5900846432889963,
      "grad_norm": 1.3212326765060425,
      "learning_rate": 3.612985456190778e-05,
      "loss": 3.0852,
      "step": 122
    },
    {
      "epoch": 0.5949214026602176,
      "grad_norm": 1.271822214126587,
      "learning_rate": 3.533749813077677e-05,
      "loss": 3.0234,
      "step": 123
    },
    {
      "epoch": 0.599758162031439,
      "grad_norm": 1.592594861984253,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 3.132,
      "step": 124
    },
    {
      "epoch": 0.6045949214026602,
      "grad_norm": 1.5006508827209473,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 3.1414,
      "step": 125
    },
    {
      "epoch": 0.6094316807738815,
      "grad_norm": 1.5481098890304565,
      "learning_rate": 3.298534127791785e-05,
      "loss": 3.339,
      "step": 126
    },
    {
      "epoch": 0.6142684401451027,
      "grad_norm": 1.4170258045196533,
      "learning_rate": 3.221030765387417e-05,
      "loss": 2.893,
      "step": 127
    },
    {
      "epoch": 0.619105199516324,
      "grad_norm": 1.6437604427337646,
      "learning_rate": 3.144013755408895e-05,
      "loss": 3.0624,
      "step": 128
    },
    {
      "epoch": 0.6239419588875453,
      "grad_norm": 1.7528533935546875,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 3.1755,
      "step": 129
    },
    {
      "epoch": 0.6287787182587666,
      "grad_norm": 1.6327184438705444,
      "learning_rate": 2.991522876735154e-05,
      "loss": 2.9975,
      "step": 130
    },
    {
      "epoch": 0.6336154776299879,
      "grad_norm": 1.668362021446228,
      "learning_rate": 2.916090697523549e-05,
      "loss": 3.0274,
      "step": 131
    },
    {
      "epoch": 0.6384522370012092,
      "grad_norm": 1.6013939380645752,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 2.938,
      "step": 132
    },
    {
      "epoch": 0.6432889963724304,
      "grad_norm": 1.6858307123184204,
      "learning_rate": 2.766955965735968e-05,
      "loss": 3.0892,
      "step": 133
    },
    {
      "epoch": 0.6481257557436517,
      "grad_norm": 1.8344348669052124,
      "learning_rate": 2.693294185106562e-05,
      "loss": 3.0604,
      "step": 134
    },
    {
      "epoch": 0.652962515114873,
      "grad_norm": 1.8722050189971924,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 3.1266,
      "step": 135
    },
    {
      "epoch": 0.6577992744860943,
      "grad_norm": 2.0373027324676514,
      "learning_rate": 2.547882480847461e-05,
      "loss": 2.8526,
      "step": 136
    },
    {
      "epoch": 0.6626360338573156,
      "grad_norm": 2.193730354309082,
      "learning_rate": 2.476172311325783e-05,
      "loss": 3.1906,
      "step": 137
    },
    {
      "epoch": 0.6674727932285369,
      "grad_norm": 2.14042592048645,
      "learning_rate": 2.405152131093926e-05,
      "loss": 3.1563,
      "step": 138
    },
    {
      "epoch": 0.6723095525997581,
      "grad_norm": 2.1373069286346436,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 3.0855,
      "step": 139
    },
    {
      "epoch": 0.6771463119709794,
      "grad_norm": 2.120323657989502,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 2.812,
      "step": 140
    },
    {
      "epoch": 0.6819830713422007,
      "grad_norm": 2.4252002239227295,
      "learning_rate": 2.196424713241637e-05,
      "loss": 3.0766,
      "step": 141
    },
    {
      "epoch": 0.686819830713422,
      "grad_norm": 2.4933314323425293,
      "learning_rate": 2.128356686585282e-05,
      "loss": 3.2087,
      "step": 142
    },
    {
      "epoch": 0.6916565900846433,
      "grad_norm": 2.5807833671569824,
      "learning_rate": 2.061073738537635e-05,
      "loss": 3.0237,
      "step": 143
    },
    {
      "epoch": 0.6964933494558646,
      "grad_norm": 2.662776231765747,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 3.1306,
      "step": 144
    },
    {
      "epoch": 0.7013301088270859,
      "grad_norm": 2.741943597793579,
      "learning_rate": 1.928936436551661e-05,
      "loss": 3.225,
      "step": 145
    },
    {
      "epoch": 0.7061668681983071,
      "grad_norm": 3.0866432189941406,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 3.1821,
      "step": 146
    },
    {
      "epoch": 0.7110036275695284,
      "grad_norm": 3.255917549133301,
      "learning_rate": 1.800157297483417e-05,
      "loss": 2.95,
      "step": 147
    },
    {
      "epoch": 0.7158403869407497,
      "grad_norm": 3.4765303134918213,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 3.0419,
      "step": 148
    },
    {
      "epoch": 0.720677146311971,
      "grad_norm": 4.259628772735596,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 2.5354,
      "step": 149
    },
    {
      "epoch": 0.7255139056831923,
      "grad_norm": 7.090859413146973,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 2.7795,
      "step": 150
    },
    {
      "epoch": 0.7255139056831923,
      "eval_loss": 3.137526750564575,
      "eval_runtime": 25.0486,
      "eval_samples_per_second": 13.933,
      "eval_steps_per_second": 3.513,
      "step": 150
    },
    {
      "epoch": 0.7303506650544136,
      "grad_norm": 0.8134868144989014,
      "learning_rate": 1.553232954407171e-05,
      "loss": 3.1718,
      "step": 151
    },
    {
      "epoch": 0.7351874244256348,
      "grad_norm": 0.8495203256607056,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 2.7118,
      "step": 152
    },
    {
      "epoch": 0.7400241837968561,
      "grad_norm": 1.0264036655426025,
      "learning_rate": 1.435357758543015e-05,
      "loss": 2.8775,
      "step": 153
    },
    {
      "epoch": 0.7448609431680774,
      "grad_norm": 1.05451500415802,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 3.1271,
      "step": 154
    },
    {
      "epoch": 0.7496977025392987,
      "grad_norm": 1.0686537027359009,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 2.9956,
      "step": 155
    },
    {
      "epoch": 0.75453446191052,
      "grad_norm": 1.030872106552124,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 3.2569,
      "step": 156
    },
    {
      "epoch": 0.7593712212817413,
      "grad_norm": 1.2028578519821167,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 3.2105,
      "step": 157
    },
    {
      "epoch": 0.7642079806529625,
      "grad_norm": 1.147829294204712,
      "learning_rate": 1.157994445715706e-05,
      "loss": 3.2371,
      "step": 158
    },
    {
      "epoch": 0.7690447400241838,
      "grad_norm": 1.1731163263320923,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 3.2199,
      "step": 159
    },
    {
      "epoch": 0.7738814993954051,
      "grad_norm": 1.1449629068374634,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 3.2715,
      "step": 160
    },
    {
      "epoch": 0.7787182587666264,
      "grad_norm": 1.0971086025238037,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 3.0647,
      "step": 161
    },
    {
      "epoch": 0.7835550181378477,
      "grad_norm": 1.144055724143982,
      "learning_rate": 9.549150281252633e-06,
      "loss": 3.151,
      "step": 162
    },
    {
      "epoch": 0.788391777509069,
      "grad_norm": 1.2181543111801147,
      "learning_rate": 9.068759265665384e-06,
      "loss": 3.2958,
      "step": 163
    },
    {
      "epoch": 0.7932285368802902,
      "grad_norm": 1.1231603622436523,
      "learning_rate": 8.599558442598998e-06,
      "loss": 2.9442,
      "step": 164
    },
    {
      "epoch": 0.7980652962515115,
      "grad_norm": 1.1957298517227173,
      "learning_rate": 8.141676086873572e-06,
      "loss": 3.2905,
      "step": 165
    },
    {
      "epoch": 0.8029020556227328,
      "grad_norm": 1.2335866689682007,
      "learning_rate": 7.695237378953223e-06,
      "loss": 3.1266,
      "step": 166
    },
    {
      "epoch": 0.8077388149939541,
      "grad_norm": 1.1873971223831177,
      "learning_rate": 7.260364370723044e-06,
      "loss": 3.062,
      "step": 167
    },
    {
      "epoch": 0.8125755743651754,
      "grad_norm": 1.3015683889389038,
      "learning_rate": 6.837175952121306e-06,
      "loss": 3.0249,
      "step": 168
    },
    {
      "epoch": 0.8174123337363967,
      "grad_norm": 1.2348692417144775,
      "learning_rate": 6.425787818636131e-06,
      "loss": 3.0331,
      "step": 169
    },
    {
      "epoch": 0.8222490931076178,
      "grad_norm": 1.3564093112945557,
      "learning_rate": 6.026312439675552e-06,
      "loss": 3.2086,
      "step": 170
    },
    {
      "epoch": 0.8270858524788391,
      "grad_norm": 1.2808902263641357,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 3.063,
      "step": 171
    },
    {
      "epoch": 0.8319226118500604,
      "grad_norm": 1.3627151250839233,
      "learning_rate": 5.263533508961827e-06,
      "loss": 3.0703,
      "step": 172
    },
    {
      "epoch": 0.8367593712212817,
      "grad_norm": 1.4379006624221802,
      "learning_rate": 4.900438493352055e-06,
      "loss": 3.0107,
      "step": 173
    },
    {
      "epoch": 0.841596130592503,
      "grad_norm": 1.52776300907135,
      "learning_rate": 4.549673247541875e-06,
      "loss": 3.3435,
      "step": 174
    },
    {
      "epoch": 0.8464328899637243,
      "grad_norm": 1.4576001167297363,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 3.1124,
      "step": 175
    },
    {
      "epoch": 0.8512696493349455,
      "grad_norm": 1.4880417585372925,
      "learning_rate": 3.885512251130763e-06,
      "loss": 2.947,
      "step": 176
    },
    {
      "epoch": 0.8561064087061668,
      "grad_norm": 1.5251941680908203,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 3.2259,
      "step": 177
    },
    {
      "epoch": 0.8609431680773881,
      "grad_norm": 1.5730133056640625,
      "learning_rate": 3.271776770026963e-06,
      "loss": 2.8914,
      "step": 178
    },
    {
      "epoch": 0.8657799274486094,
      "grad_norm": 1.546715497970581,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 3.1025,
      "step": 179
    },
    {
      "epoch": 0.8706166868198307,
      "grad_norm": 1.5666130781173706,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 2.9883,
      "step": 180
    },
    {
      "epoch": 0.875453446191052,
      "grad_norm": 1.5871328115463257,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 2.8285,
      "step": 181
    },
    {
      "epoch": 0.8802902055622733,
      "grad_norm": 1.6730198860168457,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 2.824,
      "step": 182
    },
    {
      "epoch": 0.8851269649334945,
      "grad_norm": 1.844699740409851,
      "learning_rate": 1.962316193157593e-06,
      "loss": 3.0453,
      "step": 183
    },
    {
      "epoch": 0.8899637243047158,
      "grad_norm": 1.7764073610305786,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 2.8193,
      "step": 184
    },
    {
      "epoch": 0.8948004836759371,
      "grad_norm": 1.825858235359192,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 2.9249,
      "step": 185
    },
    {
      "epoch": 0.8996372430471584,
      "grad_norm": 1.9319663047790527,
      "learning_rate": 1.333670137599713e-06,
      "loss": 2.9025,
      "step": 186
    },
    {
      "epoch": 0.9044740024183797,
      "grad_norm": 1.8836370706558228,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 2.7866,
      "step": 187
    },
    {
      "epoch": 0.909310761789601,
      "grad_norm": 2.32501220703125,
      "learning_rate": 9.810017062595322e-07,
      "loss": 3.2003,
      "step": 188
    },
    {
      "epoch": 0.9141475211608222,
      "grad_norm": 1.9631175994873047,
      "learning_rate": 8.247462563808817e-07,
      "loss": 2.9501,
      "step": 189
    },
    {
      "epoch": 0.9189842805320435,
      "grad_norm": 2.4365735054016113,
      "learning_rate": 6.819348298638839e-07,
      "loss": 3.3653,
      "step": 190
    },
    {
      "epoch": 0.9238210399032648,
      "grad_norm": 2.2507407665252686,
      "learning_rate": 5.526064699265753e-07,
      "loss": 2.9847,
      "step": 191
    },
    {
      "epoch": 0.9286577992744861,
      "grad_norm": 2.3665904998779297,
      "learning_rate": 4.367965336512403e-07,
      "loss": 3.1292,
      "step": 192
    },
    {
      "epoch": 0.9334945586457074,
      "grad_norm": 2.545895576477051,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 3.2018,
      "step": 193
    },
    {
      "epoch": 0.9383313180169287,
      "grad_norm": 2.556004285812378,
      "learning_rate": 2.458548727494292e-07,
      "loss": 3.3089,
      "step": 194
    },
    {
      "epoch": 0.9431680773881499,
      "grad_norm": 2.8826961517333984,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 3.3045,
      "step": 195
    },
    {
      "epoch": 0.9480048367593712,
      "grad_norm": 3.1025009155273438,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 2.8794,
      "step": 196
    },
    {
      "epoch": 0.9528415961305925,
      "grad_norm": 3.2988035678863525,
      "learning_rate": 6.150154258476315e-08,
      "loss": 2.8794,
      "step": 197
    },
    {
      "epoch": 0.9576783555018138,
      "grad_norm": 3.581357955932617,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 2.6826,
      "step": 198
    },
    {
      "epoch": 0.9625151148730351,
      "grad_norm": 4.3072381019592285,
      "learning_rate": 6.834750376549792e-09,
      "loss": 2.9599,
      "step": 199
    },
    {
      "epoch": 0.9673518742442564,
      "grad_norm": 7.317171096801758,
      "learning_rate": 0.0,
      "loss": 2.6062,
      "step": 200
    },
    {
      "epoch": 0.9673518742442564,
      "eval_loss": 3.096618890762329,
      "eval_runtime": 25.047,
      "eval_samples_per_second": 13.934,
      "eval_steps_per_second": 3.513,
      "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": 2.8651057009655808e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}