{
  "best_metric": 1.2668575048446655,
  "best_model_checkpoint": "miner_id_24/checkpoint-150",
  "epoch": 2.73037542662116,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.013651877133105802,
      "grad_norm": 1.719310998916626,
      "learning_rate": 1e-05,
      "loss": 2.9932,
      "step": 1
    },
    {
      "epoch": 0.013651877133105802,
      "eval_loss": 4.401421546936035,
      "eval_runtime": 14.3934,
      "eval_samples_per_second": 8.615,
      "eval_steps_per_second": 2.154,
      "step": 1
    },
    {
      "epoch": 0.027303754266211604,
      "grad_norm": 2.238464593887329,
      "learning_rate": 2e-05,
      "loss": 3.7395,
      "step": 2
    },
    {
      "epoch": 0.040955631399317405,
      "grad_norm": 2.4502103328704834,
      "learning_rate": 3e-05,
      "loss": 4.3517,
      "step": 3
    },
    {
      "epoch": 0.05460750853242321,
      "grad_norm": 3.0267837047576904,
      "learning_rate": 4e-05,
      "loss": 4.7774,
      "step": 4
    },
    {
      "epoch": 0.06825938566552901,
      "grad_norm": 3.3979783058166504,
      "learning_rate": 5e-05,
      "loss": 5.1628,
      "step": 5
    },
    {
      "epoch": 0.08191126279863481,
      "grad_norm": 5.054656028747559,
      "learning_rate": 6e-05,
      "loss": 5.1449,
      "step": 6
    },
    {
      "epoch": 0.09556313993174062,
      "grad_norm": 4.114360809326172,
      "learning_rate": 7e-05,
      "loss": 5.5108,
      "step": 7
    },
    {
      "epoch": 0.10921501706484642,
      "grad_norm": 4.146178722381592,
      "learning_rate": 8e-05,
      "loss": 5.8428,
      "step": 8
    },
    {
      "epoch": 0.12286689419795221,
      "grad_norm": 5.803869247436523,
      "learning_rate": 9e-05,
      "loss": 8.1564,
      "step": 9
    },
    {
      "epoch": 0.13651877133105803,
      "grad_norm": 3.166918992996216,
      "learning_rate": 0.0001,
      "loss": 2.6081,
      "step": 10
    },
    {
      "epoch": 0.15017064846416384,
      "grad_norm": 2.4443671703338623,
      "learning_rate": 9.999316524962345e-05,
      "loss": 2.0239,
      "step": 11
    },
    {
      "epoch": 0.16382252559726962,
      "grad_norm": 2.46828031539917,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.8438,
      "step": 12
    },
    {
      "epoch": 0.17747440273037543,
      "grad_norm": 2.6492178440093994,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.7638,
      "step": 13
    },
    {
      "epoch": 0.19112627986348124,
      "grad_norm": 2.8727335929870605,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.6679,
      "step": 14
    },
    {
      "epoch": 0.20477815699658702,
      "grad_norm": 2.14016056060791,
      "learning_rate": 9.98292246503335e-05,
      "loss": 0.9907,
      "step": 15
    },
    {
      "epoch": 0.21843003412969283,
      "grad_norm": 2.197594165802002,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.0472,
      "step": 16
    },
    {
      "epoch": 0.23208191126279865,
      "grad_norm": 2.2688441276550293,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.189,
      "step": 17
    },
    {
      "epoch": 0.24573378839590443,
      "grad_norm": 1.764783263206482,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.1757,
      "step": 18
    },
    {
      "epoch": 0.2593856655290102,
      "grad_norm": 1.7969028949737549,
      "learning_rate": 9.944739353007344e-05,
      "loss": 2.3791,
      "step": 19
    },
    {
      "epoch": 0.27303754266211605,
      "grad_norm": 2.2887489795684814,
      "learning_rate": 9.931806517013612e-05,
      "loss": 2.5468,
      "step": 20
    },
    {
      "epoch": 0.28668941979522183,
      "grad_norm": 2.3017189502716064,
      "learning_rate": 9.917525374361912e-05,
      "loss": 2.5925,
      "step": 21
    },
    {
      "epoch": 0.3003412969283277,
      "grad_norm": 2.335562229156494,
      "learning_rate": 9.901899829374047e-05,
      "loss": 2.3268,
      "step": 22
    },
    {
      "epoch": 0.31399317406143346,
      "grad_norm": 2.3432226181030273,
      "learning_rate": 9.884934153917997e-05,
      "loss": 2.4406,
      "step": 23
    },
    {
      "epoch": 0.32764505119453924,
      "grad_norm": 4.281378269195557,
      "learning_rate": 9.86663298624003e-05,
      "loss": 2.9583,
      "step": 24
    },
    {
      "epoch": 0.3412969283276451,
      "grad_norm": 3.466893434524536,
      "learning_rate": 9.847001329696653e-05,
      "loss": 2.9325,
      "step": 25
    },
    {
      "epoch": 0.35494880546075086,
      "grad_norm": 4.426032066345215,
      "learning_rate": 9.826044551386744e-05,
      "loss": 3.2662,
      "step": 26
    },
    {
      "epoch": 0.36860068259385664,
      "grad_norm": 4.625338077545166,
      "learning_rate": 9.803768380684242e-05,
      "loss": 3.6393,
      "step": 27
    },
    {
      "epoch": 0.3822525597269625,
      "grad_norm": 3.043260097503662,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.281,
      "step": 28
    },
    {
      "epoch": 0.39590443686006827,
      "grad_norm": 3.151878833770752,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.0022,
      "step": 29
    },
    {
      "epoch": 0.40955631399317405,
      "grad_norm": 2.018608808517456,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.9311,
      "step": 30
    },
    {
      "epoch": 0.4232081911262799,
      "grad_norm": 1.840545415878296,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.926,
      "step": 31
    },
    {
      "epoch": 0.43686006825938567,
      "grad_norm": 1.8817676305770874,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.1601,
      "step": 32
    },
    {
      "epoch": 0.45051194539249145,
      "grad_norm": 1.687111496925354,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.8272,
      "step": 33
    },
    {
      "epoch": 0.4641638225255973,
      "grad_norm": 1.5064914226531982,
      "learning_rate": 9.611448774886924e-05,
      "loss": 0.7478,
      "step": 34
    },
    {
      "epoch": 0.4778156996587031,
      "grad_norm": 1.4334555864334106,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.9874,
      "step": 35
    },
    {
      "epoch": 0.49146757679180886,
      "grad_norm": 1.3956588506698608,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.0269,
      "step": 36
    },
    {
      "epoch": 0.5051194539249146,
      "grad_norm": 1.1524420976638794,
      "learning_rate": 9.509956150664796e-05,
      "loss": 2.0262,
      "step": 37
    },
    {
      "epoch": 0.5187713310580204,
      "grad_norm": 1.6388391256332397,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.8871,
      "step": 38
    },
    {
      "epoch": 0.5324232081911263,
      "grad_norm": 2.863281488418579,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.6705,
      "step": 39
    },
    {
      "epoch": 0.5460750853242321,
      "grad_norm": 2.1617486476898193,
      "learning_rate": 9.397368756032445e-05,
      "loss": 2.5542,
      "step": 40
    },
    {
      "epoch": 0.5597269624573379,
      "grad_norm": 2.1348841190338135,
      "learning_rate": 9.357421218136386e-05,
      "loss": 2.2504,
      "step": 41
    },
    {
      "epoch": 0.5733788395904437,
      "grad_norm": 2.580812931060791,
      "learning_rate": 9.316282404787871e-05,
      "loss": 2.2669,
      "step": 42
    },
    {
      "epoch": 0.5870307167235495,
      "grad_norm": 2.951887607574463,
      "learning_rate": 9.273963562927695e-05,
      "loss": 2.2174,
      "step": 43
    },
    {
      "epoch": 0.6006825938566553,
      "grad_norm": 3.2826178073883057,
      "learning_rate": 9.230476262104677e-05,
      "loss": 2.3888,
      "step": 44
    },
    {
      "epoch": 0.6143344709897611,
      "grad_norm": 5.060423374176025,
      "learning_rate": 9.185832391312644e-05,
      "loss": 2.3709,
      "step": 45
    },
    {
      "epoch": 0.6279863481228669,
      "grad_norm": 3.734896659851074,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.4683,
      "step": 46
    },
    {
      "epoch": 0.6416382252559727,
      "grad_norm": 5.408026218414307,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.962,
      "step": 47
    },
    {
      "epoch": 0.6552901023890785,
      "grad_norm": 2.589977979660034,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.0165,
      "step": 48
    },
    {
      "epoch": 0.6689419795221843,
      "grad_norm": 1.7457537651062012,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.6151,
      "step": 49
    },
    {
      "epoch": 0.6825938566552902,
      "grad_norm": 1.6913813352584839,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.9228,
      "step": 50
    },
    {
      "epoch": 0.6825938566552902,
      "eval_loss": 1.516947865486145,
      "eval_runtime": 14.4752,
      "eval_samples_per_second": 8.566,
      "eval_steps_per_second": 2.142,
      "step": 50
    },
    {
      "epoch": 0.6962457337883959,
      "grad_norm": 1.5579912662506104,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.8021,
      "step": 51
    },
    {
      "epoch": 0.7098976109215017,
      "grad_norm": 1.6826483011245728,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.005,
      "step": 52
    },
    {
      "epoch": 0.7235494880546075,
      "grad_norm": 1.3245519399642944,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.8833,
      "step": 53
    },
    {
      "epoch": 0.7372013651877133,
      "grad_norm": 1.642776370048523,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.1467,
      "step": 54
    },
    {
      "epoch": 0.7508532423208191,
      "grad_norm": 1.4104365110397339,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.5755,
      "step": 55
    },
    {
      "epoch": 0.764505119453925,
      "grad_norm": 1.4757987260818481,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.9944,
      "step": 56
    },
    {
      "epoch": 0.7781569965870307,
      "grad_norm": 1.7735190391540527,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.4657,
      "step": 57
    },
    {
      "epoch": 0.7918088737201365,
      "grad_norm": 2.1073620319366455,
      "learning_rate": 8.506183921362443e-05,
      "loss": 2.0537,
      "step": 58
    },
    {
      "epoch": 0.8054607508532423,
      "grad_norm": 2.783012866973877,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.8992,
      "step": 59
    },
    {
      "epoch": 0.8191126279863481,
      "grad_norm": 2.4432077407836914,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.8746,
      "step": 60
    },
    {
      "epoch": 0.8327645051194539,
      "grad_norm": 2.731022834777832,
      "learning_rate": 8.32512286056924e-05,
      "loss": 2.4562,
      "step": 61
    },
    {
      "epoch": 0.8464163822525598,
      "grad_norm": 3.184903383255005,
      "learning_rate": 8.262928807620843e-05,
      "loss": 2.0082,
      "step": 62
    },
    {
      "epoch": 0.8600682593856656,
      "grad_norm": 7.827123165130615,
      "learning_rate": 8.199842702516583e-05,
      "loss": 2.4323,
      "step": 63
    },
    {
      "epoch": 0.8737201365187713,
      "grad_norm": 2.6870064735412598,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.8885,
      "step": 64
    },
    {
      "epoch": 0.8873720136518771,
      "grad_norm": 1.4525294303894043,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.7316,
      "step": 65
    },
    {
      "epoch": 0.9010238907849829,
      "grad_norm": 1.6883432865142822,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.7486,
      "step": 66
    },
    {
      "epoch": 0.9146757679180887,
      "grad_norm": 1.0709468126296997,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.7072,
      "step": 67
    },
    {
      "epoch": 0.9283276450511946,
      "grad_norm": 1.5337930917739868,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.8322,
      "step": 68
    },
    {
      "epoch": 0.9419795221843004,
      "grad_norm": 1.6072499752044678,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.8798,
      "step": 69
    },
    {
      "epoch": 0.9556313993174061,
      "grad_norm": 1.292209267616272,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.7413,
      "step": 70
    },
    {
      "epoch": 0.9692832764505119,
      "grad_norm": 1.7525714635849,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.1242,
      "step": 71
    },
    {
      "epoch": 0.9829351535836177,
      "grad_norm": 1.1664561033248901,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.8623,
      "step": 72
    },
    {
      "epoch": 0.9965870307167235,
      "grad_norm": 1.6809195280075073,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.7342,
      "step": 73
    },
    {
      "epoch": 1.0102389078498293,
      "grad_norm": 4.731753826141357,
      "learning_rate": 7.452117519152542e-05,
      "loss": 2.1651,
      "step": 74
    },
    {
      "epoch": 1.023890784982935,
      "grad_norm": 1.324560284614563,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.3404,
      "step": 75
    },
    {
      "epoch": 1.0375426621160408,
      "grad_norm": 1.5524150133132935,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.4117,
      "step": 76
    },
    {
      "epoch": 1.0511945392491469,
      "grad_norm": 1.4553591012954712,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.2991,
      "step": 77
    },
    {
      "epoch": 1.0648464163822526,
      "grad_norm": 1.4772950410842896,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.3542,
      "step": 78
    },
    {
      "epoch": 1.0784982935153584,
      "grad_norm": 1.9716956615447998,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.1237,
      "step": 79
    },
    {
      "epoch": 1.0921501706484642,
      "grad_norm": 2.163557767868042,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.6688,
      "step": 80
    },
    {
      "epoch": 1.10580204778157,
      "grad_norm": 2.486646890640259,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.1999,
      "step": 81
    },
    {
      "epoch": 1.1194539249146758,
      "grad_norm": 3.223355770111084,
      "learning_rate": 6.855986244591104e-05,
      "loss": 2.2653,
      "step": 82
    },
    {
      "epoch": 1.1331058020477816,
      "grad_norm": 2.090512275695801,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.8192,
      "step": 83
    },
    {
      "epoch": 1.1467576791808873,
      "grad_norm": 1.1039810180664062,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.6106,
      "step": 84
    },
    {
      "epoch": 1.1604095563139931,
      "grad_norm": 1.094523549079895,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.5174,
      "step": 85
    },
    {
      "epoch": 1.174061433447099,
      "grad_norm": 1.1173726320266724,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.5748,
      "step": 86
    },
    {
      "epoch": 1.1877133105802047,
      "grad_norm": 1.275300145149231,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.7178,
      "step": 87
    },
    {
      "epoch": 1.2013651877133107,
      "grad_norm": 0.9684453010559082,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.3651,
      "step": 88
    },
    {
      "epoch": 1.2150170648464165,
      "grad_norm": 0.9446448683738708,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.4854,
      "step": 89
    },
    {
      "epoch": 1.2286689419795223,
      "grad_norm": 0.8939529657363892,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.4723,
      "step": 90
    },
    {
      "epoch": 1.242320819112628,
      "grad_norm": 0.9787300229072571,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.5207,
      "step": 91
    },
    {
      "epoch": 1.2559726962457338,
      "grad_norm": 1.1003116369247437,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.9599,
      "step": 92
    },
    {
      "epoch": 1.2696245733788396,
      "grad_norm": 1.5936921834945679,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.1448,
      "step": 93
    },
    {
      "epoch": 1.2832764505119454,
      "grad_norm": 1.4547646045684814,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.9992,
      "step": 94
    },
    {
      "epoch": 1.2969283276450512,
      "grad_norm": 1.9140379428863525,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.657,
      "step": 95
    },
    {
      "epoch": 1.310580204778157,
      "grad_norm": 2.2299554347991943,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.8714,
      "step": 96
    },
    {
      "epoch": 1.3242320819112627,
      "grad_norm": 2.0495800971984863,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.1038,
      "step": 97
    },
    {
      "epoch": 1.3378839590443685,
      "grad_norm": 2.7487759590148926,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.4194,
      "step": 98
    },
    {
      "epoch": 1.3515358361774745,
      "grad_norm": 2.7785885334014893,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.2243,
      "step": 99
    },
    {
      "epoch": 1.36518771331058,
      "grad_norm": 5.104328155517578,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 2.1085,
      "step": 100
    },
    {
      "epoch": 1.36518771331058,
      "eval_loss": 1.3428871631622314,
      "eval_runtime": 14.7444,
      "eval_samples_per_second": 8.41,
      "eval_steps_per_second": 2.102,
      "step": 100
    },
    {
      "epoch": 1.378839590443686,
      "grad_norm": 2.784961700439453,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.1989,
      "step": 101
    },
    {
      "epoch": 1.3924914675767919,
      "grad_norm": 1.3277562856674194,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.5914,
      "step": 102
    },
    {
      "epoch": 1.4061433447098977,
      "grad_norm": 1.4193967580795288,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.6386,
      "step": 103
    },
    {
      "epoch": 1.4197952218430034,
      "grad_norm": 1.231856346130371,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.3774,
      "step": 104
    },
    {
      "epoch": 1.4334470989761092,
      "grad_norm": 1.3934533596038818,
      "learning_rate": 5e-05,
      "loss": 0.5349,
      "step": 105
    },
    {
      "epoch": 1.447098976109215,
      "grad_norm": 2.485903024673462,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.693,
      "step": 106
    },
    {
      "epoch": 1.4607508532423208,
      "grad_norm": 1.2066667079925537,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.3854,
      "step": 107
    },
    {
      "epoch": 1.4744027303754266,
      "grad_norm": 1.1464177370071411,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.4949,
      "step": 108
    },
    {
      "epoch": 1.4880546075085324,
      "grad_norm": 1.5378305912017822,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.6986,
      "step": 109
    },
    {
      "epoch": 1.5017064846416384,
      "grad_norm": 1.3141582012176514,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.3167,
      "step": 110
    },
    {
      "epoch": 1.515358361774744,
      "grad_norm": 1.590103268623352,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.2284,
      "step": 111
    },
    {
      "epoch": 1.52901023890785,
      "grad_norm": 2.363316059112549,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.1963,
      "step": 112
    },
    {
      "epoch": 1.5426621160409555,
      "grad_norm": 1.634173035621643,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.9362,
      "step": 113
    },
    {
      "epoch": 1.5563139931740615,
      "grad_norm": 2.561746597290039,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.2717,
      "step": 114
    },
    {
      "epoch": 1.5699658703071673,
      "grad_norm": 1.815083384513855,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.8425,
      "step": 115
    },
    {
      "epoch": 1.583617747440273,
      "grad_norm": 2.6442713737487793,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.5452,
      "step": 116
    },
    {
      "epoch": 1.5972696245733788,
      "grad_norm": 3.2571229934692383,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.3657,
      "step": 117
    },
    {
      "epoch": 1.6109215017064846,
      "grad_norm": 3.1047441959381104,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.4363,
      "step": 118
    },
    {
      "epoch": 1.6245733788395904,
      "grad_norm": 2.843052387237549,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.2048,
      "step": 119
    },
    {
      "epoch": 1.6382252559726962,
      "grad_norm": 1.7302285432815552,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.3767,
      "step": 120
    },
    {
      "epoch": 1.6518771331058022,
      "grad_norm": 1.5493896007537842,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.7396,
      "step": 121
    },
    {
      "epoch": 1.6655290102389078,
      "grad_norm": 1.06333589553833,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.3545,
      "step": 122
    },
    {
      "epoch": 1.6791808873720138,
      "grad_norm": 1.2839938402175903,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.5186,
      "step": 123
    },
    {
      "epoch": 1.6928327645051193,
      "grad_norm": 1.1351975202560425,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.5084,
      "step": 124
    },
    {
      "epoch": 1.7064846416382253,
      "grad_norm": 1.094598650932312,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.3189,
      "step": 125
    },
    {
      "epoch": 1.7201365187713311,
      "grad_norm": 1.8149356842041016,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.6204,
      "step": 126
    },
    {
      "epoch": 1.733788395904437,
      "grad_norm": 1.4751465320587158,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.6011,
      "step": 127
    },
    {
      "epoch": 1.7474402730375427,
      "grad_norm": 1.3222554922103882,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.3059,
      "step": 128
    },
    {
      "epoch": 1.7610921501706485,
      "grad_norm": 1.7212960720062256,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.3679,
      "step": 129
    },
    {
      "epoch": 1.7747440273037542,
      "grad_norm": 2.1338675022125244,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.3326,
      "step": 130
    },
    {
      "epoch": 1.78839590443686,
      "grad_norm": 1.7768455743789673,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.9252,
      "step": 131
    },
    {
      "epoch": 1.802047781569966,
      "grad_norm": 2.179547071456909,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.4933,
      "step": 132
    },
    {
      "epoch": 1.8156996587030716,
      "grad_norm": 2.231325149536133,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.1236,
      "step": 133
    },
    {
      "epoch": 1.8293515358361776,
      "grad_norm": 2.3787450790405273,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.2358,
      "step": 134
    },
    {
      "epoch": 1.8430034129692832,
      "grad_norm": 2.809525728225708,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.0868,
      "step": 135
    },
    {
      "epoch": 1.8566552901023892,
      "grad_norm": 3.231095314025879,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.4506,
      "step": 136
    },
    {
      "epoch": 1.8703071672354947,
      "grad_norm": 3.0768117904663086,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.8524,
      "step": 137
    },
    {
      "epoch": 1.8839590443686007,
      "grad_norm": 1.898419976234436,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.5237,
      "step": 138
    },
    {
      "epoch": 1.8976109215017065,
      "grad_norm": 1.5133816003799438,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.4064,
      "step": 139
    },
    {
      "epoch": 1.9112627986348123,
      "grad_norm": 1.5006005764007568,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.403,
      "step": 140
    },
    {
      "epoch": 1.924914675767918,
      "grad_norm": 1.6118359565734863,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.5631,
      "step": 141
    },
    {
      "epoch": 1.9385665529010239,
      "grad_norm": 1.1373324394226074,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.2933,
      "step": 142
    },
    {
      "epoch": 1.9522184300341296,
      "grad_norm": 0.9513692855834961,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.3874,
      "step": 143
    },
    {
      "epoch": 1.9658703071672354,
      "grad_norm": 1.295910120010376,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.4531,
      "step": 144
    },
    {
      "epoch": 1.9795221843003414,
      "grad_norm": 1.5424376726150513,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.7297,
      "step": 145
    },
    {
      "epoch": 1.993174061433447,
      "grad_norm": 1.933988094329834,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.8683,
      "step": 146
    },
    {
      "epoch": 2.006825938566553,
      "grad_norm": 2.8978970050811768,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.31,
      "step": 147
    },
    {
      "epoch": 2.0204778156996586,
      "grad_norm": 1.1956030130386353,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.8997,
      "step": 148
    },
    {
      "epoch": 2.0341296928327646,
      "grad_norm": 1.3279848098754883,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.6323,
      "step": 149
    },
    {
      "epoch": 2.04778156996587,
      "grad_norm": 3.2569780349731445,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.0525,
      "step": 150
    },
    {
      "epoch": 2.04778156996587,
      "eval_loss": 1.2668575048446655,
      "eval_runtime": 14.7496,
      "eval_samples_per_second": 8.407,
      "eval_steps_per_second": 2.102,
      "step": 150
    },
    {
      "epoch": 2.061433447098976,
      "grad_norm": 1.5874288082122803,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.7044,
      "step": 151
    },
    {
      "epoch": 2.0750853242320817,
      "grad_norm": 1.622295618057251,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.9022,
      "step": 152
    },
    {
      "epoch": 2.0887372013651877,
      "grad_norm": 2.4951493740081787,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.7067,
      "step": 153
    },
    {
      "epoch": 2.1023890784982937,
      "grad_norm": 1.790401816368103,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.6197,
      "step": 154
    },
    {
      "epoch": 2.1160409556313993,
      "grad_norm": 2.8261895179748535,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.7072,
      "step": 155
    },
    {
      "epoch": 2.1296928327645053,
      "grad_norm": 2.6220691204071045,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.8324,
      "step": 156
    },
    {
      "epoch": 2.143344709897611,
      "grad_norm": 1.7422630786895752,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.4189,
      "step": 157
    },
    {
      "epoch": 2.156996587030717,
      "grad_norm": 1.2428183555603027,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.4257,
      "step": 158
    },
    {
      "epoch": 2.1706484641638224,
      "grad_norm": 0.6868012547492981,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.1315,
      "step": 159
    },
    {
      "epoch": 2.1843003412969284,
      "grad_norm": 0.9342251420021057,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.2073,
      "step": 160
    },
    {
      "epoch": 2.197952218430034,
      "grad_norm": 1.017391324043274,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.2174,
      "step": 161
    },
    {
      "epoch": 2.21160409556314,
      "grad_norm": 0.9734142422676086,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.2015,
      "step": 162
    },
    {
      "epoch": 2.2252559726962455,
      "grad_norm": 0.9567795395851135,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.2733,
      "step": 163
    },
    {
      "epoch": 2.2389078498293515,
      "grad_norm": 1.4026451110839844,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.4957,
      "step": 164
    },
    {
      "epoch": 2.252559726962457,
      "grad_norm": 1.0274150371551514,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.6015,
      "step": 165
    },
    {
      "epoch": 2.266211604095563,
      "grad_norm": 1.5300015211105347,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.855,
      "step": 166
    },
    {
      "epoch": 2.279863481228669,
      "grad_norm": 2.4075684547424316,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.7729,
      "step": 167
    },
    {
      "epoch": 2.2935153583617747,
      "grad_norm": 1.639745831489563,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.5757,
      "step": 168
    },
    {
      "epoch": 2.3071672354948807,
      "grad_norm": 1.9001551866531372,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.8153,
      "step": 169
    },
    {
      "epoch": 2.3208191126279862,
      "grad_norm": 2.063537120819092,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.6746,
      "step": 170
    },
    {
      "epoch": 2.3344709897610922,
      "grad_norm": 2.424978494644165,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.7172,
      "step": 171
    },
    {
      "epoch": 2.348122866894198,
      "grad_norm": 2.2778947353363037,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.5389,
      "step": 172
    },
    {
      "epoch": 2.361774744027304,
      "grad_norm": 3.422502279281616,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.1104,
      "step": 173
    },
    {
      "epoch": 2.3754266211604094,
      "grad_norm": 3.3360912799835205,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.6589,
      "step": 174
    },
    {
      "epoch": 2.3890784982935154,
      "grad_norm": 1.3084882497787476,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.3136,
      "step": 175
    },
    {
      "epoch": 2.4027303754266214,
      "grad_norm": 1.3405717611312866,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.3244,
      "step": 176
    },
    {
      "epoch": 2.416382252559727,
      "grad_norm": 0.8116306066513062,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.1371,
      "step": 177
    },
    {
      "epoch": 2.430034129692833,
      "grad_norm": 1.3822133541107178,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.4692,
      "step": 178
    },
    {
      "epoch": 2.4436860068259385,
      "grad_norm": 1.094436526298523,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.23,
      "step": 179
    },
    {
      "epoch": 2.4573378839590445,
      "grad_norm": 0.9031869769096375,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.1919,
      "step": 180
    },
    {
      "epoch": 2.47098976109215,
      "grad_norm": 1.222503662109375,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.2703,
      "step": 181
    },
    {
      "epoch": 2.484641638225256,
      "grad_norm": 0.9252071976661682,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.1403,
      "step": 182
    },
    {
      "epoch": 2.4982935153583616,
      "grad_norm": 1.456536054611206,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.6979,
      "step": 183
    },
    {
      "epoch": 2.5119453924914676,
      "grad_norm": 1.8548897504806519,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.0622,
      "step": 184
    },
    {
      "epoch": 2.5255972696245736,
      "grad_norm": 2.509766101837158,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.7123,
      "step": 185
    },
    {
      "epoch": 2.539249146757679,
      "grad_norm": 2.0521974563598633,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.7055,
      "step": 186
    },
    {
      "epoch": 2.5529010238907848,
      "grad_norm": 2.069265604019165,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.6965,
      "step": 187
    },
    {
      "epoch": 2.5665529010238908,
      "grad_norm": 2.298844337463379,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.7607,
      "step": 188
    },
    {
      "epoch": 2.580204778156997,
      "grad_norm": 2.759459972381592,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.8986,
      "step": 189
    },
    {
      "epoch": 2.5938566552901023,
      "grad_norm": 2.7197718620300293,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.707,
      "step": 190
    },
    {
      "epoch": 2.6075085324232083,
      "grad_norm": 3.737642765045166,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.0091,
      "step": 191
    },
    {
      "epoch": 2.621160409556314,
      "grad_norm": 5.674477577209473,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.7568,
      "step": 192
    },
    {
      "epoch": 2.63481228668942,
      "grad_norm": 1.6524834632873535,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.2302,
      "step": 193
    },
    {
      "epoch": 2.6484641638225255,
      "grad_norm": 1.3406412601470947,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.3325,
      "step": 194
    },
    {
      "epoch": 2.6621160409556315,
      "grad_norm": 0.9953098893165588,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.1551,
      "step": 195
    },
    {
      "epoch": 2.675767918088737,
      "grad_norm": 1.3571770191192627,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.3437,
      "step": 196
    },
    {
      "epoch": 2.689419795221843,
      "grad_norm": 1.2093820571899414,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.3354,
      "step": 197
    },
    {
      "epoch": 2.703071672354949,
      "grad_norm": 1.0174179077148438,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.2715,
      "step": 198
    },
    {
      "epoch": 2.7167235494880546,
      "grad_norm": 0.8516709208488464,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.1394,
      "step": 199
    },
    {
      "epoch": 2.73037542662116,
      "grad_norm": 1.463282585144043,
      "learning_rate": 0.0,
      "loss": 0.3191,
      "step": 200
    },
    {
      "epoch": 2.73037542662116,
      "eval_loss": 1.4357415437698364,
      "eval_runtime": 14.4843,
      "eval_samples_per_second": 8.561,
      "eval_steps_per_second": 2.14,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 1
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 4.341203892436992e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}