{
  "best_metric": 1.4081310033798218,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 2.2905982905982905,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.011396011396011397,
      "grad_norm": 0.7800776362419128,
      "learning_rate": 1e-05,
      "loss": 6.7971,
      "step": 1
    },
    {
      "epoch": 0.011396011396011397,
      "eval_loss": 2.4783096313476562,
      "eval_runtime": 15.0317,
      "eval_samples_per_second": 9.846,
      "eval_steps_per_second": 2.461,
      "step": 1
    },
    {
      "epoch": 0.022792022792022793,
      "grad_norm": 1.0287824869155884,
      "learning_rate": 2e-05,
      "loss": 7.6062,
      "step": 2
    },
    {
      "epoch": 0.03418803418803419,
      "grad_norm": 1.1188607215881348,
      "learning_rate": 3e-05,
      "loss": 8.3986,
      "step": 3
    },
    {
      "epoch": 0.045584045584045586,
      "grad_norm": 1.0409597158432007,
      "learning_rate": 4e-05,
      "loss": 7.8685,
      "step": 4
    },
    {
      "epoch": 0.05698005698005698,
      "grad_norm": 1.2380366325378418,
      "learning_rate": 5e-05,
      "loss": 8.1156,
      "step": 5
    },
    {
      "epoch": 0.06837606837606838,
      "grad_norm": 1.2219274044036865,
      "learning_rate": 6e-05,
      "loss": 8.0964,
      "step": 6
    },
    {
      "epoch": 0.07977207977207977,
      "grad_norm": 1.3197040557861328,
      "learning_rate": 7e-05,
      "loss": 8.2776,
      "step": 7
    },
    {
      "epoch": 0.09116809116809117,
      "grad_norm": 1.4048258066177368,
      "learning_rate": 8e-05,
      "loss": 8.0699,
      "step": 8
    },
    {
      "epoch": 0.10256410256410256,
      "grad_norm": 1.4832782745361328,
      "learning_rate": 9e-05,
      "loss": 8.6843,
      "step": 9
    },
    {
      "epoch": 0.11396011396011396,
      "grad_norm": 1.489786148071289,
      "learning_rate": 0.0001,
      "loss": 7.8548,
      "step": 10
    },
    {
      "epoch": 0.12535612535612536,
      "grad_norm": 1.5326049327850342,
      "learning_rate": 9.999316524962345e-05,
      "loss": 7.1396,
      "step": 11
    },
    {
      "epoch": 0.13675213675213677,
      "grad_norm": 1.5600051879882812,
      "learning_rate": 9.997266286704631e-05,
      "loss": 7.5341,
      "step": 12
    },
    {
      "epoch": 0.14814814814814814,
      "grad_norm": 2.1242332458496094,
      "learning_rate": 9.993849845741524e-05,
      "loss": 7.4737,
      "step": 13
    },
    {
      "epoch": 0.15954415954415954,
      "grad_norm": 2.2768564224243164,
      "learning_rate": 9.989068136093873e-05,
      "loss": 6.9102,
      "step": 14
    },
    {
      "epoch": 0.17094017094017094,
      "grad_norm": 2.8336424827575684,
      "learning_rate": 9.98292246503335e-05,
      "loss": 7.2374,
      "step": 15
    },
    {
      "epoch": 0.18233618233618235,
      "grad_norm": 2.9529685974121094,
      "learning_rate": 9.975414512725057e-05,
      "loss": 7.6385,
      "step": 16
    },
    {
      "epoch": 0.19373219373219372,
      "grad_norm": 3.0211853981018066,
      "learning_rate": 9.966546331768191e-05,
      "loss": 7.2637,
      "step": 17
    },
    {
      "epoch": 0.20512820512820512,
      "grad_norm": 2.8280179500579834,
      "learning_rate": 9.956320346634876e-05,
      "loss": 7.4542,
      "step": 18
    },
    {
      "epoch": 0.21652421652421652,
      "grad_norm": 3.124953508377075,
      "learning_rate": 9.944739353007344e-05,
      "loss": 7.5336,
      "step": 19
    },
    {
      "epoch": 0.22792022792022792,
      "grad_norm": 3.0875401496887207,
      "learning_rate": 9.931806517013612e-05,
      "loss": 7.9075,
      "step": 20
    },
    {
      "epoch": 0.23931623931623933,
      "grad_norm": 4.2555155754089355,
      "learning_rate": 9.917525374361912e-05,
      "loss": 9.1171,
      "step": 21
    },
    {
      "epoch": 0.25071225071225073,
      "grad_norm": 1.7275997400283813,
      "learning_rate": 9.901899829374047e-05,
      "loss": 5.6106,
      "step": 22
    },
    {
      "epoch": 0.2621082621082621,
      "grad_norm": 1.3996858596801758,
      "learning_rate": 9.884934153917997e-05,
      "loss": 5.6704,
      "step": 23
    },
    {
      "epoch": 0.27350427350427353,
      "grad_norm": 1.4265042543411255,
      "learning_rate": 9.86663298624003e-05,
      "loss": 5.9568,
      "step": 24
    },
    {
      "epoch": 0.2849002849002849,
      "grad_norm": 1.4421463012695312,
      "learning_rate": 9.847001329696653e-05,
      "loss": 5.845,
      "step": 25
    },
    {
      "epoch": 0.2962962962962963,
      "grad_norm": 1.4597059488296509,
      "learning_rate": 9.826044551386744e-05,
      "loss": 5.7813,
      "step": 26
    },
    {
      "epoch": 0.3076923076923077,
      "grad_norm": 1.6328085660934448,
      "learning_rate": 9.803768380684242e-05,
      "loss": 6.1604,
      "step": 27
    },
    {
      "epoch": 0.3190883190883191,
      "grad_norm": 1.641039490699768,
      "learning_rate": 9.780178907671789e-05,
      "loss": 5.7038,
      "step": 28
    },
    {
      "epoch": 0.33048433048433046,
      "grad_norm": 1.4992609024047852,
      "learning_rate": 9.755282581475769e-05,
      "loss": 5.7989,
      "step": 29
    },
    {
      "epoch": 0.3418803418803419,
      "grad_norm": 1.755906581878662,
      "learning_rate": 9.729086208503174e-05,
      "loss": 5.9431,
      "step": 30
    },
    {
      "epoch": 0.35327635327635326,
      "grad_norm": 1.9734288454055786,
      "learning_rate": 9.701596950580806e-05,
      "loss": 6.0303,
      "step": 31
    },
    {
      "epoch": 0.3646723646723647,
      "grad_norm": 1.957170009613037,
      "learning_rate": 9.672822322997305e-05,
      "loss": 6.2432,
      "step": 32
    },
    {
      "epoch": 0.37606837606837606,
      "grad_norm": 2.221837282180786,
      "learning_rate": 9.642770192448536e-05,
      "loss": 6.2897,
      "step": 33
    },
    {
      "epoch": 0.38746438746438744,
      "grad_norm": 2.3104963302612305,
      "learning_rate": 9.611448774886924e-05,
      "loss": 5.7418,
      "step": 34
    },
    {
      "epoch": 0.39886039886039887,
      "grad_norm": 2.7960598468780518,
      "learning_rate": 9.578866633275288e-05,
      "loss": 6.5385,
      "step": 35
    },
    {
      "epoch": 0.41025641025641024,
      "grad_norm": 3.1290805339813232,
      "learning_rate": 9.545032675245813e-05,
      "loss": 6.755,
      "step": 36
    },
    {
      "epoch": 0.42165242165242167,
      "grad_norm": 3.145620822906494,
      "learning_rate": 9.509956150664796e-05,
      "loss": 6.5514,
      "step": 37
    },
    {
      "epoch": 0.43304843304843305,
      "grad_norm": 3.0899741649627686,
      "learning_rate": 9.473646649103818e-05,
      "loss": 6.4334,
      "step": 38
    },
    {
      "epoch": 0.4444444444444444,
      "grad_norm": 3.6191301345825195,
      "learning_rate": 9.43611409721806e-05,
      "loss": 7.3383,
      "step": 39
    },
    {
      "epoch": 0.45584045584045585,
      "grad_norm": 3.4292280673980713,
      "learning_rate": 9.397368756032445e-05,
      "loss": 6.9495,
      "step": 40
    },
    {
      "epoch": 0.4672364672364672,
      "grad_norm": 3.5323736667633057,
      "learning_rate": 9.357421218136386e-05,
      "loss": 7.0832,
      "step": 41
    },
    {
      "epoch": 0.47863247863247865,
      "grad_norm": 4.316342353820801,
      "learning_rate": 9.316282404787871e-05,
      "loss": 7.9894,
      "step": 42
    },
    {
      "epoch": 0.49002849002849,
      "grad_norm": 1.7970763444900513,
      "learning_rate": 9.273963562927695e-05,
      "loss": 5.5176,
      "step": 43
    },
    {
      "epoch": 0.5014245014245015,
      "grad_norm": 1.9848921298980713,
      "learning_rate": 9.230476262104677e-05,
      "loss": 5.2071,
      "step": 44
    },
    {
      "epoch": 0.5128205128205128,
      "grad_norm": 2.0237419605255127,
      "learning_rate": 9.185832391312644e-05,
      "loss": 5.3054,
      "step": 45
    },
    {
      "epoch": 0.5242165242165242,
      "grad_norm": 1.8416670560836792,
      "learning_rate": 9.140044155740101e-05,
      "loss": 4.8872,
      "step": 46
    },
    {
      "epoch": 0.5356125356125356,
      "grad_norm": 1.8425672054290771,
      "learning_rate": 9.093124073433463e-05,
      "loss": 5.3802,
      "step": 47
    },
    {
      "epoch": 0.5470085470085471,
      "grad_norm": 1.806503415107727,
      "learning_rate": 9.045084971874738e-05,
      "loss": 5.6238,
      "step": 48
    },
    {
      "epoch": 0.5584045584045584,
      "grad_norm": 1.8732221126556396,
      "learning_rate": 8.995939984474624e-05,
      "loss": 5.4535,
      "step": 49
    },
    {
      "epoch": 0.5698005698005698,
      "grad_norm": 2.0408949851989746,
      "learning_rate": 8.945702546981969e-05,
      "loss": 5.8758,
      "step": 50
    },
    {
      "epoch": 0.5698005698005698,
      "eval_loss": 1.539510726928711,
      "eval_runtime": 15.3095,
      "eval_samples_per_second": 9.667,
      "eval_steps_per_second": 2.417,
      "step": 50
    },
    {
      "epoch": 0.5811965811965812,
      "grad_norm": 1.8603322505950928,
      "learning_rate": 8.894386393810563e-05,
      "loss": 5.6762,
      "step": 51
    },
    {
      "epoch": 0.5925925925925926,
      "grad_norm": 2.074678421020508,
      "learning_rate": 8.842005554284296e-05,
      "loss": 5.5585,
      "step": 52
    },
    {
      "epoch": 0.603988603988604,
      "grad_norm": 2.094933271408081,
      "learning_rate": 8.788574348801675e-05,
      "loss": 5.3675,
      "step": 53
    },
    {
      "epoch": 0.6153846153846154,
      "grad_norm": 2.75547194480896,
      "learning_rate": 8.73410738492077e-05,
      "loss": 6.5333,
      "step": 54
    },
    {
      "epoch": 0.6267806267806267,
      "grad_norm": 2.6254167556762695,
      "learning_rate": 8.678619553365659e-05,
      "loss": 5.8712,
      "step": 55
    },
    {
      "epoch": 0.6381766381766382,
      "grad_norm": 2.7161309719085693,
      "learning_rate": 8.622126023955446e-05,
      "loss": 5.845,
      "step": 56
    },
    {
      "epoch": 0.6495726495726496,
      "grad_norm": 3.127302885055542,
      "learning_rate": 8.564642241456986e-05,
      "loss": 6.3484,
      "step": 57
    },
    {
      "epoch": 0.6609686609686609,
      "grad_norm": 3.096575975418091,
      "learning_rate": 8.506183921362443e-05,
      "loss": 5.9578,
      "step": 58
    },
    {
      "epoch": 0.6723646723646723,
      "grad_norm": 3.587453842163086,
      "learning_rate": 8.44676704559283e-05,
      "loss": 6.3445,
      "step": 59
    },
    {
      "epoch": 0.6837606837606838,
      "grad_norm": 3.5581254959106445,
      "learning_rate": 8.386407858128706e-05,
      "loss": 6.3916,
      "step": 60
    },
    {
      "epoch": 0.6951566951566952,
      "grad_norm": 3.776169538497925,
      "learning_rate": 8.32512286056924e-05,
      "loss": 6.3217,
      "step": 61
    },
    {
      "epoch": 0.7065527065527065,
      "grad_norm": 3.76227068901062,
      "learning_rate": 8.262928807620843e-05,
      "loss": 6.2981,
      "step": 62
    },
    {
      "epoch": 0.717948717948718,
      "grad_norm": 5.723356246948242,
      "learning_rate": 8.199842702516583e-05,
      "loss": 7.3963,
      "step": 63
    },
    {
      "epoch": 0.7293447293447294,
      "grad_norm": 1.6600637435913086,
      "learning_rate": 8.135881792367686e-05,
      "loss": 4.9299,
      "step": 64
    },
    {
      "epoch": 0.7407407407407407,
      "grad_norm": 2.031022787094116,
      "learning_rate": 8.07106356344834e-05,
      "loss": 4.6597,
      "step": 65
    },
    {
      "epoch": 0.7521367521367521,
      "grad_norm": 2.0800983905792236,
      "learning_rate": 8.005405736415126e-05,
      "loss": 4.809,
      "step": 66
    },
    {
      "epoch": 0.7635327635327636,
      "grad_norm": 1.9453895092010498,
      "learning_rate": 7.938926261462366e-05,
      "loss": 4.7113,
      "step": 67
    },
    {
      "epoch": 0.7749287749287749,
      "grad_norm": 2.313082218170166,
      "learning_rate": 7.871643313414718e-05,
      "loss": 5.2085,
      "step": 68
    },
    {
      "epoch": 0.7863247863247863,
      "grad_norm": 2.1648385524749756,
      "learning_rate": 7.803575286758364e-05,
      "loss": 4.9098,
      "step": 69
    },
    {
      "epoch": 0.7977207977207977,
      "grad_norm": 2.59464693069458,
      "learning_rate": 7.734740790612136e-05,
      "loss": 5.3112,
      "step": 70
    },
    {
      "epoch": 0.8091168091168092,
      "grad_norm": 2.403724193572998,
      "learning_rate": 7.66515864363997e-05,
      "loss": 5.3834,
      "step": 71
    },
    {
      "epoch": 0.8205128205128205,
      "grad_norm": 2.4704670906066895,
      "learning_rate": 7.594847868906076e-05,
      "loss": 5.5864,
      "step": 72
    },
    {
      "epoch": 0.8319088319088319,
      "grad_norm": 2.5529868602752686,
      "learning_rate": 7.52382768867422e-05,
      "loss": 5.7712,
      "step": 73
    },
    {
      "epoch": 0.8433048433048433,
      "grad_norm": 2.5005476474761963,
      "learning_rate": 7.452117519152542e-05,
      "loss": 5.4132,
      "step": 74
    },
    {
      "epoch": 0.8547008547008547,
      "grad_norm": 2.534529685974121,
      "learning_rate": 7.379736965185368e-05,
      "loss": 5.5437,
      "step": 75
    },
    {
      "epoch": 0.8660968660968661,
      "grad_norm": 2.9444375038146973,
      "learning_rate": 7.30670581489344e-05,
      "loss": 5.856,
      "step": 76
    },
    {
      "epoch": 0.8774928774928775,
      "grad_norm": 2.935525417327881,
      "learning_rate": 7.233044034264034e-05,
      "loss": 6.0189,
      "step": 77
    },
    {
      "epoch": 0.8888888888888888,
      "grad_norm": 3.1442325115203857,
      "learning_rate": 7.158771761692464e-05,
      "loss": 5.5968,
      "step": 78
    },
    {
      "epoch": 0.9002849002849003,
      "grad_norm": 3.431941509246826,
      "learning_rate": 7.083909302476453e-05,
      "loss": 5.4577,
      "step": 79
    },
    {
      "epoch": 0.9116809116809117,
      "grad_norm": 3.3542709350585938,
      "learning_rate": 7.008477123264848e-05,
      "loss": 5.9354,
      "step": 80
    },
    {
      "epoch": 0.9230769230769231,
      "grad_norm": 3.749603748321533,
      "learning_rate": 6.932495846462261e-05,
      "loss": 6.2203,
      "step": 81
    },
    {
      "epoch": 0.9344729344729344,
      "grad_norm": 4.117802143096924,
      "learning_rate": 6.855986244591104e-05,
      "loss": 5.8782,
      "step": 82
    },
    {
      "epoch": 0.9458689458689459,
      "grad_norm": 4.240657806396484,
      "learning_rate": 6.778969234612584e-05,
      "loss": 6.3362,
      "step": 83
    },
    {
      "epoch": 0.9572649572649573,
      "grad_norm": 4.9266276359558105,
      "learning_rate": 6.701465872208216e-05,
      "loss": 6.9487,
      "step": 84
    },
    {
      "epoch": 0.9686609686609686,
      "grad_norm": 2.111222743988037,
      "learning_rate": 6.623497346023418e-05,
      "loss": 4.9511,
      "step": 85
    },
    {
      "epoch": 0.98005698005698,
      "grad_norm": 2.8679890632629395,
      "learning_rate": 6.545084971874738e-05,
      "loss": 4.901,
      "step": 86
    },
    {
      "epoch": 0.9914529914529915,
      "grad_norm": 3.5176985263824463,
      "learning_rate": 6.466250186922325e-05,
      "loss": 5.9415,
      "step": 87
    },
    {
      "epoch": 1.0085470085470085,
      "grad_norm": 2.208169937133789,
      "learning_rate": 6.387014543809223e-05,
      "loss": 4.8643,
      "step": 88
    },
    {
      "epoch": 1.01994301994302,
      "grad_norm": 1.5939043760299683,
      "learning_rate": 6.307399704769099e-05,
      "loss": 4.0825,
      "step": 89
    },
    {
      "epoch": 1.0313390313390314,
      "grad_norm": 1.7511546611785889,
      "learning_rate": 6.227427435703997e-05,
      "loss": 4.3576,
      "step": 90
    },
    {
      "epoch": 1.0427350427350428,
      "grad_norm": 1.9381104707717896,
      "learning_rate": 6.147119600233758e-05,
      "loss": 3.9595,
      "step": 91
    },
    {
      "epoch": 1.0541310541310542,
      "grad_norm": 2.012346029281616,
      "learning_rate": 6.066498153718735e-05,
      "loss": 4.4896,
      "step": 92
    },
    {
      "epoch": 1.0655270655270654,
      "grad_norm": 2.2163355350494385,
      "learning_rate": 5.985585137257401e-05,
      "loss": 4.4412,
      "step": 93
    },
    {
      "epoch": 1.0769230769230769,
      "grad_norm": 2.5285067558288574,
      "learning_rate": 5.90440267166055e-05,
      "loss": 4.4152,
      "step": 94
    },
    {
      "epoch": 1.0883190883190883,
      "grad_norm": 2.401972532272339,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 4.3998,
      "step": 95
    },
    {
      "epoch": 1.0997150997150997,
      "grad_norm": 2.8148128986358643,
      "learning_rate": 5.74131823855921e-05,
      "loss": 4.3649,
      "step": 96
    },
    {
      "epoch": 1.1111111111111112,
      "grad_norm": 2.7760674953460693,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 4.4393,
      "step": 97
    },
    {
      "epoch": 1.1225071225071226,
      "grad_norm": 2.9436051845550537,
      "learning_rate": 5.577423184847932e-05,
      "loss": 4.4825,
      "step": 98
    },
    {
      "epoch": 1.133903133903134,
      "grad_norm": 3.12250018119812,
      "learning_rate": 5.495227651252315e-05,
      "loss": 4.7238,
      "step": 99
    },
    {
      "epoch": 1.1452991452991452,
      "grad_norm": 3.8211746215820312,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 4.3247,
      "step": 100
    },
    {
      "epoch": 1.1452991452991452,
      "eval_loss": 1.4336222410202026,
      "eval_runtime": 15.2996,
      "eval_samples_per_second": 9.673,
      "eval_steps_per_second": 2.418,
      "step": 100
    },
    {
      "epoch": 1.1566951566951567,
      "grad_norm": 3.5345652103424072,
      "learning_rate": 5.330452921628497e-05,
      "loss": 4.5452,
      "step": 101
    },
    {
      "epoch": 1.168091168091168,
      "grad_norm": 3.6309292316436768,
      "learning_rate": 5.247918773366112e-05,
      "loss": 4.6139,
      "step": 102
    },
    {
      "epoch": 1.1794871794871795,
      "grad_norm": 3.7351112365722656,
      "learning_rate": 5.165316846586541e-05,
      "loss": 4.0948,
      "step": 103
    },
    {
      "epoch": 1.190883190883191,
      "grad_norm": 4.135587215423584,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 4.3645,
      "step": 104
    },
    {
      "epoch": 1.2022792022792024,
      "grad_norm": 4.397957801818848,
      "learning_rate": 5e-05,
      "loss": 4.9212,
      "step": 105
    },
    {
      "epoch": 1.2136752136752136,
      "grad_norm": 5.19450044631958,
      "learning_rate": 4.917330276168208e-05,
      "loss": 4.4998,
      "step": 106
    },
    {
      "epoch": 1.225071225071225,
      "grad_norm": 5.325326442718506,
      "learning_rate": 4.834683153413459e-05,
      "loss": 4.6552,
      "step": 107
    },
    {
      "epoch": 1.2364672364672364,
      "grad_norm": 7.047455787658691,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 5.4832,
      "step": 108
    },
    {
      "epoch": 1.2478632478632479,
      "grad_norm": 4.013706207275391,
      "learning_rate": 4.669547078371504e-05,
      "loss": 4.3359,
      "step": 109
    },
    {
      "epoch": 1.2592592592592593,
      "grad_norm": 2.7769999504089355,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 4.5682,
      "step": 110
    },
    {
      "epoch": 1.2706552706552707,
      "grad_norm": 3.2002756595611572,
      "learning_rate": 4.504772348747687e-05,
      "loss": 4.3197,
      "step": 111
    },
    {
      "epoch": 1.282051282051282,
      "grad_norm": 3.232100486755371,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 3.9719,
      "step": 112
    },
    {
      "epoch": 1.2934472934472934,
      "grad_norm": 3.339895486831665,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 4.3998,
      "step": 113
    },
    {
      "epoch": 1.3048433048433048,
      "grad_norm": 3.3506736755371094,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 4.0521,
      "step": 114
    },
    {
      "epoch": 1.3162393162393162,
      "grad_norm": 3.607597827911377,
      "learning_rate": 4.17702704859633e-05,
      "loss": 3.972,
      "step": 115
    },
    {
      "epoch": 1.3276353276353277,
      "grad_norm": 4.277341365814209,
      "learning_rate": 4.095597328339452e-05,
      "loss": 4.735,
      "step": 116
    },
    {
      "epoch": 1.339031339031339,
      "grad_norm": 3.293569564819336,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 4.4544,
      "step": 117
    },
    {
      "epoch": 1.3504273504273505,
      "grad_norm": 3.6193156242370605,
      "learning_rate": 3.933501846281267e-05,
      "loss": 4.3191,
      "step": 118
    },
    {
      "epoch": 1.3618233618233617,
      "grad_norm": 3.550133466720581,
      "learning_rate": 3.852880399766243e-05,
      "loss": 4.0343,
      "step": 119
    },
    {
      "epoch": 1.3732193732193732,
      "grad_norm": 3.585671901702881,
      "learning_rate": 3.772572564296005e-05,
      "loss": 4.5684,
      "step": 120
    },
    {
      "epoch": 1.3846153846153846,
      "grad_norm": 3.9105350971221924,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 4.3923,
      "step": 121
    },
    {
      "epoch": 1.396011396011396,
      "grad_norm": 4.082669734954834,
      "learning_rate": 3.612985456190778e-05,
      "loss": 4.3241,
      "step": 122
    },
    {
      "epoch": 1.4074074074074074,
      "grad_norm": 4.256656169891357,
      "learning_rate": 3.533749813077677e-05,
      "loss": 4.296,
      "step": 123
    },
    {
      "epoch": 1.4188034188034189,
      "grad_norm": 3.9480738639831543,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 4.1513,
      "step": 124
    },
    {
      "epoch": 1.4301994301994303,
      "grad_norm": 4.2875075340271,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 4.0438,
      "step": 125
    },
    {
      "epoch": 1.4415954415954415,
      "grad_norm": 4.951995372772217,
      "learning_rate": 3.298534127791785e-05,
      "loss": 4.2859,
      "step": 126
    },
    {
      "epoch": 1.452991452991453,
      "grad_norm": 5.24794340133667,
      "learning_rate": 3.221030765387417e-05,
      "loss": 4.3501,
      "step": 127
    },
    {
      "epoch": 1.4643874643874644,
      "grad_norm": 5.447783470153809,
      "learning_rate": 3.144013755408895e-05,
      "loss": 4.3088,
      "step": 128
    },
    {
      "epoch": 1.4757834757834758,
      "grad_norm": 6.2533488273620605,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 4.5398,
      "step": 129
    },
    {
      "epoch": 1.4871794871794872,
      "grad_norm": 4.588549613952637,
      "learning_rate": 2.991522876735154e-05,
      "loss": 4.6695,
      "step": 130
    },
    {
      "epoch": 1.4985754985754987,
      "grad_norm": 3.1861557960510254,
      "learning_rate": 2.916090697523549e-05,
      "loss": 4.2024,
      "step": 131
    },
    {
      "epoch": 1.50997150997151,
      "grad_norm": 3.2797012329101562,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 4.2605,
      "step": 132
    },
    {
      "epoch": 1.5213675213675213,
      "grad_norm": 3.2831523418426514,
      "learning_rate": 2.766955965735968e-05,
      "loss": 4.1959,
      "step": 133
    },
    {
      "epoch": 1.5327635327635327,
      "grad_norm": 3.3353328704833984,
      "learning_rate": 2.693294185106562e-05,
      "loss": 3.9844,
      "step": 134
    },
    {
      "epoch": 1.5441595441595442,
      "grad_norm": 3.5544533729553223,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 4.2347,
      "step": 135
    },
    {
      "epoch": 1.5555555555555556,
      "grad_norm": 4.138932228088379,
      "learning_rate": 2.547882480847461e-05,
      "loss": 4.0827,
      "step": 136
    },
    {
      "epoch": 1.566951566951567,
      "grad_norm": 4.110633373260498,
      "learning_rate": 2.476172311325783e-05,
      "loss": 4.6006,
      "step": 137
    },
    {
      "epoch": 1.5783475783475782,
      "grad_norm": 4.300783157348633,
      "learning_rate": 2.405152131093926e-05,
      "loss": 4.1072,
      "step": 138
    },
    {
      "epoch": 1.5897435897435899,
      "grad_norm": 4.127902030944824,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 4.0142,
      "step": 139
    },
    {
      "epoch": 1.601139601139601,
      "grad_norm": 3.928335428237915,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 3.9498,
      "step": 140
    },
    {
      "epoch": 1.6125356125356125,
      "grad_norm": 4.345295429229736,
      "learning_rate": 2.196424713241637e-05,
      "loss": 4.0087,
      "step": 141
    },
    {
      "epoch": 1.623931623931624,
      "grad_norm": 4.230784893035889,
      "learning_rate": 2.128356686585282e-05,
      "loss": 3.8495,
      "step": 142
    },
    {
      "epoch": 1.6353276353276354,
      "grad_norm": 4.621153831481934,
      "learning_rate": 2.061073738537635e-05,
      "loss": 4.1292,
      "step": 143
    },
    {
      "epoch": 1.6467236467236468,
      "grad_norm": 5.0727033615112305,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 4.6743,
      "step": 144
    },
    {
      "epoch": 1.658119658119658,
      "grad_norm": 5.030786991119385,
      "learning_rate": 1.928936436551661e-05,
      "loss": 4.2299,
      "step": 145
    },
    {
      "epoch": 1.6695156695156697,
      "grad_norm": 4.871549606323242,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 3.9124,
      "step": 146
    },
    {
      "epoch": 1.6809116809116809,
      "grad_norm": 5.950706958770752,
      "learning_rate": 1.800157297483417e-05,
      "loss": 4.3492,
      "step": 147
    },
    {
      "epoch": 1.6923076923076923,
      "grad_norm": 5.761005401611328,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 4.1399,
      "step": 148
    },
    {
      "epoch": 1.7037037037037037,
      "grad_norm": 6.105738162994385,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 4.1512,
      "step": 149
    },
    {
      "epoch": 1.7150997150997151,
      "grad_norm": 7.922430038452148,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 5.3958,
      "step": 150
    },
    {
      "epoch": 1.7150997150997151,
      "eval_loss": 1.4163821935653687,
      "eval_runtime": 15.3078,
      "eval_samples_per_second": 9.668,
      "eval_steps_per_second": 2.417,
      "step": 150
    },
    {
      "epoch": 1.7264957264957266,
      "grad_norm": 4.421020984649658,
      "learning_rate": 1.553232954407171e-05,
      "loss": 4.8287,
      "step": 151
    },
    {
      "epoch": 1.7378917378917378,
      "grad_norm": 2.5518932342529297,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 4.1319,
      "step": 152
    },
    {
      "epoch": 1.7492877492877494,
      "grad_norm": 3.3472037315368652,
      "learning_rate": 1.435357758543015e-05,
      "loss": 3.9143,
      "step": 153
    },
    {
      "epoch": 1.7606837606837606,
      "grad_norm": 3.0143449306488037,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 3.7723,
      "step": 154
    },
    {
      "epoch": 1.772079772079772,
      "grad_norm": 3.4963533878326416,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 4.1917,
      "step": 155
    },
    {
      "epoch": 1.7834757834757835,
      "grad_norm": 3.3898444175720215,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 3.8471,
      "step": 156
    },
    {
      "epoch": 1.7948717948717947,
      "grad_norm": 4.001238822937012,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 4.1722,
      "step": 157
    },
    {
      "epoch": 1.8062678062678064,
      "grad_norm": 3.774350881576538,
      "learning_rate": 1.157994445715706e-05,
      "loss": 4.0202,
      "step": 158
    },
    {
      "epoch": 1.8176638176638176,
      "grad_norm": 3.82694935798645,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 4.3685,
      "step": 159
    },
    {
      "epoch": 1.8290598290598292,
      "grad_norm": 4.0727972984313965,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 4.0844,
      "step": 160
    },
    {
      "epoch": 1.8404558404558404,
      "grad_norm": 4.442108154296875,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 4.3282,
      "step": 161
    },
    {
      "epoch": 1.8518518518518519,
      "grad_norm": 4.567256927490234,
      "learning_rate": 9.549150281252633e-06,
      "loss": 4.2291,
      "step": 162
    },
    {
      "epoch": 1.8632478632478633,
      "grad_norm": 4.891412734985352,
      "learning_rate": 9.068759265665384e-06,
      "loss": 4.2728,
      "step": 163
    },
    {
      "epoch": 1.8746438746438745,
      "grad_norm": 4.913301944732666,
      "learning_rate": 8.599558442598998e-06,
      "loss": 3.8642,
      "step": 164
    },
    {
      "epoch": 1.8860398860398861,
      "grad_norm": 5.405317306518555,
      "learning_rate": 8.141676086873572e-06,
      "loss": 4.5023,
      "step": 165
    },
    {
      "epoch": 1.8974358974358974,
      "grad_norm": 5.18913459777832,
      "learning_rate": 7.695237378953223e-06,
      "loss": 3.965,
      "step": 166
    },
    {
      "epoch": 1.9088319088319088,
      "grad_norm": 5.323193550109863,
      "learning_rate": 7.260364370723044e-06,
      "loss": 4.3581,
      "step": 167
    },
    {
      "epoch": 1.9202279202279202,
      "grad_norm": 5.793462753295898,
      "learning_rate": 6.837175952121306e-06,
      "loss": 4.2465,
      "step": 168
    },
    {
      "epoch": 1.9316239316239316,
      "grad_norm": 5.612836837768555,
      "learning_rate": 6.425787818636131e-06,
      "loss": 3.9352,
      "step": 169
    },
    {
      "epoch": 1.943019943019943,
      "grad_norm": 6.385334014892578,
      "learning_rate": 6.026312439675552e-06,
      "loss": 4.3491,
      "step": 170
    },
    {
      "epoch": 1.9544159544159543,
      "grad_norm": 6.797797679901123,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 4.0942,
      "step": 171
    },
    {
      "epoch": 1.965811965811966,
      "grad_norm": 4.70216703414917,
      "learning_rate": 5.263533508961827e-06,
      "loss": 4.4347,
      "step": 172
    },
    {
      "epoch": 1.9772079772079771,
      "grad_norm": 3.5790581703186035,
      "learning_rate": 4.900438493352055e-06,
      "loss": 3.7052,
      "step": 173
    },
    {
      "epoch": 1.9886039886039886,
      "grad_norm": 4.768205165863037,
      "learning_rate": 4.549673247541875e-06,
      "loss": 4.2112,
      "step": 174
    },
    {
      "epoch": 2.005698005698006,
      "grad_norm": 3.689624547958374,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 4.0812,
      "step": 175
    },
    {
      "epoch": 2.017094017094017,
      "grad_norm": 2.281919479370117,
      "learning_rate": 3.885512251130763e-06,
      "loss": 3.8929,
      "step": 176
    },
    {
      "epoch": 2.0284900284900287,
      "grad_norm": 2.6691129207611084,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 3.4994,
      "step": 177
    },
    {
      "epoch": 2.03988603988604,
      "grad_norm": 2.6085894107818604,
      "learning_rate": 3.271776770026963e-06,
      "loss": 3.0444,
      "step": 178
    },
    {
      "epoch": 2.051282051282051,
      "grad_norm": 2.897402286529541,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 3.0561,
      "step": 179
    },
    {
      "epoch": 2.0626780626780628,
      "grad_norm": 2.9752447605133057,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 3.3909,
      "step": 180
    },
    {
      "epoch": 2.074074074074074,
      "grad_norm": 3.2343292236328125,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 3.4835,
      "step": 181
    },
    {
      "epoch": 2.0854700854700856,
      "grad_norm": 3.5127015113830566,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 3.5321,
      "step": 182
    },
    {
      "epoch": 2.096866096866097,
      "grad_norm": 3.42910099029541,
      "learning_rate": 1.962316193157593e-06,
      "loss": 3.7008,
      "step": 183
    },
    {
      "epoch": 2.1082621082621085,
      "grad_norm": 3.5284929275512695,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 3.5783,
      "step": 184
    },
    {
      "epoch": 2.1196581196581197,
      "grad_norm": 3.772944927215576,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 3.8424,
      "step": 185
    },
    {
      "epoch": 2.131054131054131,
      "grad_norm": 3.4069995880126953,
      "learning_rate": 1.333670137599713e-06,
      "loss": 3.4327,
      "step": 186
    },
    {
      "epoch": 2.1424501424501425,
      "grad_norm": 3.828049421310425,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 3.5472,
      "step": 187
    },
    {
      "epoch": 2.1538461538461537,
      "grad_norm": 4.0321245193481445,
      "learning_rate": 9.810017062595322e-07,
      "loss": 3.5372,
      "step": 188
    },
    {
      "epoch": 2.1652421652421654,
      "grad_norm": 4.3415703773498535,
      "learning_rate": 8.247462563808817e-07,
      "loss": 3.3885,
      "step": 189
    },
    {
      "epoch": 2.1766381766381766,
      "grad_norm": 4.1906914710998535,
      "learning_rate": 6.819348298638839e-07,
      "loss": 3.2704,
      "step": 190
    },
    {
      "epoch": 2.1880341880341883,
      "grad_norm": 4.669715881347656,
      "learning_rate": 5.526064699265753e-07,
      "loss": 3.5131,
      "step": 191
    },
    {
      "epoch": 2.1994301994301995,
      "grad_norm": 5.2640814781188965,
      "learning_rate": 4.367965336512403e-07,
      "loss": 3.7865,
      "step": 192
    },
    {
      "epoch": 2.2108262108262107,
      "grad_norm": 5.3306193351745605,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 3.7096,
      "step": 193
    },
    {
      "epoch": 2.2222222222222223,
      "grad_norm": 5.284204006195068,
      "learning_rate": 2.458548727494292e-07,
      "loss": 3.6549,
      "step": 194
    },
    {
      "epoch": 2.2336182336182335,
      "grad_norm": 6.041280746459961,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 3.5195,
      "step": 195
    },
    {
      "epoch": 2.245014245014245,
      "grad_norm": 5.608485698699951,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 4.346,
      "step": 196
    },
    {
      "epoch": 2.2564102564102564,
      "grad_norm": 2.6626598834991455,
      "learning_rate": 6.150154258476315e-08,
      "loss": 3.5209,
      "step": 197
    },
    {
      "epoch": 2.267806267806268,
      "grad_norm": 3.21748423576355,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 3.8922,
      "step": 198
    },
    {
      "epoch": 2.2792022792022792,
      "grad_norm": 3.1684672832489014,
      "learning_rate": 6.834750376549792e-09,
      "loss": 3.5994,
      "step": 199
    },
    {
      "epoch": 2.2905982905982905,
      "grad_norm": 3.193687915802002,
      "learning_rate": 0.0,
      "loss": 3.3862,
      "step": 200
    },
    {
      "epoch": 2.2905982905982905,
      "eval_loss": 1.4081310033798218,
      "eval_runtime": 15.2931,
      "eval_samples_per_second": 9.678,
      "eval_steps_per_second": 2.419,
      "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": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5.09187566075904e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}