{
  "best_metric": 1.3650026321411133,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.020255215718047396,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00010127607859023699,
      "grad_norm": 31.278818130493164,
      "learning_rate": 1e-05,
      "loss": 5.6035,
      "step": 1
    },
    {
      "epoch": 0.00010127607859023699,
      "eval_loss": 6.345576286315918,
      "eval_runtime": 1564.2303,
      "eval_samples_per_second": 10.631,
      "eval_steps_per_second": 2.658,
      "step": 1
    },
    {
      "epoch": 0.00020255215718047398,
      "grad_norm": 23.076887130737305,
      "learning_rate": 2e-05,
      "loss": 5.6016,
      "step": 2
    },
    {
      "epoch": 0.00030382823577071097,
      "grad_norm": 30.049488067626953,
      "learning_rate": 3e-05,
      "loss": 5.0056,
      "step": 3
    },
    {
      "epoch": 0.00040510431436094796,
      "grad_norm": 21.293628692626953,
      "learning_rate": 4e-05,
      "loss": 4.0928,
      "step": 4
    },
    {
      "epoch": 0.0005063803929511849,
      "grad_norm": 51.72479248046875,
      "learning_rate": 5e-05,
      "loss": 3.268,
      "step": 5
    },
    {
      "epoch": 0.0006076564715414219,
      "grad_norm": 69.23747253417969,
      "learning_rate": 6e-05,
      "loss": 2.4954,
      "step": 6
    },
    {
      "epoch": 0.0007089325501316589,
      "grad_norm": 12.229398727416992,
      "learning_rate": 7e-05,
      "loss": 1.7844,
      "step": 7
    },
    {
      "epoch": 0.0008102086287218959,
      "grad_norm": 14.901517868041992,
      "learning_rate": 8e-05,
      "loss": 1.5585,
      "step": 8
    },
    {
      "epoch": 0.0009114847073121329,
      "grad_norm": 9.97788143157959,
      "learning_rate": 9e-05,
      "loss": 1.5229,
      "step": 9
    },
    {
      "epoch": 0.0010127607859023698,
      "grad_norm": 17.119428634643555,
      "learning_rate": 0.0001,
      "loss": 1.5665,
      "step": 10
    },
    {
      "epoch": 0.001114036864492607,
      "grad_norm": 8.448174476623535,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.2061,
      "step": 11
    },
    {
      "epoch": 0.0012153129430828439,
      "grad_norm": 6.501319408416748,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.2982,
      "step": 12
    },
    {
      "epoch": 0.0013165890216730808,
      "grad_norm": 6.863130569458008,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.4555,
      "step": 13
    },
    {
      "epoch": 0.0014178651002633178,
      "grad_norm": 7.778407096862793,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.4297,
      "step": 14
    },
    {
      "epoch": 0.0015191411788535547,
      "grad_norm": 4.9213385581970215,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.478,
      "step": 15
    },
    {
      "epoch": 0.0016204172574437918,
      "grad_norm": 6.808879375457764,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.6191,
      "step": 16
    },
    {
      "epoch": 0.0017216933360340288,
      "grad_norm": 4.654536724090576,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.3514,
      "step": 17
    },
    {
      "epoch": 0.0018229694146242657,
      "grad_norm": 5.336677551269531,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.3981,
      "step": 18
    },
    {
      "epoch": 0.0019242454932145027,
      "grad_norm": 5.827622413635254,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.3458,
      "step": 19
    },
    {
      "epoch": 0.0020255215718047396,
      "grad_norm": 6.260201454162598,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.7247,
      "step": 20
    },
    {
      "epoch": 0.0021267976503949767,
      "grad_norm": 9.512893676757812,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.7886,
      "step": 21
    },
    {
      "epoch": 0.002228073728985214,
      "grad_norm": 6.11754846572876,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.6451,
      "step": 22
    },
    {
      "epoch": 0.0023293498075754506,
      "grad_norm": 5.766722679138184,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.506,
      "step": 23
    },
    {
      "epoch": 0.0024306258861656878,
      "grad_norm": 6.981493949890137,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.822,
      "step": 24
    },
    {
      "epoch": 0.0025319019647559245,
      "grad_norm": 5.49668025970459,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.6161,
      "step": 25
    },
    {
      "epoch": 0.0026331780433461616,
      "grad_norm": 6.683679580688477,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.6799,
      "step": 26
    },
    {
      "epoch": 0.002734454121936399,
      "grad_norm": 6.836819648742676,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.5681,
      "step": 27
    },
    {
      "epoch": 0.0028357302005266355,
      "grad_norm": 5.405344486236572,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.603,
      "step": 28
    },
    {
      "epoch": 0.0029370062791168727,
      "grad_norm": 5.259994029998779,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.6561,
      "step": 29
    },
    {
      "epoch": 0.0030382823577071094,
      "grad_norm": 4.639139175415039,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.7005,
      "step": 30
    },
    {
      "epoch": 0.0031395584362973465,
      "grad_norm": 8.204778671264648,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.2015,
      "step": 31
    },
    {
      "epoch": 0.0032408345148875837,
      "grad_norm": 5.750695705413818,
      "learning_rate": 9.672822322997305e-05,
      "loss": 2.1686,
      "step": 32
    },
    {
      "epoch": 0.0033421105934778204,
      "grad_norm": 4.413594722747803,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.9129,
      "step": 33
    },
    {
      "epoch": 0.0034433866720680576,
      "grad_norm": 3.6709742546081543,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.7818,
      "step": 34
    },
    {
      "epoch": 0.0035446627506582947,
      "grad_norm": 3.3729166984558105,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.7133,
      "step": 35
    },
    {
      "epoch": 0.0036459388292485314,
      "grad_norm": 3.6017041206359863,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.5988,
      "step": 36
    },
    {
      "epoch": 0.0037472149078387686,
      "grad_norm": 2.8339967727661133,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.7085,
      "step": 37
    },
    {
      "epoch": 0.0038484909864290053,
      "grad_norm": 5.1753644943237305,
      "learning_rate": 9.473646649103818e-05,
      "loss": 2.3217,
      "step": 38
    },
    {
      "epoch": 0.0039497670650192425,
      "grad_norm": 8.048284530639648,
      "learning_rate": 9.43611409721806e-05,
      "loss": 2.868,
      "step": 39
    },
    {
      "epoch": 0.004051043143609479,
      "grad_norm": 8.418875694274902,
      "learning_rate": 9.397368756032445e-05,
      "loss": 3.1661,
      "step": 40
    },
    {
      "epoch": 0.004152319222199717,
      "grad_norm": 7.274316310882568,
      "learning_rate": 9.357421218136386e-05,
      "loss": 2.6893,
      "step": 41
    },
    {
      "epoch": 0.0042535953007899535,
      "grad_norm": 5.224125385284424,
      "learning_rate": 9.316282404787871e-05,
      "loss": 2.3319,
      "step": 42
    },
    {
      "epoch": 0.00435487137938019,
      "grad_norm": 6.756013870239258,
      "learning_rate": 9.273963562927695e-05,
      "loss": 2.4331,
      "step": 43
    },
    {
      "epoch": 0.004456147457970428,
      "grad_norm": 4.957799911499023,
      "learning_rate": 9.230476262104677e-05,
      "loss": 2.3517,
      "step": 44
    },
    {
      "epoch": 0.0045574235365606645,
      "grad_norm": 5.128690719604492,
      "learning_rate": 9.185832391312644e-05,
      "loss": 2.304,
      "step": 45
    },
    {
      "epoch": 0.004658699615150901,
      "grad_norm": 8.41048526763916,
      "learning_rate": 9.140044155740101e-05,
      "loss": 2.8535,
      "step": 46
    },
    {
      "epoch": 0.004759975693741138,
      "grad_norm": 6.968003273010254,
      "learning_rate": 9.093124073433463e-05,
      "loss": 2.6746,
      "step": 47
    },
    {
      "epoch": 0.0048612517723313755,
      "grad_norm": 5.947181224822998,
      "learning_rate": 9.045084971874738e-05,
      "loss": 2.7757,
      "step": 48
    },
    {
      "epoch": 0.004962527850921612,
      "grad_norm": 4.515350818634033,
      "learning_rate": 8.995939984474624e-05,
      "loss": 2.4865,
      "step": 49
    },
    {
      "epoch": 0.005063803929511849,
      "grad_norm": 5.1472296714782715,
      "learning_rate": 8.945702546981969e-05,
      "loss": 2.5788,
      "step": 50
    },
    {
      "epoch": 0.005063803929511849,
      "eval_loss": 2.3614883422851562,
      "eval_runtime": 1571.6178,
      "eval_samples_per_second": 10.581,
      "eval_steps_per_second": 2.646,
      "step": 50
    },
    {
      "epoch": 0.0051650800081020866,
      "grad_norm": 10.651284217834473,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.9172,
      "step": 51
    },
    {
      "epoch": 0.005266356086692323,
      "grad_norm": 7.893736362457275,
      "learning_rate": 8.842005554284296e-05,
      "loss": 2.101,
      "step": 52
    },
    {
      "epoch": 0.00536763216528256,
      "grad_norm": 5.897404670715332,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.436,
      "step": 53
    },
    {
      "epoch": 0.005468908243872798,
      "grad_norm": 4.449448585510254,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.2535,
      "step": 54
    },
    {
      "epoch": 0.005570184322463034,
      "grad_norm": 4.3071699142456055,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.4453,
      "step": 55
    },
    {
      "epoch": 0.005671460401053271,
      "grad_norm": 3.9876794815063477,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.2031,
      "step": 56
    },
    {
      "epoch": 0.005772736479643509,
      "grad_norm": 3.1994035243988037,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.2472,
      "step": 57
    },
    {
      "epoch": 0.005874012558233745,
      "grad_norm": 2.906554937362671,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.5271,
      "step": 58
    },
    {
      "epoch": 0.005975288636823982,
      "grad_norm": 3.6116013526916504,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.8218,
      "step": 59
    },
    {
      "epoch": 0.006076564715414219,
      "grad_norm": 4.342185020446777,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.2505,
      "step": 60
    },
    {
      "epoch": 0.006177840794004456,
      "grad_norm": 13.123361587524414,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.0744,
      "step": 61
    },
    {
      "epoch": 0.006279116872594693,
      "grad_norm": 6.584843158721924,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.2059,
      "step": 62
    },
    {
      "epoch": 0.00638039295118493,
      "grad_norm": 5.268929958343506,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.137,
      "step": 63
    },
    {
      "epoch": 0.006481669029775167,
      "grad_norm": 3.412992477416992,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.0627,
      "step": 64
    },
    {
      "epoch": 0.006582945108365404,
      "grad_norm": 3.735121250152588,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.1707,
      "step": 65
    },
    {
      "epoch": 0.006684221186955641,
      "grad_norm": 2.984276056289673,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.8477,
      "step": 66
    },
    {
      "epoch": 0.006785497265545878,
      "grad_norm": 7.2863359451293945,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.1729,
      "step": 67
    },
    {
      "epoch": 0.006886773344136115,
      "grad_norm": 6.308945178985596,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.1328,
      "step": 68
    },
    {
      "epoch": 0.006988049422726352,
      "grad_norm": 4.090075969696045,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.3419,
      "step": 69
    },
    {
      "epoch": 0.007089325501316589,
      "grad_norm": 3.4586753845214844,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.3241,
      "step": 70
    },
    {
      "epoch": 0.007190601579906826,
      "grad_norm": 4.972193241119385,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.5905,
      "step": 71
    },
    {
      "epoch": 0.007291877658497063,
      "grad_norm": 4.1412434577941895,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.3276,
      "step": 72
    },
    {
      "epoch": 0.0073931537370873,
      "grad_norm": 3.7608914375305176,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.5672,
      "step": 73
    },
    {
      "epoch": 0.007494429815677537,
      "grad_norm": 4.152456760406494,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.4999,
      "step": 74
    },
    {
      "epoch": 0.007595705894267774,
      "grad_norm": 3.5097086429595947,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.532,
      "step": 75
    },
    {
      "epoch": 0.007696981972858011,
      "grad_norm": 4.236787796020508,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.8479,
      "step": 76
    },
    {
      "epoch": 0.007798258051448248,
      "grad_norm": 4.112372875213623,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.3158,
      "step": 77
    },
    {
      "epoch": 0.007899534130038485,
      "grad_norm": 3.9809229373931885,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.357,
      "step": 78
    },
    {
      "epoch": 0.008000810208628722,
      "grad_norm": 4.142438888549805,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.371,
      "step": 79
    },
    {
      "epoch": 0.008102086287218958,
      "grad_norm": 6.03681755065918,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.9426,
      "step": 80
    },
    {
      "epoch": 0.008203362365809196,
      "grad_norm": 4.4276628494262695,
      "learning_rate": 6.932495846462261e-05,
      "loss": 2.033,
      "step": 81
    },
    {
      "epoch": 0.008304638444399434,
      "grad_norm": 3.3350882530212402,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.7404,
      "step": 82
    },
    {
      "epoch": 0.00840591452298967,
      "grad_norm": 2.6922543048858643,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.5807,
      "step": 83
    },
    {
      "epoch": 0.008507190601579907,
      "grad_norm": 2.3181540966033936,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.3461,
      "step": 84
    },
    {
      "epoch": 0.008608466680170145,
      "grad_norm": 2.83367919921875,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.4557,
      "step": 85
    },
    {
      "epoch": 0.00870974275876038,
      "grad_norm": 3.4856133460998535,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.8367,
      "step": 86
    },
    {
      "epoch": 0.008811018837350618,
      "grad_norm": 2.253290891647339,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.4891,
      "step": 87
    },
    {
      "epoch": 0.008912294915940856,
      "grad_norm": 3.215484857559204,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.881,
      "step": 88
    },
    {
      "epoch": 0.009013570994531091,
      "grad_norm": 4.4721360206604,
      "learning_rate": 6.307399704769099e-05,
      "loss": 2.3519,
      "step": 89
    },
    {
      "epoch": 0.009114847073121329,
      "grad_norm": 4.574413299560547,
      "learning_rate": 6.227427435703997e-05,
      "loss": 2.4677,
      "step": 90
    },
    {
      "epoch": 0.009216123151711565,
      "grad_norm": 5.168091773986816,
      "learning_rate": 6.147119600233758e-05,
      "loss": 2.6011,
      "step": 91
    },
    {
      "epoch": 0.009317399230301802,
      "grad_norm": 6.12443208694458,
      "learning_rate": 6.066498153718735e-05,
      "loss": 2.3665,
      "step": 92
    },
    {
      "epoch": 0.00941867530889204,
      "grad_norm": 4.022878170013428,
      "learning_rate": 5.985585137257401e-05,
      "loss": 2.5555,
      "step": 93
    },
    {
      "epoch": 0.009519951387482276,
      "grad_norm": 3.3305866718292236,
      "learning_rate": 5.90440267166055e-05,
      "loss": 2.0126,
      "step": 94
    },
    {
      "epoch": 0.009621227466072513,
      "grad_norm": 3.748392343521118,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 2.4115,
      "step": 95
    },
    {
      "epoch": 0.009722503544662751,
      "grad_norm": 4.195624828338623,
      "learning_rate": 5.74131823855921e-05,
      "loss": 2.2669,
      "step": 96
    },
    {
      "epoch": 0.009823779623252987,
      "grad_norm": 5.135915756225586,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 2.433,
      "step": 97
    },
    {
      "epoch": 0.009925055701843224,
      "grad_norm": 4.754821300506592,
      "learning_rate": 5.577423184847932e-05,
      "loss": 2.4903,
      "step": 98
    },
    {
      "epoch": 0.010026331780433462,
      "grad_norm": 6.451806545257568,
      "learning_rate": 5.495227651252315e-05,
      "loss": 2.83,
      "step": 99
    },
    {
      "epoch": 0.010127607859023698,
      "grad_norm": 4.6385650634765625,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 2.4094,
      "step": 100
    },
    {
      "epoch": 0.010127607859023698,
      "eval_loss": 2.240602731704712,
      "eval_runtime": 1571.5596,
      "eval_samples_per_second": 10.582,
      "eval_steps_per_second": 2.646,
      "step": 100
    },
    {
      "epoch": 0.010228883937613936,
      "grad_norm": 6.583124160766602,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.5519,
      "step": 101
    },
    {
      "epoch": 0.010330160016204173,
      "grad_norm": 7.981141567230225,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.6663,
      "step": 102
    },
    {
      "epoch": 0.010431436094794409,
      "grad_norm": 5.900326251983643,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.4544,
      "step": 103
    },
    {
      "epoch": 0.010532712173384647,
      "grad_norm": 7.027140140533447,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.5907,
      "step": 104
    },
    {
      "epoch": 0.010633988251974884,
      "grad_norm": 4.8545756340026855,
      "learning_rate": 5e-05,
      "loss": 1.0845,
      "step": 105
    },
    {
      "epoch": 0.01073526433056512,
      "grad_norm": 3.0874416828155518,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.0303,
      "step": 106
    },
    {
      "epoch": 0.010836540409155358,
      "grad_norm": 2.4116547107696533,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.6618,
      "step": 107
    },
    {
      "epoch": 0.010937816487745595,
      "grad_norm": 2.3724544048309326,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.8639,
      "step": 108
    },
    {
      "epoch": 0.011039092566335831,
      "grad_norm": 4.351852893829346,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.1216,
      "step": 109
    },
    {
      "epoch": 0.011140368644926069,
      "grad_norm": 4.345721244812012,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.9437,
      "step": 110
    },
    {
      "epoch": 0.011241644723516306,
      "grad_norm": 4.805398464202881,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.1012,
      "step": 111
    },
    {
      "epoch": 0.011342920802106542,
      "grad_norm": 2.533634662628174,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.9933,
      "step": 112
    },
    {
      "epoch": 0.01144419688069678,
      "grad_norm": 4.556289196014404,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.1198,
      "step": 113
    },
    {
      "epoch": 0.011545472959287017,
      "grad_norm": 2.5026047229766846,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.9707,
      "step": 114
    },
    {
      "epoch": 0.011646749037877253,
      "grad_norm": 3.4550931453704834,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.9138,
      "step": 115
    },
    {
      "epoch": 0.01174802511646749,
      "grad_norm": 4.361387729644775,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.0032,
      "step": 116
    },
    {
      "epoch": 0.011849301195057726,
      "grad_norm": 3.028383493423462,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.9695,
      "step": 117
    },
    {
      "epoch": 0.011950577273647964,
      "grad_norm": 2.9003260135650635,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.0773,
      "step": 118
    },
    {
      "epoch": 0.012051853352238202,
      "grad_norm": 3.9012155532836914,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.0103,
      "step": 119
    },
    {
      "epoch": 0.012153129430828438,
      "grad_norm": 4.370852947235107,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.7424,
      "step": 120
    },
    {
      "epoch": 0.012254405509418675,
      "grad_norm": 3.128495931625366,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.2989,
      "step": 121
    },
    {
      "epoch": 0.012355681588008913,
      "grad_norm": 3.703368902206421,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.3236,
      "step": 122
    },
    {
      "epoch": 0.012456957666599149,
      "grad_norm": 2.874260425567627,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.4198,
      "step": 123
    },
    {
      "epoch": 0.012558233745189386,
      "grad_norm": 3.027625322341919,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.5042,
      "step": 124
    },
    {
      "epoch": 0.012659509823779624,
      "grad_norm": 2.950544595718384,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.8291,
      "step": 125
    },
    {
      "epoch": 0.01276078590236986,
      "grad_norm": 3.659226655960083,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.3674,
      "step": 126
    },
    {
      "epoch": 0.012862061980960097,
      "grad_norm": 3.4479410648345947,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.6669,
      "step": 127
    },
    {
      "epoch": 0.012963338059550335,
      "grad_norm": 2.3538930416107178,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.0548,
      "step": 128
    },
    {
      "epoch": 0.01306461413814057,
      "grad_norm": 3.0581886768341064,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.319,
      "step": 129
    },
    {
      "epoch": 0.013165890216730808,
      "grad_norm": 3.2363953590393066,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.2731,
      "step": 130
    },
    {
      "epoch": 0.013267166295321046,
      "grad_norm": 3.3476040363311768,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.5927,
      "step": 131
    },
    {
      "epoch": 0.013368442373911282,
      "grad_norm": 2.697167158126831,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.326,
      "step": 132
    },
    {
      "epoch": 0.01346971845250152,
      "grad_norm": 3.1736996173858643,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.4011,
      "step": 133
    },
    {
      "epoch": 0.013570994531091757,
      "grad_norm": 2.6347978115081787,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.5996,
      "step": 134
    },
    {
      "epoch": 0.013672270609681993,
      "grad_norm": 2.117367744445801,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.4682,
      "step": 135
    },
    {
      "epoch": 0.01377354668827223,
      "grad_norm": 3.113630771636963,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.3594,
      "step": 136
    },
    {
      "epoch": 0.013874822766862468,
      "grad_norm": 3.4249775409698486,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.6358,
      "step": 137
    },
    {
      "epoch": 0.013976098845452704,
      "grad_norm": 2.141242504119873,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.439,
      "step": 138
    },
    {
      "epoch": 0.014077374924042941,
      "grad_norm": 2.9997174739837646,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 2.0737,
      "step": 139
    },
    {
      "epoch": 0.014178651002633179,
      "grad_norm": 3.2043118476867676,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 2.4137,
      "step": 140
    },
    {
      "epoch": 0.014279927081223415,
      "grad_norm": 3.2861759662628174,
      "learning_rate": 2.196424713241637e-05,
      "loss": 2.2806,
      "step": 141
    },
    {
      "epoch": 0.014381203159813652,
      "grad_norm": 4.050476551055908,
      "learning_rate": 2.128356686585282e-05,
      "loss": 2.3203,
      "step": 142
    },
    {
      "epoch": 0.014482479238403888,
      "grad_norm": 2.97379469871521,
      "learning_rate": 2.061073738537635e-05,
      "loss": 2.1789,
      "step": 143
    },
    {
      "epoch": 0.014583755316994126,
      "grad_norm": 3.7634708881378174,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 2.2709,
      "step": 144
    },
    {
      "epoch": 0.014685031395584363,
      "grad_norm": 3.99497389793396,
      "learning_rate": 1.928936436551661e-05,
      "loss": 2.4362,
      "step": 145
    },
    {
      "epoch": 0.0147863074741746,
      "grad_norm": 3.6659791469573975,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.898,
      "step": 146
    },
    {
      "epoch": 0.014887583552764837,
      "grad_norm": 4.151111125946045,
      "learning_rate": 1.800157297483417e-05,
      "loss": 2.3185,
      "step": 147
    },
    {
      "epoch": 0.014988859631355074,
      "grad_norm": 3.7545299530029297,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 2.254,
      "step": 148
    },
    {
      "epoch": 0.01509013570994531,
      "grad_norm": 3.2946999073028564,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 2.195,
      "step": 149
    },
    {
      "epoch": 0.015191411788535548,
      "grad_norm": 3.872542142868042,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 2.2813,
      "step": 150
    },
    {
      "epoch": 0.015191411788535548,
      "eval_loss": 1.4213684797286987,
      "eval_runtime": 1571.7297,
      "eval_samples_per_second": 10.581,
      "eval_steps_per_second": 2.645,
      "step": 150
    },
    {
      "epoch": 0.015292687867125785,
      "grad_norm": 4.454233169555664,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.9667,
      "step": 151
    },
    {
      "epoch": 0.015393963945716021,
      "grad_norm": 3.2457315921783447,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.133,
      "step": 152
    },
    {
      "epoch": 0.015495240024306259,
      "grad_norm": 1.9486116170883179,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.672,
      "step": 153
    },
    {
      "epoch": 0.015596516102896496,
      "grad_norm": 2.5552918910980225,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.8465,
      "step": 154
    },
    {
      "epoch": 0.015697792181486734,
      "grad_norm": 2.387343168258667,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.9506,
      "step": 155
    },
    {
      "epoch": 0.01579906826007697,
      "grad_norm": 1.7938592433929443,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.9771,
      "step": 156
    },
    {
      "epoch": 0.015900344338667206,
      "grad_norm": 3.2489726543426514,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.5909,
      "step": 157
    },
    {
      "epoch": 0.016001620417257445,
      "grad_norm": 2.3328659534454346,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.7564,
      "step": 158
    },
    {
      "epoch": 0.01610289649584768,
      "grad_norm": 2.8104710578918457,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.7603,
      "step": 159
    },
    {
      "epoch": 0.016204172574437917,
      "grad_norm": 2.808901309967041,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.6635,
      "step": 160
    },
    {
      "epoch": 0.016305448653028156,
      "grad_norm": 3.6609928607940674,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.2593,
      "step": 161
    },
    {
      "epoch": 0.016406724731618392,
      "grad_norm": 2.648317337036133,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.0744,
      "step": 162
    },
    {
      "epoch": 0.016508000810208628,
      "grad_norm": 2.719637870788574,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.844,
      "step": 163
    },
    {
      "epoch": 0.016609276888798867,
      "grad_norm": 4.580341815948486,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.0398,
      "step": 164
    },
    {
      "epoch": 0.016710552967389103,
      "grad_norm": 2.197094202041626,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.7417,
      "step": 165
    },
    {
      "epoch": 0.01681182904597934,
      "grad_norm": 2.474677324295044,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.9719,
      "step": 166
    },
    {
      "epoch": 0.016913105124569578,
      "grad_norm": 4.165216445922852,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.8895,
      "step": 167
    },
    {
      "epoch": 0.017014381203159814,
      "grad_norm": 2.8705661296844482,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.2317,
      "step": 168
    },
    {
      "epoch": 0.01711565728175005,
      "grad_norm": 2.839369773864746,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.2891,
      "step": 169
    },
    {
      "epoch": 0.01721693336034029,
      "grad_norm": 3.5295631885528564,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.314,
      "step": 170
    },
    {
      "epoch": 0.017318209438930525,
      "grad_norm": 3.5073323249816895,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.566,
      "step": 171
    },
    {
      "epoch": 0.01741948551752076,
      "grad_norm": 3.479749917984009,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.2162,
      "step": 172
    },
    {
      "epoch": 0.017520761596111,
      "grad_norm": 2.6487128734588623,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.1712,
      "step": 173
    },
    {
      "epoch": 0.017622037674701236,
      "grad_norm": 3.034012794494629,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.4889,
      "step": 174
    },
    {
      "epoch": 0.017723313753291472,
      "grad_norm": 2.9881718158721924,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.1848,
      "step": 175
    },
    {
      "epoch": 0.01782458983188171,
      "grad_norm": 3.1335384845733643,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.4917,
      "step": 176
    },
    {
      "epoch": 0.017925865910471947,
      "grad_norm": 2.802159070968628,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.1624,
      "step": 177
    },
    {
      "epoch": 0.018027141989062183,
      "grad_norm": 2.91914701461792,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.166,
      "step": 178
    },
    {
      "epoch": 0.018128418067652422,
      "grad_norm": 3.416661262512207,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.1668,
      "step": 179
    },
    {
      "epoch": 0.018229694146242658,
      "grad_norm": 3.1717991828918457,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.4008,
      "step": 180
    },
    {
      "epoch": 0.018330970224832894,
      "grad_norm": 2.290952682495117,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.2574,
      "step": 181
    },
    {
      "epoch": 0.01843224630342313,
      "grad_norm": 2.1434242725372314,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.2813,
      "step": 182
    },
    {
      "epoch": 0.01853352238201337,
      "grad_norm": 2.1748228073120117,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.1821,
      "step": 183
    },
    {
      "epoch": 0.018634798460603605,
      "grad_norm": 2.5101327896118164,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.3385,
      "step": 184
    },
    {
      "epoch": 0.01873607453919384,
      "grad_norm": 2.1042656898498535,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.1635,
      "step": 185
    },
    {
      "epoch": 0.01883735061778408,
      "grad_norm": 3.1028172969818115,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.7765,
      "step": 186
    },
    {
      "epoch": 0.018938626696374316,
      "grad_norm": 1.9531184434890747,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.1414,
      "step": 187
    },
    {
      "epoch": 0.019039902774964552,
      "grad_norm": 2.500147581100464,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.4207,
      "step": 188
    },
    {
      "epoch": 0.01914117885355479,
      "grad_norm": 4.331672668457031,
      "learning_rate": 8.247462563808817e-07,
      "loss": 2.1343,
      "step": 189
    },
    {
      "epoch": 0.019242454932145027,
      "grad_norm": 4.009225368499756,
      "learning_rate": 6.819348298638839e-07,
      "loss": 2.1703,
      "step": 190
    },
    {
      "epoch": 0.019343731010735263,
      "grad_norm": 3.9454243183135986,
      "learning_rate": 5.526064699265753e-07,
      "loss": 2.2049,
      "step": 191
    },
    {
      "epoch": 0.019445007089325502,
      "grad_norm": 4.177087783813477,
      "learning_rate": 4.367965336512403e-07,
      "loss": 2.4599,
      "step": 192
    },
    {
      "epoch": 0.019546283167915738,
      "grad_norm": 4.1772003173828125,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.7791,
      "step": 193
    },
    {
      "epoch": 0.019647559246505974,
      "grad_norm": 2.9933736324310303,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.8294,
      "step": 194
    },
    {
      "epoch": 0.019748835325096213,
      "grad_norm": 3.9825005531311035,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.943,
      "step": 195
    },
    {
      "epoch": 0.01985011140368645,
      "grad_norm": 3.3033840656280518,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 2.1107,
      "step": 196
    },
    {
      "epoch": 0.019951387482276685,
      "grad_norm": 4.105139255523682,
      "learning_rate": 6.150154258476315e-08,
      "loss": 2.1399,
      "step": 197
    },
    {
      "epoch": 0.020052663560866924,
      "grad_norm": 3.5981619358062744,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 2.0967,
      "step": 198
    },
    {
      "epoch": 0.02015393963945716,
      "grad_norm": 3.259740114212036,
      "learning_rate": 6.834750376549792e-09,
      "loss": 2.1803,
      "step": 199
    },
    {
      "epoch": 0.020255215718047396,
      "grad_norm": 4.242725372314453,
      "learning_rate": 0.0,
      "loss": 2.2505,
      "step": 200
    },
    {
      "epoch": 0.020255215718047396,
      "eval_loss": 1.3650026321411133,
      "eval_runtime": 1572.2308,
      "eval_samples_per_second": 10.577,
      "eval_steps_per_second": 2.645,
      "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": 3.454443806785536e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}