{
  "best_metric": 2.577688217163086,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.8743169398907104,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004371584699453552,
      "grad_norm": 4.126193046569824,
      "learning_rate": 1e-05,
      "loss": 3.5196,
      "step": 1
    },
    {
      "epoch": 0.004371584699453552,
      "eval_loss": 4.559682369232178,
      "eval_runtime": 6.2865,
      "eval_samples_per_second": 61.401,
      "eval_steps_per_second": 15.43,
      "step": 1
    },
    {
      "epoch": 0.008743169398907104,
      "grad_norm": 6.943887710571289,
      "learning_rate": 2e-05,
      "loss": 3.952,
      "step": 2
    },
    {
      "epoch": 0.013114754098360656,
      "grad_norm": 6.852543354034424,
      "learning_rate": 3e-05,
      "loss": 3.6652,
      "step": 3
    },
    {
      "epoch": 0.017486338797814208,
      "grad_norm": 6.262274265289307,
      "learning_rate": 4e-05,
      "loss": 3.8174,
      "step": 4
    },
    {
      "epoch": 0.02185792349726776,
      "grad_norm": 5.588265419006348,
      "learning_rate": 5e-05,
      "loss": 3.5521,
      "step": 5
    },
    {
      "epoch": 0.02622950819672131,
      "grad_norm": 5.240731716156006,
      "learning_rate": 6e-05,
      "loss": 3.5265,
      "step": 6
    },
    {
      "epoch": 0.030601092896174863,
      "grad_norm": 4.693352222442627,
      "learning_rate": 7e-05,
      "loss": 3.4053,
      "step": 7
    },
    {
      "epoch": 0.034972677595628415,
      "grad_norm": 4.592766761779785,
      "learning_rate": 8e-05,
      "loss": 3.4041,
      "step": 8
    },
    {
      "epoch": 0.03934426229508197,
      "grad_norm": 4.271055698394775,
      "learning_rate": 9e-05,
      "loss": 3.248,
      "step": 9
    },
    {
      "epoch": 0.04371584699453552,
      "grad_norm": 4.057760715484619,
      "learning_rate": 0.0001,
      "loss": 3.0703,
      "step": 10
    },
    {
      "epoch": 0.048087431693989074,
      "grad_norm": 3.9648282527923584,
      "learning_rate": 9.999316524962345e-05,
      "loss": 3.1942,
      "step": 11
    },
    {
      "epoch": 0.05245901639344262,
      "grad_norm": 3.366271495819092,
      "learning_rate": 9.997266286704631e-05,
      "loss": 3.0367,
      "step": 12
    },
    {
      "epoch": 0.05683060109289618,
      "grad_norm": 4.005966663360596,
      "learning_rate": 9.993849845741524e-05,
      "loss": 2.6963,
      "step": 13
    },
    {
      "epoch": 0.06120218579234973,
      "grad_norm": 3.822646379470825,
      "learning_rate": 9.989068136093873e-05,
      "loss": 2.9447,
      "step": 14
    },
    {
      "epoch": 0.06557377049180328,
      "grad_norm": 3.927497148513794,
      "learning_rate": 9.98292246503335e-05,
      "loss": 3.0044,
      "step": 15
    },
    {
      "epoch": 0.06994535519125683,
      "grad_norm": 3.667811155319214,
      "learning_rate": 9.975414512725057e-05,
      "loss": 2.6057,
      "step": 16
    },
    {
      "epoch": 0.07431693989071038,
      "grad_norm": 3.5979390144348145,
      "learning_rate": 9.966546331768191e-05,
      "loss": 3.0311,
      "step": 17
    },
    {
      "epoch": 0.07868852459016394,
      "grad_norm": 3.7359302043914795,
      "learning_rate": 9.956320346634876e-05,
      "loss": 2.8641,
      "step": 18
    },
    {
      "epoch": 0.08306010928961749,
      "grad_norm": 3.7435851097106934,
      "learning_rate": 9.944739353007344e-05,
      "loss": 3.124,
      "step": 19
    },
    {
      "epoch": 0.08743169398907104,
      "grad_norm": 3.6843996047973633,
      "learning_rate": 9.931806517013612e-05,
      "loss": 2.6124,
      "step": 20
    },
    {
      "epoch": 0.09180327868852459,
      "grad_norm": 3.8037829399108887,
      "learning_rate": 9.917525374361912e-05,
      "loss": 3.0047,
      "step": 21
    },
    {
      "epoch": 0.09617486338797815,
      "grad_norm": 3.4682741165161133,
      "learning_rate": 9.901899829374047e-05,
      "loss": 2.7574,
      "step": 22
    },
    {
      "epoch": 0.1005464480874317,
      "grad_norm": 3.982518434524536,
      "learning_rate": 9.884934153917997e-05,
      "loss": 3.3102,
      "step": 23
    },
    {
      "epoch": 0.10491803278688525,
      "grad_norm": 3.4155569076538086,
      "learning_rate": 9.86663298624003e-05,
      "loss": 2.7225,
      "step": 24
    },
    {
      "epoch": 0.1092896174863388,
      "grad_norm": 3.415499210357666,
      "learning_rate": 9.847001329696653e-05,
      "loss": 2.8517,
      "step": 25
    },
    {
      "epoch": 0.11366120218579236,
      "grad_norm": 3.5038135051727295,
      "learning_rate": 9.826044551386744e-05,
      "loss": 2.7497,
      "step": 26
    },
    {
      "epoch": 0.1180327868852459,
      "grad_norm": 3.3416104316711426,
      "learning_rate": 9.803768380684242e-05,
      "loss": 2.7564,
      "step": 27
    },
    {
      "epoch": 0.12240437158469945,
      "grad_norm": 3.915132761001587,
      "learning_rate": 9.780178907671789e-05,
      "loss": 3.0684,
      "step": 28
    },
    {
      "epoch": 0.126775956284153,
      "grad_norm": 3.8346669673919678,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.8902,
      "step": 29
    },
    {
      "epoch": 0.13114754098360656,
      "grad_norm": 3.334312677383423,
      "learning_rate": 9.729086208503174e-05,
      "loss": 2.9412,
      "step": 30
    },
    {
      "epoch": 0.1355191256830601,
      "grad_norm": 4.0946044921875,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.924,
      "step": 31
    },
    {
      "epoch": 0.13989071038251366,
      "grad_norm": 3.4741051197052,
      "learning_rate": 9.672822322997305e-05,
      "loss": 2.7817,
      "step": 32
    },
    {
      "epoch": 0.14426229508196722,
      "grad_norm": 3.407179594039917,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.8444,
      "step": 33
    },
    {
      "epoch": 0.14863387978142076,
      "grad_norm": 3.3746161460876465,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.5191,
      "step": 34
    },
    {
      "epoch": 0.15300546448087432,
      "grad_norm": 3.8201727867126465,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.9822,
      "step": 35
    },
    {
      "epoch": 0.15737704918032788,
      "grad_norm": 3.5378103256225586,
      "learning_rate": 9.545032675245813e-05,
      "loss": 2.6514,
      "step": 36
    },
    {
      "epoch": 0.16174863387978142,
      "grad_norm": 3.944777011871338,
      "learning_rate": 9.509956150664796e-05,
      "loss": 3.1401,
      "step": 37
    },
    {
      "epoch": 0.16612021857923498,
      "grad_norm": 3.4765994548797607,
      "learning_rate": 9.473646649103818e-05,
      "loss": 2.7413,
      "step": 38
    },
    {
      "epoch": 0.17049180327868851,
      "grad_norm": 3.572260618209839,
      "learning_rate": 9.43611409721806e-05,
      "loss": 2.8709,
      "step": 39
    },
    {
      "epoch": 0.17486338797814208,
      "grad_norm": 3.784226655960083,
      "learning_rate": 9.397368756032445e-05,
      "loss": 2.9777,
      "step": 40
    },
    {
      "epoch": 0.17923497267759564,
      "grad_norm": 3.696209192276001,
      "learning_rate": 9.357421218136386e-05,
      "loss": 2.5847,
      "step": 41
    },
    {
      "epoch": 0.18360655737704917,
      "grad_norm": 3.1744725704193115,
      "learning_rate": 9.316282404787871e-05,
      "loss": 2.4249,
      "step": 42
    },
    {
      "epoch": 0.18797814207650274,
      "grad_norm": 3.39530873298645,
      "learning_rate": 9.273963562927695e-05,
      "loss": 2.7218,
      "step": 43
    },
    {
      "epoch": 0.1923497267759563,
      "grad_norm": 3.955327272415161,
      "learning_rate": 9.230476262104677e-05,
      "loss": 2.7967,
      "step": 44
    },
    {
      "epoch": 0.19672131147540983,
      "grad_norm": 3.8752644062042236,
      "learning_rate": 9.185832391312644e-05,
      "loss": 2.5713,
      "step": 45
    },
    {
      "epoch": 0.2010928961748634,
      "grad_norm": 3.940387725830078,
      "learning_rate": 9.140044155740101e-05,
      "loss": 2.3542,
      "step": 46
    },
    {
      "epoch": 0.20546448087431693,
      "grad_norm": 3.9077024459838867,
      "learning_rate": 9.093124073433463e-05,
      "loss": 2.3798,
      "step": 47
    },
    {
      "epoch": 0.2098360655737705,
      "grad_norm": 3.9562506675720215,
      "learning_rate": 9.045084971874738e-05,
      "loss": 2.8128,
      "step": 48
    },
    {
      "epoch": 0.21420765027322405,
      "grad_norm": 4.585770606994629,
      "learning_rate": 8.995939984474624e-05,
      "loss": 2.9016,
      "step": 49
    },
    {
      "epoch": 0.2185792349726776,
      "grad_norm": 5.764311790466309,
      "learning_rate": 8.945702546981969e-05,
      "loss": 3.0644,
      "step": 50
    },
    {
      "epoch": 0.2185792349726776,
      "eval_loss": 2.8919804096221924,
      "eval_runtime": 6.2479,
      "eval_samples_per_second": 61.781,
      "eval_steps_per_second": 15.525,
      "step": 50
    },
    {
      "epoch": 0.22295081967213115,
      "grad_norm": 6.3447585105896,
      "learning_rate": 8.894386393810563e-05,
      "loss": 3.2161,
      "step": 51
    },
    {
      "epoch": 0.2273224043715847,
      "grad_norm": 6.504059791564941,
      "learning_rate": 8.842005554284296e-05,
      "loss": 3.2594,
      "step": 52
    },
    {
      "epoch": 0.23169398907103825,
      "grad_norm": 5.788021087646484,
      "learning_rate": 8.788574348801675e-05,
      "loss": 3.0391,
      "step": 53
    },
    {
      "epoch": 0.2360655737704918,
      "grad_norm": 5.123461723327637,
      "learning_rate": 8.73410738492077e-05,
      "loss": 2.9697,
      "step": 54
    },
    {
      "epoch": 0.24043715846994534,
      "grad_norm": 4.111291408538818,
      "learning_rate": 8.678619553365659e-05,
      "loss": 3.0358,
      "step": 55
    },
    {
      "epoch": 0.2448087431693989,
      "grad_norm": 3.5588459968566895,
      "learning_rate": 8.622126023955446e-05,
      "loss": 2.9169,
      "step": 56
    },
    {
      "epoch": 0.24918032786885247,
      "grad_norm": 2.9790215492248535,
      "learning_rate": 8.564642241456986e-05,
      "loss": 2.8494,
      "step": 57
    },
    {
      "epoch": 0.253551912568306,
      "grad_norm": 2.991408109664917,
      "learning_rate": 8.506183921362443e-05,
      "loss": 2.7641,
      "step": 58
    },
    {
      "epoch": 0.25792349726775954,
      "grad_norm": 2.850606918334961,
      "learning_rate": 8.44676704559283e-05,
      "loss": 2.8361,
      "step": 59
    },
    {
      "epoch": 0.26229508196721313,
      "grad_norm": 2.8943991661071777,
      "learning_rate": 8.386407858128706e-05,
      "loss": 2.8989,
      "step": 60
    },
    {
      "epoch": 0.26666666666666666,
      "grad_norm": 2.745375871658325,
      "learning_rate": 8.32512286056924e-05,
      "loss": 2.6493,
      "step": 61
    },
    {
      "epoch": 0.2710382513661202,
      "grad_norm": 2.604910373687744,
      "learning_rate": 8.262928807620843e-05,
      "loss": 2.4587,
      "step": 62
    },
    {
      "epoch": 0.2754098360655738,
      "grad_norm": 2.986539602279663,
      "learning_rate": 8.199842702516583e-05,
      "loss": 2.7807,
      "step": 63
    },
    {
      "epoch": 0.2797814207650273,
      "grad_norm": 2.827967643737793,
      "learning_rate": 8.135881792367686e-05,
      "loss": 2.7345,
      "step": 64
    },
    {
      "epoch": 0.28415300546448086,
      "grad_norm": 3.115030288696289,
      "learning_rate": 8.07106356344834e-05,
      "loss": 2.8969,
      "step": 65
    },
    {
      "epoch": 0.28852459016393445,
      "grad_norm": 2.7786996364593506,
      "learning_rate": 8.005405736415126e-05,
      "loss": 2.6046,
      "step": 66
    },
    {
      "epoch": 0.292896174863388,
      "grad_norm": 2.654737710952759,
      "learning_rate": 7.938926261462366e-05,
      "loss": 2.9504,
      "step": 67
    },
    {
      "epoch": 0.2972677595628415,
      "grad_norm": 2.599520444869995,
      "learning_rate": 7.871643313414718e-05,
      "loss": 2.638,
      "step": 68
    },
    {
      "epoch": 0.3016393442622951,
      "grad_norm": 3.2816624641418457,
      "learning_rate": 7.803575286758364e-05,
      "loss": 2.7567,
      "step": 69
    },
    {
      "epoch": 0.30601092896174864,
      "grad_norm": 3.0041587352752686,
      "learning_rate": 7.734740790612136e-05,
      "loss": 3.0367,
      "step": 70
    },
    {
      "epoch": 0.3103825136612022,
      "grad_norm": 2.7854859828948975,
      "learning_rate": 7.66515864363997e-05,
      "loss": 2.5757,
      "step": 71
    },
    {
      "epoch": 0.31475409836065577,
      "grad_norm": 2.885754346847534,
      "learning_rate": 7.594847868906076e-05,
      "loss": 2.5267,
      "step": 72
    },
    {
      "epoch": 0.3191256830601093,
      "grad_norm": 2.771885395050049,
      "learning_rate": 7.52382768867422e-05,
      "loss": 2.6094,
      "step": 73
    },
    {
      "epoch": 0.32349726775956283,
      "grad_norm": 2.8158531188964844,
      "learning_rate": 7.452117519152542e-05,
      "loss": 2.5808,
      "step": 74
    },
    {
      "epoch": 0.32786885245901637,
      "grad_norm": 2.928452491760254,
      "learning_rate": 7.379736965185368e-05,
      "loss": 2.6154,
      "step": 75
    },
    {
      "epoch": 0.33224043715846996,
      "grad_norm": 3.154531478881836,
      "learning_rate": 7.30670581489344e-05,
      "loss": 3.019,
      "step": 76
    },
    {
      "epoch": 0.3366120218579235,
      "grad_norm": 3.2921383380889893,
      "learning_rate": 7.233044034264034e-05,
      "loss": 2.7771,
      "step": 77
    },
    {
      "epoch": 0.34098360655737703,
      "grad_norm": 2.65779709815979,
      "learning_rate": 7.158771761692464e-05,
      "loss": 2.4068,
      "step": 78
    },
    {
      "epoch": 0.3453551912568306,
      "grad_norm": 3.1646389961242676,
      "learning_rate": 7.083909302476453e-05,
      "loss": 2.9349,
      "step": 79
    },
    {
      "epoch": 0.34972677595628415,
      "grad_norm": 2.7925400733947754,
      "learning_rate": 7.008477123264848e-05,
      "loss": 2.9227,
      "step": 80
    },
    {
      "epoch": 0.3540983606557377,
      "grad_norm": 3.1522791385650635,
      "learning_rate": 6.932495846462261e-05,
      "loss": 2.654,
      "step": 81
    },
    {
      "epoch": 0.3584699453551913,
      "grad_norm": 3.1256582736968994,
      "learning_rate": 6.855986244591104e-05,
      "loss": 2.8973,
      "step": 82
    },
    {
      "epoch": 0.3628415300546448,
      "grad_norm": 2.7471134662628174,
      "learning_rate": 6.778969234612584e-05,
      "loss": 2.6907,
      "step": 83
    },
    {
      "epoch": 0.36721311475409835,
      "grad_norm": 2.9811859130859375,
      "learning_rate": 6.701465872208216e-05,
      "loss": 2.7087,
      "step": 84
    },
    {
      "epoch": 0.37158469945355194,
      "grad_norm": 2.7266488075256348,
      "learning_rate": 6.623497346023418e-05,
      "loss": 2.5972,
      "step": 85
    },
    {
      "epoch": 0.37595628415300547,
      "grad_norm": 3.1146697998046875,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.6129,
      "step": 86
    },
    {
      "epoch": 0.380327868852459,
      "grad_norm": 2.780282497406006,
      "learning_rate": 6.466250186922325e-05,
      "loss": 2.6278,
      "step": 87
    },
    {
      "epoch": 0.3846994535519126,
      "grad_norm": 3.0733914375305176,
      "learning_rate": 6.387014543809223e-05,
      "loss": 2.558,
      "step": 88
    },
    {
      "epoch": 0.38907103825136613,
      "grad_norm": 2.821974515914917,
      "learning_rate": 6.307399704769099e-05,
      "loss": 2.5018,
      "step": 89
    },
    {
      "epoch": 0.39344262295081966,
      "grad_norm": 3.0181190967559814,
      "learning_rate": 6.227427435703997e-05,
      "loss": 2.6537,
      "step": 90
    },
    {
      "epoch": 0.3978142076502732,
      "grad_norm": 3.044379472732544,
      "learning_rate": 6.147119600233758e-05,
      "loss": 2.6091,
      "step": 91
    },
    {
      "epoch": 0.4021857923497268,
      "grad_norm": 3.092782735824585,
      "learning_rate": 6.066498153718735e-05,
      "loss": 2.7654,
      "step": 92
    },
    {
      "epoch": 0.4065573770491803,
      "grad_norm": 2.96266770362854,
      "learning_rate": 5.985585137257401e-05,
      "loss": 2.6814,
      "step": 93
    },
    {
      "epoch": 0.41092896174863386,
      "grad_norm": 3.3606293201446533,
      "learning_rate": 5.90440267166055e-05,
      "loss": 2.8824,
      "step": 94
    },
    {
      "epoch": 0.41530054644808745,
      "grad_norm": 3.2069389820098877,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 2.6191,
      "step": 95
    },
    {
      "epoch": 0.419672131147541,
      "grad_norm": 3.4668021202087402,
      "learning_rate": 5.74131823855921e-05,
      "loss": 2.7656,
      "step": 96
    },
    {
      "epoch": 0.4240437158469945,
      "grad_norm": 3.754943609237671,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 2.751,
      "step": 97
    },
    {
      "epoch": 0.4284153005464481,
      "grad_norm": 3.9282078742980957,
      "learning_rate": 5.577423184847932e-05,
      "loss": 2.916,
      "step": 98
    },
    {
      "epoch": 0.43278688524590164,
      "grad_norm": 3.681250810623169,
      "learning_rate": 5.495227651252315e-05,
      "loss": 2.574,
      "step": 99
    },
    {
      "epoch": 0.4371584699453552,
      "grad_norm": 4.507355690002441,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 2.4579,
      "step": 100
    },
    {
      "epoch": 0.4371584699453552,
      "eval_loss": 2.7137081623077393,
      "eval_runtime": 6.252,
      "eval_samples_per_second": 61.74,
      "eval_steps_per_second": 15.515,
      "step": 100
    },
    {
      "epoch": 0.44153005464480877,
      "grad_norm": 3.6674201488494873,
      "learning_rate": 5.330452921628497e-05,
      "loss": 3.0462,
      "step": 101
    },
    {
      "epoch": 0.4459016393442623,
      "grad_norm": 3.6958484649658203,
      "learning_rate": 5.247918773366112e-05,
      "loss": 2.9499,
      "step": 102
    },
    {
      "epoch": 0.45027322404371584,
      "grad_norm": 3.729661703109741,
      "learning_rate": 5.165316846586541e-05,
      "loss": 2.8639,
      "step": 103
    },
    {
      "epoch": 0.4546448087431694,
      "grad_norm": 3.8177297115325928,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 3.1011,
      "step": 104
    },
    {
      "epoch": 0.45901639344262296,
      "grad_norm": 3.2442538738250732,
      "learning_rate": 5e-05,
      "loss": 2.7323,
      "step": 105
    },
    {
      "epoch": 0.4633879781420765,
      "grad_norm": 3.5539169311523438,
      "learning_rate": 4.917330276168208e-05,
      "loss": 2.7662,
      "step": 106
    },
    {
      "epoch": 0.46775956284153003,
      "grad_norm": 3.732706308364868,
      "learning_rate": 4.834683153413459e-05,
      "loss": 3.1186,
      "step": 107
    },
    {
      "epoch": 0.4721311475409836,
      "grad_norm": 2.835399627685547,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 2.6593,
      "step": 108
    },
    {
      "epoch": 0.47650273224043715,
      "grad_norm": 2.702887773513794,
      "learning_rate": 4.669547078371504e-05,
      "loss": 2.5559,
      "step": 109
    },
    {
      "epoch": 0.4808743169398907,
      "grad_norm": 2.5662825107574463,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 2.4342,
      "step": 110
    },
    {
      "epoch": 0.4852459016393443,
      "grad_norm": 2.879852771759033,
      "learning_rate": 4.504772348747687e-05,
      "loss": 2.8159,
      "step": 111
    },
    {
      "epoch": 0.4896174863387978,
      "grad_norm": 2.3322107791900635,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 2.69,
      "step": 112
    },
    {
      "epoch": 0.49398907103825135,
      "grad_norm": 2.4512860774993896,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 2.6321,
      "step": 113
    },
    {
      "epoch": 0.49836065573770494,
      "grad_norm": 2.4820594787597656,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 2.6554,
      "step": 114
    },
    {
      "epoch": 0.5027322404371585,
      "grad_norm": 2.387237548828125,
      "learning_rate": 4.17702704859633e-05,
      "loss": 2.563,
      "step": 115
    },
    {
      "epoch": 0.507103825136612,
      "grad_norm": 2.6377415657043457,
      "learning_rate": 4.095597328339452e-05,
      "loss": 2.754,
      "step": 116
    },
    {
      "epoch": 0.5114754098360655,
      "grad_norm": 2.5735561847686768,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 2.8631,
      "step": 117
    },
    {
      "epoch": 0.5158469945355191,
      "grad_norm": 2.5990209579467773,
      "learning_rate": 3.933501846281267e-05,
      "loss": 2.5718,
      "step": 118
    },
    {
      "epoch": 0.5202185792349727,
      "grad_norm": 2.685863733291626,
      "learning_rate": 3.852880399766243e-05,
      "loss": 2.9478,
      "step": 119
    },
    {
      "epoch": 0.5245901639344263,
      "grad_norm": 2.7012393474578857,
      "learning_rate": 3.772572564296005e-05,
      "loss": 2.6173,
      "step": 120
    },
    {
      "epoch": 0.5289617486338798,
      "grad_norm": 2.47837233543396,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 2.8273,
      "step": 121
    },
    {
      "epoch": 0.5333333333333333,
      "grad_norm": 2.8187203407287598,
      "learning_rate": 3.612985456190778e-05,
      "loss": 2.6816,
      "step": 122
    },
    {
      "epoch": 0.5377049180327869,
      "grad_norm": 2.7357730865478516,
      "learning_rate": 3.533749813077677e-05,
      "loss": 2.7524,
      "step": 123
    },
    {
      "epoch": 0.5420765027322404,
      "grad_norm": 2.868358850479126,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 2.8465,
      "step": 124
    },
    {
      "epoch": 0.546448087431694,
      "grad_norm": 2.652200937271118,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 2.8772,
      "step": 125
    },
    {
      "epoch": 0.5508196721311476,
      "grad_norm": 2.764561653137207,
      "learning_rate": 3.298534127791785e-05,
      "loss": 2.7359,
      "step": 126
    },
    {
      "epoch": 0.5551912568306011,
      "grad_norm": 3.1617627143859863,
      "learning_rate": 3.221030765387417e-05,
      "loss": 2.6818,
      "step": 127
    },
    {
      "epoch": 0.5595628415300546,
      "grad_norm": 2.701509952545166,
      "learning_rate": 3.144013755408895e-05,
      "loss": 2.9375,
      "step": 128
    },
    {
      "epoch": 0.5639344262295082,
      "grad_norm": 2.542893886566162,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 2.438,
      "step": 129
    },
    {
      "epoch": 0.5683060109289617,
      "grad_norm": 2.793731212615967,
      "learning_rate": 2.991522876735154e-05,
      "loss": 2.7625,
      "step": 130
    },
    {
      "epoch": 0.5726775956284152,
      "grad_norm": 2.736274480819702,
      "learning_rate": 2.916090697523549e-05,
      "loss": 2.7501,
      "step": 131
    },
    {
      "epoch": 0.5770491803278689,
      "grad_norm": 3.0499179363250732,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 2.9075,
      "step": 132
    },
    {
      "epoch": 0.5814207650273224,
      "grad_norm": 2.8469018936157227,
      "learning_rate": 2.766955965735968e-05,
      "loss": 2.7627,
      "step": 133
    },
    {
      "epoch": 0.585792349726776,
      "grad_norm": 2.7116096019744873,
      "learning_rate": 2.693294185106562e-05,
      "loss": 2.4971,
      "step": 134
    },
    {
      "epoch": 0.5901639344262295,
      "grad_norm": 2.695939540863037,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 2.5254,
      "step": 135
    },
    {
      "epoch": 0.594535519125683,
      "grad_norm": 2.9823148250579834,
      "learning_rate": 2.547882480847461e-05,
      "loss": 2.8118,
      "step": 136
    },
    {
      "epoch": 0.5989071038251366,
      "grad_norm": 3.03371000289917,
      "learning_rate": 2.476172311325783e-05,
      "loss": 2.6509,
      "step": 137
    },
    {
      "epoch": 0.6032786885245902,
      "grad_norm": 2.901766061782837,
      "learning_rate": 2.405152131093926e-05,
      "loss": 2.8283,
      "step": 138
    },
    {
      "epoch": 0.6076502732240437,
      "grad_norm": 2.5338025093078613,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 2.2989,
      "step": 139
    },
    {
      "epoch": 0.6120218579234973,
      "grad_norm": 2.924435615539551,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 2.8503,
      "step": 140
    },
    {
      "epoch": 0.6163934426229508,
      "grad_norm": 2.920508623123169,
      "learning_rate": 2.196424713241637e-05,
      "loss": 2.6653,
      "step": 141
    },
    {
      "epoch": 0.6207650273224044,
      "grad_norm": 2.804220199584961,
      "learning_rate": 2.128356686585282e-05,
      "loss": 2.445,
      "step": 142
    },
    {
      "epoch": 0.6251366120218579,
      "grad_norm": 2.881897449493408,
      "learning_rate": 2.061073738537635e-05,
      "loss": 2.2941,
      "step": 143
    },
    {
      "epoch": 0.6295081967213115,
      "grad_norm": 3.2550201416015625,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 2.4319,
      "step": 144
    },
    {
      "epoch": 0.6338797814207651,
      "grad_norm": 3.1499900817871094,
      "learning_rate": 1.928936436551661e-05,
      "loss": 2.7472,
      "step": 145
    },
    {
      "epoch": 0.6382513661202186,
      "grad_norm": 3.791355609893799,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 2.7002,
      "step": 146
    },
    {
      "epoch": 0.6426229508196721,
      "grad_norm": 3.142332077026367,
      "learning_rate": 1.800157297483417e-05,
      "loss": 2.4231,
      "step": 147
    },
    {
      "epoch": 0.6469945355191257,
      "grad_norm": 3.307037830352783,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 2.5866,
      "step": 148
    },
    {
      "epoch": 0.6513661202185792,
      "grad_norm": 3.90175199508667,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 2.5459,
      "step": 149
    },
    {
      "epoch": 0.6557377049180327,
      "grad_norm": 4.2901129722595215,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 2.6595,
      "step": 150
    },
    {
      "epoch": 0.6557377049180327,
      "eval_loss": 2.595000743865967,
      "eval_runtime": 6.2478,
      "eval_samples_per_second": 61.782,
      "eval_steps_per_second": 15.525,
      "step": 150
    },
    {
      "epoch": 0.6601092896174864,
      "grad_norm": 2.328075647354126,
      "learning_rate": 1.553232954407171e-05,
      "loss": 2.6545,
      "step": 151
    },
    {
      "epoch": 0.6644808743169399,
      "grad_norm": 2.3307573795318604,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 2.5771,
      "step": 152
    },
    {
      "epoch": 0.6688524590163935,
      "grad_norm": 2.4141488075256348,
      "learning_rate": 1.435357758543015e-05,
      "loss": 2.404,
      "step": 153
    },
    {
      "epoch": 0.673224043715847,
      "grad_norm": 2.554847002029419,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 2.5138,
      "step": 154
    },
    {
      "epoch": 0.6775956284153005,
      "grad_norm": 2.672499418258667,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 2.4998,
      "step": 155
    },
    {
      "epoch": 0.6819672131147541,
      "grad_norm": 2.756692409515381,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 2.6056,
      "step": 156
    },
    {
      "epoch": 0.6863387978142077,
      "grad_norm": 2.573209047317505,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 2.4626,
      "step": 157
    },
    {
      "epoch": 0.6907103825136612,
      "grad_norm": 2.6111252307891846,
      "learning_rate": 1.157994445715706e-05,
      "loss": 2.4958,
      "step": 158
    },
    {
      "epoch": 0.6950819672131148,
      "grad_norm": 2.6481621265411377,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 2.4732,
      "step": 159
    },
    {
      "epoch": 0.6994535519125683,
      "grad_norm": 2.672646999359131,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 2.7222,
      "step": 160
    },
    {
      "epoch": 0.7038251366120218,
      "grad_norm": 2.8730673789978027,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 2.5666,
      "step": 161
    },
    {
      "epoch": 0.7081967213114754,
      "grad_norm": 2.826026201248169,
      "learning_rate": 9.549150281252633e-06,
      "loss": 2.8777,
      "step": 162
    },
    {
      "epoch": 0.7125683060109289,
      "grad_norm": 2.426215171813965,
      "learning_rate": 9.068759265665384e-06,
      "loss": 2.4276,
      "step": 163
    },
    {
      "epoch": 0.7169398907103826,
      "grad_norm": 2.6274406909942627,
      "learning_rate": 8.599558442598998e-06,
      "loss": 2.7816,
      "step": 164
    },
    {
      "epoch": 0.7213114754098361,
      "grad_norm": 2.7676916122436523,
      "learning_rate": 8.141676086873572e-06,
      "loss": 2.7727,
      "step": 165
    },
    {
      "epoch": 0.7256830601092896,
      "grad_norm": 2.4770750999450684,
      "learning_rate": 7.695237378953223e-06,
      "loss": 2.6207,
      "step": 166
    },
    {
      "epoch": 0.7300546448087432,
      "grad_norm": 2.5767009258270264,
      "learning_rate": 7.260364370723044e-06,
      "loss": 2.2808,
      "step": 167
    },
    {
      "epoch": 0.7344262295081967,
      "grad_norm": 2.7475545406341553,
      "learning_rate": 6.837175952121306e-06,
      "loss": 2.683,
      "step": 168
    },
    {
      "epoch": 0.7387978142076502,
      "grad_norm": 2.733898639678955,
      "learning_rate": 6.425787818636131e-06,
      "loss": 2.745,
      "step": 169
    },
    {
      "epoch": 0.7431693989071039,
      "grad_norm": 2.9545092582702637,
      "learning_rate": 6.026312439675552e-06,
      "loss": 2.7705,
      "step": 170
    },
    {
      "epoch": 0.7475409836065574,
      "grad_norm": 2.800896406173706,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 2.639,
      "step": 171
    },
    {
      "epoch": 0.7519125683060109,
      "grad_norm": 2.4532346725463867,
      "learning_rate": 5.263533508961827e-06,
      "loss": 2.4557,
      "step": 172
    },
    {
      "epoch": 0.7562841530054645,
      "grad_norm": 2.5168747901916504,
      "learning_rate": 4.900438493352055e-06,
      "loss": 2.6409,
      "step": 173
    },
    {
      "epoch": 0.760655737704918,
      "grad_norm": 2.6835062503814697,
      "learning_rate": 4.549673247541875e-06,
      "loss": 2.6317,
      "step": 174
    },
    {
      "epoch": 0.7650273224043715,
      "grad_norm": 2.4492502212524414,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 2.5361,
      "step": 175
    },
    {
      "epoch": 0.7693989071038252,
      "grad_norm": 2.343942403793335,
      "learning_rate": 3.885512251130763e-06,
      "loss": 2.4797,
      "step": 176
    },
    {
      "epoch": 0.7737704918032787,
      "grad_norm": 2.6233112812042236,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 2.5932,
      "step": 177
    },
    {
      "epoch": 0.7781420765027323,
      "grad_norm": 2.7487967014312744,
      "learning_rate": 3.271776770026963e-06,
      "loss": 2.6665,
      "step": 178
    },
    {
      "epoch": 0.7825136612021858,
      "grad_norm": 2.8340024948120117,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 2.4697,
      "step": 179
    },
    {
      "epoch": 0.7868852459016393,
      "grad_norm": 2.7985870838165283,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 2.7148,
      "step": 180
    },
    {
      "epoch": 0.7912568306010929,
      "grad_norm": 2.6471643447875977,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 2.5548,
      "step": 181
    },
    {
      "epoch": 0.7956284153005464,
      "grad_norm": 3.110170364379883,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 2.4558,
      "step": 182
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.8638408184051514,
      "learning_rate": 1.962316193157593e-06,
      "loss": 2.4847,
      "step": 183
    },
    {
      "epoch": 0.8043715846994536,
      "grad_norm": 3.0267791748046875,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 2.831,
      "step": 184
    },
    {
      "epoch": 0.8087431693989071,
      "grad_norm": 2.809739112854004,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 2.3306,
      "step": 185
    },
    {
      "epoch": 0.8131147540983606,
      "grad_norm": 2.982239246368408,
      "learning_rate": 1.333670137599713e-06,
      "loss": 2.6445,
      "step": 186
    },
    {
      "epoch": 0.8174863387978142,
      "grad_norm": 3.3002889156341553,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 2.8025,
      "step": 187
    },
    {
      "epoch": 0.8218579234972677,
      "grad_norm": 3.0541188716888428,
      "learning_rate": 9.810017062595322e-07,
      "loss": 2.7641,
      "step": 188
    },
    {
      "epoch": 0.8262295081967214,
      "grad_norm": 3.0663740634918213,
      "learning_rate": 8.247462563808817e-07,
      "loss": 2.9265,
      "step": 189
    },
    {
      "epoch": 0.8306010928961749,
      "grad_norm": 2.663666009902954,
      "learning_rate": 6.819348298638839e-07,
      "loss": 2.3198,
      "step": 190
    },
    {
      "epoch": 0.8349726775956284,
      "grad_norm": 3.159210443496704,
      "learning_rate": 5.526064699265753e-07,
      "loss": 2.6147,
      "step": 191
    },
    {
      "epoch": 0.839344262295082,
      "grad_norm": 2.8222692012786865,
      "learning_rate": 4.367965336512403e-07,
      "loss": 2.563,
      "step": 192
    },
    {
      "epoch": 0.8437158469945355,
      "grad_norm": 3.365823745727539,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 2.6577,
      "step": 193
    },
    {
      "epoch": 0.848087431693989,
      "grad_norm": 2.962089776992798,
      "learning_rate": 2.458548727494292e-07,
      "loss": 2.7274,
      "step": 194
    },
    {
      "epoch": 0.8524590163934426,
      "grad_norm": 3.3423912525177,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 2.5187,
      "step": 195
    },
    {
      "epoch": 0.8568306010928962,
      "grad_norm": 2.9570553302764893,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 2.5091,
      "step": 196
    },
    {
      "epoch": 0.8612021857923498,
      "grad_norm": 2.900071144104004,
      "learning_rate": 6.150154258476315e-08,
      "loss": 2.662,
      "step": 197
    },
    {
      "epoch": 0.8655737704918033,
      "grad_norm": 3.2200496196746826,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 2.54,
      "step": 198
    },
    {
      "epoch": 0.8699453551912568,
      "grad_norm": 3.7401468753814697,
      "learning_rate": 6.834750376549792e-09,
      "loss": 2.4818,
      "step": 199
    },
    {
      "epoch": 0.8743169398907104,
      "grad_norm": 4.200563907623291,
      "learning_rate": 0.0,
      "loss": 2.9021,
      "step": 200
    },
    {
      "epoch": 0.8743169398907104,
      "eval_loss": 2.577688217163086,
      "eval_runtime": 6.2476,
      "eval_samples_per_second": 61.784,
      "eval_steps_per_second": 15.526,
      "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": 1.3317552734208e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}