{
  "best_metric": 0.4371192753314972,
  "best_model_checkpoint": "miner_id_24/checkpoint-150",
  "epoch": 1.834862385321101,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.009174311926605505,
      "grad_norm": 0.2309478372335434,
      "learning_rate": 1e-05,
      "loss": 0.7218,
      "step": 1
    },
    {
      "epoch": 0.009174311926605505,
      "eval_loss": 0.9211078882217407,
      "eval_runtime": 5.5712,
      "eval_samples_per_second": 33.027,
      "eval_steps_per_second": 8.257,
      "step": 1
    },
    {
      "epoch": 0.01834862385321101,
      "grad_norm": 0.2558485269546509,
      "learning_rate": 2e-05,
      "loss": 0.3937,
      "step": 2
    },
    {
      "epoch": 0.027522935779816515,
      "grad_norm": 0.26436418294906616,
      "learning_rate": 3e-05,
      "loss": 0.446,
      "step": 3
    },
    {
      "epoch": 0.03669724770642202,
      "grad_norm": 0.27746403217315674,
      "learning_rate": 4e-05,
      "loss": 0.6881,
      "step": 4
    },
    {
      "epoch": 0.045871559633027525,
      "grad_norm": 0.270297646522522,
      "learning_rate": 5e-05,
      "loss": 0.6016,
      "step": 5
    },
    {
      "epoch": 0.05504587155963303,
      "grad_norm": 0.316300630569458,
      "learning_rate": 6e-05,
      "loss": 0.4676,
      "step": 6
    },
    {
      "epoch": 0.06422018348623854,
      "grad_norm": 0.34652575850486755,
      "learning_rate": 7e-05,
      "loss": 0.5995,
      "step": 7
    },
    {
      "epoch": 0.07339449541284404,
      "grad_norm": 0.28786805272102356,
      "learning_rate": 8e-05,
      "loss": 0.6142,
      "step": 8
    },
    {
      "epoch": 0.08256880733944955,
      "grad_norm": 0.25441649556159973,
      "learning_rate": 9e-05,
      "loss": 0.5177,
      "step": 9
    },
    {
      "epoch": 0.09174311926605505,
      "grad_norm": 0.276737242937088,
      "learning_rate": 0.0001,
      "loss": 0.706,
      "step": 10
    },
    {
      "epoch": 0.10091743119266056,
      "grad_norm": 0.242123544216156,
      "learning_rate": 9.999316524962345e-05,
      "loss": 0.416,
      "step": 11
    },
    {
      "epoch": 0.11009174311926606,
      "grad_norm": 0.2807343304157257,
      "learning_rate": 9.997266286704631e-05,
      "loss": 0.5332,
      "step": 12
    },
    {
      "epoch": 0.11926605504587157,
      "grad_norm": 0.3464376628398895,
      "learning_rate": 9.993849845741524e-05,
      "loss": 0.6402,
      "step": 13
    },
    {
      "epoch": 0.12844036697247707,
      "grad_norm": 0.31960099935531616,
      "learning_rate": 9.989068136093873e-05,
      "loss": 0.591,
      "step": 14
    },
    {
      "epoch": 0.13761467889908258,
      "grad_norm": 0.38411182165145874,
      "learning_rate": 9.98292246503335e-05,
      "loss": 0.6984,
      "step": 15
    },
    {
      "epoch": 0.14678899082568808,
      "grad_norm": 0.3537607491016388,
      "learning_rate": 9.975414512725057e-05,
      "loss": 0.6213,
      "step": 16
    },
    {
      "epoch": 0.1559633027522936,
      "grad_norm": 0.34811100363731384,
      "learning_rate": 9.966546331768191e-05,
      "loss": 0.6437,
      "step": 17
    },
    {
      "epoch": 0.1651376146788991,
      "grad_norm": 0.31534284353256226,
      "learning_rate": 9.956320346634876e-05,
      "loss": 0.7355,
      "step": 18
    },
    {
      "epoch": 0.1743119266055046,
      "grad_norm": 0.38784778118133545,
      "learning_rate": 9.944739353007344e-05,
      "loss": 0.7483,
      "step": 19
    },
    {
      "epoch": 0.1834862385321101,
      "grad_norm": 0.36261245608329773,
      "learning_rate": 9.931806517013612e-05,
      "loss": 0.7511,
      "step": 20
    },
    {
      "epoch": 0.1926605504587156,
      "grad_norm": 0.3402577340602875,
      "learning_rate": 9.917525374361912e-05,
      "loss": 0.6167,
      "step": 21
    },
    {
      "epoch": 0.2018348623853211,
      "grad_norm": 0.38130655884742737,
      "learning_rate": 9.901899829374047e-05,
      "loss": 0.7665,
      "step": 22
    },
    {
      "epoch": 0.21100917431192662,
      "grad_norm": 0.3944738805294037,
      "learning_rate": 9.884934153917997e-05,
      "loss": 0.7471,
      "step": 23
    },
    {
      "epoch": 0.22018348623853212,
      "grad_norm": 0.41142916679382324,
      "learning_rate": 9.86663298624003e-05,
      "loss": 0.5941,
      "step": 24
    },
    {
      "epoch": 0.22935779816513763,
      "grad_norm": 0.4905121922492981,
      "learning_rate": 9.847001329696653e-05,
      "loss": 0.7661,
      "step": 25
    },
    {
      "epoch": 0.23853211009174313,
      "grad_norm": 0.5909762382507324,
      "learning_rate": 9.826044551386744e-05,
      "loss": 0.8668,
      "step": 26
    },
    {
      "epoch": 0.24770642201834864,
      "grad_norm": 0.9485186338424683,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.3186,
      "step": 27
    },
    {
      "epoch": 0.25688073394495414,
      "grad_norm": 0.26398882269859314,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.3755,
      "step": 28
    },
    {
      "epoch": 0.26605504587155965,
      "grad_norm": 0.2771925926208496,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.2588,
      "step": 29
    },
    {
      "epoch": 0.27522935779816515,
      "grad_norm": 0.27546319365501404,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.2342,
      "step": 30
    },
    {
      "epoch": 0.28440366972477066,
      "grad_norm": 0.31774288415908813,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.4869,
      "step": 31
    },
    {
      "epoch": 0.29357798165137616,
      "grad_norm": 0.2786838114261627,
      "learning_rate": 9.672822322997305e-05,
      "loss": 0.4508,
      "step": 32
    },
    {
      "epoch": 0.30275229357798167,
      "grad_norm": 0.2284572869539261,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.4029,
      "step": 33
    },
    {
      "epoch": 0.3119266055045872,
      "grad_norm": 0.23630881309509277,
      "learning_rate": 9.611448774886924e-05,
      "loss": 0.4647,
      "step": 34
    },
    {
      "epoch": 0.3211009174311927,
      "grad_norm": 0.21207673847675323,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.4569,
      "step": 35
    },
    {
      "epoch": 0.3302752293577982,
      "grad_norm": 0.2219172716140747,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.4848,
      "step": 36
    },
    {
      "epoch": 0.3394495412844037,
      "grad_norm": 0.24454958736896515,
      "learning_rate": 9.509956150664796e-05,
      "loss": 0.5733,
      "step": 37
    },
    {
      "epoch": 0.3486238532110092,
      "grad_norm": 0.21839550137519836,
      "learning_rate": 9.473646649103818e-05,
      "loss": 0.4542,
      "step": 38
    },
    {
      "epoch": 0.3577981651376147,
      "grad_norm": 0.23129194974899292,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.3962,
      "step": 39
    },
    {
      "epoch": 0.3669724770642202,
      "grad_norm": 0.30784866213798523,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.6942,
      "step": 40
    },
    {
      "epoch": 0.3761467889908257,
      "grad_norm": 0.26374784111976624,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.5208,
      "step": 41
    },
    {
      "epoch": 0.3853211009174312,
      "grad_norm": 0.23250478506088257,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.4836,
      "step": 42
    },
    {
      "epoch": 0.3944954128440367,
      "grad_norm": 0.2499152570962906,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.5889,
      "step": 43
    },
    {
      "epoch": 0.4036697247706422,
      "grad_norm": 0.2545929253101349,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.5097,
      "step": 44
    },
    {
      "epoch": 0.41284403669724773,
      "grad_norm": 0.27340060472488403,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.5573,
      "step": 45
    },
    {
      "epoch": 0.42201834862385323,
      "grad_norm": 0.2989652752876282,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.6495,
      "step": 46
    },
    {
      "epoch": 0.43119266055045874,
      "grad_norm": 0.31596639752388,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.6203,
      "step": 47
    },
    {
      "epoch": 0.44036697247706424,
      "grad_norm": 0.2849104702472687,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.5877,
      "step": 48
    },
    {
      "epoch": 0.44954128440366975,
      "grad_norm": 0.315329909324646,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.5392,
      "step": 49
    },
    {
      "epoch": 0.45871559633027525,
      "grad_norm": 0.2864372134208679,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.4789,
      "step": 50
    },
    {
      "epoch": 0.45871559633027525,
      "eval_loss": 0.5081414580345154,
      "eval_runtime": 5.6397,
      "eval_samples_per_second": 32.626,
      "eval_steps_per_second": 8.156,
      "step": 50
    },
    {
      "epoch": 0.46788990825688076,
      "grad_norm": 0.413135290145874,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.7594,
      "step": 51
    },
    {
      "epoch": 0.47706422018348627,
      "grad_norm": 0.3986172080039978,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.6837,
      "step": 52
    },
    {
      "epoch": 0.48623853211009177,
      "grad_norm": 0.4236753284931183,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.6169,
      "step": 53
    },
    {
      "epoch": 0.4954128440366973,
      "grad_norm": 0.7464210987091064,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.8942,
      "step": 54
    },
    {
      "epoch": 0.5045871559633027,
      "grad_norm": 0.12521986663341522,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.207,
      "step": 55
    },
    {
      "epoch": 0.5137614678899083,
      "grad_norm": 0.16913723945617676,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.3694,
      "step": 56
    },
    {
      "epoch": 0.5229357798165137,
      "grad_norm": 0.17396824061870575,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.2562,
      "step": 57
    },
    {
      "epoch": 0.5321100917431193,
      "grad_norm": 0.19990207254886627,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.3281,
      "step": 58
    },
    {
      "epoch": 0.5412844036697247,
      "grad_norm": 0.20998096466064453,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.2928,
      "step": 59
    },
    {
      "epoch": 0.5504587155963303,
      "grad_norm": 0.20039355754852295,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.4086,
      "step": 60
    },
    {
      "epoch": 0.5596330275229358,
      "grad_norm": 0.20050932466983795,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.3363,
      "step": 61
    },
    {
      "epoch": 0.5688073394495413,
      "grad_norm": 0.20360559225082397,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.4509,
      "step": 62
    },
    {
      "epoch": 0.5779816513761468,
      "grad_norm": 0.18123404681682587,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.2623,
      "step": 63
    },
    {
      "epoch": 0.5871559633027523,
      "grad_norm": 0.18353700637817383,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.3231,
      "step": 64
    },
    {
      "epoch": 0.5963302752293578,
      "grad_norm": 0.21949049830436707,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.4079,
      "step": 65
    },
    {
      "epoch": 0.6055045871559633,
      "grad_norm": 0.2185269594192505,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.4271,
      "step": 66
    },
    {
      "epoch": 0.6146788990825688,
      "grad_norm": 0.27815037965774536,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.5174,
      "step": 67
    },
    {
      "epoch": 0.6238532110091743,
      "grad_norm": 0.2754443883895874,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.6151,
      "step": 68
    },
    {
      "epoch": 0.6330275229357798,
      "grad_norm": 0.303902804851532,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.6629,
      "step": 69
    },
    {
      "epoch": 0.6422018348623854,
      "grad_norm": 0.30774471163749695,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.6654,
      "step": 70
    },
    {
      "epoch": 0.6513761467889908,
      "grad_norm": 0.21824637055397034,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.3602,
      "step": 71
    },
    {
      "epoch": 0.6605504587155964,
      "grad_norm": 0.3259966969490051,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.5904,
      "step": 72
    },
    {
      "epoch": 0.6697247706422018,
      "grad_norm": 0.2658520042896271,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.4526,
      "step": 73
    },
    {
      "epoch": 0.6788990825688074,
      "grad_norm": 0.3060029447078705,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.5391,
      "step": 74
    },
    {
      "epoch": 0.6880733944954128,
      "grad_norm": 0.4056127965450287,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.5666,
      "step": 75
    },
    {
      "epoch": 0.6972477064220184,
      "grad_norm": 0.2942802906036377,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.4573,
      "step": 76
    },
    {
      "epoch": 0.7064220183486238,
      "grad_norm": 0.34859296679496765,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.6152,
      "step": 77
    },
    {
      "epoch": 0.7155963302752294,
      "grad_norm": 0.36043593287467957,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.7046,
      "step": 78
    },
    {
      "epoch": 0.7247706422018348,
      "grad_norm": 0.38596341013908386,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.6234,
      "step": 79
    },
    {
      "epoch": 0.7339449541284404,
      "grad_norm": 0.4440882205963135,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.6546,
      "step": 80
    },
    {
      "epoch": 0.7431192660550459,
      "grad_norm": 0.5909287333488464,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.6696,
      "step": 81
    },
    {
      "epoch": 0.7522935779816514,
      "grad_norm": 0.13881871104240417,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.2437,
      "step": 82
    },
    {
      "epoch": 0.7614678899082569,
      "grad_norm": 0.14617930352687836,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.2884,
      "step": 83
    },
    {
      "epoch": 0.7706422018348624,
      "grad_norm": 0.17851407825946808,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.4207,
      "step": 84
    },
    {
      "epoch": 0.7798165137614679,
      "grad_norm": 0.17666982114315033,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.3388,
      "step": 85
    },
    {
      "epoch": 0.7889908256880734,
      "grad_norm": 0.20175470411777496,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.4042,
      "step": 86
    },
    {
      "epoch": 0.7981651376146789,
      "grad_norm": 0.2009691745042801,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.4041,
      "step": 87
    },
    {
      "epoch": 0.8073394495412844,
      "grad_norm": 0.23336203396320343,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.6005,
      "step": 88
    },
    {
      "epoch": 0.8165137614678899,
      "grad_norm": 0.22401797771453857,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.5363,
      "step": 89
    },
    {
      "epoch": 0.8256880733944955,
      "grad_norm": 0.2362067997455597,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.5477,
      "step": 90
    },
    {
      "epoch": 0.8348623853211009,
      "grad_norm": 0.2374013513326645,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.4734,
      "step": 91
    },
    {
      "epoch": 0.8440366972477065,
      "grad_norm": 0.29697301983833313,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.6649,
      "step": 92
    },
    {
      "epoch": 0.8532110091743119,
      "grad_norm": 0.2264970988035202,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.44,
      "step": 93
    },
    {
      "epoch": 0.8623853211009175,
      "grad_norm": 0.22832238674163818,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.4787,
      "step": 94
    },
    {
      "epoch": 0.8715596330275229,
      "grad_norm": 0.28362220525741577,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.5545,
      "step": 95
    },
    {
      "epoch": 0.8807339449541285,
      "grad_norm": 0.22183015942573547,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.3126,
      "step": 96
    },
    {
      "epoch": 0.8899082568807339,
      "grad_norm": 0.24474367499351501,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.4426,
      "step": 97
    },
    {
      "epoch": 0.8990825688073395,
      "grad_norm": 0.3317551612854004,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.627,
      "step": 98
    },
    {
      "epoch": 0.908256880733945,
      "grad_norm": 0.32000917196273804,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.5788,
      "step": 99
    },
    {
      "epoch": 0.9174311926605505,
      "grad_norm": 0.2908952832221985,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.4396,
      "step": 100
    },
    {
      "epoch": 0.9174311926605505,
      "eval_loss": 0.46243494749069214,
      "eval_runtime": 5.6372,
      "eval_samples_per_second": 32.641,
      "eval_steps_per_second": 8.16,
      "step": 100
    },
    {
      "epoch": 0.926605504587156,
      "grad_norm": 0.3079433739185333,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.5249,
      "step": 101
    },
    {
      "epoch": 0.9357798165137615,
      "grad_norm": 0.2792729139328003,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.4466,
      "step": 102
    },
    {
      "epoch": 0.944954128440367,
      "grad_norm": 0.3364558219909668,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.5365,
      "step": 103
    },
    {
      "epoch": 0.9541284403669725,
      "grad_norm": 0.3221740126609802,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.496,
      "step": 104
    },
    {
      "epoch": 0.963302752293578,
      "grad_norm": 0.3582867980003357,
      "learning_rate": 5e-05,
      "loss": 0.5445,
      "step": 105
    },
    {
      "epoch": 0.9724770642201835,
      "grad_norm": 0.44838571548461914,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.6834,
      "step": 106
    },
    {
      "epoch": 0.981651376146789,
      "grad_norm": 0.39783820509910583,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.4235,
      "step": 107
    },
    {
      "epoch": 0.9908256880733946,
      "grad_norm": 0.8808133602142334,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.985,
      "step": 108
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.2551686763763428,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.528,
      "step": 109
    },
    {
      "epoch": 1.0091743119266054,
      "grad_norm": 0.1278965026140213,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.1972,
      "step": 110
    },
    {
      "epoch": 1.018348623853211,
      "grad_norm": 0.141044482588768,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.2327,
      "step": 111
    },
    {
      "epoch": 1.0275229357798166,
      "grad_norm": 0.13628904521465302,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.1995,
      "step": 112
    },
    {
      "epoch": 1.036697247706422,
      "grad_norm": 0.18226638436317444,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.3852,
      "step": 113
    },
    {
      "epoch": 1.0458715596330275,
      "grad_norm": 0.22630174458026886,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.463,
      "step": 114
    },
    {
      "epoch": 1.0550458715596331,
      "grad_norm": 0.18771854043006897,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.2828,
      "step": 115
    },
    {
      "epoch": 1.0642201834862386,
      "grad_norm": 0.21680407226085663,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.4297,
      "step": 116
    },
    {
      "epoch": 1.073394495412844,
      "grad_norm": 0.17024627327919006,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.2751,
      "step": 117
    },
    {
      "epoch": 1.0825688073394495,
      "grad_norm": 0.21628738939762115,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.3745,
      "step": 118
    },
    {
      "epoch": 1.091743119266055,
      "grad_norm": 0.24735943973064423,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.4844,
      "step": 119
    },
    {
      "epoch": 1.1009174311926606,
      "grad_norm": 0.21772918105125427,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.2957,
      "step": 120
    },
    {
      "epoch": 1.110091743119266,
      "grad_norm": 0.2418142408132553,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.2949,
      "step": 121
    },
    {
      "epoch": 1.1192660550458715,
      "grad_norm": 0.25259315967559814,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.4492,
      "step": 122
    },
    {
      "epoch": 1.1284403669724772,
      "grad_norm": 0.3049667477607727,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.4814,
      "step": 123
    },
    {
      "epoch": 1.1376146788990826,
      "grad_norm": 0.30187371373176575,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.5876,
      "step": 124
    },
    {
      "epoch": 1.146788990825688,
      "grad_norm": 0.31283828616142273,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.5676,
      "step": 125
    },
    {
      "epoch": 1.1559633027522935,
      "grad_norm": 0.2810775637626648,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.4347,
      "step": 126
    },
    {
      "epoch": 1.165137614678899,
      "grad_norm": 0.23284779489040375,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.3206,
      "step": 127
    },
    {
      "epoch": 1.1743119266055047,
      "grad_norm": 0.29070916771888733,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.4457,
      "step": 128
    },
    {
      "epoch": 1.18348623853211,
      "grad_norm": 0.3149646818637848,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.4184,
      "step": 129
    },
    {
      "epoch": 1.1926605504587156,
      "grad_norm": 0.33116212487220764,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.4638,
      "step": 130
    },
    {
      "epoch": 1.2018348623853212,
      "grad_norm": 0.33836284279823303,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.472,
      "step": 131
    },
    {
      "epoch": 1.2110091743119267,
      "grad_norm": 0.3052822947502136,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.4088,
      "step": 132
    },
    {
      "epoch": 1.2201834862385321,
      "grad_norm": 0.3751441240310669,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.5534,
      "step": 133
    },
    {
      "epoch": 1.2293577981651376,
      "grad_norm": 0.4007648527622223,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.522,
      "step": 134
    },
    {
      "epoch": 1.238532110091743,
      "grad_norm": 0.4260527193546295,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.3983,
      "step": 135
    },
    {
      "epoch": 1.2477064220183487,
      "grad_norm": 0.5929573178291321,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.5811,
      "step": 136
    },
    {
      "epoch": 1.2568807339449541,
      "grad_norm": 0.1770479828119278,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.4312,
      "step": 137
    },
    {
      "epoch": 1.2660550458715596,
      "grad_norm": 0.182498037815094,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.3243,
      "step": 138
    },
    {
      "epoch": 1.2752293577981653,
      "grad_norm": 0.21625658869743347,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.3693,
      "step": 139
    },
    {
      "epoch": 1.2844036697247707,
      "grad_norm": 0.19629067182540894,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.3357,
      "step": 140
    },
    {
      "epoch": 1.2935779816513762,
      "grad_norm": 0.23945830762386322,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.4062,
      "step": 141
    },
    {
      "epoch": 1.3027522935779816,
      "grad_norm": 0.21827231347560883,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.2891,
      "step": 142
    },
    {
      "epoch": 1.311926605504587,
      "grad_norm": 0.21502086520195007,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.2466,
      "step": 143
    },
    {
      "epoch": 1.3211009174311927,
      "grad_norm": 0.28084877133369446,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.4362,
      "step": 144
    },
    {
      "epoch": 1.3302752293577982,
      "grad_norm": 0.2033384144306183,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.232,
      "step": 145
    },
    {
      "epoch": 1.3394495412844036,
      "grad_norm": 0.28219449520111084,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.423,
      "step": 146
    },
    {
      "epoch": 1.3486238532110093,
      "grad_norm": 0.24117745459079742,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.3882,
      "step": 147
    },
    {
      "epoch": 1.3577981651376148,
      "grad_norm": 0.35435786843299866,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.5143,
      "step": 148
    },
    {
      "epoch": 1.3669724770642202,
      "grad_norm": 0.2955686151981354,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.4645,
      "step": 149
    },
    {
      "epoch": 1.3761467889908257,
      "grad_norm": 0.29498574137687683,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.4583,
      "step": 150
    },
    {
      "epoch": 1.3761467889908257,
      "eval_loss": 0.4371192753314972,
      "eval_runtime": 5.6365,
      "eval_samples_per_second": 32.645,
      "eval_steps_per_second": 8.161,
      "step": 150
    },
    {
      "epoch": 1.385321100917431,
      "grad_norm": 0.2961886525154114,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.4263,
      "step": 151
    },
    {
      "epoch": 1.3944954128440368,
      "grad_norm": 0.2849169969558716,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.4266,
      "step": 152
    },
    {
      "epoch": 1.4036697247706422,
      "grad_norm": 0.35253971815109253,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.3831,
      "step": 153
    },
    {
      "epoch": 1.4128440366972477,
      "grad_norm": 0.36027321219444275,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.5065,
      "step": 154
    },
    {
      "epoch": 1.4220183486238533,
      "grad_norm": 0.29704010486602783,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.3747,
      "step": 155
    },
    {
      "epoch": 1.4311926605504588,
      "grad_norm": 0.3163253664970398,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.3458,
      "step": 156
    },
    {
      "epoch": 1.4403669724770642,
      "grad_norm": 0.35480210185050964,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.3676,
      "step": 157
    },
    {
      "epoch": 1.4495412844036697,
      "grad_norm": 0.36188286542892456,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.3767,
      "step": 158
    },
    {
      "epoch": 1.4587155963302751,
      "grad_norm": 0.43707719445228577,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.5595,
      "step": 159
    },
    {
      "epoch": 1.4678899082568808,
      "grad_norm": 0.42985525727272034,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.4369,
      "step": 160
    },
    {
      "epoch": 1.4770642201834863,
      "grad_norm": 0.4681553840637207,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.4487,
      "step": 161
    },
    {
      "epoch": 1.4862385321100917,
      "grad_norm": 0.4328238070011139,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.4018,
      "step": 162
    },
    {
      "epoch": 1.4954128440366974,
      "grad_norm": 0.6042909622192383,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.5553,
      "step": 163
    },
    {
      "epoch": 1.5045871559633026,
      "grad_norm": 0.1732768416404724,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.3188,
      "step": 164
    },
    {
      "epoch": 1.5137614678899083,
      "grad_norm": 0.15223254263401031,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.1591,
      "step": 165
    },
    {
      "epoch": 1.5229357798165137,
      "grad_norm": 0.1565789133310318,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.1634,
      "step": 166
    },
    {
      "epoch": 1.5321100917431192,
      "grad_norm": 0.2163132131099701,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.3584,
      "step": 167
    },
    {
      "epoch": 1.5412844036697249,
      "grad_norm": 0.20443567633628845,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.2952,
      "step": 168
    },
    {
      "epoch": 1.5504587155963303,
      "grad_norm": 0.21337220072746277,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.2751,
      "step": 169
    },
    {
      "epoch": 1.5596330275229358,
      "grad_norm": 0.21843741834163666,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.3128,
      "step": 170
    },
    {
      "epoch": 1.5688073394495414,
      "grad_norm": 0.24943356215953827,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.3791,
      "step": 171
    },
    {
      "epoch": 1.5779816513761467,
      "grad_norm": 0.2663065493106842,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.4073,
      "step": 172
    },
    {
      "epoch": 1.5871559633027523,
      "grad_norm": 0.26393768191337585,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.4341,
      "step": 173
    },
    {
      "epoch": 1.5963302752293578,
      "grad_norm": 0.2898072302341461,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.4263,
      "step": 174
    },
    {
      "epoch": 1.6055045871559632,
      "grad_norm": 0.2792298197746277,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.3971,
      "step": 175
    },
    {
      "epoch": 1.614678899082569,
      "grad_norm": 0.2888216972351074,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.3595,
      "step": 176
    },
    {
      "epoch": 1.6238532110091743,
      "grad_norm": 0.2951292097568512,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.4329,
      "step": 177
    },
    {
      "epoch": 1.6330275229357798,
      "grad_norm": 0.293870747089386,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.3425,
      "step": 178
    },
    {
      "epoch": 1.6422018348623855,
      "grad_norm": 0.2852609157562256,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.3408,
      "step": 179
    },
    {
      "epoch": 1.6513761467889907,
      "grad_norm": 0.3356017470359802,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.4425,
      "step": 180
    },
    {
      "epoch": 1.6605504587155964,
      "grad_norm": 0.3327827453613281,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.3166,
      "step": 181
    },
    {
      "epoch": 1.6697247706422018,
      "grad_norm": 0.35221245884895325,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.4511,
      "step": 182
    },
    {
      "epoch": 1.6788990825688073,
      "grad_norm": 0.4240897297859192,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.6051,
      "step": 183
    },
    {
      "epoch": 1.688073394495413,
      "grad_norm": 0.3719145357608795,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.4049,
      "step": 184
    },
    {
      "epoch": 1.6972477064220184,
      "grad_norm": 0.4433523714542389,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.5137,
      "step": 185
    },
    {
      "epoch": 1.7064220183486238,
      "grad_norm": 0.3403400480747223,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.3654,
      "step": 186
    },
    {
      "epoch": 1.7155963302752295,
      "grad_norm": 0.44516339898109436,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.5133,
      "step": 187
    },
    {
      "epoch": 1.7247706422018347,
      "grad_norm": 0.4940146505832672,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.506,
      "step": 188
    },
    {
      "epoch": 1.7339449541284404,
      "grad_norm": 0.6156914234161377,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.5122,
      "step": 189
    },
    {
      "epoch": 1.7431192660550459,
      "grad_norm": 0.90720534324646,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.6268,
      "step": 190
    },
    {
      "epoch": 1.7522935779816513,
      "grad_norm": 0.1722932606935501,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.2719,
      "step": 191
    },
    {
      "epoch": 1.761467889908257,
      "grad_norm": 0.1498982012271881,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.1906,
      "step": 192
    },
    {
      "epoch": 1.7706422018348624,
      "grad_norm": 0.17369672656059265,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.2508,
      "step": 193
    },
    {
      "epoch": 1.7798165137614679,
      "grad_norm": 0.26676490902900696,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.4323,
      "step": 194
    },
    {
      "epoch": 1.7889908256880735,
      "grad_norm": 0.17191696166992188,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.2093,
      "step": 195
    },
    {
      "epoch": 1.7981651376146788,
      "grad_norm": 0.20608827471733093,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.2698,
      "step": 196
    },
    {
      "epoch": 1.8073394495412844,
      "grad_norm": 0.28117769956588745,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.4748,
      "step": 197
    },
    {
      "epoch": 1.81651376146789,
      "grad_norm": 0.26510024070739746,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.3991,
      "step": 198
    },
    {
      "epoch": 1.8256880733944953,
      "grad_norm": 0.2825140655040741,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.4276,
      "step": 199
    },
    {
      "epoch": 1.834862385321101,
      "grad_norm": 0.2518618404865265,
      "learning_rate": 0.0,
      "loss": 0.3283,
      "step": 200
    },
    {
      "epoch": 1.834862385321101,
      "eval_loss": 0.43783652782440186,
      "eval_runtime": 5.6364,
      "eval_samples_per_second": 32.645,
      "eval_steps_per_second": 8.161,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 1
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 6.61808020979712e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}