{
  "best_metric": 1.339722990989685,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.2809975412715139,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0014049877063575693,
      "grad_norm": 0.9972674250602722,
      "learning_rate": 1e-05,
      "loss": 2.9055,
      "step": 1
    },
    {
      "epoch": 0.0014049877063575693,
      "eval_loss": 4.420830726623535,
      "eval_runtime": 83.3308,
      "eval_samples_per_second": 14.388,
      "eval_steps_per_second": 3.6,
      "step": 1
    },
    {
      "epoch": 0.0028099754127151387,
      "grad_norm": 1.1112552881240845,
      "learning_rate": 2e-05,
      "loss": 3.0491,
      "step": 2
    },
    {
      "epoch": 0.004214963119072708,
      "grad_norm": 1.0404009819030762,
      "learning_rate": 3e-05,
      "loss": 2.9953,
      "step": 3
    },
    {
      "epoch": 0.005619950825430277,
      "grad_norm": 1.1601241827011108,
      "learning_rate": 4e-05,
      "loss": 3.0966,
      "step": 4
    },
    {
      "epoch": 0.007024938531787847,
      "grad_norm": 1.3235045671463013,
      "learning_rate": 5e-05,
      "loss": 3.3946,
      "step": 5
    },
    {
      "epoch": 0.008429926238145416,
      "grad_norm": 1.2510929107666016,
      "learning_rate": 6e-05,
      "loss": 3.2366,
      "step": 6
    },
    {
      "epoch": 0.009834913944502985,
      "grad_norm": 1.294310450553894,
      "learning_rate": 7e-05,
      "loss": 3.1938,
      "step": 7
    },
    {
      "epoch": 0.011239901650860555,
      "grad_norm": 1.5444555282592773,
      "learning_rate": 8e-05,
      "loss": 3.2229,
      "step": 8
    },
    {
      "epoch": 0.012644889357218124,
      "grad_norm": 1.3435684442520142,
      "learning_rate": 9e-05,
      "loss": 2.9153,
      "step": 9
    },
    {
      "epoch": 0.014049877063575694,
      "grad_norm": 1.2428182363510132,
      "learning_rate": 0.0001,
      "loss": 2.7069,
      "step": 10
    },
    {
      "epoch": 0.015454864769933263,
      "grad_norm": 1.3734021186828613,
      "learning_rate": 9.999316524962345e-05,
      "loss": 2.5861,
      "step": 11
    },
    {
      "epoch": 0.01685985247629083,
      "grad_norm": 1.3884536027908325,
      "learning_rate": 9.997266286704631e-05,
      "loss": 2.3073,
      "step": 12
    },
    {
      "epoch": 0.0182648401826484,
      "grad_norm": 1.833651065826416,
      "learning_rate": 9.993849845741524e-05,
      "loss": 2.3279,
      "step": 13
    },
    {
      "epoch": 0.01966982788900597,
      "grad_norm": 2.1904890537261963,
      "learning_rate": 9.989068136093873e-05,
      "loss": 2.0808,
      "step": 14
    },
    {
      "epoch": 0.02107481559536354,
      "grad_norm": 1.6890671253204346,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.9595,
      "step": 15
    },
    {
      "epoch": 0.02247980330172111,
      "grad_norm": 1.8512685298919678,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.7353,
      "step": 16
    },
    {
      "epoch": 0.02388479100807868,
      "grad_norm": 1.4279896020889282,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.7225,
      "step": 17
    },
    {
      "epoch": 0.02528977871443625,
      "grad_norm": 1.1936205625534058,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.6935,
      "step": 18
    },
    {
      "epoch": 0.026694766420793818,
      "grad_norm": 1.207401990890503,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.6212,
      "step": 19
    },
    {
      "epoch": 0.028099754127151388,
      "grad_norm": 1.2050328254699707,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.6166,
      "step": 20
    },
    {
      "epoch": 0.029504741833508957,
      "grad_norm": 1.1021475791931152,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.6601,
      "step": 21
    },
    {
      "epoch": 0.030909729539866527,
      "grad_norm": 1.0891368389129639,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.4044,
      "step": 22
    },
    {
      "epoch": 0.032314717246224096,
      "grad_norm": 1.040321707725525,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.5733,
      "step": 23
    },
    {
      "epoch": 0.03371970495258166,
      "grad_norm": 1.0636420249938965,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.6717,
      "step": 24
    },
    {
      "epoch": 0.035124692658939236,
      "grad_norm": 1.026659369468689,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.6393,
      "step": 25
    },
    {
      "epoch": 0.0365296803652968,
      "grad_norm": 0.9542109370231628,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.4407,
      "step": 26
    },
    {
      "epoch": 0.037934668071654375,
      "grad_norm": 1.0161194801330566,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.486,
      "step": 27
    },
    {
      "epoch": 0.03933965577801194,
      "grad_norm": 1.0536093711853027,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.4646,
      "step": 28
    },
    {
      "epoch": 0.040744643484369514,
      "grad_norm": 0.9943352341651917,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.4431,
      "step": 29
    },
    {
      "epoch": 0.04214963119072708,
      "grad_norm": 1.1126229763031006,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.6247,
      "step": 30
    },
    {
      "epoch": 0.04355461889708465,
      "grad_norm": 1.010001301765442,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.2922,
      "step": 31
    },
    {
      "epoch": 0.04495960660344222,
      "grad_norm": 0.9857097864151001,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.196,
      "step": 32
    },
    {
      "epoch": 0.04636459430979979,
      "grad_norm": 0.9724398851394653,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.3649,
      "step": 33
    },
    {
      "epoch": 0.04776958201615736,
      "grad_norm": 1.0950841903686523,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.5287,
      "step": 34
    },
    {
      "epoch": 0.04917456972251493,
      "grad_norm": 0.9695940017700195,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.0857,
      "step": 35
    },
    {
      "epoch": 0.0505795574288725,
      "grad_norm": 1.477586269378662,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.1775,
      "step": 36
    },
    {
      "epoch": 0.05198454513523006,
      "grad_norm": 1.0694470405578613,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.4015,
      "step": 37
    },
    {
      "epoch": 0.053389532841587636,
      "grad_norm": 1.1835302114486694,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.4238,
      "step": 38
    },
    {
      "epoch": 0.0547945205479452,
      "grad_norm": 1.1791400909423828,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.415,
      "step": 39
    },
    {
      "epoch": 0.056199508254302775,
      "grad_norm": 1.2953441143035889,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.4913,
      "step": 40
    },
    {
      "epoch": 0.05760449596066034,
      "grad_norm": 1.400848150253296,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.3347,
      "step": 41
    },
    {
      "epoch": 0.059009483667017915,
      "grad_norm": 1.141520380973816,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.2636,
      "step": 42
    },
    {
      "epoch": 0.06041447137337548,
      "grad_norm": 1.1916424036026,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.3949,
      "step": 43
    },
    {
      "epoch": 0.061819459079733054,
      "grad_norm": 1.3091131448745728,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.5263,
      "step": 44
    },
    {
      "epoch": 0.06322444678609063,
      "grad_norm": 1.147409200668335,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.382,
      "step": 45
    },
    {
      "epoch": 0.06462943449244819,
      "grad_norm": 1.256136178970337,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.2304,
      "step": 46
    },
    {
      "epoch": 0.06603442219880576,
      "grad_norm": 1.1283365488052368,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.1921,
      "step": 47
    },
    {
      "epoch": 0.06743940990516333,
      "grad_norm": 2.8419339656829834,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.2642,
      "step": 48
    },
    {
      "epoch": 0.0688443976115209,
      "grad_norm": 1.434983491897583,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.3072,
      "step": 49
    },
    {
      "epoch": 0.07024938531787847,
      "grad_norm": 4.590909481048584,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.0,
      "step": 50
    },
    {
      "epoch": 0.07024938531787847,
      "eval_loss": 1.5483478307724,
      "eval_runtime": 84.4917,
      "eval_samples_per_second": 14.191,
      "eval_steps_per_second": 3.551,
      "step": 50
    },
    {
      "epoch": 0.07165437302423604,
      "grad_norm": 1.3938902616500854,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.9708,
      "step": 51
    },
    {
      "epoch": 0.0730593607305936,
      "grad_norm": 1.2834646701812744,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.7901,
      "step": 52
    },
    {
      "epoch": 0.07446434843695118,
      "grad_norm": 0.989642858505249,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.7908,
      "step": 53
    },
    {
      "epoch": 0.07586933614330875,
      "grad_norm": 0.8171737790107727,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.6761,
      "step": 54
    },
    {
      "epoch": 0.07727432384966632,
      "grad_norm": 0.829910159111023,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.8174,
      "step": 55
    },
    {
      "epoch": 0.07867931155602388,
      "grad_norm": 0.7820039391517639,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.7537,
      "step": 56
    },
    {
      "epoch": 0.08008429926238145,
      "grad_norm": 0.8432946801185608,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.7614,
      "step": 57
    },
    {
      "epoch": 0.08148928696873903,
      "grad_norm": 0.9533034563064575,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.7421,
      "step": 58
    },
    {
      "epoch": 0.0828942746750966,
      "grad_norm": 0.8141330480575562,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.5224,
      "step": 59
    },
    {
      "epoch": 0.08429926238145416,
      "grad_norm": 0.8789393901824951,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.611,
      "step": 60
    },
    {
      "epoch": 0.08570425008781173,
      "grad_norm": 1.016982913017273,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.5215,
      "step": 61
    },
    {
      "epoch": 0.0871092377941693,
      "grad_norm": 0.8757651448249817,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.5735,
      "step": 62
    },
    {
      "epoch": 0.08851422550052687,
      "grad_norm": 0.8604658246040344,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.5364,
      "step": 63
    },
    {
      "epoch": 0.08991921320688444,
      "grad_norm": 0.760289192199707,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.4356,
      "step": 64
    },
    {
      "epoch": 0.091324200913242,
      "grad_norm": 0.7767530083656311,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.4073,
      "step": 65
    },
    {
      "epoch": 0.09272918861959958,
      "grad_norm": 0.8706574440002441,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.7314,
      "step": 66
    },
    {
      "epoch": 0.09413417632595715,
      "grad_norm": 0.7042254209518433,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.5041,
      "step": 67
    },
    {
      "epoch": 0.09553916403231472,
      "grad_norm": 1.035076379776001,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.7077,
      "step": 68
    },
    {
      "epoch": 0.09694415173867228,
      "grad_norm": 0.7429786920547485,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.3884,
      "step": 69
    },
    {
      "epoch": 0.09834913944502986,
      "grad_norm": 0.8116068243980408,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.55,
      "step": 70
    },
    {
      "epoch": 0.09975412715138743,
      "grad_norm": 0.8268506526947021,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.5021,
      "step": 71
    },
    {
      "epoch": 0.101159114857745,
      "grad_norm": 0.8867709636688232,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.4448,
      "step": 72
    },
    {
      "epoch": 0.10256410256410256,
      "grad_norm": 0.9326731562614441,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.4763,
      "step": 73
    },
    {
      "epoch": 0.10396909027046013,
      "grad_norm": 0.8769791126251221,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.4656,
      "step": 74
    },
    {
      "epoch": 0.1053740779768177,
      "grad_norm": 0.8183702230453491,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.4242,
      "step": 75
    },
    {
      "epoch": 0.10677906568317527,
      "grad_norm": 0.9858356714248657,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.6275,
      "step": 76
    },
    {
      "epoch": 0.10818405338953284,
      "grad_norm": 0.8036893010139465,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.319,
      "step": 77
    },
    {
      "epoch": 0.1095890410958904,
      "grad_norm": 0.8129367232322693,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.2911,
      "step": 78
    },
    {
      "epoch": 0.11099402880224798,
      "grad_norm": 0.7879998683929443,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.3329,
      "step": 79
    },
    {
      "epoch": 0.11239901650860555,
      "grad_norm": 0.8973305821418762,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.3244,
      "step": 80
    },
    {
      "epoch": 0.11380400421496312,
      "grad_norm": 0.8386538028717041,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.2091,
      "step": 81
    },
    {
      "epoch": 0.11520899192132068,
      "grad_norm": 0.9214687347412109,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.3991,
      "step": 82
    },
    {
      "epoch": 0.11661397962767826,
      "grad_norm": 0.927963376045227,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.3907,
      "step": 83
    },
    {
      "epoch": 0.11801896733403583,
      "grad_norm": 0.8660138845443726,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.3108,
      "step": 84
    },
    {
      "epoch": 0.1194239550403934,
      "grad_norm": 0.9522641897201538,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.3538,
      "step": 85
    },
    {
      "epoch": 0.12082894274675096,
      "grad_norm": 0.9129881262779236,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.4466,
      "step": 86
    },
    {
      "epoch": 0.12223393045310854,
      "grad_norm": 0.8134028315544128,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.043,
      "step": 87
    },
    {
      "epoch": 0.12363891815946611,
      "grad_norm": 1.0265110731124878,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.3019,
      "step": 88
    },
    {
      "epoch": 0.12504390586582367,
      "grad_norm": 0.9077882170677185,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.1888,
      "step": 89
    },
    {
      "epoch": 0.12644889357218125,
      "grad_norm": 1.0472909212112427,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.3583,
      "step": 90
    },
    {
      "epoch": 0.1278538812785388,
      "grad_norm": 1.0520256757736206,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.2427,
      "step": 91
    },
    {
      "epoch": 0.12925886898489639,
      "grad_norm": 1.0810383558273315,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.4557,
      "step": 92
    },
    {
      "epoch": 0.13066385669125394,
      "grad_norm": 1.0170336961746216,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.1115,
      "step": 93
    },
    {
      "epoch": 0.13206884439761152,
      "grad_norm": 0.9999778270721436,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.9091,
      "step": 94
    },
    {
      "epoch": 0.1334738321039691,
      "grad_norm": 1.2446062564849854,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.3555,
      "step": 95
    },
    {
      "epoch": 0.13487881981032665,
      "grad_norm": 1.0751291513442993,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.1834,
      "step": 96
    },
    {
      "epoch": 0.13628380751668423,
      "grad_norm": 1.0852233171463013,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.9316,
      "step": 97
    },
    {
      "epoch": 0.1376887952230418,
      "grad_norm": 1.0455330610275269,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.9053,
      "step": 98
    },
    {
      "epoch": 0.13909378292939936,
      "grad_norm": 1.2128297090530396,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.9725,
      "step": 99
    },
    {
      "epoch": 0.14049877063575694,
      "grad_norm": 1.4873278141021729,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.1628,
      "step": 100
    },
    {
      "epoch": 0.14049877063575694,
      "eval_loss": 1.4107543230056763,
      "eval_runtime": 84.4714,
      "eval_samples_per_second": 14.194,
      "eval_steps_per_second": 3.551,
      "step": 100
    },
    {
      "epoch": 0.1419037583421145,
      "grad_norm": 0.8266401886940002,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.8173,
      "step": 101
    },
    {
      "epoch": 0.14330874604847207,
      "grad_norm": 0.7520312666893005,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.7358,
      "step": 102
    },
    {
      "epoch": 0.14471373375482965,
      "grad_norm": 0.7603346705436707,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.695,
      "step": 103
    },
    {
      "epoch": 0.1461187214611872,
      "grad_norm": 0.8081402778625488,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.7236,
      "step": 104
    },
    {
      "epoch": 0.1475237091675448,
      "grad_norm": 0.6830846071243286,
      "learning_rate": 5e-05,
      "loss": 1.6068,
      "step": 105
    },
    {
      "epoch": 0.14892869687390237,
      "grad_norm": 0.7592134475708008,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.6881,
      "step": 106
    },
    {
      "epoch": 0.15033368458025992,
      "grad_norm": 0.7273774743080139,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.6521,
      "step": 107
    },
    {
      "epoch": 0.1517386722866175,
      "grad_norm": 0.7609537243843079,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.6649,
      "step": 108
    },
    {
      "epoch": 0.15314365999297505,
      "grad_norm": 0.6720927357673645,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.4932,
      "step": 109
    },
    {
      "epoch": 0.15454864769933263,
      "grad_norm": 0.65521240234375,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.4279,
      "step": 110
    },
    {
      "epoch": 0.1559536354056902,
      "grad_norm": 0.7774723172187805,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.5391,
      "step": 111
    },
    {
      "epoch": 0.15735862311204776,
      "grad_norm": 0.6253892183303833,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.2865,
      "step": 112
    },
    {
      "epoch": 0.15876361081840534,
      "grad_norm": 0.7566734552383423,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.5871,
      "step": 113
    },
    {
      "epoch": 0.1601685985247629,
      "grad_norm": 0.86098712682724,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.5117,
      "step": 114
    },
    {
      "epoch": 0.16157358623112048,
      "grad_norm": 0.7550438046455383,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.5728,
      "step": 115
    },
    {
      "epoch": 0.16297857393747806,
      "grad_norm": 0.8277958035469055,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.3947,
      "step": 116
    },
    {
      "epoch": 0.1643835616438356,
      "grad_norm": 0.7309897541999817,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.3241,
      "step": 117
    },
    {
      "epoch": 0.1657885493501932,
      "grad_norm": 0.8135260343551636,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.3944,
      "step": 118
    },
    {
      "epoch": 0.16719353705655077,
      "grad_norm": 0.785655677318573,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.4975,
      "step": 119
    },
    {
      "epoch": 0.16859852476290832,
      "grad_norm": 0.8995600342750549,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.3514,
      "step": 120
    },
    {
      "epoch": 0.1700035124692659,
      "grad_norm": 0.7944087386131287,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.5605,
      "step": 121
    },
    {
      "epoch": 0.17140850017562345,
      "grad_norm": 0.8236826658248901,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.6369,
      "step": 122
    },
    {
      "epoch": 0.17281348788198103,
      "grad_norm": 0.8151525855064392,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.2519,
      "step": 123
    },
    {
      "epoch": 0.1742184755883386,
      "grad_norm": 0.7470896244049072,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.4297,
      "step": 124
    },
    {
      "epoch": 0.17562346329469616,
      "grad_norm": 0.8376610279083252,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.3551,
      "step": 125
    },
    {
      "epoch": 0.17702845100105374,
      "grad_norm": 0.8496617674827576,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.2804,
      "step": 126
    },
    {
      "epoch": 0.17843343870741132,
      "grad_norm": 0.7745879292488098,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.142,
      "step": 127
    },
    {
      "epoch": 0.17983842641376888,
      "grad_norm": 0.8647757172584534,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.4648,
      "step": 128
    },
    {
      "epoch": 0.18124341412012646,
      "grad_norm": 0.9334298968315125,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.4795,
      "step": 129
    },
    {
      "epoch": 0.182648401826484,
      "grad_norm": 0.8354178071022034,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.2782,
      "step": 130
    },
    {
      "epoch": 0.1840533895328416,
      "grad_norm": 0.8976739645004272,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.3562,
      "step": 131
    },
    {
      "epoch": 0.18545837723919917,
      "grad_norm": 0.9169337749481201,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.3939,
      "step": 132
    },
    {
      "epoch": 0.18686336494555672,
      "grad_norm": 0.9184233546257019,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.4194,
      "step": 133
    },
    {
      "epoch": 0.1882683526519143,
      "grad_norm": 0.8388327360153198,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.2844,
      "step": 134
    },
    {
      "epoch": 0.18967334035827185,
      "grad_norm": 0.9228142499923706,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.4828,
      "step": 135
    },
    {
      "epoch": 0.19107832806462943,
      "grad_norm": 0.8032878041267395,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.1321,
      "step": 136
    },
    {
      "epoch": 0.192483315770987,
      "grad_norm": 0.9350921511650085,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.2751,
      "step": 137
    },
    {
      "epoch": 0.19388830347734456,
      "grad_norm": 1.0890933275222778,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.267,
      "step": 138
    },
    {
      "epoch": 0.19529329118370214,
      "grad_norm": 1.130176305770874,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.2198,
      "step": 139
    },
    {
      "epoch": 0.19669827889005972,
      "grad_norm": 1.1094470024108887,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.1142,
      "step": 140
    },
    {
      "epoch": 0.19810326659641728,
      "grad_norm": 1.092997670173645,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.3614,
      "step": 141
    },
    {
      "epoch": 0.19950825430277486,
      "grad_norm": 1.2220638990402222,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.2653,
      "step": 142
    },
    {
      "epoch": 0.2009132420091324,
      "grad_norm": 1.0458637475967407,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.1072,
      "step": 143
    },
    {
      "epoch": 0.20231822971549,
      "grad_norm": 1.0568877458572388,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.2813,
      "step": 144
    },
    {
      "epoch": 0.20372321742184757,
      "grad_norm": 1.074480652809143,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.1229,
      "step": 145
    },
    {
      "epoch": 0.20512820512820512,
      "grad_norm": 0.9594824314117432,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.9672,
      "step": 146
    },
    {
      "epoch": 0.2065331928345627,
      "grad_norm": 1.0882999897003174,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.8887,
      "step": 147
    },
    {
      "epoch": 0.20793818054092025,
      "grad_norm": 1.2830413579940796,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.1222,
      "step": 148
    },
    {
      "epoch": 0.20934316824727783,
      "grad_norm": 1.4997934103012085,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.1711,
      "step": 149
    },
    {
      "epoch": 0.2107481559536354,
      "grad_norm": 1.4325093030929565,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.181,
      "step": 150
    },
    {
      "epoch": 0.2107481559536354,
      "eval_loss": 1.3457560539245605,
      "eval_runtime": 84.4618,
      "eval_samples_per_second": 14.196,
      "eval_steps_per_second": 3.552,
      "step": 150
    },
    {
      "epoch": 0.21215314365999297,
      "grad_norm": 0.5772650837898254,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.552,
      "step": 151
    },
    {
      "epoch": 0.21355813136635055,
      "grad_norm": 0.6446923017501831,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.6053,
      "step": 152
    },
    {
      "epoch": 0.21496311907270813,
      "grad_norm": 0.6607129573822021,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.6874,
      "step": 153
    },
    {
      "epoch": 0.21636810677906568,
      "grad_norm": 0.648309051990509,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.5613,
      "step": 154
    },
    {
      "epoch": 0.21777309448542326,
      "grad_norm": 0.6242180466651917,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.4462,
      "step": 155
    },
    {
      "epoch": 0.2191780821917808,
      "grad_norm": 0.7166935801506042,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.6288,
      "step": 156
    },
    {
      "epoch": 0.2205830698981384,
      "grad_norm": 0.7276003956794739,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.4788,
      "step": 157
    },
    {
      "epoch": 0.22198805760449597,
      "grad_norm": 0.7129059433937073,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.526,
      "step": 158
    },
    {
      "epoch": 0.22339304531085352,
      "grad_norm": 0.6585018038749695,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.3963,
      "step": 159
    },
    {
      "epoch": 0.2247980330172111,
      "grad_norm": 0.773478090763092,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.4844,
      "step": 160
    },
    {
      "epoch": 0.22620302072356868,
      "grad_norm": 0.666933536529541,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.4999,
      "step": 161
    },
    {
      "epoch": 0.22760800842992623,
      "grad_norm": 0.7217544913291931,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.5581,
      "step": 162
    },
    {
      "epoch": 0.22901299613628381,
      "grad_norm": 0.6706897616386414,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.3775,
      "step": 163
    },
    {
      "epoch": 0.23041798384264137,
      "grad_norm": 0.7253771424293518,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.4909,
      "step": 164
    },
    {
      "epoch": 0.23182297154899895,
      "grad_norm": 0.7991068959236145,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.4129,
      "step": 165
    },
    {
      "epoch": 0.23322795925535653,
      "grad_norm": 0.712374210357666,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.3779,
      "step": 166
    },
    {
      "epoch": 0.23463294696171408,
      "grad_norm": 0.7540154457092285,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.5628,
      "step": 167
    },
    {
      "epoch": 0.23603793466807166,
      "grad_norm": 0.7488724589347839,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.5358,
      "step": 168
    },
    {
      "epoch": 0.2374429223744292,
      "grad_norm": 0.7723095417022705,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.4974,
      "step": 169
    },
    {
      "epoch": 0.2388479100807868,
      "grad_norm": 0.7994570136070251,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.6147,
      "step": 170
    },
    {
      "epoch": 0.24025289778714437,
      "grad_norm": 0.8343621492385864,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.5065,
      "step": 171
    },
    {
      "epoch": 0.24165788549350192,
      "grad_norm": 0.724969208240509,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.3455,
      "step": 172
    },
    {
      "epoch": 0.2430628731998595,
      "grad_norm": 0.7027336359024048,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.258,
      "step": 173
    },
    {
      "epoch": 0.24446786090621708,
      "grad_norm": 0.8356617093086243,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.4175,
      "step": 174
    },
    {
      "epoch": 0.24587284861257463,
      "grad_norm": 0.8144857883453369,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.4302,
      "step": 175
    },
    {
      "epoch": 0.24727783631893221,
      "grad_norm": 0.7722607254981995,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.156,
      "step": 176
    },
    {
      "epoch": 0.24868282402528977,
      "grad_norm": 0.7871822118759155,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.2404,
      "step": 177
    },
    {
      "epoch": 0.25008781173164735,
      "grad_norm": 0.9211126565933228,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.3861,
      "step": 178
    },
    {
      "epoch": 0.2514927994380049,
      "grad_norm": 0.9086484313011169,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.4015,
      "step": 179
    },
    {
      "epoch": 0.2528977871443625,
      "grad_norm": 0.8126121163368225,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.354,
      "step": 180
    },
    {
      "epoch": 0.25430277485072006,
      "grad_norm": 0.8340513110160828,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.2059,
      "step": 181
    },
    {
      "epoch": 0.2557077625570776,
      "grad_norm": 0.9006524085998535,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.1969,
      "step": 182
    },
    {
      "epoch": 0.2571127502634352,
      "grad_norm": 0.9551442265510559,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.3814,
      "step": 183
    },
    {
      "epoch": 0.25851773796979277,
      "grad_norm": 0.8320140242576599,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.1613,
      "step": 184
    },
    {
      "epoch": 0.2599227256761503,
      "grad_norm": 0.9696279764175415,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.3156,
      "step": 185
    },
    {
      "epoch": 0.2613277133825079,
      "grad_norm": 0.8337900042533875,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.2951,
      "step": 186
    },
    {
      "epoch": 0.2627327010888655,
      "grad_norm": 1.931960940361023,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.3444,
      "step": 187
    },
    {
      "epoch": 0.26413768879522304,
      "grad_norm": 0.9092104434967041,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.1115,
      "step": 188
    },
    {
      "epoch": 0.2655426765015806,
      "grad_norm": 0.9779451489448547,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.2397,
      "step": 189
    },
    {
      "epoch": 0.2669476642079382,
      "grad_norm": 0.942230224609375,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.0104,
      "step": 190
    },
    {
      "epoch": 0.26835265191429575,
      "grad_norm": 1.0627790689468384,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.1445,
      "step": 191
    },
    {
      "epoch": 0.2697576396206533,
      "grad_norm": 0.9928768873214722,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.0991,
      "step": 192
    },
    {
      "epoch": 0.2711626273270109,
      "grad_norm": 1.006177306175232,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.1929,
      "step": 193
    },
    {
      "epoch": 0.27256761503336846,
      "grad_norm": 1.1792783737182617,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.9793,
      "step": 194
    },
    {
      "epoch": 0.273972602739726,
      "grad_norm": 1.1026349067687988,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.1345,
      "step": 195
    },
    {
      "epoch": 0.2753775904460836,
      "grad_norm": 1.2108447551727295,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.3772,
      "step": 196
    },
    {
      "epoch": 0.27678257815244117,
      "grad_norm": 1.1245533227920532,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.9868,
      "step": 197
    },
    {
      "epoch": 0.2781875658587987,
      "grad_norm": 1.133299469947815,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.1552,
      "step": 198
    },
    {
      "epoch": 0.27959255356515633,
      "grad_norm": 1.241783618927002,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.9261,
      "step": 199
    },
    {
      "epoch": 0.2809975412715139,
      "grad_norm": 1.8515037298202515,
      "learning_rate": 0.0,
      "loss": 1.2411,
      "step": 200
    },
    {
      "epoch": 0.2809975412715139,
      "eval_loss": 1.339722990989685,
      "eval_runtime": 84.5468,
      "eval_samples_per_second": 14.181,
      "eval_steps_per_second": 3.548,
      "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": 2.661403865382912e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}