{
  "best_metric": 1.9404832124710083,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.03937589211005562,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0001968794605502781,
      "grad_norm": 4.722285270690918,
      "learning_rate": 1e-05,
      "loss": 4.3603,
      "step": 1
    },
    {
      "epoch": 0.0001968794605502781,
      "eval_loss": 5.185981750488281,
      "eval_runtime": 991.8236,
      "eval_samples_per_second": 8.626,
      "eval_steps_per_second": 2.157,
      "step": 1
    },
    {
      "epoch": 0.0003937589211005562,
      "grad_norm": 5.517726421356201,
      "learning_rate": 2e-05,
      "loss": 4.4805,
      "step": 2
    },
    {
      "epoch": 0.0005906383816508343,
      "grad_norm": 5.287935256958008,
      "learning_rate": 3e-05,
      "loss": 4.4314,
      "step": 3
    },
    {
      "epoch": 0.0007875178422011124,
      "grad_norm": 4.986377716064453,
      "learning_rate": 4e-05,
      "loss": 4.3941,
      "step": 4
    },
    {
      "epoch": 0.0009843973027513905,
      "grad_norm": 5.3017072677612305,
      "learning_rate": 5e-05,
      "loss": 4.2773,
      "step": 5
    },
    {
      "epoch": 0.0011812767633016686,
      "grad_norm": 5.697668075561523,
      "learning_rate": 6e-05,
      "loss": 3.9134,
      "step": 6
    },
    {
      "epoch": 0.0013781562238519467,
      "grad_norm": 5.562041759490967,
      "learning_rate": 7e-05,
      "loss": 3.9943,
      "step": 7
    },
    {
      "epoch": 0.0015750356844022248,
      "grad_norm": 4.46408748626709,
      "learning_rate": 8e-05,
      "loss": 3.3548,
      "step": 8
    },
    {
      "epoch": 0.001771915144952503,
      "grad_norm": 5.244144439697266,
      "learning_rate": 9e-05,
      "loss": 3.0075,
      "step": 9
    },
    {
      "epoch": 0.001968794605502781,
      "grad_norm": 3.6670258045196533,
      "learning_rate": 0.0001,
      "loss": 2.6177,
      "step": 10
    },
    {
      "epoch": 0.002165674066053059,
      "grad_norm": 4.358741760253906,
      "learning_rate": 9.999316524962345e-05,
      "loss": 2.4993,
      "step": 11
    },
    {
      "epoch": 0.002362553526603337,
      "grad_norm": 2.7043681144714355,
      "learning_rate": 9.997266286704631e-05,
      "loss": 2.5074,
      "step": 12
    },
    {
      "epoch": 0.0025594329871536153,
      "grad_norm": 4.361660480499268,
      "learning_rate": 9.993849845741524e-05,
      "loss": 2.5137,
      "step": 13
    },
    {
      "epoch": 0.0027563124477038934,
      "grad_norm": 3.5170817375183105,
      "learning_rate": 9.989068136093873e-05,
      "loss": 2.3223,
      "step": 14
    },
    {
      "epoch": 0.0029531919082541715,
      "grad_norm": 3.0592331886291504,
      "learning_rate": 9.98292246503335e-05,
      "loss": 2.2921,
      "step": 15
    },
    {
      "epoch": 0.0031500713688044496,
      "grad_norm": 2.7591488361358643,
      "learning_rate": 9.975414512725057e-05,
      "loss": 2.2832,
      "step": 16
    },
    {
      "epoch": 0.0033469508293547277,
      "grad_norm": 2.473106861114502,
      "learning_rate": 9.966546331768191e-05,
      "loss": 2.1306,
      "step": 17
    },
    {
      "epoch": 0.003543830289905006,
      "grad_norm": 2.901942491531372,
      "learning_rate": 9.956320346634876e-05,
      "loss": 2.3132,
      "step": 18
    },
    {
      "epoch": 0.003740709750455284,
      "grad_norm": 2.7212612628936768,
      "learning_rate": 9.944739353007344e-05,
      "loss": 2.1009,
      "step": 19
    },
    {
      "epoch": 0.003937589211005562,
      "grad_norm": 2.115468740463257,
      "learning_rate": 9.931806517013612e-05,
      "loss": 2.1042,
      "step": 20
    },
    {
      "epoch": 0.00413446867155584,
      "grad_norm": 2.501617193222046,
      "learning_rate": 9.917525374361912e-05,
      "loss": 2.0805,
      "step": 21
    },
    {
      "epoch": 0.004331348132106118,
      "grad_norm": 2.2306952476501465,
      "learning_rate": 9.901899829374047e-05,
      "loss": 2.0436,
      "step": 22
    },
    {
      "epoch": 0.004528227592656396,
      "grad_norm": 2.614072561264038,
      "learning_rate": 9.884934153917997e-05,
      "loss": 2.054,
      "step": 23
    },
    {
      "epoch": 0.004725107053206674,
      "grad_norm": 2.2532169818878174,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.9873,
      "step": 24
    },
    {
      "epoch": 0.0049219865137569525,
      "grad_norm": 3.412367343902588,
      "learning_rate": 9.847001329696653e-05,
      "loss": 2.1435,
      "step": 25
    },
    {
      "epoch": 0.005118865974307231,
      "grad_norm": 2.6151926517486572,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.9335,
      "step": 26
    },
    {
      "epoch": 0.005315745434857509,
      "grad_norm": 2.999908685684204,
      "learning_rate": 9.803768380684242e-05,
      "loss": 2.1296,
      "step": 27
    },
    {
      "epoch": 0.005512624895407787,
      "grad_norm": 2.8248817920684814,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.8962,
      "step": 28
    },
    {
      "epoch": 0.005709504355958065,
      "grad_norm": 2.4659695625305176,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.9723,
      "step": 29
    },
    {
      "epoch": 0.005906383816508343,
      "grad_norm": 2.144352436065674,
      "learning_rate": 9.729086208503174e-05,
      "loss": 2.1925,
      "step": 30
    },
    {
      "epoch": 0.006103263277058621,
      "grad_norm": 2.27182936668396,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.0143,
      "step": 31
    },
    {
      "epoch": 0.006300142737608899,
      "grad_norm": 1.9760721921920776,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.9399,
      "step": 32
    },
    {
      "epoch": 0.006497022198159177,
      "grad_norm": 1.9209957122802734,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.0524,
      "step": 33
    },
    {
      "epoch": 0.006693901658709455,
      "grad_norm": 2.1844968795776367,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.1256,
      "step": 34
    },
    {
      "epoch": 0.0068907811192597335,
      "grad_norm": 2.4915807247161865,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.042,
      "step": 35
    },
    {
      "epoch": 0.007087660579810012,
      "grad_norm": 2.3629589080810547,
      "learning_rate": 9.545032675245813e-05,
      "loss": 2.1509,
      "step": 36
    },
    {
      "epoch": 0.00728454004036029,
      "grad_norm": 3.1022274494171143,
      "learning_rate": 9.509956150664796e-05,
      "loss": 2.0537,
      "step": 37
    },
    {
      "epoch": 0.007481419500910568,
      "grad_norm": 2.714031934738159,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.7935,
      "step": 38
    },
    {
      "epoch": 0.007678298961460846,
      "grad_norm": 2.287045478820801,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.8456,
      "step": 39
    },
    {
      "epoch": 0.007875178422011124,
      "grad_norm": 2.3300211429595947,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.9945,
      "step": 40
    },
    {
      "epoch": 0.008072057882561401,
      "grad_norm": 2.420959949493408,
      "learning_rate": 9.357421218136386e-05,
      "loss": 2.1356,
      "step": 41
    },
    {
      "epoch": 0.00826893734311168,
      "grad_norm": 3.8930702209472656,
      "learning_rate": 9.316282404787871e-05,
      "loss": 2.0933,
      "step": 42
    },
    {
      "epoch": 0.008465816803661957,
      "grad_norm": 2.3729233741760254,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.9622,
      "step": 43
    },
    {
      "epoch": 0.008662696264212236,
      "grad_norm": 2.014636278152466,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.6516,
      "step": 44
    },
    {
      "epoch": 0.008859575724762514,
      "grad_norm": 3.216425895690918,
      "learning_rate": 9.185832391312644e-05,
      "loss": 2.0221,
      "step": 45
    },
    {
      "epoch": 0.009056455185312793,
      "grad_norm": 2.529404401779175,
      "learning_rate": 9.140044155740101e-05,
      "loss": 2.0313,
      "step": 46
    },
    {
      "epoch": 0.00925333464586307,
      "grad_norm": 2.8222763538360596,
      "learning_rate": 9.093124073433463e-05,
      "loss": 2.0187,
      "step": 47
    },
    {
      "epoch": 0.009450214106413349,
      "grad_norm": 5.74941349029541,
      "learning_rate": 9.045084971874738e-05,
      "loss": 2.0414,
      "step": 48
    },
    {
      "epoch": 0.009647093566963626,
      "grad_norm": 3.829970121383667,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.7407,
      "step": 49
    },
    {
      "epoch": 0.009843973027513905,
      "grad_norm": 6.1021728515625,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.6955,
      "step": 50
    },
    {
      "epoch": 0.009843973027513905,
      "eval_loss": 2.121528387069702,
      "eval_runtime": 997.4855,
      "eval_samples_per_second": 8.577,
      "eval_steps_per_second": 2.144,
      "step": 50
    },
    {
      "epoch": 0.010040852488064182,
      "grad_norm": 2.033238649368286,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.9538,
      "step": 51
    },
    {
      "epoch": 0.010237731948614461,
      "grad_norm": 2.1589324474334717,
      "learning_rate": 8.842005554284296e-05,
      "loss": 2.077,
      "step": 52
    },
    {
      "epoch": 0.010434611409164738,
      "grad_norm": 2.2278130054473877,
      "learning_rate": 8.788574348801675e-05,
      "loss": 2.0108,
      "step": 53
    },
    {
      "epoch": 0.010631490869715017,
      "grad_norm": 2.2761881351470947,
      "learning_rate": 8.73410738492077e-05,
      "loss": 2.146,
      "step": 54
    },
    {
      "epoch": 0.010828370330265295,
      "grad_norm": 2.1180636882781982,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.9721,
      "step": 55
    },
    {
      "epoch": 0.011025249790815574,
      "grad_norm": 1.9664140939712524,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.9298,
      "step": 56
    },
    {
      "epoch": 0.01122212925136585,
      "grad_norm": 2.442275047302246,
      "learning_rate": 8.564642241456986e-05,
      "loss": 2.3,
      "step": 57
    },
    {
      "epoch": 0.01141900871191613,
      "grad_norm": 1.992834448814392,
      "learning_rate": 8.506183921362443e-05,
      "loss": 2.108,
      "step": 58
    },
    {
      "epoch": 0.011615888172466407,
      "grad_norm": 2.0596256256103516,
      "learning_rate": 8.44676704559283e-05,
      "loss": 2.132,
      "step": 59
    },
    {
      "epoch": 0.011812767633016686,
      "grad_norm": 1.9328343868255615,
      "learning_rate": 8.386407858128706e-05,
      "loss": 2.0588,
      "step": 60
    },
    {
      "epoch": 0.012009647093566963,
      "grad_norm": 1.9313613176345825,
      "learning_rate": 8.32512286056924e-05,
      "loss": 2.1842,
      "step": 61
    },
    {
      "epoch": 0.012206526554117242,
      "grad_norm": 1.897388219833374,
      "learning_rate": 8.262928807620843e-05,
      "loss": 2.1686,
      "step": 62
    },
    {
      "epoch": 0.01240340601466752,
      "grad_norm": 1.9466983079910278,
      "learning_rate": 8.199842702516583e-05,
      "loss": 2.1302,
      "step": 63
    },
    {
      "epoch": 0.012600285475217798,
      "grad_norm": 1.7845702171325684,
      "learning_rate": 8.135881792367686e-05,
      "loss": 2.0203,
      "step": 64
    },
    {
      "epoch": 0.012797164935768076,
      "grad_norm": 1.8993163108825684,
      "learning_rate": 8.07106356344834e-05,
      "loss": 2.1231,
      "step": 65
    },
    {
      "epoch": 0.012994044396318355,
      "grad_norm": 1.7987322807312012,
      "learning_rate": 8.005405736415126e-05,
      "loss": 2.0143,
      "step": 66
    },
    {
      "epoch": 0.013190923856868632,
      "grad_norm": 1.692613124847412,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.892,
      "step": 67
    },
    {
      "epoch": 0.01338780331741891,
      "grad_norm": 1.9041842222213745,
      "learning_rate": 7.871643313414718e-05,
      "loss": 2.1199,
      "step": 68
    },
    {
      "epoch": 0.013584682777969188,
      "grad_norm": 2.28175687789917,
      "learning_rate": 7.803575286758364e-05,
      "loss": 2.1138,
      "step": 69
    },
    {
      "epoch": 0.013781562238519467,
      "grad_norm": 1.9009627103805542,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.9233,
      "step": 70
    },
    {
      "epoch": 0.013978441699069744,
      "grad_norm": 1.9802814722061157,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.9362,
      "step": 71
    },
    {
      "epoch": 0.014175321159620023,
      "grad_norm": 1.9521552324295044,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.9364,
      "step": 72
    },
    {
      "epoch": 0.0143722006201703,
      "grad_norm": 1.8512924909591675,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.9446,
      "step": 73
    },
    {
      "epoch": 0.01456908008072058,
      "grad_norm": 1.8273296356201172,
      "learning_rate": 7.452117519152542e-05,
      "loss": 2.0599,
      "step": 74
    },
    {
      "epoch": 0.014765959541270857,
      "grad_norm": 1.9407672882080078,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.9647,
      "step": 75
    },
    {
      "epoch": 0.014962839001821136,
      "grad_norm": 1.73810613155365,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.7585,
      "step": 76
    },
    {
      "epoch": 0.015159718462371413,
      "grad_norm": 2.0903658866882324,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.9544,
      "step": 77
    },
    {
      "epoch": 0.015356597922921692,
      "grad_norm": 1.9962186813354492,
      "learning_rate": 7.158771761692464e-05,
      "loss": 2.043,
      "step": 78
    },
    {
      "epoch": 0.015553477383471969,
      "grad_norm": 1.6904006004333496,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.9195,
      "step": 79
    },
    {
      "epoch": 0.015750356844022248,
      "grad_norm": 2.076584815979004,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.8485,
      "step": 80
    },
    {
      "epoch": 0.015947236304572527,
      "grad_norm": 2.0259509086608887,
      "learning_rate": 6.932495846462261e-05,
      "loss": 2.2198,
      "step": 81
    },
    {
      "epoch": 0.016144115765122802,
      "grad_norm": 1.8992706537246704,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.8143,
      "step": 82
    },
    {
      "epoch": 0.01634099522567308,
      "grad_norm": 2.1401584148406982,
      "learning_rate": 6.778969234612584e-05,
      "loss": 2.1014,
      "step": 83
    },
    {
      "epoch": 0.01653787468622336,
      "grad_norm": 1.8982717990875244,
      "learning_rate": 6.701465872208216e-05,
      "loss": 2.0747,
      "step": 84
    },
    {
      "epoch": 0.01673475414677364,
      "grad_norm": 1.8973262310028076,
      "learning_rate": 6.623497346023418e-05,
      "loss": 2.0424,
      "step": 85
    },
    {
      "epoch": 0.016931633607323915,
      "grad_norm": 1.8528293371200562,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.9253,
      "step": 86
    },
    {
      "epoch": 0.017128513067874194,
      "grad_norm": 1.9229885339736938,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.9506,
      "step": 87
    },
    {
      "epoch": 0.017325392528424473,
      "grad_norm": 1.9180034399032593,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.8687,
      "step": 88
    },
    {
      "epoch": 0.017522271988974752,
      "grad_norm": 1.7185838222503662,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.8805,
      "step": 89
    },
    {
      "epoch": 0.017719151449525027,
      "grad_norm": 1.8684093952178955,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.991,
      "step": 90
    },
    {
      "epoch": 0.017916030910075306,
      "grad_norm": 1.9489952325820923,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.9663,
      "step": 91
    },
    {
      "epoch": 0.018112910370625585,
      "grad_norm": 2.1701080799102783,
      "learning_rate": 6.066498153718735e-05,
      "loss": 2.1342,
      "step": 92
    },
    {
      "epoch": 0.018309789831175864,
      "grad_norm": 2.0452468395233154,
      "learning_rate": 5.985585137257401e-05,
      "loss": 2.0369,
      "step": 93
    },
    {
      "epoch": 0.01850666929172614,
      "grad_norm": 2.1710257530212402,
      "learning_rate": 5.90440267166055e-05,
      "loss": 2.0067,
      "step": 94
    },
    {
      "epoch": 0.01870354875227642,
      "grad_norm": 2.162400007247925,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 2.0096,
      "step": 95
    },
    {
      "epoch": 0.018900428212826698,
      "grad_norm": 2.511369466781616,
      "learning_rate": 5.74131823855921e-05,
      "loss": 2.1659,
      "step": 96
    },
    {
      "epoch": 0.019097307673376977,
      "grad_norm": 2.212477445602417,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 2.0527,
      "step": 97
    },
    {
      "epoch": 0.019294187133927252,
      "grad_norm": 1.8975532054901123,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.7164,
      "step": 98
    },
    {
      "epoch": 0.01949106659447753,
      "grad_norm": 2.514016628265381,
      "learning_rate": 5.495227651252315e-05,
      "loss": 2.0455,
      "step": 99
    },
    {
      "epoch": 0.01968794605502781,
      "grad_norm": 2.2222836017608643,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.7905,
      "step": 100
    },
    {
      "epoch": 0.01968794605502781,
      "eval_loss": 2.2839598655700684,
      "eval_runtime": 994.4579,
      "eval_samples_per_second": 8.603,
      "eval_steps_per_second": 2.151,
      "step": 100
    },
    {
      "epoch": 0.01988482551557809,
      "grad_norm": 4.464885711669922,
      "learning_rate": 5.330452921628497e-05,
      "loss": 2.2882,
      "step": 101
    },
    {
      "epoch": 0.020081704976128364,
      "grad_norm": 4.608921051025391,
      "learning_rate": 5.247918773366112e-05,
      "loss": 2.3429,
      "step": 102
    },
    {
      "epoch": 0.020278584436678643,
      "grad_norm": 3.279172658920288,
      "learning_rate": 5.165316846586541e-05,
      "loss": 2.3008,
      "step": 103
    },
    {
      "epoch": 0.020475463897228922,
      "grad_norm": 2.4048798084259033,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.9456,
      "step": 104
    },
    {
      "epoch": 0.0206723433577792,
      "grad_norm": 1.8811132907867432,
      "learning_rate": 5e-05,
      "loss": 1.9413,
      "step": 105
    },
    {
      "epoch": 0.020869222818329477,
      "grad_norm": 1.7084625959396362,
      "learning_rate": 4.917330276168208e-05,
      "loss": 2.1956,
      "step": 106
    },
    {
      "epoch": 0.021066102278879756,
      "grad_norm": 1.6710535287857056,
      "learning_rate": 4.834683153413459e-05,
      "loss": 2.1052,
      "step": 107
    },
    {
      "epoch": 0.021262981739430035,
      "grad_norm": 1.763252854347229,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.9297,
      "step": 108
    },
    {
      "epoch": 0.021459861199980314,
      "grad_norm": 1.7069237232208252,
      "learning_rate": 4.669547078371504e-05,
      "loss": 2.0104,
      "step": 109
    },
    {
      "epoch": 0.02165674066053059,
      "grad_norm": 2.005214214324951,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 2.1658,
      "step": 110
    },
    {
      "epoch": 0.021853620121080868,
      "grad_norm": 2.3560802936553955,
      "learning_rate": 4.504772348747687e-05,
      "loss": 2.2162,
      "step": 111
    },
    {
      "epoch": 0.022050499581631147,
      "grad_norm": 1.742017388343811,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.8804,
      "step": 112
    },
    {
      "epoch": 0.022247379042181423,
      "grad_norm": 1.8132730722427368,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 2.0396,
      "step": 113
    },
    {
      "epoch": 0.0224442585027317,
      "grad_norm": 1.8238927125930786,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 2.0578,
      "step": 114
    },
    {
      "epoch": 0.02264113796328198,
      "grad_norm": 1.8106420040130615,
      "learning_rate": 4.17702704859633e-05,
      "loss": 2.0014,
      "step": 115
    },
    {
      "epoch": 0.02283801742383226,
      "grad_norm": 1.6731736660003662,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.8581,
      "step": 116
    },
    {
      "epoch": 0.023034896884382535,
      "grad_norm": 1.6497639417648315,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.9013,
      "step": 117
    },
    {
      "epoch": 0.023231776344932814,
      "grad_norm": 1.7323822975158691,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.7526,
      "step": 118
    },
    {
      "epoch": 0.023428655805483093,
      "grad_norm": 1.758415699005127,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.9007,
      "step": 119
    },
    {
      "epoch": 0.023625535266033372,
      "grad_norm": 1.8373111486434937,
      "learning_rate": 3.772572564296005e-05,
      "loss": 2.1219,
      "step": 120
    },
    {
      "epoch": 0.023822414726583648,
      "grad_norm": 1.7346097230911255,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.9534,
      "step": 121
    },
    {
      "epoch": 0.024019294187133926,
      "grad_norm": 1.764580488204956,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.9987,
      "step": 122
    },
    {
      "epoch": 0.024216173647684205,
      "grad_norm": 1.750520944595337,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.9747,
      "step": 123
    },
    {
      "epoch": 0.024413053108234484,
      "grad_norm": 1.7164589166641235,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.9209,
      "step": 124
    },
    {
      "epoch": 0.02460993256878476,
      "grad_norm": 1.7653440237045288,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.8264,
      "step": 125
    },
    {
      "epoch": 0.02480681202933504,
      "grad_norm": 1.8179305791854858,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.8391,
      "step": 126
    },
    {
      "epoch": 0.025003691489885318,
      "grad_norm": 1.7005746364593506,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.9018,
      "step": 127
    },
    {
      "epoch": 0.025200570950435597,
      "grad_norm": 1.7188222408294678,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.8897,
      "step": 128
    },
    {
      "epoch": 0.025397450410985872,
      "grad_norm": 1.712591290473938,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.7492,
      "step": 129
    },
    {
      "epoch": 0.02559432987153615,
      "grad_norm": 1.9163239002227783,
      "learning_rate": 2.991522876735154e-05,
      "loss": 2.0165,
      "step": 130
    },
    {
      "epoch": 0.02579120933208643,
      "grad_norm": 1.9214532375335693,
      "learning_rate": 2.916090697523549e-05,
      "loss": 2.0788,
      "step": 131
    },
    {
      "epoch": 0.02598808879263671,
      "grad_norm": 1.896920919418335,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.9214,
      "step": 132
    },
    {
      "epoch": 0.026184968253186985,
      "grad_norm": 1.6200294494628906,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.744,
      "step": 133
    },
    {
      "epoch": 0.026381847713737264,
      "grad_norm": 1.626323938369751,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.8176,
      "step": 134
    },
    {
      "epoch": 0.026578727174287543,
      "grad_norm": 1.9877479076385498,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 2.0644,
      "step": 135
    },
    {
      "epoch": 0.02677560663483782,
      "grad_norm": 1.8600821495056152,
      "learning_rate": 2.547882480847461e-05,
      "loss": 2.003,
      "step": 136
    },
    {
      "epoch": 0.026972486095388097,
      "grad_norm": 1.773910641670227,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.9803,
      "step": 137
    },
    {
      "epoch": 0.027169365555938376,
      "grad_norm": 1.6655477285385132,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.7294,
      "step": 138
    },
    {
      "epoch": 0.027366245016488655,
      "grad_norm": 1.8801628351211548,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.8536,
      "step": 139
    },
    {
      "epoch": 0.027563124477038934,
      "grad_norm": 1.880556344985962,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.924,
      "step": 140
    },
    {
      "epoch": 0.02776000393758921,
      "grad_norm": 1.9137473106384277,
      "learning_rate": 2.196424713241637e-05,
      "loss": 2.0049,
      "step": 141
    },
    {
      "epoch": 0.02795688339813949,
      "grad_norm": 1.8949143886566162,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.8792,
      "step": 142
    },
    {
      "epoch": 0.028153762858689767,
      "grad_norm": 1.9418630599975586,
      "learning_rate": 2.061073738537635e-05,
      "loss": 2.05,
      "step": 143
    },
    {
      "epoch": 0.028350642319240046,
      "grad_norm": 2.0430312156677246,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 2.0717,
      "step": 144
    },
    {
      "epoch": 0.028547521779790322,
      "grad_norm": 1.998457908630371,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.9202,
      "step": 145
    },
    {
      "epoch": 0.0287444012403406,
      "grad_norm": 2.010180950164795,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.979,
      "step": 146
    },
    {
      "epoch": 0.02894128070089088,
      "grad_norm": 2.290452241897583,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.8416,
      "step": 147
    },
    {
      "epoch": 0.02913816016144116,
      "grad_norm": 2.139098644256592,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 2.0033,
      "step": 148
    },
    {
      "epoch": 0.029335039621991434,
      "grad_norm": 2.1091532707214355,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.7023,
      "step": 149
    },
    {
      "epoch": 0.029531919082541713,
      "grad_norm": 2.765894651412964,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.7839,
      "step": 150
    },
    {
      "epoch": 0.029531919082541713,
      "eval_loss": 1.9811595678329468,
      "eval_runtime": 995.7334,
      "eval_samples_per_second": 8.592,
      "eval_steps_per_second": 2.148,
      "step": 150
    },
    {
      "epoch": 0.029728798543091992,
      "grad_norm": 2.3368406295776367,
      "learning_rate": 1.553232954407171e-05,
      "loss": 2.2091,
      "step": 151
    },
    {
      "epoch": 0.02992567800364227,
      "grad_norm": 2.4661142826080322,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 2.1848,
      "step": 152
    },
    {
      "epoch": 0.030122557464192547,
      "grad_norm": 2.4162588119506836,
      "learning_rate": 1.435357758543015e-05,
      "loss": 2.2599,
      "step": 153
    },
    {
      "epoch": 0.030319436924742826,
      "grad_norm": 2.4898364543914795,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 2.259,
      "step": 154
    },
    {
      "epoch": 0.030516316385293105,
      "grad_norm": 2.370333194732666,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 2.3746,
      "step": 155
    },
    {
      "epoch": 0.030713195845843384,
      "grad_norm": 2.2642598152160645,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 2.0568,
      "step": 156
    },
    {
      "epoch": 0.03091007530639366,
      "grad_norm": 2.4862935543060303,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 2.1699,
      "step": 157
    },
    {
      "epoch": 0.031106954766943938,
      "grad_norm": 1.9254343509674072,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.9791,
      "step": 158
    },
    {
      "epoch": 0.031303834227494214,
      "grad_norm": 2.4063682556152344,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 2.0675,
      "step": 159
    },
    {
      "epoch": 0.031500713688044496,
      "grad_norm": 2.0526909828186035,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 2.0096,
      "step": 160
    },
    {
      "epoch": 0.03169759314859477,
      "grad_norm": 2.0006186962127686,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 2.0546,
      "step": 161
    },
    {
      "epoch": 0.031894472609145054,
      "grad_norm": 1.79495370388031,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.8905,
      "step": 162
    },
    {
      "epoch": 0.03209135206969533,
      "grad_norm": 1.9229059219360352,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.8656,
      "step": 163
    },
    {
      "epoch": 0.032288231530245605,
      "grad_norm": 1.7289212942123413,
      "learning_rate": 8.599558442598998e-06,
      "loss": 2.082,
      "step": 164
    },
    {
      "epoch": 0.03248511099079589,
      "grad_norm": 1.7867295742034912,
      "learning_rate": 8.141676086873572e-06,
      "loss": 2.0735,
      "step": 165
    },
    {
      "epoch": 0.03268199045134616,
      "grad_norm": 1.7193914651870728,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.9228,
      "step": 166
    },
    {
      "epoch": 0.03287886991189644,
      "grad_norm": 1.7749696969985962,
      "learning_rate": 7.260364370723044e-06,
      "loss": 2.1935,
      "step": 167
    },
    {
      "epoch": 0.03307574937244672,
      "grad_norm": 1.6653289794921875,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.947,
      "step": 168
    },
    {
      "epoch": 0.033272628832996996,
      "grad_norm": 1.7008565664291382,
      "learning_rate": 6.425787818636131e-06,
      "loss": 2.0406,
      "step": 169
    },
    {
      "epoch": 0.03346950829354728,
      "grad_norm": 1.6552019119262695,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.882,
      "step": 170
    },
    {
      "epoch": 0.033666387754097554,
      "grad_norm": 1.4661345481872559,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.6179,
      "step": 171
    },
    {
      "epoch": 0.03386326721464783,
      "grad_norm": 1.7197890281677246,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.9525,
      "step": 172
    },
    {
      "epoch": 0.03406014667519811,
      "grad_norm": 1.7171134948730469,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.8843,
      "step": 173
    },
    {
      "epoch": 0.03425702613574839,
      "grad_norm": 1.7397480010986328,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.8908,
      "step": 174
    },
    {
      "epoch": 0.03445390559629866,
      "grad_norm": 1.6638896465301514,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.794,
      "step": 175
    },
    {
      "epoch": 0.034650785056848946,
      "grad_norm": 1.7965015172958374,
      "learning_rate": 3.885512251130763e-06,
      "loss": 2.048,
      "step": 176
    },
    {
      "epoch": 0.03484766451739922,
      "grad_norm": 1.6608469486236572,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.9697,
      "step": 177
    },
    {
      "epoch": 0.035044543977949504,
      "grad_norm": 1.6930960416793823,
      "learning_rate": 3.271776770026963e-06,
      "loss": 2.0323,
      "step": 178
    },
    {
      "epoch": 0.03524142343849978,
      "grad_norm": 1.6391199827194214,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.8504,
      "step": 179
    },
    {
      "epoch": 0.035438302899050055,
      "grad_norm": 1.7861101627349854,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.8967,
      "step": 180
    },
    {
      "epoch": 0.03563518235960034,
      "grad_norm": 1.645735740661621,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.773,
      "step": 181
    },
    {
      "epoch": 0.03583206182015061,
      "grad_norm": 1.6818723678588867,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 2.1578,
      "step": 182
    },
    {
      "epoch": 0.03602894128070089,
      "grad_norm": 1.6487213373184204,
      "learning_rate": 1.962316193157593e-06,
      "loss": 2.0459,
      "step": 183
    },
    {
      "epoch": 0.03622582074125117,
      "grad_norm": 1.7500289678573608,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.9734,
      "step": 184
    },
    {
      "epoch": 0.036422700201801446,
      "grad_norm": 1.6304092407226562,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.9591,
      "step": 185
    },
    {
      "epoch": 0.03661957966235173,
      "grad_norm": 1.9713987112045288,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.9772,
      "step": 186
    },
    {
      "epoch": 0.036816459122902004,
      "grad_norm": 1.6248691082000732,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.7881,
      "step": 187
    },
    {
      "epoch": 0.03701333858345228,
      "grad_norm": 1.7883880138397217,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.9583,
      "step": 188
    },
    {
      "epoch": 0.03721021804400256,
      "grad_norm": 1.8968380689620972,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.9857,
      "step": 189
    },
    {
      "epoch": 0.03740709750455284,
      "grad_norm": 1.7393633127212524,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.7641,
      "step": 190
    },
    {
      "epoch": 0.03760397696510311,
      "grad_norm": 1.9815459251403809,
      "learning_rate": 5.526064699265753e-07,
      "loss": 2.0591,
      "step": 191
    },
    {
      "epoch": 0.037800856425653395,
      "grad_norm": 1.5676066875457764,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.8335,
      "step": 192
    },
    {
      "epoch": 0.03799773588620367,
      "grad_norm": 1.8990899324417114,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.8151,
      "step": 193
    },
    {
      "epoch": 0.03819461534675395,
      "grad_norm": 1.8243780136108398,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.9152,
      "step": 194
    },
    {
      "epoch": 0.03839149480730423,
      "grad_norm": 2.0028982162475586,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.7972,
      "step": 195
    },
    {
      "epoch": 0.038588374267854504,
      "grad_norm": 2.096224308013916,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.6984,
      "step": 196
    },
    {
      "epoch": 0.03878525372840479,
      "grad_norm": 2.4684019088745117,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.8951,
      "step": 197
    },
    {
      "epoch": 0.03898213318895506,
      "grad_norm": 2.1698174476623535,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.84,
      "step": 198
    },
    {
      "epoch": 0.03917901264950534,
      "grad_norm": 2.8406214714050293,
      "learning_rate": 6.834750376549792e-09,
      "loss": 2.1935,
      "step": 199
    },
    {
      "epoch": 0.03937589211005562,
      "grad_norm": 2.8411245346069336,
      "learning_rate": 0.0,
      "loss": 1.8803,
      "step": 200
    },
    {
      "epoch": 0.03937589211005562,
      "eval_loss": 1.9404832124710083,
      "eval_runtime": 997.0806,
      "eval_samples_per_second": 8.58,
      "eval_steps_per_second": 2.145,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3.017078697099264e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}