{
  "best_metric": 0.9804362058639526,
  "best_model_checkpoint": "miner_id_24/checkpoint-100",
  "epoch": 1.8475750577367207,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.009237875288683603,
      "grad_norm": 0.9806708097457886,
      "learning_rate": 1e-05,
      "loss": 1.6953,
      "step": 1
    },
    {
      "epoch": 0.009237875288683603,
      "eval_loss": 1.5332071781158447,
      "eval_runtime": 13.9898,
      "eval_samples_per_second": 13.01,
      "eval_steps_per_second": 3.288,
      "step": 1
    },
    {
      "epoch": 0.018475750577367205,
      "grad_norm": 1.1749752759933472,
      "learning_rate": 2e-05,
      "loss": 1.6017,
      "step": 2
    },
    {
      "epoch": 0.02771362586605081,
      "grad_norm": 1.3498163223266602,
      "learning_rate": 3e-05,
      "loss": 1.6898,
      "step": 3
    },
    {
      "epoch": 0.03695150115473441,
      "grad_norm": 1.3697495460510254,
      "learning_rate": 4e-05,
      "loss": 1.4683,
      "step": 4
    },
    {
      "epoch": 0.046189376443418015,
      "grad_norm": 1.1013462543487549,
      "learning_rate": 5e-05,
      "loss": 1.598,
      "step": 5
    },
    {
      "epoch": 0.05542725173210162,
      "grad_norm": 1.1153621673583984,
      "learning_rate": 6e-05,
      "loss": 1.472,
      "step": 6
    },
    {
      "epoch": 0.06466512702078522,
      "grad_norm": 1.2484103441238403,
      "learning_rate": 7e-05,
      "loss": 1.5789,
      "step": 7
    },
    {
      "epoch": 0.07390300230946882,
      "grad_norm": 1.1237813234329224,
      "learning_rate": 8e-05,
      "loss": 1.3814,
      "step": 8
    },
    {
      "epoch": 0.08314087759815242,
      "grad_norm": 0.8995471000671387,
      "learning_rate": 9e-05,
      "loss": 1.1743,
      "step": 9
    },
    {
      "epoch": 0.09237875288683603,
      "grad_norm": 0.7138063311576843,
      "learning_rate": 0.0001,
      "loss": 1.2305,
      "step": 10
    },
    {
      "epoch": 0.10161662817551963,
      "grad_norm": 0.7228544354438782,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.0903,
      "step": 11
    },
    {
      "epoch": 0.11085450346420324,
      "grad_norm": 0.7282606363296509,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.2087,
      "step": 12
    },
    {
      "epoch": 0.12009237875288684,
      "grad_norm": 0.86252760887146,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.3266,
      "step": 13
    },
    {
      "epoch": 0.12933025404157045,
      "grad_norm": 0.6563358902931213,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.1544,
      "step": 14
    },
    {
      "epoch": 0.13856812933025403,
      "grad_norm": 0.6294015645980835,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.1182,
      "step": 15
    },
    {
      "epoch": 0.14780600461893764,
      "grad_norm": 0.6522502303123474,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.0447,
      "step": 16
    },
    {
      "epoch": 0.15704387990762125,
      "grad_norm": 0.6219693422317505,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.1154,
      "step": 17
    },
    {
      "epoch": 0.16628175519630484,
      "grad_norm": 0.604991614818573,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.0856,
      "step": 18
    },
    {
      "epoch": 0.17551963048498845,
      "grad_norm": 0.6020306944847107,
      "learning_rate": 9.944739353007344e-05,
      "loss": 0.9799,
      "step": 19
    },
    {
      "epoch": 0.18475750577367206,
      "grad_norm": 0.6264792680740356,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.0722,
      "step": 20
    },
    {
      "epoch": 0.19399538106235567,
      "grad_norm": 0.5752476453781128,
      "learning_rate": 9.917525374361912e-05,
      "loss": 0.9899,
      "step": 21
    },
    {
      "epoch": 0.20323325635103925,
      "grad_norm": 0.595992386341095,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.0695,
      "step": 22
    },
    {
      "epoch": 0.21247113163972287,
      "grad_norm": 0.7354147434234619,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.162,
      "step": 23
    },
    {
      "epoch": 0.22170900692840648,
      "grad_norm": 0.660594642162323,
      "learning_rate": 9.86663298624003e-05,
      "loss": 0.8381,
      "step": 24
    },
    {
      "epoch": 0.23094688221709006,
      "grad_norm": 0.6191681027412415,
      "learning_rate": 9.847001329696653e-05,
      "loss": 0.798,
      "step": 25
    },
    {
      "epoch": 0.24018475750577367,
      "grad_norm": 0.5531579256057739,
      "learning_rate": 9.826044551386744e-05,
      "loss": 0.5592,
      "step": 26
    },
    {
      "epoch": 0.24942263279445728,
      "grad_norm": 0.6260628700256348,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.8096,
      "step": 27
    },
    {
      "epoch": 0.2586605080831409,
      "grad_norm": 0.6584234237670898,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.262,
      "step": 28
    },
    {
      "epoch": 0.2678983833718245,
      "grad_norm": 0.539679765701294,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.0146,
      "step": 29
    },
    {
      "epoch": 0.27713625866050806,
      "grad_norm": 0.5727636218070984,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.0946,
      "step": 30
    },
    {
      "epoch": 0.2863741339491917,
      "grad_norm": 0.5342716574668884,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.1785,
      "step": 31
    },
    {
      "epoch": 0.2956120092378753,
      "grad_norm": 0.46868839859962463,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.0631,
      "step": 32
    },
    {
      "epoch": 0.30484988452655887,
      "grad_norm": 0.6001697778701782,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.1004,
      "step": 33
    },
    {
      "epoch": 0.3140877598152425,
      "grad_norm": 0.6160048246383667,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.2023,
      "step": 34
    },
    {
      "epoch": 0.3233256351039261,
      "grad_norm": 0.49292078614234924,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.1693,
      "step": 35
    },
    {
      "epoch": 0.3325635103926097,
      "grad_norm": 0.5050878524780273,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.0462,
      "step": 36
    },
    {
      "epoch": 0.3418013856812933,
      "grad_norm": 0.4972296357154846,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.1851,
      "step": 37
    },
    {
      "epoch": 0.3510392609699769,
      "grad_norm": 0.5008955001831055,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.1472,
      "step": 38
    },
    {
      "epoch": 0.36027713625866054,
      "grad_norm": 0.5147765278816223,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.1505,
      "step": 39
    },
    {
      "epoch": 0.3695150115473441,
      "grad_norm": 0.48647770285606384,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.0636,
      "step": 40
    },
    {
      "epoch": 0.3787528868360277,
      "grad_norm": 0.5242158770561218,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.0655,
      "step": 41
    },
    {
      "epoch": 0.38799076212471134,
      "grad_norm": 0.5062406659126282,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.0745,
      "step": 42
    },
    {
      "epoch": 0.3972286374133949,
      "grad_norm": 0.47710514068603516,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.066,
      "step": 43
    },
    {
      "epoch": 0.4064665127020785,
      "grad_norm": 0.5392140746116638,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.0351,
      "step": 44
    },
    {
      "epoch": 0.41570438799076215,
      "grad_norm": 0.5057617425918579,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.1135,
      "step": 45
    },
    {
      "epoch": 0.42494226327944573,
      "grad_norm": 0.5254191160202026,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.0757,
      "step": 46
    },
    {
      "epoch": 0.4341801385681293,
      "grad_norm": 0.5257070064544678,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.0442,
      "step": 47
    },
    {
      "epoch": 0.44341801385681295,
      "grad_norm": 0.5173420310020447,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.9856,
      "step": 48
    },
    {
      "epoch": 0.45265588914549654,
      "grad_norm": 0.543683648109436,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.9677,
      "step": 49
    },
    {
      "epoch": 0.4618937644341801,
      "grad_norm": 0.5509938597679138,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.8849,
      "step": 50
    },
    {
      "epoch": 0.4618937644341801,
      "eval_loss": 0.9942653775215149,
      "eval_runtime": 14.2459,
      "eval_samples_per_second": 12.776,
      "eval_steps_per_second": 3.229,
      "step": 50
    },
    {
      "epoch": 0.47113163972286376,
      "grad_norm": 0.5746851563453674,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.8847,
      "step": 51
    },
    {
      "epoch": 0.48036951501154734,
      "grad_norm": 0.5114031434059143,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.6295,
      "step": 52
    },
    {
      "epoch": 0.4896073903002309,
      "grad_norm": 0.5455335378646851,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.6957,
      "step": 53
    },
    {
      "epoch": 0.49884526558891457,
      "grad_norm": 0.579850435256958,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.7315,
      "step": 54
    },
    {
      "epoch": 0.5080831408775982,
      "grad_norm": 0.5195852518081665,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.0362,
      "step": 55
    },
    {
      "epoch": 0.5173210161662818,
      "grad_norm": 0.49305659532546997,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.9826,
      "step": 56
    },
    {
      "epoch": 0.5265588914549654,
      "grad_norm": 0.48710188269615173,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.9681,
      "step": 57
    },
    {
      "epoch": 0.535796766743649,
      "grad_norm": 0.48450198769569397,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.0281,
      "step": 58
    },
    {
      "epoch": 0.5450346420323325,
      "grad_norm": 0.44846394658088684,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.0718,
      "step": 59
    },
    {
      "epoch": 0.5542725173210161,
      "grad_norm": 0.4434879720211029,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.1324,
      "step": 60
    },
    {
      "epoch": 0.5635103926096998,
      "grad_norm": 0.47181084752082825,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.1549,
      "step": 61
    },
    {
      "epoch": 0.5727482678983834,
      "grad_norm": 0.5568523406982422,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.1637,
      "step": 62
    },
    {
      "epoch": 0.581986143187067,
      "grad_norm": 0.4398137032985687,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.0788,
      "step": 63
    },
    {
      "epoch": 0.5912240184757506,
      "grad_norm": 0.4948345720767975,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.1497,
      "step": 64
    },
    {
      "epoch": 0.6004618937644342,
      "grad_norm": 0.4466833174228668,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.094,
      "step": 65
    },
    {
      "epoch": 0.6096997690531177,
      "grad_norm": 0.48202162981033325,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.0487,
      "step": 66
    },
    {
      "epoch": 0.6189376443418014,
      "grad_norm": 0.48863211274147034,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.0532,
      "step": 67
    },
    {
      "epoch": 0.628175519630485,
      "grad_norm": 0.5070480108261108,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.0793,
      "step": 68
    },
    {
      "epoch": 0.6374133949191686,
      "grad_norm": 0.4808672070503235,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.0188,
      "step": 69
    },
    {
      "epoch": 0.6466512702078522,
      "grad_norm": 0.505221962928772,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.1293,
      "step": 70
    },
    {
      "epoch": 0.6558891454965358,
      "grad_norm": 0.47788479924201965,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.0201,
      "step": 71
    },
    {
      "epoch": 0.6651270207852193,
      "grad_norm": 0.5395892858505249,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.0304,
      "step": 72
    },
    {
      "epoch": 0.674364896073903,
      "grad_norm": 0.5172073245048523,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.9421,
      "step": 73
    },
    {
      "epoch": 0.6836027713625866,
      "grad_norm": 0.576568603515625,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.078,
      "step": 74
    },
    {
      "epoch": 0.6928406466512702,
      "grad_norm": 0.5167725086212158,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.129,
      "step": 75
    },
    {
      "epoch": 0.7020785219399538,
      "grad_norm": 0.5086294412612915,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.9559,
      "step": 76
    },
    {
      "epoch": 0.7113163972286374,
      "grad_norm": 0.5268338918685913,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.0093,
      "step": 77
    },
    {
      "epoch": 0.7205542725173211,
      "grad_norm": 0.5083650350570679,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.8693,
      "step": 78
    },
    {
      "epoch": 0.7297921478060047,
      "grad_norm": 0.5479512214660645,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.8753,
      "step": 79
    },
    {
      "epoch": 0.7390300230946882,
      "grad_norm": 0.5435703992843628,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.6777,
      "step": 80
    },
    {
      "epoch": 0.7482678983833718,
      "grad_norm": 0.6067147254943848,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.6897,
      "step": 81
    },
    {
      "epoch": 0.7575057736720554,
      "grad_norm": 0.49712732434272766,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.0738,
      "step": 82
    },
    {
      "epoch": 0.766743648960739,
      "grad_norm": 0.4505629539489746,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.9474,
      "step": 83
    },
    {
      "epoch": 0.7759815242494227,
      "grad_norm": 0.48265042901039124,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.1468,
      "step": 84
    },
    {
      "epoch": 0.7852193995381063,
      "grad_norm": 0.4400395154953003,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.0387,
      "step": 85
    },
    {
      "epoch": 0.7944572748267898,
      "grad_norm": 0.43379443883895874,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.0801,
      "step": 86
    },
    {
      "epoch": 0.8036951501154734,
      "grad_norm": 0.4374670684337616,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.0116,
      "step": 87
    },
    {
      "epoch": 0.812933025404157,
      "grad_norm": 0.4649677872657776,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.0474,
      "step": 88
    },
    {
      "epoch": 0.8221709006928406,
      "grad_norm": 0.5176208019256592,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.0435,
      "step": 89
    },
    {
      "epoch": 0.8314087759815243,
      "grad_norm": 0.5411157011985779,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.029,
      "step": 90
    },
    {
      "epoch": 0.8406466512702079,
      "grad_norm": 0.46420204639434814,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.9366,
      "step": 91
    },
    {
      "epoch": 0.8498845265588915,
      "grad_norm": 0.46111324429512024,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.9799,
      "step": 92
    },
    {
      "epoch": 0.859122401847575,
      "grad_norm": 0.47871342301368713,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.0859,
      "step": 93
    },
    {
      "epoch": 0.8683602771362586,
      "grad_norm": 0.470443993806839,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.0841,
      "step": 94
    },
    {
      "epoch": 0.8775981524249422,
      "grad_norm": 0.49334949254989624,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.0523,
      "step": 95
    },
    {
      "epoch": 0.8868360277136259,
      "grad_norm": 0.5240926146507263,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.0805,
      "step": 96
    },
    {
      "epoch": 0.8960739030023095,
      "grad_norm": 0.4867502748966217,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.0895,
      "step": 97
    },
    {
      "epoch": 0.9053117782909931,
      "grad_norm": 0.47729411721229553,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.9414,
      "step": 98
    },
    {
      "epoch": 0.9145496535796767,
      "grad_norm": 0.45714375376701355,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.9619,
      "step": 99
    },
    {
      "epoch": 0.9237875288683602,
      "grad_norm": 0.4967270493507385,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.0907,
      "step": 100
    },
    {
      "epoch": 0.9237875288683602,
      "eval_loss": 0.9804362058639526,
      "eval_runtime": 14.2492,
      "eval_samples_per_second": 12.773,
      "eval_steps_per_second": 3.228,
      "step": 100
    },
    {
      "epoch": 0.9330254041570438,
      "grad_norm": 0.48148465156555176,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.9965,
      "step": 101
    },
    {
      "epoch": 0.9422632794457275,
      "grad_norm": 0.5082510113716125,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.0027,
      "step": 102
    },
    {
      "epoch": 0.9515011547344111,
      "grad_norm": 0.4773678779602051,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.9008,
      "step": 103
    },
    {
      "epoch": 0.9607390300230947,
      "grad_norm": 0.48190367221832275,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.8234,
      "step": 104
    },
    {
      "epoch": 0.9699769053117783,
      "grad_norm": 0.5185900926589966,
      "learning_rate": 5e-05,
      "loss": 0.9037,
      "step": 105
    },
    {
      "epoch": 0.9792147806004619,
      "grad_norm": 0.5802286267280579,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.9145,
      "step": 106
    },
    {
      "epoch": 0.9884526558891455,
      "grad_norm": 0.5321520566940308,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.5947,
      "step": 107
    },
    {
      "epoch": 0.9976905311778291,
      "grad_norm": 0.6070680022239685,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.5773,
      "step": 108
    },
    {
      "epoch": 1.0069284064665127,
      "grad_norm": 1.7495030164718628,
      "learning_rate": 4.669547078371504e-05,
      "loss": 2.3347,
      "step": 109
    },
    {
      "epoch": 1.0161662817551964,
      "grad_norm": 0.4606659710407257,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.8723,
      "step": 110
    },
    {
      "epoch": 1.0254041570438799,
      "grad_norm": 0.48735034465789795,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.9338,
      "step": 111
    },
    {
      "epoch": 1.0346420323325636,
      "grad_norm": 0.45384830236434937,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.9132,
      "step": 112
    },
    {
      "epoch": 1.043879907621247,
      "grad_norm": 0.4299413561820984,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.9486,
      "step": 113
    },
    {
      "epoch": 1.0531177829099307,
      "grad_norm": 0.47057202458381653,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.0253,
      "step": 114
    },
    {
      "epoch": 1.0623556581986142,
      "grad_norm": 0.44118061661720276,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.9094,
      "step": 115
    },
    {
      "epoch": 1.071593533487298,
      "grad_norm": 0.4594104588031769,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.9597,
      "step": 116
    },
    {
      "epoch": 1.0808314087759816,
      "grad_norm": 0.4826590418815613,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.0089,
      "step": 117
    },
    {
      "epoch": 1.090069284064665,
      "grad_norm": 0.4662156403064728,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.8262,
      "step": 118
    },
    {
      "epoch": 1.0993071593533488,
      "grad_norm": 0.4533056914806366,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.9388,
      "step": 119
    },
    {
      "epoch": 1.1085450346420322,
      "grad_norm": 0.4630964696407318,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.8342,
      "step": 120
    },
    {
      "epoch": 1.117782909930716,
      "grad_norm": 0.5027814507484436,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.9256,
      "step": 121
    },
    {
      "epoch": 1.1270207852193996,
      "grad_norm": 0.5189921855926514,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.9944,
      "step": 122
    },
    {
      "epoch": 1.136258660508083,
      "grad_norm": 0.49351710081100464,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.8448,
      "step": 123
    },
    {
      "epoch": 1.1454965357967668,
      "grad_norm": 0.49378764629364014,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.8795,
      "step": 124
    },
    {
      "epoch": 1.1547344110854503,
      "grad_norm": 0.5376058220863342,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.9479,
      "step": 125
    },
    {
      "epoch": 1.163972286374134,
      "grad_norm": 0.5019356608390808,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.7882,
      "step": 126
    },
    {
      "epoch": 1.1732101616628174,
      "grad_norm": 0.49805977940559387,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.7039,
      "step": 127
    },
    {
      "epoch": 1.1824480369515011,
      "grad_norm": 0.5588062405586243,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.853,
      "step": 128
    },
    {
      "epoch": 1.1916859122401848,
      "grad_norm": 0.5479720234870911,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.7947,
      "step": 129
    },
    {
      "epoch": 1.2009237875288683,
      "grad_norm": 0.5878831744194031,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.8535,
      "step": 130
    },
    {
      "epoch": 1.210161662817552,
      "grad_norm": 0.5415154695510864,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.7087,
      "step": 131
    },
    {
      "epoch": 1.2193995381062355,
      "grad_norm": 0.5652676224708557,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.6225,
      "step": 132
    },
    {
      "epoch": 1.2286374133949192,
      "grad_norm": 0.5161898732185364,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.486,
      "step": 133
    },
    {
      "epoch": 1.2378752886836029,
      "grad_norm": 0.5891221761703491,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.537,
      "step": 134
    },
    {
      "epoch": 1.2471131639722863,
      "grad_norm": 0.49396812915802,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.4155,
      "step": 135
    },
    {
      "epoch": 1.25635103926097,
      "grad_norm": 0.5404485464096069,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.7553,
      "step": 136
    },
    {
      "epoch": 1.2655889145496535,
      "grad_norm": 0.5885382890701294,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.9317,
      "step": 137
    },
    {
      "epoch": 1.2748267898383372,
      "grad_norm": 0.5917038321495056,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.9376,
      "step": 138
    },
    {
      "epoch": 1.2840646651270209,
      "grad_norm": 0.5704668164253235,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.9635,
      "step": 139
    },
    {
      "epoch": 1.2933025404157044,
      "grad_norm": 0.5660439729690552,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.9408,
      "step": 140
    },
    {
      "epoch": 1.302540415704388,
      "grad_norm": 0.5617368817329407,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.9484,
      "step": 141
    },
    {
      "epoch": 1.3117782909930715,
      "grad_norm": 0.5856223106384277,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.105,
      "step": 142
    },
    {
      "epoch": 1.3210161662817552,
      "grad_norm": 0.5251100063323975,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.9021,
      "step": 143
    },
    {
      "epoch": 1.3302540415704387,
      "grad_norm": 0.5774368047714233,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.9709,
      "step": 144
    },
    {
      "epoch": 1.3394919168591224,
      "grad_norm": 0.5094186067581177,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.8995,
      "step": 145
    },
    {
      "epoch": 1.348729792147806,
      "grad_norm": 0.5760278701782227,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.951,
      "step": 146
    },
    {
      "epoch": 1.3579676674364896,
      "grad_norm": 0.5219050645828247,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.88,
      "step": 147
    },
    {
      "epoch": 1.3672055427251733,
      "grad_norm": 0.5386266112327576,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.9435,
      "step": 148
    },
    {
      "epoch": 1.376443418013857,
      "grad_norm": 0.5403474569320679,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.8793,
      "step": 149
    },
    {
      "epoch": 1.3856812933025404,
      "grad_norm": 0.5808868408203125,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.8653,
      "step": 150
    },
    {
      "epoch": 1.3856812933025404,
      "eval_loss": 0.9949983954429626,
      "eval_runtime": 13.9797,
      "eval_samples_per_second": 13.019,
      "eval_steps_per_second": 3.29,
      "step": 150
    },
    {
      "epoch": 1.394919168591224,
      "grad_norm": 0.556486189365387,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.8824,
      "step": 151
    },
    {
      "epoch": 1.4041570438799076,
      "grad_norm": 0.5511496663093567,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.8475,
      "step": 152
    },
    {
      "epoch": 1.4133949191685913,
      "grad_norm": 0.5870489478111267,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.8244,
      "step": 153
    },
    {
      "epoch": 1.4226327944572748,
      "grad_norm": 0.5970893502235413,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.8569,
      "step": 154
    },
    {
      "epoch": 1.4318706697459584,
      "grad_norm": 0.5708542466163635,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.854,
      "step": 155
    },
    {
      "epoch": 1.4411085450346421,
      "grad_norm": 0.6128978133201599,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.8596,
      "step": 156
    },
    {
      "epoch": 1.4503464203233256,
      "grad_norm": 0.5593998432159424,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.8191,
      "step": 157
    },
    {
      "epoch": 1.4595842956120093,
      "grad_norm": 0.6363524198532104,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.7603,
      "step": 158
    },
    {
      "epoch": 1.4688221709006928,
      "grad_norm": 0.6085260510444641,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.7047,
      "step": 159
    },
    {
      "epoch": 1.4780600461893765,
      "grad_norm": 0.69758141040802,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.7252,
      "step": 160
    },
    {
      "epoch": 1.48729792147806,
      "grad_norm": 0.5065174102783203,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.3189,
      "step": 161
    },
    {
      "epoch": 1.4965357967667436,
      "grad_norm": 0.5280264019966125,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.3924,
      "step": 162
    },
    {
      "epoch": 1.5057736720554273,
      "grad_norm": 0.6419398784637451,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.9497,
      "step": 163
    },
    {
      "epoch": 1.5150115473441108,
      "grad_norm": 0.5234368443489075,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.7864,
      "step": 164
    },
    {
      "epoch": 1.5242494226327945,
      "grad_norm": 0.59024578332901,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.0156,
      "step": 165
    },
    {
      "epoch": 1.5334872979214782,
      "grad_norm": 0.5937689542770386,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.9005,
      "step": 166
    },
    {
      "epoch": 1.5427251732101617,
      "grad_norm": 0.5771698951721191,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.8969,
      "step": 167
    },
    {
      "epoch": 1.5519630484988451,
      "grad_norm": 0.6146437525749207,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.9865,
      "step": 168
    },
    {
      "epoch": 1.5612009237875288,
      "grad_norm": 0.5948532819747925,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.8952,
      "step": 169
    },
    {
      "epoch": 1.5704387990762125,
      "grad_norm": 0.6427422165870667,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.9981,
      "step": 170
    },
    {
      "epoch": 1.579676674364896,
      "grad_norm": 0.5856215357780457,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.8602,
      "step": 171
    },
    {
      "epoch": 1.5889145496535797,
      "grad_norm": 0.6068481206893921,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.9625,
      "step": 172
    },
    {
      "epoch": 1.5981524249422634,
      "grad_norm": 0.5857320427894592,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.9129,
      "step": 173
    },
    {
      "epoch": 1.6073903002309469,
      "grad_norm": 0.5553707480430603,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.8602,
      "step": 174
    },
    {
      "epoch": 1.6166281755196303,
      "grad_norm": 0.6019971370697021,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.931,
      "step": 175
    },
    {
      "epoch": 1.625866050808314,
      "grad_norm": 0.5924701690673828,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.9383,
      "step": 176
    },
    {
      "epoch": 1.6351039260969977,
      "grad_norm": 0.5981978178024292,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.9003,
      "step": 177
    },
    {
      "epoch": 1.6443418013856812,
      "grad_norm": 0.5562158226966858,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.7259,
      "step": 178
    },
    {
      "epoch": 1.653579676674365,
      "grad_norm": 0.5878584384918213,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.8436,
      "step": 179
    },
    {
      "epoch": 1.6628175519630486,
      "grad_norm": 0.6131454706192017,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.9112,
      "step": 180
    },
    {
      "epoch": 1.672055427251732,
      "grad_norm": 0.5964655876159668,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.7954,
      "step": 181
    },
    {
      "epoch": 1.6812933025404158,
      "grad_norm": 0.5727086067199707,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.7918,
      "step": 182
    },
    {
      "epoch": 1.6905311778290995,
      "grad_norm": 0.5908867120742798,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.7868,
      "step": 183
    },
    {
      "epoch": 1.699769053117783,
      "grad_norm": 0.6519110202789307,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.8147,
      "step": 184
    },
    {
      "epoch": 1.7090069284064664,
      "grad_norm": 0.6141422390937805,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.8075,
      "step": 185
    },
    {
      "epoch": 1.71824480369515,
      "grad_norm": 0.6434694528579712,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.7692,
      "step": 186
    },
    {
      "epoch": 1.7274826789838338,
      "grad_norm": 0.6423685550689697,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.6751,
      "step": 187
    },
    {
      "epoch": 1.7367205542725173,
      "grad_norm": 0.5378662347793579,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.4113,
      "step": 188
    },
    {
      "epoch": 1.745958429561201,
      "grad_norm": 0.608502209186554,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.4306,
      "step": 189
    },
    {
      "epoch": 1.7551963048498846,
      "grad_norm": 0.6008912324905396,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.8909,
      "step": 190
    },
    {
      "epoch": 1.7644341801385681,
      "grad_norm": 0.5199741721153259,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.8115,
      "step": 191
    },
    {
      "epoch": 1.7736720554272516,
      "grad_norm": 0.5280171632766724,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.8418,
      "step": 192
    },
    {
      "epoch": 1.7829099307159353,
      "grad_norm": 0.5399708151817322,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.9326,
      "step": 193
    },
    {
      "epoch": 1.792147806004619,
      "grad_norm": 0.5415851473808289,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.8668,
      "step": 194
    },
    {
      "epoch": 1.8013856812933025,
      "grad_norm": 0.51853346824646,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.8861,
      "step": 195
    },
    {
      "epoch": 1.8106235565819861,
      "grad_norm": 0.5716439485549927,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.0486,
      "step": 196
    },
    {
      "epoch": 1.8198614318706698,
      "grad_norm": 0.5678825378417969,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.8998,
      "step": 197
    },
    {
      "epoch": 1.8290993071593533,
      "grad_norm": 0.5393827557563782,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.8118,
      "step": 198
    },
    {
      "epoch": 1.8383371824480368,
      "grad_norm": 0.575571596622467,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.9744,
      "step": 199
    },
    {
      "epoch": 1.8475750577367207,
      "grad_norm": 0.5658337473869324,
      "learning_rate": 0.0,
      "loss": 0.9203,
      "step": 200
    },
    {
      "epoch": 1.8475750577367207,
      "eval_loss": 0.9946041703224182,
      "eval_runtime": 14.2506,
      "eval_samples_per_second": 12.771,
      "eval_steps_per_second": 3.228,
      "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": 2
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.937088815661056e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}