{
  "best_metric": 12.416987419128418,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.15270089711777057,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0007635044855888528,
      "grad_norm": 0.007160913199186325,
      "learning_rate": 1e-05,
      "loss": 12.4558,
      "step": 1
    },
    {
      "epoch": 0.0007635044855888528,
      "eval_loss": 12.45350170135498,
      "eval_runtime": 15.5103,
      "eval_samples_per_second": 142.228,
      "eval_steps_per_second": 35.589,
      "step": 1
    },
    {
      "epoch": 0.0015270089711777056,
      "grad_norm": 0.007832451723515987,
      "learning_rate": 2e-05,
      "loss": 12.4588,
      "step": 2
    },
    {
      "epoch": 0.0022905134567665587,
      "grad_norm": 0.007961795665323734,
      "learning_rate": 3e-05,
      "loss": 12.4544,
      "step": 3
    },
    {
      "epoch": 0.003054017942355411,
      "grad_norm": 0.008210752159357071,
      "learning_rate": 4e-05,
      "loss": 12.4571,
      "step": 4
    },
    {
      "epoch": 0.003817522427944264,
      "grad_norm": 0.009312192909419537,
      "learning_rate": 5e-05,
      "loss": 12.4517,
      "step": 5
    },
    {
      "epoch": 0.004581026913533117,
      "grad_norm": 0.008079441264271736,
      "learning_rate": 6e-05,
      "loss": 12.4573,
      "step": 6
    },
    {
      "epoch": 0.0053445313991219694,
      "grad_norm": 0.007759293541312218,
      "learning_rate": 7e-05,
      "loss": 12.4554,
      "step": 7
    },
    {
      "epoch": 0.006108035884710822,
      "grad_norm": 0.00820003543049097,
      "learning_rate": 8e-05,
      "loss": 12.4503,
      "step": 8
    },
    {
      "epoch": 0.006871540370299675,
      "grad_norm": 0.009337841533124447,
      "learning_rate": 9e-05,
      "loss": 12.4576,
      "step": 9
    },
    {
      "epoch": 0.007635044855888528,
      "grad_norm": 0.008671695366501808,
      "learning_rate": 0.0001,
      "loss": 12.4569,
      "step": 10
    },
    {
      "epoch": 0.008398549341477381,
      "grad_norm": 0.009205568581819534,
      "learning_rate": 9.999316524962345e-05,
      "loss": 12.4598,
      "step": 11
    },
    {
      "epoch": 0.009162053827066235,
      "grad_norm": 0.010048221796751022,
      "learning_rate": 9.997266286704631e-05,
      "loss": 12.4504,
      "step": 12
    },
    {
      "epoch": 0.009925558312655087,
      "grad_norm": 0.007624107412993908,
      "learning_rate": 9.993849845741524e-05,
      "loss": 12.4561,
      "step": 13
    },
    {
      "epoch": 0.010689062798243939,
      "grad_norm": 0.009537247009575367,
      "learning_rate": 9.989068136093873e-05,
      "loss": 12.4542,
      "step": 14
    },
    {
      "epoch": 0.011452567283832793,
      "grad_norm": 0.011066782288253307,
      "learning_rate": 9.98292246503335e-05,
      "loss": 12.4482,
      "step": 15
    },
    {
      "epoch": 0.012216071769421645,
      "grad_norm": 0.009247601963579655,
      "learning_rate": 9.975414512725057e-05,
      "loss": 12.4462,
      "step": 16
    },
    {
      "epoch": 0.012979576255010498,
      "grad_norm": 0.011382615193724632,
      "learning_rate": 9.966546331768191e-05,
      "loss": 12.4542,
      "step": 17
    },
    {
      "epoch": 0.01374308074059935,
      "grad_norm": 0.009538145735859871,
      "learning_rate": 9.956320346634876e-05,
      "loss": 12.457,
      "step": 18
    },
    {
      "epoch": 0.014506585226188204,
      "grad_norm": 0.013374575413763523,
      "learning_rate": 9.944739353007344e-05,
      "loss": 12.4518,
      "step": 19
    },
    {
      "epoch": 0.015270089711777056,
      "grad_norm": 0.011300849728286266,
      "learning_rate": 9.931806517013612e-05,
      "loss": 12.4556,
      "step": 20
    },
    {
      "epoch": 0.01603359419736591,
      "grad_norm": 0.014204300008714199,
      "learning_rate": 9.917525374361912e-05,
      "loss": 12.4499,
      "step": 21
    },
    {
      "epoch": 0.016797098682954762,
      "grad_norm": 0.01346464455127716,
      "learning_rate": 9.901899829374047e-05,
      "loss": 12.4482,
      "step": 22
    },
    {
      "epoch": 0.017560603168543614,
      "grad_norm": 0.014953072182834148,
      "learning_rate": 9.884934153917997e-05,
      "loss": 12.4475,
      "step": 23
    },
    {
      "epoch": 0.01832410765413247,
      "grad_norm": 0.016649188473820686,
      "learning_rate": 9.86663298624003e-05,
      "loss": 12.4575,
      "step": 24
    },
    {
      "epoch": 0.01908761213972132,
      "grad_norm": 0.014094348065555096,
      "learning_rate": 9.847001329696653e-05,
      "loss": 12.4603,
      "step": 25
    },
    {
      "epoch": 0.019851116625310174,
      "grad_norm": 0.016804827377200127,
      "learning_rate": 9.826044551386744e-05,
      "loss": 12.4534,
      "step": 26
    },
    {
      "epoch": 0.020614621110899026,
      "grad_norm": 0.01771470159292221,
      "learning_rate": 9.803768380684242e-05,
      "loss": 12.4494,
      "step": 27
    },
    {
      "epoch": 0.021378125596487878,
      "grad_norm": 0.015941236168146133,
      "learning_rate": 9.780178907671789e-05,
      "loss": 12.4538,
      "step": 28
    },
    {
      "epoch": 0.022141630082076733,
      "grad_norm": 0.01862405613064766,
      "learning_rate": 9.755282581475769e-05,
      "loss": 12.4502,
      "step": 29
    },
    {
      "epoch": 0.022905134567665585,
      "grad_norm": 0.01642240770161152,
      "learning_rate": 9.729086208503174e-05,
      "loss": 12.452,
      "step": 30
    },
    {
      "epoch": 0.023668639053254437,
      "grad_norm": 0.025656452402472496,
      "learning_rate": 9.701596950580806e-05,
      "loss": 12.4464,
      "step": 31
    },
    {
      "epoch": 0.02443214353884329,
      "grad_norm": 0.023164507001638412,
      "learning_rate": 9.672822322997305e-05,
      "loss": 12.4435,
      "step": 32
    },
    {
      "epoch": 0.025195648024432145,
      "grad_norm": 0.022210022434592247,
      "learning_rate": 9.642770192448536e-05,
      "loss": 12.4495,
      "step": 33
    },
    {
      "epoch": 0.025959152510020997,
      "grad_norm": 0.026153987273573875,
      "learning_rate": 9.611448774886924e-05,
      "loss": 12.4516,
      "step": 34
    },
    {
      "epoch": 0.02672265699560985,
      "grad_norm": 0.023791734129190445,
      "learning_rate": 9.578866633275288e-05,
      "loss": 12.4439,
      "step": 35
    },
    {
      "epoch": 0.0274861614811987,
      "grad_norm": 0.030438346788287163,
      "learning_rate": 9.545032675245813e-05,
      "loss": 12.461,
      "step": 36
    },
    {
      "epoch": 0.028249665966787556,
      "grad_norm": 0.036134760826826096,
      "learning_rate": 9.509956150664796e-05,
      "loss": 12.4536,
      "step": 37
    },
    {
      "epoch": 0.02901317045237641,
      "grad_norm": 0.032248642295598984,
      "learning_rate": 9.473646649103818e-05,
      "loss": 12.4505,
      "step": 38
    },
    {
      "epoch": 0.02977667493796526,
      "grad_norm": 0.03967161104083061,
      "learning_rate": 9.43611409721806e-05,
      "loss": 12.4466,
      "step": 39
    },
    {
      "epoch": 0.030540179423554113,
      "grad_norm": 0.03780242055654526,
      "learning_rate": 9.397368756032445e-05,
      "loss": 12.4551,
      "step": 40
    },
    {
      "epoch": 0.03130368390914297,
      "grad_norm": 0.030669184401631355,
      "learning_rate": 9.357421218136386e-05,
      "loss": 12.4588,
      "step": 41
    },
    {
      "epoch": 0.03206718839473182,
      "grad_norm": 0.03821001574397087,
      "learning_rate": 9.316282404787871e-05,
      "loss": 12.4527,
      "step": 42
    },
    {
      "epoch": 0.03283069288032067,
      "grad_norm": 0.049706269055604935,
      "learning_rate": 9.273963562927695e-05,
      "loss": 12.4526,
      "step": 43
    },
    {
      "epoch": 0.033594197365909524,
      "grad_norm": 0.04330707713961601,
      "learning_rate": 9.230476262104677e-05,
      "loss": 12.4484,
      "step": 44
    },
    {
      "epoch": 0.034357701851498376,
      "grad_norm": 0.04611953720450401,
      "learning_rate": 9.185832391312644e-05,
      "loss": 12.4486,
      "step": 45
    },
    {
      "epoch": 0.03512120633708723,
      "grad_norm": 0.04229813441634178,
      "learning_rate": 9.140044155740101e-05,
      "loss": 12.4483,
      "step": 46
    },
    {
      "epoch": 0.03588471082267608,
      "grad_norm": 0.05988840013742447,
      "learning_rate": 9.093124073433463e-05,
      "loss": 12.4485,
      "step": 47
    },
    {
      "epoch": 0.03664821530826494,
      "grad_norm": 0.053687188774347305,
      "learning_rate": 9.045084971874738e-05,
      "loss": 12.457,
      "step": 48
    },
    {
      "epoch": 0.03741171979385379,
      "grad_norm": 0.06279481947422028,
      "learning_rate": 8.995939984474624e-05,
      "loss": 12.4528,
      "step": 49
    },
    {
      "epoch": 0.03817522427944264,
      "grad_norm": 0.08129668235778809,
      "learning_rate": 8.945702546981969e-05,
      "loss": 12.4405,
      "step": 50
    },
    {
      "epoch": 0.03817522427944264,
      "eval_loss": 12.449444770812988,
      "eval_runtime": 15.5173,
      "eval_samples_per_second": 142.164,
      "eval_steps_per_second": 35.573,
      "step": 50
    },
    {
      "epoch": 0.038938728765031495,
      "grad_norm": 0.04444519430398941,
      "learning_rate": 8.894386393810563e-05,
      "loss": 12.4564,
      "step": 51
    },
    {
      "epoch": 0.03970223325062035,
      "grad_norm": 0.040527038276195526,
      "learning_rate": 8.842005554284296e-05,
      "loss": 12.4542,
      "step": 52
    },
    {
      "epoch": 0.0404657377362092,
      "grad_norm": 0.04578302800655365,
      "learning_rate": 8.788574348801675e-05,
      "loss": 12.4454,
      "step": 53
    },
    {
      "epoch": 0.04122924222179805,
      "grad_norm": 0.04210494086146355,
      "learning_rate": 8.73410738492077e-05,
      "loss": 12.4503,
      "step": 54
    },
    {
      "epoch": 0.0419927467073869,
      "grad_norm": 0.05325311794877052,
      "learning_rate": 8.678619553365659e-05,
      "loss": 12.4412,
      "step": 55
    },
    {
      "epoch": 0.042756251192975755,
      "grad_norm": 0.053876765072345734,
      "learning_rate": 8.622126023955446e-05,
      "loss": 12.4449,
      "step": 56
    },
    {
      "epoch": 0.043519755678564614,
      "grad_norm": 0.05875396728515625,
      "learning_rate": 8.564642241456986e-05,
      "loss": 12.4484,
      "step": 57
    },
    {
      "epoch": 0.044283260164153466,
      "grad_norm": 0.06220713630318642,
      "learning_rate": 8.506183921362443e-05,
      "loss": 12.451,
      "step": 58
    },
    {
      "epoch": 0.04504676464974232,
      "grad_norm": 0.050935592502355576,
      "learning_rate": 8.44676704559283e-05,
      "loss": 12.4538,
      "step": 59
    },
    {
      "epoch": 0.04581026913533117,
      "grad_norm": 0.0560225211083889,
      "learning_rate": 8.386407858128706e-05,
      "loss": 12.4476,
      "step": 60
    },
    {
      "epoch": 0.04657377362092002,
      "grad_norm": 0.061477672308683395,
      "learning_rate": 8.32512286056924e-05,
      "loss": 12.4451,
      "step": 61
    },
    {
      "epoch": 0.047337278106508875,
      "grad_norm": 0.05897868797183037,
      "learning_rate": 8.262928807620843e-05,
      "loss": 12.4505,
      "step": 62
    },
    {
      "epoch": 0.04810078259209773,
      "grad_norm": 0.07552457600831985,
      "learning_rate": 8.199842702516583e-05,
      "loss": 12.4441,
      "step": 63
    },
    {
      "epoch": 0.04886428707768658,
      "grad_norm": 0.0720105767250061,
      "learning_rate": 8.135881792367686e-05,
      "loss": 12.4477,
      "step": 64
    },
    {
      "epoch": 0.04962779156327544,
      "grad_norm": 0.0720081701874733,
      "learning_rate": 8.07106356344834e-05,
      "loss": 12.4549,
      "step": 65
    },
    {
      "epoch": 0.05039129604886429,
      "grad_norm": 0.06441877782344818,
      "learning_rate": 8.005405736415126e-05,
      "loss": 12.4433,
      "step": 66
    },
    {
      "epoch": 0.05115480053445314,
      "grad_norm": 0.07265959680080414,
      "learning_rate": 7.938926261462366e-05,
      "loss": 12.4427,
      "step": 67
    },
    {
      "epoch": 0.051918305020041994,
      "grad_norm": 0.08124711364507675,
      "learning_rate": 7.871643313414718e-05,
      "loss": 12.4417,
      "step": 68
    },
    {
      "epoch": 0.052681809505630846,
      "grad_norm": 0.08123979717493057,
      "learning_rate": 7.803575286758364e-05,
      "loss": 12.4478,
      "step": 69
    },
    {
      "epoch": 0.0534453139912197,
      "grad_norm": 0.07414422929286957,
      "learning_rate": 7.734740790612136e-05,
      "loss": 12.447,
      "step": 70
    },
    {
      "epoch": 0.05420881847680855,
      "grad_norm": 0.08728949725627899,
      "learning_rate": 7.66515864363997e-05,
      "loss": 12.4441,
      "step": 71
    },
    {
      "epoch": 0.0549723229623974,
      "grad_norm": 0.09869679063558578,
      "learning_rate": 7.594847868906076e-05,
      "loss": 12.4423,
      "step": 72
    },
    {
      "epoch": 0.055735827447986254,
      "grad_norm": 0.09882236272096634,
      "learning_rate": 7.52382768867422e-05,
      "loss": 12.446,
      "step": 73
    },
    {
      "epoch": 0.05649933193357511,
      "grad_norm": 0.09295084327459335,
      "learning_rate": 7.452117519152542e-05,
      "loss": 12.448,
      "step": 74
    },
    {
      "epoch": 0.057262836419163965,
      "grad_norm": 0.0962289497256279,
      "learning_rate": 7.379736965185368e-05,
      "loss": 12.4442,
      "step": 75
    },
    {
      "epoch": 0.05802634090475282,
      "grad_norm": 0.09535492956638336,
      "learning_rate": 7.30670581489344e-05,
      "loss": 12.4437,
      "step": 76
    },
    {
      "epoch": 0.05878984539034167,
      "grad_norm": 0.09165014326572418,
      "learning_rate": 7.233044034264034e-05,
      "loss": 12.4366,
      "step": 77
    },
    {
      "epoch": 0.05955334987593052,
      "grad_norm": 0.11812420934438705,
      "learning_rate": 7.158771761692464e-05,
      "loss": 12.436,
      "step": 78
    },
    {
      "epoch": 0.06031685436151937,
      "grad_norm": 0.09278896450996399,
      "learning_rate": 7.083909302476453e-05,
      "loss": 12.4443,
      "step": 79
    },
    {
      "epoch": 0.061080358847108225,
      "grad_norm": 0.09456734359264374,
      "learning_rate": 7.008477123264848e-05,
      "loss": 12.4392,
      "step": 80
    },
    {
      "epoch": 0.06184386333269708,
      "grad_norm": 0.100967638194561,
      "learning_rate": 6.932495846462261e-05,
      "loss": 12.4334,
      "step": 81
    },
    {
      "epoch": 0.06260736781828594,
      "grad_norm": 0.09592237323522568,
      "learning_rate": 6.855986244591104e-05,
      "loss": 12.4359,
      "step": 82
    },
    {
      "epoch": 0.06337087230387478,
      "grad_norm": 0.10887469351291656,
      "learning_rate": 6.778969234612584e-05,
      "loss": 12.4376,
      "step": 83
    },
    {
      "epoch": 0.06413437678946364,
      "grad_norm": 0.10575887560844421,
      "learning_rate": 6.701465872208216e-05,
      "loss": 12.4324,
      "step": 84
    },
    {
      "epoch": 0.06489788127505249,
      "grad_norm": 0.11523894965648651,
      "learning_rate": 6.623497346023418e-05,
      "loss": 12.438,
      "step": 85
    },
    {
      "epoch": 0.06566138576064134,
      "grad_norm": 0.11139199137687683,
      "learning_rate": 6.545084971874738e-05,
      "loss": 12.4305,
      "step": 86
    },
    {
      "epoch": 0.0664248902462302,
      "grad_norm": 0.1084304004907608,
      "learning_rate": 6.466250186922325e-05,
      "loss": 12.4326,
      "step": 87
    },
    {
      "epoch": 0.06718839473181905,
      "grad_norm": 0.1049765795469284,
      "learning_rate": 6.387014543809223e-05,
      "loss": 12.4281,
      "step": 88
    },
    {
      "epoch": 0.06795189921740791,
      "grad_norm": 0.10682480037212372,
      "learning_rate": 6.307399704769099e-05,
      "loss": 12.433,
      "step": 89
    },
    {
      "epoch": 0.06871540370299675,
      "grad_norm": 0.1162189170718193,
      "learning_rate": 6.227427435703997e-05,
      "loss": 12.431,
      "step": 90
    },
    {
      "epoch": 0.06947890818858561,
      "grad_norm": 0.10560418665409088,
      "learning_rate": 6.147119600233758e-05,
      "loss": 12.434,
      "step": 91
    },
    {
      "epoch": 0.07024241267417446,
      "grad_norm": 0.12470274418592453,
      "learning_rate": 6.066498153718735e-05,
      "loss": 12.4261,
      "step": 92
    },
    {
      "epoch": 0.07100591715976332,
      "grad_norm": 0.11700281500816345,
      "learning_rate": 5.985585137257401e-05,
      "loss": 12.4364,
      "step": 93
    },
    {
      "epoch": 0.07176942164535216,
      "grad_norm": 0.10540790855884552,
      "learning_rate": 5.90440267166055e-05,
      "loss": 12.4269,
      "step": 94
    },
    {
      "epoch": 0.07253292613094102,
      "grad_norm": 0.11336833238601685,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 12.4348,
      "step": 95
    },
    {
      "epoch": 0.07329643061652988,
      "grad_norm": 0.1218598410487175,
      "learning_rate": 5.74131823855921e-05,
      "loss": 12.4313,
      "step": 96
    },
    {
      "epoch": 0.07405993510211872,
      "grad_norm": 0.13793238997459412,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 12.423,
      "step": 97
    },
    {
      "epoch": 0.07482343958770758,
      "grad_norm": 0.15503503382205963,
      "learning_rate": 5.577423184847932e-05,
      "loss": 12.4225,
      "step": 98
    },
    {
      "epoch": 0.07558694407329643,
      "grad_norm": 0.14302343130111694,
      "learning_rate": 5.495227651252315e-05,
      "loss": 12.4263,
      "step": 99
    },
    {
      "epoch": 0.07635044855888529,
      "grad_norm": 0.17625275254249573,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 12.4151,
      "step": 100
    },
    {
      "epoch": 0.07635044855888529,
      "eval_loss": 12.428129196166992,
      "eval_runtime": 15.4753,
      "eval_samples_per_second": 142.55,
      "eval_steps_per_second": 35.67,
      "step": 100
    },
    {
      "epoch": 0.07711395304447413,
      "grad_norm": 0.08658638596534729,
      "learning_rate": 5.330452921628497e-05,
      "loss": 12.4347,
      "step": 101
    },
    {
      "epoch": 0.07787745753006299,
      "grad_norm": 0.10176920145750046,
      "learning_rate": 5.247918773366112e-05,
      "loss": 12.4322,
      "step": 102
    },
    {
      "epoch": 0.07864096201565184,
      "grad_norm": 0.10068823397159576,
      "learning_rate": 5.165316846586541e-05,
      "loss": 12.4319,
      "step": 103
    },
    {
      "epoch": 0.0794044665012407,
      "grad_norm": 0.09565402567386627,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 12.4327,
      "step": 104
    },
    {
      "epoch": 0.08016797098682955,
      "grad_norm": 0.116241954267025,
      "learning_rate": 5e-05,
      "loss": 12.43,
      "step": 105
    },
    {
      "epoch": 0.0809314754724184,
      "grad_norm": 0.08978600054979324,
      "learning_rate": 4.917330276168208e-05,
      "loss": 12.4311,
      "step": 106
    },
    {
      "epoch": 0.08169497995800726,
      "grad_norm": 0.08738912642002106,
      "learning_rate": 4.834683153413459e-05,
      "loss": 12.4289,
      "step": 107
    },
    {
      "epoch": 0.0824584844435961,
      "grad_norm": 0.11887920647859573,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 12.4258,
      "step": 108
    },
    {
      "epoch": 0.08322198892918496,
      "grad_norm": 0.1026986688375473,
      "learning_rate": 4.669547078371504e-05,
      "loss": 12.4327,
      "step": 109
    },
    {
      "epoch": 0.0839854934147738,
      "grad_norm": 0.13835877180099487,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 12.4291,
      "step": 110
    },
    {
      "epoch": 0.08474899790036267,
      "grad_norm": 0.07262998074293137,
      "learning_rate": 4.504772348747687e-05,
      "loss": 12.4239,
      "step": 111
    },
    {
      "epoch": 0.08551250238595151,
      "grad_norm": 0.09426653385162354,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 12.4323,
      "step": 112
    },
    {
      "epoch": 0.08627600687154037,
      "grad_norm": 0.11951074004173279,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 12.4242,
      "step": 113
    },
    {
      "epoch": 0.08703951135712923,
      "grad_norm": 0.09134028106927872,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 12.4342,
      "step": 114
    },
    {
      "epoch": 0.08780301584271807,
      "grad_norm": 0.10953062772750854,
      "learning_rate": 4.17702704859633e-05,
      "loss": 12.4261,
      "step": 115
    },
    {
      "epoch": 0.08856652032830693,
      "grad_norm": 0.10271038115024567,
      "learning_rate": 4.095597328339452e-05,
      "loss": 12.4225,
      "step": 116
    },
    {
      "epoch": 0.08933002481389578,
      "grad_norm": 0.10300962626934052,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 12.4248,
      "step": 117
    },
    {
      "epoch": 0.09009352929948464,
      "grad_norm": 0.09402592480182648,
      "learning_rate": 3.933501846281267e-05,
      "loss": 12.428,
      "step": 118
    },
    {
      "epoch": 0.09085703378507348,
      "grad_norm": 0.10680106282234192,
      "learning_rate": 3.852880399766243e-05,
      "loss": 12.4218,
      "step": 119
    },
    {
      "epoch": 0.09162053827066234,
      "grad_norm": 0.12042912095785141,
      "learning_rate": 3.772572564296005e-05,
      "loss": 12.4134,
      "step": 120
    },
    {
      "epoch": 0.09238404275625119,
      "grad_norm": 0.0977160856127739,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 12.4243,
      "step": 121
    },
    {
      "epoch": 0.09314754724184005,
      "grad_norm": 0.11174153536558151,
      "learning_rate": 3.612985456190778e-05,
      "loss": 12.4279,
      "step": 122
    },
    {
      "epoch": 0.0939110517274289,
      "grad_norm": 0.11660799384117126,
      "learning_rate": 3.533749813077677e-05,
      "loss": 12.424,
      "step": 123
    },
    {
      "epoch": 0.09467455621301775,
      "grad_norm": 0.09053874760866165,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 12.4218,
      "step": 124
    },
    {
      "epoch": 0.09543806069860661,
      "grad_norm": 0.10263125598430634,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 12.4221,
      "step": 125
    },
    {
      "epoch": 0.09620156518419545,
      "grad_norm": 0.09530821442604065,
      "learning_rate": 3.298534127791785e-05,
      "loss": 12.4188,
      "step": 126
    },
    {
      "epoch": 0.09696506966978431,
      "grad_norm": 0.1392821967601776,
      "learning_rate": 3.221030765387417e-05,
      "loss": 12.4219,
      "step": 127
    },
    {
      "epoch": 0.09772857415537316,
      "grad_norm": 0.09401389956474304,
      "learning_rate": 3.144013755408895e-05,
      "loss": 12.4279,
      "step": 128
    },
    {
      "epoch": 0.09849207864096202,
      "grad_norm": 0.08276765048503876,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 12.4312,
      "step": 129
    },
    {
      "epoch": 0.09925558312655088,
      "grad_norm": 0.08019660413265228,
      "learning_rate": 2.991522876735154e-05,
      "loss": 12.4252,
      "step": 130
    },
    {
      "epoch": 0.10001908761213972,
      "grad_norm": 0.09544660896062851,
      "learning_rate": 2.916090697523549e-05,
      "loss": 12.4324,
      "step": 131
    },
    {
      "epoch": 0.10078259209772858,
      "grad_norm": 0.07803331315517426,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 12.422,
      "step": 132
    },
    {
      "epoch": 0.10154609658331742,
      "grad_norm": 0.08372296392917633,
      "learning_rate": 2.766955965735968e-05,
      "loss": 12.4197,
      "step": 133
    },
    {
      "epoch": 0.10230960106890628,
      "grad_norm": 0.08578063547611237,
      "learning_rate": 2.693294185106562e-05,
      "loss": 12.4224,
      "step": 134
    },
    {
      "epoch": 0.10307310555449513,
      "grad_norm": 0.10194368660449982,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 12.4256,
      "step": 135
    },
    {
      "epoch": 0.10383661004008399,
      "grad_norm": 0.11023808270692825,
      "learning_rate": 2.547882480847461e-05,
      "loss": 12.4237,
      "step": 136
    },
    {
      "epoch": 0.10460011452567283,
      "grad_norm": 0.0821557492017746,
      "learning_rate": 2.476172311325783e-05,
      "loss": 12.4238,
      "step": 137
    },
    {
      "epoch": 0.10536361901126169,
      "grad_norm": 0.08240222930908203,
      "learning_rate": 2.405152131093926e-05,
      "loss": 12.4204,
      "step": 138
    },
    {
      "epoch": 0.10612712349685055,
      "grad_norm": 0.0834621861577034,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 12.4169,
      "step": 139
    },
    {
      "epoch": 0.1068906279824394,
      "grad_norm": 0.1026996597647667,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 12.4243,
      "step": 140
    },
    {
      "epoch": 0.10765413246802825,
      "grad_norm": 0.08680897951126099,
      "learning_rate": 2.196424713241637e-05,
      "loss": 12.4217,
      "step": 141
    },
    {
      "epoch": 0.1084176369536171,
      "grad_norm": 0.09690609574317932,
      "learning_rate": 2.128356686585282e-05,
      "loss": 12.41,
      "step": 142
    },
    {
      "epoch": 0.10918114143920596,
      "grad_norm": 0.11314906179904938,
      "learning_rate": 2.061073738537635e-05,
      "loss": 12.4134,
      "step": 143
    },
    {
      "epoch": 0.1099446459247948,
      "grad_norm": 0.09332529455423355,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 12.4202,
      "step": 144
    },
    {
      "epoch": 0.11070815041038366,
      "grad_norm": 0.08546222001314163,
      "learning_rate": 1.928936436551661e-05,
      "loss": 12.4167,
      "step": 145
    },
    {
      "epoch": 0.11147165489597251,
      "grad_norm": 0.0985637679696083,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 12.416,
      "step": 146
    },
    {
      "epoch": 0.11223515938156137,
      "grad_norm": 0.09000271558761597,
      "learning_rate": 1.800157297483417e-05,
      "loss": 12.4173,
      "step": 147
    },
    {
      "epoch": 0.11299866386715023,
      "grad_norm": 0.08960099518299103,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 12.4116,
      "step": 148
    },
    {
      "epoch": 0.11376216835273907,
      "grad_norm": 0.10776525735855103,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 12.4184,
      "step": 149
    },
    {
      "epoch": 0.11452567283832793,
      "grad_norm": 0.1175934448838234,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 12.4054,
      "step": 150
    },
    {
      "epoch": 0.11452567283832793,
      "eval_loss": 12.41812515258789,
      "eval_runtime": 15.4741,
      "eval_samples_per_second": 142.561,
      "eval_steps_per_second": 35.673,
      "step": 150
    },
    {
      "epoch": 0.11528917732391677,
      "grad_norm": 0.10583053529262543,
      "learning_rate": 1.553232954407171e-05,
      "loss": 12.422,
      "step": 151
    },
    {
      "epoch": 0.11605268180950563,
      "grad_norm": 0.0869411826133728,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 12.4249,
      "step": 152
    },
    {
      "epoch": 0.11681618629509448,
      "grad_norm": 0.08589145541191101,
      "learning_rate": 1.435357758543015e-05,
      "loss": 12.4202,
      "step": 153
    },
    {
      "epoch": 0.11757969078068334,
      "grad_norm": 0.0950031653046608,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 12.421,
      "step": 154
    },
    {
      "epoch": 0.11834319526627218,
      "grad_norm": 0.08891143649816513,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 12.4146,
      "step": 155
    },
    {
      "epoch": 0.11910669975186104,
      "grad_norm": 0.09742388129234314,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 12.4211,
      "step": 156
    },
    {
      "epoch": 0.1198702042374499,
      "grad_norm": 0.08353932946920395,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 12.4229,
      "step": 157
    },
    {
      "epoch": 0.12063370872303875,
      "grad_norm": 0.08882568776607513,
      "learning_rate": 1.157994445715706e-05,
      "loss": 12.4252,
      "step": 158
    },
    {
      "epoch": 0.1213972132086276,
      "grad_norm": 0.0746406838297844,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 12.4262,
      "step": 159
    },
    {
      "epoch": 0.12216071769421645,
      "grad_norm": 0.07782018929719925,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 12.4302,
      "step": 160
    },
    {
      "epoch": 0.12292422217980531,
      "grad_norm": 0.08453496545553207,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 12.4241,
      "step": 161
    },
    {
      "epoch": 0.12368772666539415,
      "grad_norm": 0.10034674406051636,
      "learning_rate": 9.549150281252633e-06,
      "loss": 12.416,
      "step": 162
    },
    {
      "epoch": 0.12445123115098301,
      "grad_norm": 0.0753684714436531,
      "learning_rate": 9.068759265665384e-06,
      "loss": 12.4148,
      "step": 163
    },
    {
      "epoch": 0.12521473563657187,
      "grad_norm": 0.11089301109313965,
      "learning_rate": 8.599558442598998e-06,
      "loss": 12.417,
      "step": 164
    },
    {
      "epoch": 0.12597824012216072,
      "grad_norm": 0.07608238607645035,
      "learning_rate": 8.141676086873572e-06,
      "loss": 12.4255,
      "step": 165
    },
    {
      "epoch": 0.12674174460774956,
      "grad_norm": 0.07675814628601074,
      "learning_rate": 7.695237378953223e-06,
      "loss": 12.4192,
      "step": 166
    },
    {
      "epoch": 0.12750524909333844,
      "grad_norm": 0.10014967620372772,
      "learning_rate": 7.260364370723044e-06,
      "loss": 12.4222,
      "step": 167
    },
    {
      "epoch": 0.12826875357892728,
      "grad_norm": 0.07093244045972824,
      "learning_rate": 6.837175952121306e-06,
      "loss": 12.4204,
      "step": 168
    },
    {
      "epoch": 0.12903225806451613,
      "grad_norm": 0.10388762503862381,
      "learning_rate": 6.425787818636131e-06,
      "loss": 12.4219,
      "step": 169
    },
    {
      "epoch": 0.12979576255010497,
      "grad_norm": 0.07800211757421494,
      "learning_rate": 6.026312439675552e-06,
      "loss": 12.4153,
      "step": 170
    },
    {
      "epoch": 0.13055926703569384,
      "grad_norm": 0.08672941476106644,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 12.4221,
      "step": 171
    },
    {
      "epoch": 0.1313227715212827,
      "grad_norm": 0.10953150689601898,
      "learning_rate": 5.263533508961827e-06,
      "loss": 12.4127,
      "step": 172
    },
    {
      "epoch": 0.13208627600687153,
      "grad_norm": 0.08263696730136871,
      "learning_rate": 4.900438493352055e-06,
      "loss": 12.4167,
      "step": 173
    },
    {
      "epoch": 0.1328497804924604,
      "grad_norm": 0.11236874014139175,
      "learning_rate": 4.549673247541875e-06,
      "loss": 12.4213,
      "step": 174
    },
    {
      "epoch": 0.13361328497804925,
      "grad_norm": 0.0768594816327095,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 12.4249,
      "step": 175
    },
    {
      "epoch": 0.1343767894636381,
      "grad_norm": 0.08035300672054291,
      "learning_rate": 3.885512251130763e-06,
      "loss": 12.4174,
      "step": 176
    },
    {
      "epoch": 0.13514029394922694,
      "grad_norm": 0.10196048021316528,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 12.4163,
      "step": 177
    },
    {
      "epoch": 0.13590379843481581,
      "grad_norm": 0.08112561702728271,
      "learning_rate": 3.271776770026963e-06,
      "loss": 12.4163,
      "step": 178
    },
    {
      "epoch": 0.13666730292040466,
      "grad_norm": 0.09913696348667145,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 12.4249,
      "step": 179
    },
    {
      "epoch": 0.1374308074059935,
      "grad_norm": 0.08228102326393127,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 12.4138,
      "step": 180
    },
    {
      "epoch": 0.13819431189158235,
      "grad_norm": 0.08755149692296982,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 12.416,
      "step": 181
    },
    {
      "epoch": 0.13895781637717122,
      "grad_norm": 0.08788346499204636,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 12.4195,
      "step": 182
    },
    {
      "epoch": 0.13972132086276007,
      "grad_norm": 0.09075617045164108,
      "learning_rate": 1.962316193157593e-06,
      "loss": 12.4146,
      "step": 183
    },
    {
      "epoch": 0.1404848253483489,
      "grad_norm": 0.09291182458400726,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 12.4252,
      "step": 184
    },
    {
      "epoch": 0.14124832983393779,
      "grad_norm": 0.08399947732686996,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 12.4174,
      "step": 185
    },
    {
      "epoch": 0.14201183431952663,
      "grad_norm": 0.08799798041582108,
      "learning_rate": 1.333670137599713e-06,
      "loss": 12.418,
      "step": 186
    },
    {
      "epoch": 0.14277533880511548,
      "grad_norm": 0.0802348181605339,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 12.4187,
      "step": 187
    },
    {
      "epoch": 0.14353884329070432,
      "grad_norm": 0.08826827257871628,
      "learning_rate": 9.810017062595322e-07,
      "loss": 12.4189,
      "step": 188
    },
    {
      "epoch": 0.1443023477762932,
      "grad_norm": 0.08281069248914719,
      "learning_rate": 8.247462563808817e-07,
      "loss": 12.4103,
      "step": 189
    },
    {
      "epoch": 0.14506585226188204,
      "grad_norm": 0.10556374490261078,
      "learning_rate": 6.819348298638839e-07,
      "loss": 12.4259,
      "step": 190
    },
    {
      "epoch": 0.14582935674747088,
      "grad_norm": 0.09609169512987137,
      "learning_rate": 5.526064699265753e-07,
      "loss": 12.4076,
      "step": 191
    },
    {
      "epoch": 0.14659286123305976,
      "grad_norm": 0.07798979431390762,
      "learning_rate": 4.367965336512403e-07,
      "loss": 12.4095,
      "step": 192
    },
    {
      "epoch": 0.1473563657186486,
      "grad_norm": 0.08129992336034775,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 12.4117,
      "step": 193
    },
    {
      "epoch": 0.14811987020423745,
      "grad_norm": 0.07610326260328293,
      "learning_rate": 2.458548727494292e-07,
      "loss": 12.4112,
      "step": 194
    },
    {
      "epoch": 0.1488833746898263,
      "grad_norm": 0.07794073969125748,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 12.419,
      "step": 195
    },
    {
      "epoch": 0.14964687917541517,
      "grad_norm": 0.09371713548898697,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 12.4181,
      "step": 196
    },
    {
      "epoch": 0.150410383661004,
      "grad_norm": 0.10581816732883453,
      "learning_rate": 6.150154258476315e-08,
      "loss": 12.414,
      "step": 197
    },
    {
      "epoch": 0.15117388814659286,
      "grad_norm": 0.08147760480642319,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 12.4088,
      "step": 198
    },
    {
      "epoch": 0.1519373926321817,
      "grad_norm": 0.09897597134113312,
      "learning_rate": 6.834750376549792e-09,
      "loss": 12.4167,
      "step": 199
    },
    {
      "epoch": 0.15270089711777057,
      "grad_norm": 0.09005050361156464,
      "learning_rate": 0.0,
      "loss": 12.4003,
      "step": 200
    },
    {
      "epoch": 0.15270089711777057,
      "eval_loss": 12.416987419128418,
      "eval_runtime": 15.5227,
      "eval_samples_per_second": 142.114,
      "eval_steps_per_second": 35.561,
      "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": 775189168128.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}