{
  "best_metric": 11.5,
  "best_model_checkpoint": "miner_id_24/checkpoint-50",
  "epoch": 0.08982708286548394,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00044913541432741973,
      "grad_norm": 1.8514176190365106e-05,
      "learning_rate": 5e-06,
      "loss": 46.0,
      "step": 1
    },
    {
      "epoch": 0.00044913541432741973,
      "eval_loss": 11.5,
      "eval_runtime": 24.0375,
      "eval_samples_per_second": 156.006,
      "eval_steps_per_second": 78.003,
      "step": 1
    },
    {
      "epoch": 0.0008982708286548395,
      "grad_norm": 1.2753680493915454e-05,
      "learning_rate": 1e-05,
      "loss": 46.0,
      "step": 2
    },
    {
      "epoch": 0.001347406242982259,
      "grad_norm": 1.82186395250028e-05,
      "learning_rate": 1.5e-05,
      "loss": 46.0,
      "step": 3
    },
    {
      "epoch": 0.001796541657309679,
      "grad_norm": 1.292844899580814e-05,
      "learning_rate": 2e-05,
      "loss": 46.0,
      "step": 4
    },
    {
      "epoch": 0.0022456770716370988,
      "grad_norm": 1.5541028915322386e-05,
      "learning_rate": 2.5e-05,
      "loss": 46.0,
      "step": 5
    },
    {
      "epoch": 0.002694812485964518,
      "grad_norm": 2.8140768336015753e-05,
      "learning_rate": 3e-05,
      "loss": 46.0,
      "step": 6
    },
    {
      "epoch": 0.003143947900291938,
      "grad_norm": 2.1085226762806997e-05,
      "learning_rate": 3.5e-05,
      "loss": 46.0,
      "step": 7
    },
    {
      "epoch": 0.003593083314619358,
      "grad_norm": 1.7194925021613017e-05,
      "learning_rate": 4e-05,
      "loss": 46.0,
      "step": 8
    },
    {
      "epoch": 0.004042218728946777,
      "grad_norm": 1.6402016626670957e-05,
      "learning_rate": 4.5e-05,
      "loss": 46.0,
      "step": 9
    },
    {
      "epoch": 0.0044913541432741975,
      "grad_norm": 1.6930052879615687e-05,
      "learning_rate": 5e-05,
      "loss": 46.0,
      "step": 10
    },
    {
      "epoch": 0.004940489557601617,
      "grad_norm": 1.7235321138286963e-05,
      "learning_rate": 5.500000000000001e-05,
      "loss": 46.0,
      "step": 11
    },
    {
      "epoch": 0.005389624971929036,
      "grad_norm": 1.6744437743909657e-05,
      "learning_rate": 6e-05,
      "loss": 46.0,
      "step": 12
    },
    {
      "epoch": 0.005838760386256457,
      "grad_norm": 1.59424507728545e-05,
      "learning_rate": 6.500000000000001e-05,
      "loss": 46.0,
      "step": 13
    },
    {
      "epoch": 0.006287895800583876,
      "grad_norm": 1.3195422980061267e-05,
      "learning_rate": 7e-05,
      "loss": 46.0,
      "step": 14
    },
    {
      "epoch": 0.0067370312149112955,
      "grad_norm": 1.5616611563018523e-05,
      "learning_rate": 7.500000000000001e-05,
      "loss": 46.0,
      "step": 15
    },
    {
      "epoch": 0.007186166629238716,
      "grad_norm": 1.7075182768166997e-05,
      "learning_rate": 8e-05,
      "loss": 46.0,
      "step": 16
    },
    {
      "epoch": 0.007635302043566135,
      "grad_norm": 1.0311639016435947e-05,
      "learning_rate": 8.5e-05,
      "loss": 46.0,
      "step": 17
    },
    {
      "epoch": 0.008084437457893555,
      "grad_norm": 1.7198648492922075e-05,
      "learning_rate": 9e-05,
      "loss": 46.0,
      "step": 18
    },
    {
      "epoch": 0.008533572872220974,
      "grad_norm": 2.8005266358377412e-05,
      "learning_rate": 9.5e-05,
      "loss": 46.0,
      "step": 19
    },
    {
      "epoch": 0.008982708286548395,
      "grad_norm": 2.2636235371464863e-05,
      "learning_rate": 0.0001,
      "loss": 46.0,
      "step": 20
    },
    {
      "epoch": 0.009431843700875815,
      "grad_norm": 1.8755303244688548e-05,
      "learning_rate": 9.999238475781957e-05,
      "loss": 46.0,
      "step": 21
    },
    {
      "epoch": 0.009880979115203234,
      "grad_norm": 1.3910874258726835e-05,
      "learning_rate": 9.99695413509548e-05,
      "loss": 46.0,
      "step": 22
    },
    {
      "epoch": 0.010330114529530653,
      "grad_norm": 2.400025914539583e-05,
      "learning_rate": 9.99314767377287e-05,
      "loss": 46.0,
      "step": 23
    },
    {
      "epoch": 0.010779249943858073,
      "grad_norm": 1.4387041119334754e-05,
      "learning_rate": 9.987820251299122e-05,
      "loss": 46.0,
      "step": 24
    },
    {
      "epoch": 0.011228385358185492,
      "grad_norm": 1.1802452718256973e-05,
      "learning_rate": 9.980973490458728e-05,
      "loss": 46.0,
      "step": 25
    },
    {
      "epoch": 0.011677520772512913,
      "grad_norm": 1.4674837984784972e-05,
      "learning_rate": 9.972609476841367e-05,
      "loss": 46.0,
      "step": 26
    },
    {
      "epoch": 0.012126656186840333,
      "grad_norm": 1.8615999579196796e-05,
      "learning_rate": 9.962730758206611e-05,
      "loss": 46.0,
      "step": 27
    },
    {
      "epoch": 0.012575791601167752,
      "grad_norm": 2.1074078176752664e-05,
      "learning_rate": 9.951340343707852e-05,
      "loss": 46.0,
      "step": 28
    },
    {
      "epoch": 0.013024927015495171,
      "grad_norm": 1.2332392543612514e-05,
      "learning_rate": 9.938441702975689e-05,
      "loss": 46.0,
      "step": 29
    },
    {
      "epoch": 0.013474062429822591,
      "grad_norm": 2.9620930945384316e-05,
      "learning_rate": 9.924038765061042e-05,
      "loss": 46.0,
      "step": 30
    },
    {
      "epoch": 0.013923197844150012,
      "grad_norm": 1.5677551346016116e-05,
      "learning_rate": 9.908135917238321e-05,
      "loss": 46.0,
      "step": 31
    },
    {
      "epoch": 0.014372333258477431,
      "grad_norm": 2.232118640677072e-05,
      "learning_rate": 9.890738003669029e-05,
      "loss": 46.0,
      "step": 32
    },
    {
      "epoch": 0.01482146867280485,
      "grad_norm": 1.6364008843083866e-05,
      "learning_rate": 9.871850323926177e-05,
      "loss": 46.0,
      "step": 33
    },
    {
      "epoch": 0.01527060408713227,
      "grad_norm": 3.767064117710106e-05,
      "learning_rate": 9.851478631379982e-05,
      "loss": 46.0,
      "step": 34
    },
    {
      "epoch": 0.01571973950145969,
      "grad_norm": 1.9007693481398746e-05,
      "learning_rate": 9.829629131445342e-05,
      "loss": 46.0,
      "step": 35
    },
    {
      "epoch": 0.01616887491578711,
      "grad_norm": 1.6770418369560502e-05,
      "learning_rate": 9.806308479691595e-05,
      "loss": 46.0,
      "step": 36
    },
    {
      "epoch": 0.01661801033011453,
      "grad_norm": 1.8554952475824393e-05,
      "learning_rate": 9.781523779815179e-05,
      "loss": 46.0,
      "step": 37
    },
    {
      "epoch": 0.017067145744441948,
      "grad_norm": 1.7014823242789134e-05,
      "learning_rate": 9.755282581475769e-05,
      "loss": 46.0,
      "step": 38
    },
    {
      "epoch": 0.017516281158769367,
      "grad_norm": 3.424166061449796e-05,
      "learning_rate": 9.727592877996585e-05,
      "loss": 46.0,
      "step": 39
    },
    {
      "epoch": 0.01796541657309679,
      "grad_norm": 1.9532062651705928e-05,
      "learning_rate": 9.698463103929542e-05,
      "loss": 46.0,
      "step": 40
    },
    {
      "epoch": 0.01841455198742421,
      "grad_norm": 1.466324829380028e-05,
      "learning_rate": 9.667902132486009e-05,
      "loss": 46.0,
      "step": 41
    },
    {
      "epoch": 0.01886368740175163,
      "grad_norm": 2.3942853658809327e-05,
      "learning_rate": 9.635919272833938e-05,
      "loss": 46.0,
      "step": 42
    },
    {
      "epoch": 0.01931282281607905,
      "grad_norm": 2.1762407413916662e-05,
      "learning_rate": 9.602524267262203e-05,
      "loss": 46.0,
      "step": 43
    },
    {
      "epoch": 0.019761958230406468,
      "grad_norm": 2.6398172849440016e-05,
      "learning_rate": 9.567727288213005e-05,
      "loss": 46.0,
      "step": 44
    },
    {
      "epoch": 0.020211093644733887,
      "grad_norm": 3.763455606531352e-05,
      "learning_rate": 9.53153893518325e-05,
      "loss": 46.0,
      "step": 45
    },
    {
      "epoch": 0.020660229059061307,
      "grad_norm": 2.3635744582861662e-05,
      "learning_rate": 9.493970231495835e-05,
      "loss": 46.0,
      "step": 46
    },
    {
      "epoch": 0.021109364473388726,
      "grad_norm": 5.192641765461303e-05,
      "learning_rate": 9.45503262094184e-05,
      "loss": 46.0,
      "step": 47
    },
    {
      "epoch": 0.021558499887716145,
      "grad_norm": 2.6016559786512516e-05,
      "learning_rate": 9.414737964294636e-05,
      "loss": 46.0,
      "step": 48
    },
    {
      "epoch": 0.022007635302043565,
      "grad_norm": 3.0972663807915524e-05,
      "learning_rate": 9.373098535696979e-05,
      "loss": 46.0,
      "step": 49
    },
    {
      "epoch": 0.022456770716370984,
      "grad_norm": 4.369992529973388e-05,
      "learning_rate": 9.330127018922194e-05,
      "loss": 46.0,
      "step": 50
    },
    {
      "epoch": 0.022456770716370984,
      "eval_loss": 11.5,
      "eval_runtime": 24.1201,
      "eval_samples_per_second": 155.472,
      "eval_steps_per_second": 77.736,
      "step": 50
    },
    {
      "epoch": 0.022905906130698407,
      "grad_norm": 1.5684625395806506e-05,
      "learning_rate": 9.285836503510562e-05,
      "loss": 46.0,
      "step": 51
    },
    {
      "epoch": 0.023355041545025827,
      "grad_norm": 1.1430732229200657e-05,
      "learning_rate": 9.24024048078213e-05,
      "loss": 46.0,
      "step": 52
    },
    {
      "epoch": 0.023804176959353246,
      "grad_norm": 2.078129182336852e-05,
      "learning_rate": 9.193352839727121e-05,
      "loss": 46.0,
      "step": 53
    },
    {
      "epoch": 0.024253312373680665,
      "grad_norm": 2.5568606361048296e-05,
      "learning_rate": 9.145187862775209e-05,
      "loss": 46.0,
      "step": 54
    },
    {
      "epoch": 0.024702447788008085,
      "grad_norm": 3.224593820050359e-05,
      "learning_rate": 9.09576022144496e-05,
      "loss": 46.0,
      "step": 55
    },
    {
      "epoch": 0.025151583202335504,
      "grad_norm": 3.624117380240932e-05,
      "learning_rate": 9.045084971874738e-05,
      "loss": 46.0,
      "step": 56
    },
    {
      "epoch": 0.025600718616662924,
      "grad_norm": 2.991443579958286e-05,
      "learning_rate": 8.993177550236464e-05,
      "loss": 46.0,
      "step": 57
    },
    {
      "epoch": 0.026049854030990343,
      "grad_norm": 2.87011826003436e-05,
      "learning_rate": 8.940053768033609e-05,
      "loss": 46.0,
      "step": 58
    },
    {
      "epoch": 0.026498989445317762,
      "grad_norm": 3.15192955895327e-05,
      "learning_rate": 8.885729807284856e-05,
      "loss": 46.0,
      "step": 59
    },
    {
      "epoch": 0.026948124859645182,
      "grad_norm": 2.3219294234877452e-05,
      "learning_rate": 8.83022221559489e-05,
      "loss": 46.0,
      "step": 60
    },
    {
      "epoch": 0.0273972602739726,
      "grad_norm": 4.23833989771083e-05,
      "learning_rate": 8.773547901113862e-05,
      "loss": 46.0,
      "step": 61
    },
    {
      "epoch": 0.027846395688300024,
      "grad_norm": 5.978866829536855e-05,
      "learning_rate": 8.715724127386972e-05,
      "loss": 46.0,
      "step": 62
    },
    {
      "epoch": 0.028295531102627444,
      "grad_norm": 2.4494316676282324e-05,
      "learning_rate": 8.656768508095853e-05,
      "loss": 46.0,
      "step": 63
    },
    {
      "epoch": 0.028744666516954863,
      "grad_norm": 2.171064625144936e-05,
      "learning_rate": 8.596699001693255e-05,
      "loss": 46.0,
      "step": 64
    },
    {
      "epoch": 0.029193801931282282,
      "grad_norm": 3.84969862352591e-05,
      "learning_rate": 8.535533905932738e-05,
      "loss": 46.0,
      "step": 65
    },
    {
      "epoch": 0.0296429373456097,
      "grad_norm": 3.173514414811507e-05,
      "learning_rate": 8.473291852294987e-05,
      "loss": 46.0,
      "step": 66
    },
    {
      "epoch": 0.03009207275993712,
      "grad_norm": 3.05586181639228e-05,
      "learning_rate": 8.409991800312493e-05,
      "loss": 46.0,
      "step": 67
    },
    {
      "epoch": 0.03054120817426454,
      "grad_norm": 3.339080649311654e-05,
      "learning_rate": 8.345653031794292e-05,
      "loss": 46.0,
      "step": 68
    },
    {
      "epoch": 0.03099034358859196,
      "grad_norm": 3.156225648126565e-05,
      "learning_rate": 8.280295144952536e-05,
      "loss": 46.0,
      "step": 69
    },
    {
      "epoch": 0.03143947900291938,
      "grad_norm": 4.123676990275271e-05,
      "learning_rate": 8.213938048432697e-05,
      "loss": 46.0,
      "step": 70
    },
    {
      "epoch": 0.0318886144172468,
      "grad_norm": 3.437064151512459e-05,
      "learning_rate": 8.146601955249188e-05,
      "loss": 46.0,
      "step": 71
    },
    {
      "epoch": 0.03233774983157422,
      "grad_norm": 4.116688796784729e-05,
      "learning_rate": 8.07830737662829e-05,
      "loss": 46.0,
      "step": 72
    },
    {
      "epoch": 0.03278688524590164,
      "grad_norm": 5.3466937970370054e-05,
      "learning_rate": 8.009075115760243e-05,
      "loss": 46.0,
      "step": 73
    },
    {
      "epoch": 0.03323602066022906,
      "grad_norm": 2.7689307898981497e-05,
      "learning_rate": 7.938926261462366e-05,
      "loss": 46.0,
      "step": 74
    },
    {
      "epoch": 0.03368515607455648,
      "grad_norm": 6.431225483538583e-05,
      "learning_rate": 7.86788218175523e-05,
      "loss": 46.0,
      "step": 75
    },
    {
      "epoch": 0.034134291488883896,
      "grad_norm": 4.751375308842398e-05,
      "learning_rate": 7.795964517353735e-05,
      "loss": 46.0,
      "step": 76
    },
    {
      "epoch": 0.03458342690321132,
      "grad_norm": 3.683446630020626e-05,
      "learning_rate": 7.723195175075136e-05,
      "loss": 46.0,
      "step": 77
    },
    {
      "epoch": 0.035032562317538735,
      "grad_norm": 8.397875353693962e-05,
      "learning_rate": 7.649596321166024e-05,
      "loss": 46.0,
      "step": 78
    },
    {
      "epoch": 0.03548169773186616,
      "grad_norm": 3.1538533221464604e-05,
      "learning_rate": 7.575190374550272e-05,
      "loss": 46.0,
      "step": 79
    },
    {
      "epoch": 0.03593083314619358,
      "grad_norm": 3.556596129783429e-05,
      "learning_rate": 7.500000000000001e-05,
      "loss": 46.0,
      "step": 80
    },
    {
      "epoch": 0.036379968560520996,
      "grad_norm": 5.4827927669975907e-05,
      "learning_rate": 7.424048101231686e-05,
      "loss": 46.0,
      "step": 81
    },
    {
      "epoch": 0.03682910397484842,
      "grad_norm": 5.89424635109026e-05,
      "learning_rate": 7.347357813929454e-05,
      "loss": 46.0,
      "step": 82
    },
    {
      "epoch": 0.037278239389175835,
      "grad_norm": 2.519952249713242e-05,
      "learning_rate": 7.269952498697734e-05,
      "loss": 46.0,
      "step": 83
    },
    {
      "epoch": 0.03772737480350326,
      "grad_norm": 5.645597047987394e-05,
      "learning_rate": 7.191855733945387e-05,
      "loss": 46.0,
      "step": 84
    },
    {
      "epoch": 0.038176510217830674,
      "grad_norm": 6.161312921904027e-05,
      "learning_rate": 7.113091308703498e-05,
      "loss": 46.0,
      "step": 85
    },
    {
      "epoch": 0.0386256456321581,
      "grad_norm": 9.112574480241165e-05,
      "learning_rate": 7.033683215379002e-05,
      "loss": 46.0,
      "step": 86
    },
    {
      "epoch": 0.03907478104648551,
      "grad_norm": 9.02749306987971e-05,
      "learning_rate": 6.953655642446368e-05,
      "loss": 46.0,
      "step": 87
    },
    {
      "epoch": 0.039523916460812936,
      "grad_norm": 9.279652294935659e-05,
      "learning_rate": 6.873032967079561e-05,
      "loss": 46.0,
      "step": 88
    },
    {
      "epoch": 0.03997305187514035,
      "grad_norm": 6.995837611611933e-05,
      "learning_rate": 6.7918397477265e-05,
      "loss": 46.0,
      "step": 89
    },
    {
      "epoch": 0.040422187289467774,
      "grad_norm": 5.376504122978076e-05,
      "learning_rate": 6.710100716628344e-05,
      "loss": 46.0,
      "step": 90
    },
    {
      "epoch": 0.0408713227037952,
      "grad_norm": 4.237423854647204e-05,
      "learning_rate": 6.627840772285784e-05,
      "loss": 46.0,
      "step": 91
    },
    {
      "epoch": 0.04132045811812261,
      "grad_norm": 7.658038521185517e-05,
      "learning_rate": 6.545084971874738e-05,
      "loss": 46.0,
      "step": 92
    },
    {
      "epoch": 0.041769593532450036,
      "grad_norm": 5.848429645993747e-05,
      "learning_rate": 6.461858523613684e-05,
      "loss": 46.0,
      "step": 93
    },
    {
      "epoch": 0.04221872894677745,
      "grad_norm": 7.184476271504536e-05,
      "learning_rate": 6.378186779084995e-05,
      "loss": 46.0,
      "step": 94
    },
    {
      "epoch": 0.042667864361104875,
      "grad_norm": 6.029445648891851e-05,
      "learning_rate": 6.294095225512603e-05,
      "loss": 46.0,
      "step": 95
    },
    {
      "epoch": 0.04311699977543229,
      "grad_norm": 7.347081555053592e-05,
      "learning_rate": 6.209609477998338e-05,
      "loss": 46.0,
      "step": 96
    },
    {
      "epoch": 0.043566135189759714,
      "grad_norm": 7.249767804751173e-05,
      "learning_rate": 6.124755271719325e-05,
      "loss": 46.0,
      "step": 97
    },
    {
      "epoch": 0.04401527060408713,
      "grad_norm": 4.366773646324873e-05,
      "learning_rate": 6.0395584540887963e-05,
      "loss": 46.0,
      "step": 98
    },
    {
      "epoch": 0.04446440601841455,
      "grad_norm": 9.102965850615874e-05,
      "learning_rate": 5.9540449768827246e-05,
      "loss": 46.0,
      "step": 99
    },
    {
      "epoch": 0.04491354143274197,
      "grad_norm": 6.002643931424245e-05,
      "learning_rate": 5.868240888334653e-05,
      "loss": 46.0,
      "step": 100
    },
    {
      "epoch": 0.04491354143274197,
      "eval_loss": 11.5,
      "eval_runtime": 24.1144,
      "eval_samples_per_second": 155.509,
      "eval_steps_per_second": 77.754,
      "step": 100
    },
    {
      "epoch": 0.04536267684706939,
      "grad_norm": 6.438219133997336e-05,
      "learning_rate": 5.782172325201155e-05,
      "loss": 46.0,
      "step": 101
    },
    {
      "epoch": 0.045811812261396814,
      "grad_norm": 8.957271347753704e-05,
      "learning_rate": 5.695865504800327e-05,
      "loss": 46.0,
      "step": 102
    },
    {
      "epoch": 0.04626094767572423,
      "grad_norm": 5.1923350838478655e-05,
      "learning_rate": 5.6093467170257374e-05,
      "loss": 46.0,
      "step": 103
    },
    {
      "epoch": 0.04671008309005165,
      "grad_norm": 5.0214355724165216e-05,
      "learning_rate": 5.522642316338268e-05,
      "loss": 46.0,
      "step": 104
    },
    {
      "epoch": 0.04715921850437907,
      "grad_norm": 0.00014862336684018373,
      "learning_rate": 5.435778713738292e-05,
      "loss": 46.0,
      "step": 105
    },
    {
      "epoch": 0.04760835391870649,
      "grad_norm": 4.384276326163672e-05,
      "learning_rate": 5.348782368720626e-05,
      "loss": 46.0,
      "step": 106
    },
    {
      "epoch": 0.04805748933303391,
      "grad_norm": 5.50606164324563e-05,
      "learning_rate": 5.26167978121472e-05,
      "loss": 46.0,
      "step": 107
    },
    {
      "epoch": 0.04850662474736133,
      "grad_norm": 5.2447150665102527e-05,
      "learning_rate": 5.174497483512506e-05,
      "loss": 46.0,
      "step": 108
    },
    {
      "epoch": 0.04895576016168875,
      "grad_norm": 0.0001907071127789095,
      "learning_rate": 5.0872620321864185e-05,
      "loss": 46.0,
      "step": 109
    },
    {
      "epoch": 0.04940489557601617,
      "grad_norm": 0.00011432369501562789,
      "learning_rate": 5e-05,
      "loss": 46.0,
      "step": 110
    },
    {
      "epoch": 0.049854030990343585,
      "grad_norm": 3.465030749794096e-05,
      "learning_rate": 4.912737967813583e-05,
      "loss": 46.0,
      "step": 111
    },
    {
      "epoch": 0.05030316640467101,
      "grad_norm": 8.806870027910918e-05,
      "learning_rate": 4.825502516487497e-05,
      "loss": 46.0,
      "step": 112
    },
    {
      "epoch": 0.05075230181899843,
      "grad_norm": 4.049712879350409e-05,
      "learning_rate": 4.738320218785281e-05,
      "loss": 46.0,
      "step": 113
    },
    {
      "epoch": 0.05120143723332585,
      "grad_norm": 7.87384487921372e-05,
      "learning_rate": 4.6512176312793736e-05,
      "loss": 46.0,
      "step": 114
    },
    {
      "epoch": 0.05165057264765327,
      "grad_norm": 3.910386294592172e-05,
      "learning_rate": 4.564221286261709e-05,
      "loss": 46.0,
      "step": 115
    },
    {
      "epoch": 0.052099708061980686,
      "grad_norm": 0.0001285390608245507,
      "learning_rate": 4.477357683661734e-05,
      "loss": 46.0,
      "step": 116
    },
    {
      "epoch": 0.05254884347630811,
      "grad_norm": 4.592915138346143e-05,
      "learning_rate": 4.390653282974264e-05,
      "loss": 46.0,
      "step": 117
    },
    {
      "epoch": 0.052997978890635525,
      "grad_norm": 3.571418710635044e-05,
      "learning_rate": 4.3041344951996746e-05,
      "loss": 46.0,
      "step": 118
    },
    {
      "epoch": 0.05344711430496295,
      "grad_norm": 0.00014030133024789393,
      "learning_rate": 4.2178276747988446e-05,
      "loss": 46.0,
      "step": 119
    },
    {
      "epoch": 0.053896249719290364,
      "grad_norm": 4.988265936844982e-05,
      "learning_rate": 4.131759111665349e-05,
      "loss": 46.0,
      "step": 120
    },
    {
      "epoch": 0.054345385133617786,
      "grad_norm": 6.856449908809736e-05,
      "learning_rate": 4.045955023117276e-05,
      "loss": 46.0,
      "step": 121
    },
    {
      "epoch": 0.0547945205479452,
      "grad_norm": 9.08086949493736e-05,
      "learning_rate": 3.960441545911204e-05,
      "loss": 46.0,
      "step": 122
    },
    {
      "epoch": 0.055243655962272625,
      "grad_norm": 7.736255065537989e-05,
      "learning_rate": 3.875244728280676e-05,
      "loss": 46.0,
      "step": 123
    },
    {
      "epoch": 0.05569279137660005,
      "grad_norm": 8.342680666828528e-05,
      "learning_rate": 3.790390522001662e-05,
      "loss": 46.0,
      "step": 124
    },
    {
      "epoch": 0.056141926790927464,
      "grad_norm": 0.00010806278442032635,
      "learning_rate": 3.705904774487396e-05,
      "loss": 46.0,
      "step": 125
    },
    {
      "epoch": 0.05659106220525489,
      "grad_norm": 8.688573871040717e-05,
      "learning_rate": 3.6218132209150045e-05,
      "loss": 46.0,
      "step": 126
    },
    {
      "epoch": 0.0570401976195823,
      "grad_norm": 8.21120775071904e-05,
      "learning_rate": 3.5381414763863166e-05,
      "loss": 46.0,
      "step": 127
    },
    {
      "epoch": 0.057489333033909726,
      "grad_norm": 8.550721395295113e-05,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 46.0,
      "step": 128
    },
    {
      "epoch": 0.05793846844823714,
      "grad_norm": 0.00010507808474358171,
      "learning_rate": 3.372159227714218e-05,
      "loss": 46.0,
      "step": 129
    },
    {
      "epoch": 0.058387603862564565,
      "grad_norm": 0.00012020936992485076,
      "learning_rate": 3.289899283371657e-05,
      "loss": 46.0,
      "step": 130
    },
    {
      "epoch": 0.05883673927689198,
      "grad_norm": 0.00010288711200701073,
      "learning_rate": 3.2081602522734986e-05,
      "loss": 46.0,
      "step": 131
    },
    {
      "epoch": 0.0592858746912194,
      "grad_norm": 0.00016522681107744575,
      "learning_rate": 3.12696703292044e-05,
      "loss": 46.0,
      "step": 132
    },
    {
      "epoch": 0.05973501010554682,
      "grad_norm": 0.0001353459374513477,
      "learning_rate": 3.046344357553632e-05,
      "loss": 46.0,
      "step": 133
    },
    {
      "epoch": 0.06018414551987424,
      "grad_norm": 8.348548726644367e-05,
      "learning_rate": 2.9663167846209998e-05,
      "loss": 46.0,
      "step": 134
    },
    {
      "epoch": 0.060633280934201665,
      "grad_norm": 0.00014159231795929372,
      "learning_rate": 2.886908691296504e-05,
      "loss": 46.0,
      "step": 135
    },
    {
      "epoch": 0.06108241634852908,
      "grad_norm": 8.296170562971383e-05,
      "learning_rate": 2.8081442660546125e-05,
      "loss": 46.0,
      "step": 136
    },
    {
      "epoch": 0.061531551762856504,
      "grad_norm": 0.00014167738845571876,
      "learning_rate": 2.7300475013022663e-05,
      "loss": 46.0,
      "step": 137
    },
    {
      "epoch": 0.06198068717718392,
      "grad_norm": 9.648062405176461e-05,
      "learning_rate": 2.6526421860705473e-05,
      "loss": 46.0,
      "step": 138
    },
    {
      "epoch": 0.06242982259151134,
      "grad_norm": 9.932834655046463e-05,
      "learning_rate": 2.575951898768315e-05,
      "loss": 46.0,
      "step": 139
    },
    {
      "epoch": 0.06287895800583876,
      "grad_norm": 7.478307816199958e-05,
      "learning_rate": 2.500000000000001e-05,
      "loss": 46.0,
      "step": 140
    },
    {
      "epoch": 0.06332809342016618,
      "grad_norm": 0.00010552546882536262,
      "learning_rate": 2.4248096254497288e-05,
      "loss": 46.0,
      "step": 141
    },
    {
      "epoch": 0.0637772288344936,
      "grad_norm": 9.900185978040099e-05,
      "learning_rate": 2.350403678833976e-05,
      "loss": 46.0,
      "step": 142
    },
    {
      "epoch": 0.06422636424882101,
      "grad_norm": 6.973473500693217e-05,
      "learning_rate": 2.2768048249248648e-05,
      "loss": 46.0,
      "step": 143
    },
    {
      "epoch": 0.06467549966314844,
      "grad_norm": 0.0001482782536186278,
      "learning_rate": 2.2040354826462668e-05,
      "loss": 46.0,
      "step": 144
    },
    {
      "epoch": 0.06512463507747586,
      "grad_norm": 8.450460154563189e-05,
      "learning_rate": 2.132117818244771e-05,
      "loss": 46.0,
      "step": 145
    },
    {
      "epoch": 0.06557377049180328,
      "grad_norm": 9.04243643162772e-05,
      "learning_rate": 2.061073738537635e-05,
      "loss": 46.0,
      "step": 146
    },
    {
      "epoch": 0.0660229059061307,
      "grad_norm": 6.251658487599343e-05,
      "learning_rate": 1.9909248842397584e-05,
      "loss": 46.0,
      "step": 147
    },
    {
      "epoch": 0.06647204132045811,
      "grad_norm": 0.0001046026882249862,
      "learning_rate": 1.9216926233717085e-05,
      "loss": 46.0,
      "step": 148
    },
    {
      "epoch": 0.06692117673478554,
      "grad_norm": 5.9992311435053125e-05,
      "learning_rate": 1.8533980447508137e-05,
      "loss": 46.0,
      "step": 149
    },
    {
      "epoch": 0.06737031214911296,
      "grad_norm": 0.00010509788990020752,
      "learning_rate": 1.7860619515673033e-05,
      "loss": 46.0,
      "step": 150
    },
    {
      "epoch": 0.06737031214911296,
      "eval_loss": 11.5,
      "eval_runtime": 24.1121,
      "eval_samples_per_second": 155.523,
      "eval_steps_per_second": 77.762,
      "step": 150
    },
    {
      "epoch": 0.06781944756344038,
      "grad_norm": 0.0001453920267522335,
      "learning_rate": 1.7197048550474643e-05,
      "loss": 46.0,
      "step": 151
    },
    {
      "epoch": 0.06826858297776779,
      "grad_norm": 7.71365812397562e-05,
      "learning_rate": 1.6543469682057106e-05,
      "loss": 46.0,
      "step": 152
    },
    {
      "epoch": 0.06871771839209521,
      "grad_norm": 9.613798465579748e-05,
      "learning_rate": 1.5900081996875083e-05,
      "loss": 46.0,
      "step": 153
    },
    {
      "epoch": 0.06916685380642264,
      "grad_norm": 5.726591552956961e-05,
      "learning_rate": 1.526708147705013e-05,
      "loss": 46.0,
      "step": 154
    },
    {
      "epoch": 0.06961598922075006,
      "grad_norm": 6.774827488698065e-05,
      "learning_rate": 1.4644660940672627e-05,
      "loss": 46.0,
      "step": 155
    },
    {
      "epoch": 0.07006512463507747,
      "grad_norm": 0.0001335082488367334,
      "learning_rate": 1.4033009983067452e-05,
      "loss": 46.0,
      "step": 156
    },
    {
      "epoch": 0.07051426004940489,
      "grad_norm": 9.170570410788059e-05,
      "learning_rate": 1.3432314919041478e-05,
      "loss": 46.0,
      "step": 157
    },
    {
      "epoch": 0.07096339546373231,
      "grad_norm": 9.498088184045628e-05,
      "learning_rate": 1.2842758726130283e-05,
      "loss": 46.0,
      "step": 158
    },
    {
      "epoch": 0.07141253087805974,
      "grad_norm": 0.00011748191900551319,
      "learning_rate": 1.22645209888614e-05,
      "loss": 46.0,
      "step": 159
    },
    {
      "epoch": 0.07186166629238716,
      "grad_norm": 0.00010865752847166732,
      "learning_rate": 1.1697777844051105e-05,
      "loss": 46.0,
      "step": 160
    },
    {
      "epoch": 0.07231080170671457,
      "grad_norm": 5.8741759858094156e-05,
      "learning_rate": 1.1142701927151456e-05,
      "loss": 46.0,
      "step": 161
    },
    {
      "epoch": 0.07275993712104199,
      "grad_norm": 7.200761319836602e-05,
      "learning_rate": 1.0599462319663905e-05,
      "loss": 46.0,
      "step": 162
    },
    {
      "epoch": 0.07320907253536942,
      "grad_norm": 6.78181168041192e-05,
      "learning_rate": 1.006822449763537e-05,
      "loss": 46.0,
      "step": 163
    },
    {
      "epoch": 0.07365820794969684,
      "grad_norm": 7.606553117511794e-05,
      "learning_rate": 9.549150281252633e-06,
      "loss": 46.0,
      "step": 164
    },
    {
      "epoch": 0.07410734336402425,
      "grad_norm": 0.0001572423498146236,
      "learning_rate": 9.042397785550405e-06,
      "loss": 46.0,
      "step": 165
    },
    {
      "epoch": 0.07455647877835167,
      "grad_norm": 8.567572513129562e-05,
      "learning_rate": 8.548121372247918e-06,
      "loss": 46.0,
      "step": 166
    },
    {
      "epoch": 0.0750056141926791,
      "grad_norm": 0.00012146326480433345,
      "learning_rate": 8.066471602728803e-06,
      "loss": 46.0,
      "step": 167
    },
    {
      "epoch": 0.07545474960700652,
      "grad_norm": 8.176415576599538e-05,
      "learning_rate": 7.597595192178702e-06,
      "loss": 46.0,
      "step": 168
    },
    {
      "epoch": 0.07590388502133394,
      "grad_norm": 5.866547871846706e-05,
      "learning_rate": 7.1416349648943894e-06,
      "loss": 46.0,
      "step": 169
    },
    {
      "epoch": 0.07635302043566135,
      "grad_norm": 0.00011597013508435339,
      "learning_rate": 6.698729810778065e-06,
      "loss": 46.0,
      "step": 170
    },
    {
      "epoch": 0.07680215584998877,
      "grad_norm": 7.273782830452546e-05,
      "learning_rate": 6.269014643030213e-06,
      "loss": 46.0,
      "step": 171
    },
    {
      "epoch": 0.0772512912643162,
      "grad_norm": 0.00013684466830454767,
      "learning_rate": 5.852620357053651e-06,
      "loss": 46.0,
      "step": 172
    },
    {
      "epoch": 0.07770042667864362,
      "grad_norm": 7.729553180979565e-05,
      "learning_rate": 5.449673790581611e-06,
      "loss": 46.0,
      "step": 173
    },
    {
      "epoch": 0.07814956209297103,
      "grad_norm": 8.208224608097225e-05,
      "learning_rate": 5.060297685041659e-06,
      "loss": 46.0,
      "step": 174
    },
    {
      "epoch": 0.07859869750729845,
      "grad_norm": 9.886892803478986e-05,
      "learning_rate": 4.684610648167503e-06,
      "loss": 46.0,
      "step": 175
    },
    {
      "epoch": 0.07904783292162587,
      "grad_norm": 6.48939167149365e-05,
      "learning_rate": 4.322727117869951e-06,
      "loss": 46.0,
      "step": 176
    },
    {
      "epoch": 0.0794969683359533,
      "grad_norm": 0.0001271312212338671,
      "learning_rate": 3.974757327377981e-06,
      "loss": 46.0,
      "step": 177
    },
    {
      "epoch": 0.0799461037502807,
      "grad_norm": 9.381349809700623e-05,
      "learning_rate": 3.6408072716606346e-06,
      "loss": 46.0,
      "step": 178
    },
    {
      "epoch": 0.08039523916460813,
      "grad_norm": 8.31894067232497e-05,
      "learning_rate": 3.3209786751399187e-06,
      "loss": 46.0,
      "step": 179
    },
    {
      "epoch": 0.08084437457893555,
      "grad_norm": 5.7315708545502275e-05,
      "learning_rate": 3.0153689607045845e-06,
      "loss": 46.0,
      "step": 180
    },
    {
      "epoch": 0.08129350999326297,
      "grad_norm": 0.00012303532275836915,
      "learning_rate": 2.724071220034158e-06,
      "loss": 46.0,
      "step": 181
    },
    {
      "epoch": 0.0817426454075904,
      "grad_norm": 6.052962635294534e-05,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 46.0,
      "step": 182
    },
    {
      "epoch": 0.0821917808219178,
      "grad_norm": 0.00014605565229430795,
      "learning_rate": 2.1847622018482283e-06,
      "loss": 46.0,
      "step": 183
    },
    {
      "epoch": 0.08264091623624523,
      "grad_norm": 0.00015781601541675627,
      "learning_rate": 1.9369152030840556e-06,
      "loss": 46.0,
      "step": 184
    },
    {
      "epoch": 0.08309005165057265,
      "grad_norm": 0.00015502123278565705,
      "learning_rate": 1.70370868554659e-06,
      "loss": 46.0,
      "step": 185
    },
    {
      "epoch": 0.08353918706490007,
      "grad_norm": 0.00013865214714314789,
      "learning_rate": 1.4852136862001764e-06,
      "loss": 46.0,
      "step": 186
    },
    {
      "epoch": 0.08398832247922748,
      "grad_norm": 5.323368532117456e-05,
      "learning_rate": 1.2814967607382432e-06,
      "loss": 46.0,
      "step": 187
    },
    {
      "epoch": 0.0844374578935549,
      "grad_norm": 0.00010967773414449766,
      "learning_rate": 1.0926199633097157e-06,
      "loss": 46.0,
      "step": 188
    },
    {
      "epoch": 0.08488659330788233,
      "grad_norm": 0.0001501618098700419,
      "learning_rate": 9.186408276168013e-07,
      "loss": 46.0,
      "step": 189
    },
    {
      "epoch": 0.08533572872220975,
      "grad_norm": 0.00011096147500211373,
      "learning_rate": 7.596123493895991e-07,
      "loss": 46.0,
      "step": 190
    },
    {
      "epoch": 0.08578486413653717,
      "grad_norm": 7.873102003941312e-05,
      "learning_rate": 6.15582970243117e-07,
      "loss": 46.0,
      "step": 191
    },
    {
      "epoch": 0.08623399955086458,
      "grad_norm": 3.765293149626814e-05,
      "learning_rate": 4.865965629214819e-07,
      "loss": 46.0,
      "step": 192
    },
    {
      "epoch": 0.086683134965192,
      "grad_norm": 0.00012500130105763674,
      "learning_rate": 3.7269241793390085e-07,
      "loss": 46.0,
      "step": 193
    },
    {
      "epoch": 0.08713227037951943,
      "grad_norm": 8.419479127041996e-05,
      "learning_rate": 2.7390523158633554e-07,
      "loss": 46.0,
      "step": 194
    },
    {
      "epoch": 0.08758140579384685,
      "grad_norm": 0.00010333299724152312,
      "learning_rate": 1.9026509541272275e-07,
      "loss": 46.0,
      "step": 195
    },
    {
      "epoch": 0.08803054120817426,
      "grad_norm": 9.654317545937374e-05,
      "learning_rate": 1.2179748700879012e-07,
      "loss": 46.0,
      "step": 196
    },
    {
      "epoch": 0.08847967662250168,
      "grad_norm": 0.0001630932092666626,
      "learning_rate": 6.852326227130834e-08,
      "loss": 46.0,
      "step": 197
    },
    {
      "epoch": 0.0889288120368291,
      "grad_norm": 0.00019368848006706685,
      "learning_rate": 3.04586490452119e-08,
      "loss": 46.0,
      "step": 198
    },
    {
      "epoch": 0.08937794745115653,
      "grad_norm": 8.711022383067757e-05,
      "learning_rate": 7.615242180436522e-09,
      "loss": 46.0,
      "step": 199
    },
    {
      "epoch": 0.08982708286548394,
      "grad_norm": 0.0001318210270255804,
      "learning_rate": 0.0,
      "loss": 46.0,
      "step": 200
    },
    {
      "epoch": 0.08982708286548394,
      "eval_loss": 11.5,
      "eval_runtime": 24.1244,
      "eval_samples_per_second": 155.444,
      "eval_steps_per_second": 77.722,
      "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": 3
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 32399651241984.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}