{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9981779291768993,
  "global_step": 126000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 4.999342470094273e-05,
      "loss": 0.0915,
      "step": 500
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998682299506193e-05,
      "loss": 0.0476,
      "step": 1000
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998022128918113e-05,
      "loss": 0.0439,
      "step": 1500
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997361958330033e-05,
      "loss": 0.0416,
      "step": 2000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996701787741953e-05,
      "loss": 0.04,
      "step": 2500
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996041617153873e-05,
      "loss": 0.0388,
      "step": 3000
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.995381446565793e-05,
      "loss": 0.0384,
      "step": 3500
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.994721275977713e-05,
      "loss": 0.037,
      "step": 4000
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.994061105389633e-05,
      "loss": 0.0362,
      "step": 4500
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.993400934801553e-05,
      "loss": 0.0359,
      "step": 5000
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9927407642134734e-05,
      "loss": 0.0353,
      "step": 5500
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.992080593625393e-05,
      "loss": 0.0349,
      "step": 6000
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.991420423037313e-05,
      "loss": 0.0348,
      "step": 6500
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.990760252449233e-05,
      "loss": 0.0338,
      "step": 7000
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.990100081861153e-05,
      "loss": 0.0332,
      "step": 7500
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.989439911273073e-05,
      "loss": 0.0332,
      "step": 8000
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9887797406849934e-05,
      "loss": 0.0331,
      "step": 8500
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9881195700969135e-05,
      "loss": 0.0325,
      "step": 9000
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.987459399508833e-05,
      "loss": 0.0322,
      "step": 9500
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.986799228920753e-05,
      "loss": 0.0321,
      "step": 10000
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.986139058332673e-05,
      "loss": 0.0318,
      "step": 10500
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.985478887744593e-05,
      "loss": 0.0317,
      "step": 11000
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9848187171565134e-05,
      "loss": 0.0313,
      "step": 11500
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.9841585465684335e-05,
      "loss": 0.0312,
      "step": 12000
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.9834983759803536e-05,
      "loss": 0.031,
      "step": 12500
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.982838205392274e-05,
      "loss": 0.0307,
      "step": 13000
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.982178034804194e-05,
      "loss": 0.0306,
      "step": 13500
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.981517864216114e-05,
      "loss": 0.0301,
      "step": 14000
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.980859013969209e-05,
      "loss": 0.0301,
      "step": 14500
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.9801988433811294e-05,
      "loss": 0.0301,
      "step": 15000
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.979539993134226e-05,
      "loss": 0.0301,
      "step": 15500
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.978879822546146e-05,
      "loss": 0.0298,
      "step": 16000
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.978219651958066e-05,
      "loss": 0.0298,
      "step": 16500
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.977559481369986e-05,
      "loss": 0.0294,
      "step": 17000
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.976899310781906e-05,
      "loss": 0.0293,
      "step": 17500
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.976239140193826e-05,
      "loss": 0.0294,
      "step": 18000
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.975578969605746e-05,
      "loss": 0.0292,
      "step": 18500
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.974918799017666e-05,
      "loss": 0.0293,
      "step": 19000
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.9742586284295864e-05,
      "loss": 0.029,
      "step": 19500
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.9735984578415065e-05,
      "loss": 0.0288,
      "step": 20000
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.9729396075946025e-05,
      "loss": 0.0286,
      "step": 20500
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.9722794370065227e-05,
      "loss": 0.0289,
      "step": 21000
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.971619266418443e-05,
      "loss": 0.0285,
      "step": 21500
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.970959095830363e-05,
      "loss": 0.0287,
      "step": 22000
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.970298925242283e-05,
      "loss": 0.0284,
      "step": 22500
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.969641395336555e-05,
      "loss": 0.0284,
      "step": 23000
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.968981224748475e-05,
      "loss": 0.0284,
      "step": 23500
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.968321054160395e-05,
      "loss": 0.0282,
      "step": 24000
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.967662203913491e-05,
      "loss": 0.0278,
      "step": 24500
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.9670020333254114e-05,
      "loss": 0.0281,
      "step": 25000
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.9663418627373315e-05,
      "loss": 0.028,
      "step": 25500
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.9656816921492516e-05,
      "loss": 0.0278,
      "step": 26000
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.965021521561172e-05,
      "loss": 0.0278,
      "step": 26500
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.964361350973092e-05,
      "loss": 0.0276,
      "step": 27000
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.963701180385012e-05,
      "loss": 0.0276,
      "step": 27500
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.963042330138108e-05,
      "loss": 0.0276,
      "step": 28000
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.962382159550028e-05,
      "loss": 0.0275,
      "step": 28500
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.961721988961948e-05,
      "loss": 0.0274,
      "step": 29000
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.961061818373868e-05,
      "loss": 0.0274,
      "step": 29500
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.960401647785788e-05,
      "loss": 0.0276,
      "step": 30000
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.959741477197708e-05,
      "loss": 0.0272,
      "step": 30500
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.9590826269508046e-05,
      "loss": 0.0272,
      "step": 31000
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.958422456362725e-05,
      "loss": 0.027,
      "step": 31500
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.957762285774644e-05,
      "loss": 0.027,
      "step": 32000
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.957102115186564e-05,
      "loss": 0.027,
      "step": 32500
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.9564419445984844e-05,
      "loss": 0.0271,
      "step": 33000
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.9557817740104045e-05,
      "loss": 0.0268,
      "step": 33500
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.9551216034223246e-05,
      "loss": 0.0272,
      "step": 34000
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.954461432834245e-05,
      "loss": 0.0269,
      "step": 34500
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.953801262246165e-05,
      "loss": 0.0268,
      "step": 35000
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.953141091658084e-05,
      "loss": 0.027,
      "step": 35500
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.9524809210700044e-05,
      "loss": 0.0269,
      "step": 36000
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.9518207504819245e-05,
      "loss": 0.0266,
      "step": 36500
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.951163220576197e-05,
      "loss": 0.0265,
      "step": 37000
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.950503049988117e-05,
      "loss": 0.0265,
      "step": 37500
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.9498428794000374e-05,
      "loss": 0.0265,
      "step": 38000
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.9491827088119575e-05,
      "loss": 0.0267,
      "step": 38500
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.948522538223877e-05,
      "loss": 0.0263,
      "step": 39000
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.947862367635797e-05,
      "loss": 0.0262,
      "step": 39500
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.947203517388894e-05,
      "loss": 0.0261,
      "step": 40000
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.946543346800814e-05,
      "loss": 0.0264,
      "step": 40500
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.945883176212733e-05,
      "loss": 0.0266,
      "step": 41000
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.9452230056246534e-05,
      "loss": 0.0262,
      "step": 41500
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.94456415537775e-05,
      "loss": 0.0263,
      "step": 42000
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.94390398478967e-05,
      "loss": 0.0261,
      "step": 42500
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.94324381420159e-05,
      "loss": 0.026,
      "step": 43000
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.94258364361351e-05,
      "loss": 0.0258,
      "step": 43500
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.94192347302543e-05,
      "loss": 0.0261,
      "step": 44000
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.941264622778527e-05,
      "loss": 0.0261,
      "step": 44500
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.940604452190446e-05,
      "loss": 0.0262,
      "step": 45000
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.939944281602366e-05,
      "loss": 0.026,
      "step": 45500
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.939284111014286e-05,
      "loss": 0.0259,
      "step": 46000
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.9386239404262064e-05,
      "loss": 0.0259,
      "step": 46500
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.9379637698381266e-05,
      "loss": 0.0258,
      "step": 47000
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.937303599250047e-05,
      "loss": 0.0257,
      "step": 47500
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.936643428661967e-05,
      "loss": 0.0257,
      "step": 48000
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.935983258073886e-05,
      "loss": 0.0254,
      "step": 48500
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.935323087485806e-05,
      "loss": 0.0255,
      "step": 49000
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.9346629168977265e-05,
      "loss": 0.0253,
      "step": 49500
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.9340027463096466e-05,
      "loss": 0.0255,
      "step": 50000
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.9333438960627426e-05,
      "loss": 0.0258,
      "step": 50500
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.932683725474663e-05,
      "loss": 0.0254,
      "step": 51000
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.932023554886583e-05,
      "loss": 0.0255,
      "step": 51500
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.931363384298503e-05,
      "loss": 0.0255,
      "step": 52000
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.930703213710423e-05,
      "loss": 0.0254,
      "step": 52500
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.930044363463519e-05,
      "loss": 0.0253,
      "step": 53000
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.929384192875439e-05,
      "loss": 0.0252,
      "step": 53500
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.9287240222873593e-05,
      "loss": 0.0251,
      "step": 54000
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.9280638516992795e-05,
      "loss": 0.0254,
      "step": 54500
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.9274036811111996e-05,
      "loss": 0.0251,
      "step": 55000
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.92674351052312e-05,
      "loss": 0.0252,
      "step": 55500
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.926083339935039e-05,
      "loss": 0.0251,
      "step": 56000
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.925424489688136e-05,
      "loss": 0.0251,
      "step": 56500
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.924764319100056e-05,
      "loss": 0.0251,
      "step": 57000
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.924104148511976e-05,
      "loss": 0.0252,
      "step": 57500
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.9234439779238955e-05,
      "loss": 0.0251,
      "step": 58000
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.922785127676992e-05,
      "loss": 0.0251,
      "step": 58500
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.9221249570889124e-05,
      "loss": 0.025,
      "step": 59000
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.9214647865008325e-05,
      "loss": 0.0249,
      "step": 59500
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.920804615912752e-05,
      "loss": 0.0249,
      "step": 60000
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.920144445324672e-05,
      "loss": 0.025,
      "step": 60500
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.919484274736592e-05,
      "loss": 0.0249,
      "step": 61000
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.918825424489689e-05,
      "loss": 0.0251,
      "step": 61500
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.918165253901608e-05,
      "loss": 0.0251,
      "step": 62000
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.9175050833135284e-05,
      "loss": 0.0249,
      "step": 62500
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.916846233066625e-05,
      "loss": 0.0247,
      "step": 63000
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.916186062478545e-05,
      "loss": 0.025,
      "step": 63500
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.915525891890465e-05,
      "loss": 0.0249,
      "step": 64000
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.914865721302385e-05,
      "loss": 0.0249,
      "step": 64500
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.914205550714305e-05,
      "loss": 0.0249,
      "step": 65000
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.913545380126225e-05,
      "loss": 0.0249,
      "step": 65500
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.912885209538145e-05,
      "loss": 0.0248,
      "step": 66000
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.912226359291241e-05,
      "loss": 0.0248,
      "step": 66500
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.911566188703161e-05,
      "loss": 0.0247,
      "step": 67000
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.9109060181150814e-05,
      "loss": 0.0247,
      "step": 67500
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.9102458475270015e-05,
      "loss": 0.0245,
      "step": 68000
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.9095856769389216e-05,
      "loss": 0.0248,
      "step": 68500
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.908926826692018e-05,
      "loss": 0.0248,
      "step": 69000
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.908267976445113e-05,
      "loss": 0.0248,
      "step": 69500
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.907607805857033e-05,
      "loss": 0.0243,
      "step": 70000
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.90694895561013e-05,
      "loss": 0.0246,
      "step": 70500
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.906290105363226e-05,
      "loss": 0.0245,
      "step": 71000
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.905629934775146e-05,
      "loss": 0.0245,
      "step": 71500
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.904969764187066e-05,
      "loss": 0.0243,
      "step": 72000
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.904309593598986e-05,
      "loss": 0.0244,
      "step": 72500
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.9036494230109064e-05,
      "loss": 0.0246,
      "step": 73000
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.902989252422826e-05,
      "loss": 0.0243,
      "step": 73500
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.902329081834746e-05,
      "loss": 0.0244,
      "step": 74000
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.901668911246666e-05,
      "loss": 0.0242,
      "step": 74500
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.901008740658586e-05,
      "loss": 0.0242,
      "step": 75000
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.900348570070506e-05,
      "loss": 0.0242,
      "step": 75500
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.8996883994824264e-05,
      "loss": 0.0244,
      "step": 76000
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.8990282288943465e-05,
      "loss": 0.0241,
      "step": 76500
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.8983680583062666e-05,
      "loss": 0.0243,
      "step": 77000
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.897707887718187e-05,
      "loss": 0.0242,
      "step": 77500
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.897047717130107e-05,
      "loss": 0.0244,
      "step": 78000
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.896388866883203e-05,
      "loss": 0.0243,
      "step": 78500
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.8957286962951224e-05,
      "loss": 0.0242,
      "step": 79000
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.8950685257070425e-05,
      "loss": 0.0243,
      "step": 79500
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.8944083551189626e-05,
      "loss": 0.0241,
      "step": 80000
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.893748184530883e-05,
      "loss": 0.0244,
      "step": 80500
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.893088013942803e-05,
      "loss": 0.024,
      "step": 81000
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.8924278433547236e-05,
      "loss": 0.0242,
      "step": 81500
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.891767672766644e-05,
      "loss": 0.0239,
      "step": 82000
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.891107502178563e-05,
      "loss": 0.0243,
      "step": 82500
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.890448651931659e-05,
      "loss": 0.024,
      "step": 83000
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.889788481343579e-05,
      "loss": 0.0239,
      "step": 83500
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.8891283107554994e-05,
      "loss": 0.024,
      "step": 84000
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.8884681401674195e-05,
      "loss": 0.024,
      "step": 84500
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.8878079695793396e-05,
      "loss": 0.0239,
      "step": 85000
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.88714779899126e-05,
      "loss": 0.0239,
      "step": 85500
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.88648762840318e-05,
      "loss": 0.0238,
      "step": 86000
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.8858274578151e-05,
      "loss": 0.0236,
      "step": 86500
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.885168607568196e-05,
      "loss": 0.024,
      "step": 87000
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.884508436980116e-05,
      "loss": 0.0239,
      "step": 87500
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.883848266392036e-05,
      "loss": 0.0239,
      "step": 88000
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.8831880958039564e-05,
      "loss": 0.0239,
      "step": 88500
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.8825279252158765e-05,
      "loss": 0.0237,
      "step": 89000
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.8818677546277966e-05,
      "loss": 0.0239,
      "step": 89500
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.881207584039716e-05,
      "loss": 0.0238,
      "step": 90000
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.880547413451636e-05,
      "loss": 0.0238,
      "step": 90500
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.879887242863556e-05,
      "loss": 0.0238,
      "step": 91000
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.8792270722754764e-05,
      "loss": 0.0237,
      "step": 91500
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.8785682220285724e-05,
      "loss": 0.0236,
      "step": 92000
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.8779080514404925e-05,
      "loss": 0.0235,
      "step": 92500
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.8772478808524127e-05,
      "loss": 0.0235,
      "step": 93000
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.876587710264333e-05,
      "loss": 0.0237,
      "step": 93500
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.875928860017429e-05,
      "loss": 0.0237,
      "step": 94000
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.875268689429349e-05,
      "loss": 0.0238,
      "step": 94500
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.874608518841269e-05,
      "loss": 0.0237,
      "step": 95000
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.873948348253189e-05,
      "loss": 0.0237,
      "step": 95500
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.873288177665109e-05,
      "loss": 0.0234,
      "step": 96000
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.872629327418205e-05,
      "loss": 0.0236,
      "step": 96500
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.8719691568301254e-05,
      "loss": 0.0235,
      "step": 97000
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.8713089862420456e-05,
      "loss": 0.0237,
      "step": 97500
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.870648815653966e-05,
      "loss": 0.0235,
      "step": 98000
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.869988645065886e-05,
      "loss": 0.0233,
      "step": 98500
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.869328474477805e-05,
      "loss": 0.0235,
      "step": 99000
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.868668303889725e-05,
      "loss": 0.0238,
      "step": 99500
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.8680081333016454e-05,
      "loss": 0.0232,
      "step": 100000
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.8673479627135656e-05,
      "loss": 0.0234,
      "step": 100500
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.8666891124666616e-05,
      "loss": 0.0235,
      "step": 101000
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.866028941878582e-05,
      "loss": 0.0237,
      "step": 101500
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.865368771290502e-05,
      "loss": 0.0234,
      "step": 102000
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.864708600702422e-05,
      "loss": 0.0234,
      "step": 102500
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.864048430114342e-05,
      "loss": 0.0237,
      "step": 103000
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.863388259526262e-05,
      "loss": 0.0233,
      "step": 103500
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.862728088938182e-05,
      "loss": 0.0236,
      "step": 104000
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.862067918350102e-05,
      "loss": 0.0233,
      "step": 104500
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.861407747762022e-05,
      "loss": 0.0236,
      "step": 105000
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.8607488975151186e-05,
      "loss": 0.0234,
      "step": 105500
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.860088726927039e-05,
      "loss": 0.0232,
      "step": 106000
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.859429876680134e-05,
      "loss": 0.0232,
      "step": 106500
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.858769706092055e-05,
      "loss": 0.023,
      "step": 107000
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.858109535503975e-05,
      "loss": 0.023,
      "step": 107500
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.857449364915895e-05,
      "loss": 0.0234,
      "step": 108000
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.8567891943278145e-05,
      "loss": 0.0232,
      "step": 108500
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.8561303440809106e-05,
      "loss": 0.0231,
      "step": 109000
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.855470173492831e-05,
      "loss": 0.0233,
      "step": 109500
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.854810002904751e-05,
      "loss": 0.0233,
      "step": 110000
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.854149832316671e-05,
      "loss": 0.0231,
      "step": 110500
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.853489661728591e-05,
      "loss": 0.0232,
      "step": 111000
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.852829491140511e-05,
      "loss": 0.0232,
      "step": 111500
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.852169320552431e-05,
      "loss": 0.0231,
      "step": 112000
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.851510470305527e-05,
      "loss": 0.023,
      "step": 112500
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.8508502997174474e-05,
      "loss": 0.023,
      "step": 113000
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.8501901291293675e-05,
      "loss": 0.0231,
      "step": 113500
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.8495299585412876e-05,
      "loss": 0.023,
      "step": 114000
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.848869787953208e-05,
      "loss": 0.0231,
      "step": 114500
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.848209617365128e-05,
      "loss": 0.0227,
      "step": 115000
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.847549446777048e-05,
      "loss": 0.0228,
      "step": 115500
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.8468892761889674e-05,
      "loss": 0.023,
      "step": 116000
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.8462304259420634e-05,
      "loss": 0.0228,
      "step": 116500
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.8455702553539836e-05,
      "loss": 0.023,
      "step": 117000
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.844910084765904e-05,
      "loss": 0.023,
      "step": 117500
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.844249914177824e-05,
      "loss": 0.0229,
      "step": 118000
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.84359106393092e-05,
      "loss": 0.023,
      "step": 118500
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.84293089334284e-05,
      "loss": 0.023,
      "step": 119000
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.84227072275476e-05,
      "loss": 0.0229,
      "step": 119500
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.84161055216668e-05,
      "loss": 0.0229,
      "step": 120000
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.8409503815786e-05,
      "loss": 0.023,
      "step": 120500
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.8402902109905204e-05,
      "loss": 0.0229,
      "step": 121000
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.8396300404024405e-05,
      "loss": 0.0228,
      "step": 121500
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.8389698698143606e-05,
      "loss": 0.0229,
      "step": 122000
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.838311019567457e-05,
      "loss": 0.023,
      "step": 122500
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.837650848979377e-05,
      "loss": 0.023,
      "step": 123000
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.836990678391297e-05,
      "loss": 0.0229,
      "step": 123500
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.836330507803217e-05,
      "loss": 0.0227,
      "step": 124000
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.8356716575563124e-05,
      "loss": 0.0231,
      "step": 124500
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.8350114869682325e-05,
      "loss": 0.0228,
      "step": 125000
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.8343513163801526e-05,
      "loss": 0.0228,
      "step": 125500
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.833691145792073e-05,
      "loss": 0.0229,
      "step": 126000
    }
  ],
  "max_steps": 3786900,
  "num_train_epochs": 30,
  "total_flos": 4.741138645915031e+18,
  "trial_name": null,
  "trial_params": null
}