{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 110445,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 4.9773642989723395e-05,
      "loss": 1.1879,
      "step": 500
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.954728597944679e-05,
      "loss": 1.0573,
      "step": 1000
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.932092896917018e-05,
      "loss": 1.0167,
      "step": 1500
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.909457195889357e-05,
      "loss": 1.0412,
      "step": 2000
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.886821494861696e-05,
      "loss": 1.0694,
      "step": 2500
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.864185793834035e-05,
      "loss": 0.9758,
      "step": 3000
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.841550092806374e-05,
      "loss": 0.9976,
      "step": 3500
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.818914391778714e-05,
      "loss": 0.984,
      "step": 4000
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.7962786907510526e-05,
      "loss": 0.9322,
      "step": 4500
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.773642989723392e-05,
      "loss": 0.9718,
      "step": 5000
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.751007288695731e-05,
      "loss": 0.9602,
      "step": 5500
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.72837158766807e-05,
      "loss": 0.9755,
      "step": 6000
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.7057358866404096e-05,
      "loss": 0.9505,
      "step": 6500
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.683100185612749e-05,
      "loss": 0.9638,
      "step": 7000
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.6604644845850873e-05,
      "loss": 0.9089,
      "step": 7500
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.6378287835574266e-05,
      "loss": 0.883,
      "step": 8000
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.6151930825297665e-05,
      "loss": 0.899,
      "step": 8500
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.592557381502106e-05,
      "loss": 0.8908,
      "step": 9000
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.569921680474444e-05,
      "loss": 0.9031,
      "step": 9500
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.5472859794467835e-05,
      "loss": 0.8931,
      "step": 10000
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.524650278419123e-05,
      "loss": 0.9001,
      "step": 10500
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.502014577391462e-05,
      "loss": 0.8865,
      "step": 11000
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.479378876363801e-05,
      "loss": 0.8692,
      "step": 11500
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.4567431753361404e-05,
      "loss": 0.8683,
      "step": 12000
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.434107474308479e-05,
      "loss": 0.8443,
      "step": 12500
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.411471773280819e-05,
      "loss": 0.8733,
      "step": 13000
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.388836072253158e-05,
      "loss": 0.8967,
      "step": 13500
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.3662003712254973e-05,
      "loss": 0.8516,
      "step": 14000
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.343564670197836e-05,
      "loss": 0.8582,
      "step": 14500
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.320928969170175e-05,
      "loss": 0.8722,
      "step": 15000
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.298293268142515e-05,
      "loss": 0.8478,
      "step": 15500
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.2756575671148536e-05,
      "loss": 0.8789,
      "step": 16000
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.253021866087193e-05,
      "loss": 0.8153,
      "step": 16500
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.230386165059532e-05,
      "loss": 0.8486,
      "step": 17000
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.207750464031871e-05,
      "loss": 0.8443,
      "step": 17500
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.1851147630042105e-05,
      "loss": 0.8261,
      "step": 18000
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.16247906197655e-05,
      "loss": 0.8534,
      "step": 18500
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.139843360948889e-05,
      "loss": 0.8342,
      "step": 19000
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.1172076599212275e-05,
      "loss": 0.8622,
      "step": 19500
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.0945719588935674e-05,
      "loss": 0.8703,
      "step": 20000
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.071936257865907e-05,
      "loss": 0.8315,
      "step": 20500
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.049300556838245e-05,
      "loss": 0.8126,
      "step": 21000
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.0266648558105845e-05,
      "loss": 0.8236,
      "step": 21500
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.004029154782924e-05,
      "loss": 0.854,
      "step": 22000
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.981393453755263e-05,
      "loss": 0.8196,
      "step": 22500
    },
    {
      "epoch": 0.62,
      "learning_rate": 3.958757752727602e-05,
      "loss": 0.7705,
      "step": 23000
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.9361220516999414e-05,
      "loss": 0.7824,
      "step": 23500
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.9134863506722806e-05,
      "loss": 0.7825,
      "step": 24000
    },
    {
      "epoch": 0.67,
      "learning_rate": 3.89085064964462e-05,
      "loss": 0.7789,
      "step": 24500
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.868214948616959e-05,
      "loss": 0.8073,
      "step": 25000
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.845579247589298e-05,
      "loss": 0.8173,
      "step": 25500
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.822943546561637e-05,
      "loss": 0.8299,
      "step": 26000
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.800307845533976e-05,
      "loss": 0.8018,
      "step": 26500
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.777672144506316e-05,
      "loss": 0.8511,
      "step": 27000
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.7550364434786546e-05,
      "loss": 0.7932,
      "step": 27500
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.732400742450994e-05,
      "loss": 0.793,
      "step": 28000
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.709765041423333e-05,
      "loss": 0.7984,
      "step": 28500
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.687129340395672e-05,
      "loss": 0.814,
      "step": 29000
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.6644936393680115e-05,
      "loss": 0.7919,
      "step": 29500
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.641857938340351e-05,
      "loss": 0.8149,
      "step": 30000
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.61922223731269e-05,
      "loss": 0.7847,
      "step": 30500
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.5965865362850285e-05,
      "loss": 0.783,
      "step": 31000
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.5739508352573684e-05,
      "loss": 0.7998,
      "step": 31500
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.5513151342297076e-05,
      "loss": 0.7888,
      "step": 32000
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.528679433202046e-05,
      "loss": 0.7895,
      "step": 32500
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.5060437321743854e-05,
      "loss": 0.796,
      "step": 33000
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.4834080311467246e-05,
      "loss": 0.7896,
      "step": 33500
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.460772330119064e-05,
      "loss": 0.8016,
      "step": 34000
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.438136629091403e-05,
      "loss": 0.7789,
      "step": 34500
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.415500928063742e-05,
      "loss": 0.7974,
      "step": 35000
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.3928652270360816e-05,
      "loss": 0.7701,
      "step": 35500
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.37022952600842e-05,
      "loss": 0.753,
      "step": 36000
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.34759382498076e-05,
      "loss": 0.7681,
      "step": 36500
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.324958123953099e-05,
      "loss": 0.7854,
      "step": 37000
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.302322422925438e-05,
      "loss": 0.7462,
      "step": 37500
    },
    {
      "epoch": 1.03,
      "learning_rate": 3.279686721897777e-05,
      "loss": 0.7028,
      "step": 38000
    },
    {
      "epoch": 1.05,
      "learning_rate": 3.257051020870116e-05,
      "loss": 0.761,
      "step": 38500
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.234415319842456e-05,
      "loss": 0.7391,
      "step": 39000
    },
    {
      "epoch": 1.07,
      "learning_rate": 3.211779618814795e-05,
      "loss": 0.7345,
      "step": 39500
    },
    {
      "epoch": 1.09,
      "learning_rate": 3.189143917787134e-05,
      "loss": 0.7191,
      "step": 40000
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.166508216759473e-05,
      "loss": 0.7528,
      "step": 40500
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.1438725157318124e-05,
      "loss": 0.7449,
      "step": 41000
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.1212368147041517e-05,
      "loss": 0.7769,
      "step": 41500
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.098601113676491e-05,
      "loss": 0.7451,
      "step": 42000
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.0759654126488294e-05,
      "loss": 0.7561,
      "step": 42500
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.053329711621169e-05,
      "loss": 0.7299,
      "step": 43000
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.0306940105935082e-05,
      "loss": 0.7381,
      "step": 43500
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.0080583095658478e-05,
      "loss": 0.7397,
      "step": 44000
    },
    {
      "epoch": 1.21,
      "learning_rate": 2.9854226085381864e-05,
      "loss": 0.691,
      "step": 44500
    },
    {
      "epoch": 1.22,
      "learning_rate": 2.962786907510526e-05,
      "loss": 0.73,
      "step": 45000
    },
    {
      "epoch": 1.24,
      "learning_rate": 2.940151206482865e-05,
      "loss": 0.7078,
      "step": 45500
    },
    {
      "epoch": 1.25,
      "learning_rate": 2.917515505455204e-05,
      "loss": 0.7254,
      "step": 46000
    },
    {
      "epoch": 1.26,
      "learning_rate": 2.8948798044275433e-05,
      "loss": 0.7223,
      "step": 46500
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.8722441033998825e-05,
      "loss": 0.7113,
      "step": 47000
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.8496084023722214e-05,
      "loss": 0.7237,
      "step": 47500
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.8269727013445606e-05,
      "loss": 0.7221,
      "step": 48000
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.8043370003169002e-05,
      "loss": 0.7242,
      "step": 48500
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.7817012992892394e-05,
      "loss": 0.7214,
      "step": 49000
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.7590655982615783e-05,
      "loss": 0.7345,
      "step": 49500
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.7364298972339176e-05,
      "loss": 0.7302,
      "step": 50000
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.7137941962062568e-05,
      "loss": 0.7494,
      "step": 50500
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.6911584951785957e-05,
      "loss": 0.7391,
      "step": 51000
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.668522794150935e-05,
      "loss": 0.6945,
      "step": 51500
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.645887093123274e-05,
      "loss": 0.7372,
      "step": 52000
    },
    {
      "epoch": 1.43,
      "learning_rate": 2.623251392095613e-05,
      "loss": 0.707,
      "step": 52500
    },
    {
      "epoch": 1.44,
      "learning_rate": 2.6006156910679523e-05,
      "loss": 0.7253,
      "step": 53000
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.577979990040292e-05,
      "loss": 0.7346,
      "step": 53500
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.555344289012631e-05,
      "loss": 0.7237,
      "step": 54000
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.53270858798497e-05,
      "loss": 0.7106,
      "step": 54500
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.5100728869573092e-05,
      "loss": 0.7043,
      "step": 55000
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.4874371859296484e-05,
      "loss": 0.7339,
      "step": 55500
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.4648014849019877e-05,
      "loss": 0.7156,
      "step": 56000
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.4421657838743266e-05,
      "loss": 0.6913,
      "step": 56500
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.4195300828466658e-05,
      "loss": 0.6981,
      "step": 57000
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.396894381819005e-05,
      "loss": 0.7133,
      "step": 57500
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.3742586807913442e-05,
      "loss": 0.7359,
      "step": 58000
    },
    {
      "epoch": 1.59,
      "learning_rate": 2.3516229797636835e-05,
      "loss": 0.7033,
      "step": 58500
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.3289872787360227e-05,
      "loss": 0.7246,
      "step": 59000
    },
    {
      "epoch": 1.62,
      "learning_rate": 2.3063515777083616e-05,
      "loss": 0.7254,
      "step": 59500
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.2837158766807008e-05,
      "loss": 0.7335,
      "step": 60000
    },
    {
      "epoch": 1.64,
      "learning_rate": 2.26108017565304e-05,
      "loss": 0.735,
      "step": 60500
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.2384444746253793e-05,
      "loss": 0.6991,
      "step": 61000
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.2158087735977185e-05,
      "loss": 0.702,
      "step": 61500
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.1931730725700574e-05,
      "loss": 0.7185,
      "step": 62000
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.170537371542397e-05,
      "loss": 0.7132,
      "step": 62500
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.147901670514736e-05,
      "loss": 0.6873,
      "step": 63000
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.125265969487075e-05,
      "loss": 0.7491,
      "step": 63500
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.1026302684594143e-05,
      "loss": 0.7064,
      "step": 64000
    },
    {
      "epoch": 1.75,
      "learning_rate": 2.0799945674317532e-05,
      "loss": 0.723,
      "step": 64500
    },
    {
      "epoch": 1.77,
      "learning_rate": 2.0573588664040928e-05,
      "loss": 0.6922,
      "step": 65000
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.0347231653764317e-05,
      "loss": 0.6727,
      "step": 65500
    },
    {
      "epoch": 1.79,
      "learning_rate": 2.0120874643487713e-05,
      "loss": 0.6876,
      "step": 66000
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.98945176332111e-05,
      "loss": 0.7399,
      "step": 66500
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.9668160622934494e-05,
      "loss": 0.7107,
      "step": 67000
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.9441803612657886e-05,
      "loss": 0.6767,
      "step": 67500
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.9215446602381275e-05,
      "loss": 0.7102,
      "step": 68000
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.898908959210467e-05,
      "loss": 0.7183,
      "step": 68500
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.876273258182806e-05,
      "loss": 0.6808,
      "step": 69000
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.8536375571551452e-05,
      "loss": 0.7165,
      "step": 69500
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.8310018561274844e-05,
      "loss": 0.7035,
      "step": 70000
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.8083661550998233e-05,
      "loss": 0.6831,
      "step": 70500
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.785730454072163e-05,
      "loss": 0.7026,
      "step": 71000
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.7630947530445018e-05,
      "loss": 0.7105,
      "step": 71500
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.740459052016841e-05,
      "loss": 0.7007,
      "step": 72000
    },
    {
      "epoch": 1.97,
      "learning_rate": 1.7178233509891802e-05,
      "loss": 0.6993,
      "step": 72500
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.6951876499615195e-05,
      "loss": 0.7108,
      "step": 73000
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.6725519489338587e-05,
      "loss": 0.71,
      "step": 73500
    },
    {
      "epoch": 2.01,
      "learning_rate": 1.6499162479061976e-05,
      "loss": 0.6825,
      "step": 74000
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.627280546878537e-05,
      "loss": 0.6683,
      "step": 74500
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.604644845850876e-05,
      "loss": 0.6797,
      "step": 75000
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.5820091448232153e-05,
      "loss": 0.6507,
      "step": 75500
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.5593734437955545e-05,
      "loss": 0.6438,
      "step": 76000
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.5367377427678938e-05,
      "loss": 0.6587,
      "step": 76500
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.5141020417402326e-05,
      "loss": 0.6977,
      "step": 77000
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.491466340712572e-05,
      "loss": 0.6654,
      "step": 77500
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.4688306396849111e-05,
      "loss": 0.6654,
      "step": 78000
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.4461949386572503e-05,
      "loss": 0.6217,
      "step": 78500
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.4235592376295894e-05,
      "loss": 0.638,
      "step": 79000
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.4009235366019285e-05,
      "loss": 0.6636,
      "step": 79500
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.3782878355742679e-05,
      "loss": 0.6635,
      "step": 80000
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.355652134546607e-05,
      "loss": 0.6802,
      "step": 80500
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.3330164335189463e-05,
      "loss": 0.6387,
      "step": 81000
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.3103807324912854e-05,
      "loss": 0.6476,
      "step": 81500
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.2877450314636244e-05,
      "loss": 0.6495,
      "step": 82000
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.2651093304359637e-05,
      "loss": 0.6372,
      "step": 82500
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.2424736294083027e-05,
      "loss": 0.6304,
      "step": 83000
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.219837928380642e-05,
      "loss": 0.6643,
      "step": 83500
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.1972022273529812e-05,
      "loss": 0.6454,
      "step": 84000
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.1745665263253204e-05,
      "loss": 0.6847,
      "step": 84500
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.1519308252976595e-05,
      "loss": 0.6396,
      "step": 85000
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.1292951242699987e-05,
      "loss": 0.638,
      "step": 85500
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.106659423242338e-05,
      "loss": 0.662,
      "step": 86000
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.084023722214677e-05,
      "loss": 0.6449,
      "step": 86500
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.0613880211870162e-05,
      "loss": 0.6452,
      "step": 87000
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0387523201593553e-05,
      "loss": 0.6658,
      "step": 87500
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.0161166191316945e-05,
      "loss": 0.6677,
      "step": 88000
    },
    {
      "epoch": 2.4,
      "learning_rate": 9.934809181040338e-06,
      "loss": 0.6543,
      "step": 88500
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.70845217076373e-06,
      "loss": 0.6425,
      "step": 89000
    },
    {
      "epoch": 2.43,
      "learning_rate": 9.48209516048712e-06,
      "loss": 0.6593,
      "step": 89500
    },
    {
      "epoch": 2.44,
      "learning_rate": 9.255738150210511e-06,
      "loss": 0.641,
      "step": 90000
    },
    {
      "epoch": 2.46,
      "learning_rate": 9.029381139933904e-06,
      "loss": 0.6167,
      "step": 90500
    },
    {
      "epoch": 2.47,
      "learning_rate": 8.803024129657296e-06,
      "loss": 0.6498,
      "step": 91000
    },
    {
      "epoch": 2.49,
      "learning_rate": 8.576667119380688e-06,
      "loss": 0.667,
      "step": 91500
    },
    {
      "epoch": 2.5,
      "learning_rate": 8.35031010910408e-06,
      "loss": 0.6352,
      "step": 92000
    },
    {
      "epoch": 2.51,
      "learning_rate": 8.123953098827471e-06,
      "loss": 0.6837,
      "step": 92500
    },
    {
      "epoch": 2.53,
      "learning_rate": 7.897596088550863e-06,
      "loss": 0.6471,
      "step": 93000
    },
    {
      "epoch": 2.54,
      "learning_rate": 7.671239078274254e-06,
      "loss": 0.6601,
      "step": 93500
    },
    {
      "epoch": 2.55,
      "learning_rate": 7.444882067997646e-06,
      "loss": 0.653,
      "step": 94000
    },
    {
      "epoch": 2.57,
      "learning_rate": 7.218525057721039e-06,
      "loss": 0.6495,
      "step": 94500
    },
    {
      "epoch": 2.58,
      "learning_rate": 6.992168047444429e-06,
      "loss": 0.63,
      "step": 95000
    },
    {
      "epoch": 2.59,
      "learning_rate": 6.7658110371678216e-06,
      "loss": 0.6371,
      "step": 95500
    },
    {
      "epoch": 2.61,
      "learning_rate": 6.539454026891213e-06,
      "loss": 0.6551,
      "step": 96000
    },
    {
      "epoch": 2.62,
      "learning_rate": 6.313097016614605e-06,
      "loss": 0.6621,
      "step": 96500
    },
    {
      "epoch": 2.63,
      "learning_rate": 6.086740006337997e-06,
      "loss": 0.6169,
      "step": 97000
    },
    {
      "epoch": 2.65,
      "learning_rate": 5.860382996061388e-06,
      "loss": 0.622,
      "step": 97500
    },
    {
      "epoch": 2.66,
      "learning_rate": 5.63402598578478e-06,
      "loss": 0.6535,
      "step": 98000
    },
    {
      "epoch": 2.68,
      "learning_rate": 5.407668975508172e-06,
      "loss": 0.6599,
      "step": 98500
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.1813119652315635e-06,
      "loss": 0.6369,
      "step": 99000
    },
    {
      "epoch": 2.7,
      "learning_rate": 4.954954954954955e-06,
      "loss": 0.6215,
      "step": 99500
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.728597944678347e-06,
      "loss": 0.6548,
      "step": 100000
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.502240934401739e-06,
      "loss": 0.6472,
      "step": 100500
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.27588392412513e-06,
      "loss": 0.6335,
      "step": 101000
    },
    {
      "epoch": 2.76,
      "learning_rate": 4.049526913848522e-06,
      "loss": 0.6288,
      "step": 101500
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.823169903571914e-06,
      "loss": 0.6547,
      "step": 102000
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.596812893295305e-06,
      "loss": 0.6538,
      "step": 102500
    },
    {
      "epoch": 2.8,
      "learning_rate": 3.3704558830186973e-06,
      "loss": 0.6365,
      "step": 103000
    },
    {
      "epoch": 2.81,
      "learning_rate": 3.144098872742089e-06,
      "loss": 0.6731,
      "step": 103500
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.9177418624654807e-06,
      "loss": 0.6494,
      "step": 104000
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.6913848521888725e-06,
      "loss": 0.6325,
      "step": 104500
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.465027841912264e-06,
      "loss": 0.6237,
      "step": 105000
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.238670831635656e-06,
      "loss": 0.649,
      "step": 105500
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.0123138213590474e-06,
      "loss": 0.6477,
      "step": 106000
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.7859568110824392e-06,
      "loss": 0.6642,
      "step": 106500
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.559599800805831e-06,
      "loss": 0.6515,
      "step": 107000
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.3332427905292228e-06,
      "loss": 0.655,
      "step": 107500
    },
    {
      "epoch": 2.93,
      "learning_rate": 1.1068857802526145e-06,
      "loss": 0.6645,
      "step": 108000
    },
    {
      "epoch": 2.95,
      "learning_rate": 8.805287699760061e-07,
      "loss": 0.6654,
      "step": 108500
    },
    {
      "epoch": 2.96,
      "learning_rate": 6.541717596993979e-07,
      "loss": 0.6457,
      "step": 109000
    },
    {
      "epoch": 2.97,
      "learning_rate": 4.278147494227896e-07,
      "loss": 0.669,
      "step": 109500
    },
    {
      "epoch": 2.99,
      "learning_rate": 2.0145773914618135e-07,
      "loss": 0.6577,
      "step": 110000
    },
    {
      "epoch": 3.0,
      "step": 110445,
      "total_flos": 4.0396527424641024e+17,
      "train_loss": 0.744481670575304,
      "train_runtime": 62011.6805,
      "train_samples_per_second": 10.686,
      "train_steps_per_second": 1.781
    }
  ],
  "max_steps": 110445,
  "num_train_epochs": 3,
  "total_flos": 4.0396527424641024e+17,
  "trial_name": null,
  "trial_params": null
}