{
  "best_metric": 0.6417234539985657,
  "best_model_checkpoint": "./enko_mbartLarge_100p_sup2/checkpoint-129072",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 129072,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 9.940000000000001e-06,
      "loss": 2.2901,
      "step": 500
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.992e-05,
      "loss": 1.6084,
      "step": 1000
    },
    {
      "epoch": 0.03,
      "learning_rate": 2.9920000000000005e-05,
      "loss": 1.449,
      "step": 1500
    },
    {
      "epoch": 0.05,
      "learning_rate": 3.9920000000000004e-05,
      "loss": 1.3874,
      "step": 2000
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.992e-05,
      "loss": 1.3485,
      "step": 2500
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.996142239367825e-05,
      "loss": 1.2926,
      "step": 3000
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9922533677628105e-05,
      "loss": 1.248,
      "step": 3500
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.988364496157795e-05,
      "loss": 1.1931,
      "step": 4000
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.98447562455278e-05,
      "loss": 1.1677,
      "step": 4500
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.980594530690975e-05,
      "loss": 1.142,
      "step": 5000
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.97670565908596e-05,
      "loss": 1.122,
      "step": 5500
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.972816787480945e-05,
      "loss": 1.0855,
      "step": 6000
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.968927915875929e-05,
      "loss": 1.0719,
      "step": 6500
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.965039044270914e-05,
      "loss": 1.053,
      "step": 7000
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.9611501726658995e-05,
      "loss": 1.0359,
      "step": 7500
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.957261301060884e-05,
      "loss": 1.0232,
      "step": 8000
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.953380207199079e-05,
      "loss": 1.0169,
      "step": 8500
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.949491335594064e-05,
      "loss": 1.0141,
      "step": 9000
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.945602463989049e-05,
      "loss": 0.9855,
      "step": 9500
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.941721370127244e-05,
      "loss": 0.9932,
      "step": 10000
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.9378324985222293e-05,
      "loss": 0.9656,
      "step": 10500
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.933943626917214e-05,
      "loss": 0.9694,
      "step": 11000
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.930054755312199e-05,
      "loss": 0.9609,
      "step": 11500
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.926165883707184e-05,
      "loss": 0.9526,
      "step": 12000
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.9222770121021686e-05,
      "loss": 0.9359,
      "step": 12500
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.9183959182403636e-05,
      "loss": 0.9324,
      "step": 13000
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.914507046635349e-05,
      "loss": 0.9251,
      "step": 13500
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.910625952773543e-05,
      "loss": 0.9216,
      "step": 14000
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.906737081168528e-05,
      "loss": 0.9181,
      "step": 14500
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.902855987306723e-05,
      "loss": 0.9098,
      "step": 15000
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.898967115701708e-05,
      "loss": 0.9067,
      "step": 15500
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.895078244096693e-05,
      "loss": 0.8931,
      "step": 16000
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.891189372491678e-05,
      "loss": 0.8923,
      "step": 16500
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.887300500886663e-05,
      "loss": 0.8703,
      "step": 17000
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.8834116292816475e-05,
      "loss": 0.8861,
      "step": 17500
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.879522757676633e-05,
      "loss": 0.8864,
      "step": 18000
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.875633886071618e-05,
      "loss": 0.886,
      "step": 18500
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.871752792209813e-05,
      "loss": 0.8737,
      "step": 19000
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.867863920604798e-05,
      "loss": 0.8708,
      "step": 19500
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.8639750489997824e-05,
      "loss": 0.865,
      "step": 20000
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.8600861773947676e-05,
      "loss": 0.8571,
      "step": 20500
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.856197305789753e-05,
      "loss": 0.8621,
      "step": 21000
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.852308434184737e-05,
      "loss": 0.8607,
      "step": 21500
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.848419562579722e-05,
      "loss": 0.8519,
      "step": 22000
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.844530690974707e-05,
      "loss": 0.8472,
      "step": 22500
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.840641819369692e-05,
      "loss": 0.8381,
      "step": 23000
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.8367529477646765e-05,
      "loss": 0.8329,
      "step": 23500
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.8328640761596616e-05,
      "loss": 0.8425,
      "step": 24000
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.828975204554647e-05,
      "loss": 0.8227,
      "step": 24500
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.825086332949631e-05,
      "loss": 0.8349,
      "step": 25000
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.8211974613446164e-05,
      "loss": 0.8235,
      "step": 25500
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.8173085897396015e-05,
      "loss": 0.8116,
      "step": 26000
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.813419718134587e-05,
      "loss": 0.8367,
      "step": 26500
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.809530846529571e-05,
      "loss": 0.8219,
      "step": 27000
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.805641974924556e-05,
      "loss": 0.8206,
      "step": 27500
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.8017531033195415e-05,
      "loss": 0.8199,
      "step": 28000
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.797864231714526e-05,
      "loss": 0.816,
      "step": 28500
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.793983137852721e-05,
      "loss": 0.8144,
      "step": 29000
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.790102043990916e-05,
      "loss": 0.806,
      "step": 29500
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.786220950129111e-05,
      "loss": 0.8134,
      "step": 30000
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.782332078524095e-05,
      "loss": 0.8109,
      "step": 30500
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.7784432069190805e-05,
      "loss": 0.792,
      "step": 31000
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.7745543353140656e-05,
      "loss": 0.8046,
      "step": 31500
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.77066546370905e-05,
      "loss": 0.7983,
      "step": 32000
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.766776592104035e-05,
      "loss": 0.7937,
      "step": 32500
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.7628877204990204e-05,
      "loss": 0.7978,
      "step": 33000
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.758998848894005e-05,
      "loss": 0.7961,
      "step": 33500
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.75510997728899e-05,
      "loss": 0.7854,
      "step": 34000
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.751221105683975e-05,
      "loss": 0.8006,
      "step": 34500
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.7473322340789597e-05,
      "loss": 0.7857,
      "step": 35000
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.743443362473945e-05,
      "loss": 0.7857,
      "step": 35500
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.73956226861214e-05,
      "loss": 0.7874,
      "step": 36000
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.735673397007125e-05,
      "loss": 0.7826,
      "step": 36500
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.73178452540211e-05,
      "loss": 0.7902,
      "step": 37000
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.727895653797094e-05,
      "loss": 0.7695,
      "step": 37500
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.724006782192079e-05,
      "loss": 0.7789,
      "step": 38000
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.720117910587064e-05,
      "loss": 0.7739,
      "step": 38500
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.716229038982049e-05,
      "loss": 0.7726,
      "step": 39000
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.712340167377034e-05,
      "loss": 0.7722,
      "step": 39500
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.708451295772019e-05,
      "loss": 0.7656,
      "step": 40000
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.704562424167004e-05,
      "loss": 0.77,
      "step": 40500
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.700689108048409e-05,
      "loss": 0.7814,
      "step": 41000
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.696800236443394e-05,
      "loss": 0.7704,
      "step": 41500
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.692919142581589e-05,
      "loss": 0.7611,
      "step": 42000
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.689038048719784e-05,
      "loss": 0.7629,
      "step": 42500
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.6851491771147685e-05,
      "loss": 0.7676,
      "step": 43000
    },
    {
      "epoch": 1.0,
      "eval_bleu": 55.2526,
      "eval_gen_len": 16.382,
      "eval_loss": 0.7125306129455566,
      "eval_runtime": 7352.67,
      "eval_samples_per_second": 11.685,
      "eval_steps_per_second": 1.461,
      "step": 43024
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.6812603055097536e-05,
      "loss": 0.6434,
      "step": 43500
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.6773792116479486e-05,
      "loss": 0.6436,
      "step": 44000
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.673490340042934e-05,
      "loss": 0.651,
      "step": 44500
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.669601468437918e-05,
      "loss": 0.6517,
      "step": 45000
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.665712596832903e-05,
      "loss": 0.6416,
      "step": 45500
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.661823725227888e-05,
      "loss": 0.6517,
      "step": 46000
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.657934853622873e-05,
      "loss": 0.6539,
      "step": 46500
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.654045982017858e-05,
      "loss": 0.6587,
      "step": 47000
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.6501571104128426e-05,
      "loss": 0.6518,
      "step": 47500
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.646268238807828e-05,
      "loss": 0.6465,
      "step": 48000
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.642379367202813e-05,
      "loss": 0.6437,
      "step": 48500
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.6384904955977974e-05,
      "loss": 0.6458,
      "step": 49000
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.6346016239927825e-05,
      "loss": 0.6476,
      "step": 49500
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.630712752387768e-05,
      "loss": 0.6587,
      "step": 50000
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.626823880782752e-05,
      "loss": 0.6392,
      "step": 50500
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.622935009177737e-05,
      "loss": 0.6547,
      "step": 51000
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.6190461375727225e-05,
      "loss": 0.6511,
      "step": 51500
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.615157265967707e-05,
      "loss": 0.6522,
      "step": 52000
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.611268394362692e-05,
      "loss": 0.6555,
      "step": 52500
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.607395078244097e-05,
      "loss": 0.6461,
      "step": 53000
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.6035062066390814e-05,
      "loss": 0.6471,
      "step": 53500
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.5996173350340665e-05,
      "loss": 0.6563,
      "step": 54000
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.5957284634290517e-05,
      "loss": 0.6447,
      "step": 54500
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.591839591824036e-05,
      "loss": 0.6483,
      "step": 55000
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.587958497962231e-05,
      "loss": 0.6475,
      "step": 55500
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.584069626357216e-05,
      "loss": 0.6546,
      "step": 56000
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.5801807547522014e-05,
      "loss": 0.6528,
      "step": 56500
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.5762918831471866e-05,
      "loss": 0.6538,
      "step": 57000
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.572403011542171e-05,
      "loss": 0.6511,
      "step": 57500
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.568521917680366e-05,
      "loss": 0.6457,
      "step": 58000
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.564633046075351e-05,
      "loss": 0.659,
      "step": 58500
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.560744174470336e-05,
      "loss": 0.6358,
      "step": 59000
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.556855302865321e-05,
      "loss": 0.6425,
      "step": 59500
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.552966431260306e-05,
      "loss": 0.644,
      "step": 60000
    },
    {
      "epoch": 1.41,
      "learning_rate": 4.5490775596552904e-05,
      "loss": 0.6537,
      "step": 60500
    },
    {
      "epoch": 1.42,
      "learning_rate": 4.5451886880502756e-05,
      "loss": 0.6529,
      "step": 61000
    },
    {
      "epoch": 1.43,
      "learning_rate": 4.5413075941884705e-05,
      "loss": 0.6547,
      "step": 61500
    },
    {
      "epoch": 1.44,
      "learning_rate": 4.537418722583455e-05,
      "loss": 0.6443,
      "step": 62000
    },
    {
      "epoch": 1.45,
      "learning_rate": 4.53352985097844e-05,
      "loss": 0.6516,
      "step": 62500
    },
    {
      "epoch": 1.46,
      "learning_rate": 4.529640979373425e-05,
      "loss": 0.6439,
      "step": 63000
    },
    {
      "epoch": 1.48,
      "learning_rate": 4.52575210776841e-05,
      "loss": 0.6401,
      "step": 63500
    },
    {
      "epoch": 1.49,
      "learning_rate": 4.521863236163395e-05,
      "loss": 0.6545,
      "step": 64000
    },
    {
      "epoch": 1.5,
      "learning_rate": 4.51797436455838e-05,
      "loss": 0.6433,
      "step": 64500
    },
    {
      "epoch": 1.51,
      "learning_rate": 4.514085492953365e-05,
      "loss": 0.6538,
      "step": 65000
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.51019662134835e-05,
      "loss": 0.6548,
      "step": 65500
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.506307749743335e-05,
      "loss": 0.6458,
      "step": 66000
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.50241887813832e-05,
      "loss": 0.649,
      "step": 66500
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.4985300065333045e-05,
      "loss": 0.6471,
      "step": 67000
    },
    {
      "epoch": 1.57,
      "learning_rate": 4.4946411349282896e-05,
      "loss": 0.6526,
      "step": 67500
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.4907600410664846e-05,
      "loss": 0.646,
      "step": 68000
    },
    {
      "epoch": 1.59,
      "learning_rate": 4.486878947204679e-05,
      "loss": 0.6491,
      "step": 68500
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.482990075599664e-05,
      "loss": 0.6484,
      "step": 69000
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.479101203994649e-05,
      "loss": 0.6421,
      "step": 69500
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.475220110132844e-05,
      "loss": 0.6498,
      "step": 70000
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.4713312385278286e-05,
      "loss": 0.6497,
      "step": 70500
    },
    {
      "epoch": 1.65,
      "learning_rate": 4.467442366922814e-05,
      "loss": 0.6503,
      "step": 71000
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.463553495317799e-05,
      "loss": 0.6314,
      "step": 71500
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.459672401455994e-05,
      "loss": 0.6495,
      "step": 72000
    },
    {
      "epoch": 1.69,
      "learning_rate": 4.4557835298509784e-05,
      "loss": 0.643,
      "step": 72500
    },
    {
      "epoch": 1.7,
      "learning_rate": 4.4518946582459635e-05,
      "loss": 0.6542,
      "step": 73000
    },
    {
      "epoch": 1.71,
      "learning_rate": 4.448005786640949e-05,
      "loss": 0.6522,
      "step": 73500
    },
    {
      "epoch": 1.72,
      "learning_rate": 4.444116915035934e-05,
      "loss": 0.6476,
      "step": 74000
    },
    {
      "epoch": 1.73,
      "learning_rate": 4.440228043430918e-05,
      "loss": 0.6337,
      "step": 74500
    },
    {
      "epoch": 1.74,
      "learning_rate": 4.4363391718259035e-05,
      "loss": 0.6546,
      "step": 75000
    },
    {
      "epoch": 1.75,
      "learning_rate": 4.4324503002208886e-05,
      "loss": 0.647,
      "step": 75500
    },
    {
      "epoch": 1.77,
      "learning_rate": 4.4285614286158724e-05,
      "loss": 0.6467,
      "step": 76000
    },
    {
      "epoch": 1.78,
      "learning_rate": 4.4246725570108576e-05,
      "loss": 0.6432,
      "step": 76500
    },
    {
      "epoch": 1.79,
      "learning_rate": 4.420783685405843e-05,
      "loss": 0.6399,
      "step": 77000
    },
    {
      "epoch": 1.8,
      "learning_rate": 4.416894813800828e-05,
      "loss": 0.6382,
      "step": 77500
    },
    {
      "epoch": 1.81,
      "learning_rate": 4.4130059421958123e-05,
      "loss": 0.649,
      "step": 78000
    },
    {
      "epoch": 1.82,
      "learning_rate": 4.409124848334007e-05,
      "loss": 0.6385,
      "step": 78500
    },
    {
      "epoch": 1.84,
      "learning_rate": 4.4052359767289925e-05,
      "loss": 0.6448,
      "step": 79000
    },
    {
      "epoch": 1.85,
      "learning_rate": 4.4013471051239776e-05,
      "loss": 0.638,
      "step": 79500
    },
    {
      "epoch": 1.86,
      "learning_rate": 4.397458233518962e-05,
      "loss": 0.6317,
      "step": 80000
    },
    {
      "epoch": 1.87,
      "learning_rate": 4.393569361913947e-05,
      "loss": 0.6338,
      "step": 80500
    },
    {
      "epoch": 1.88,
      "learning_rate": 4.3896804903089324e-05,
      "loss": 0.6406,
      "step": 81000
    },
    {
      "epoch": 1.89,
      "learning_rate": 4.385791618703917e-05,
      "loss": 0.6363,
      "step": 81500
    },
    {
      "epoch": 1.91,
      "learning_rate": 4.381902747098902e-05,
      "loss": 0.6381,
      "step": 82000
    },
    {
      "epoch": 1.92,
      "learning_rate": 4.378021653237097e-05,
      "loss": 0.6383,
      "step": 82500
    },
    {
      "epoch": 1.93,
      "learning_rate": 4.374140559375292e-05,
      "loss": 0.6351,
      "step": 83000
    },
    {
      "epoch": 1.94,
      "learning_rate": 4.370251687770277e-05,
      "loss": 0.642,
      "step": 83500
    },
    {
      "epoch": 1.95,
      "learning_rate": 4.3663705939084714e-05,
      "loss": 0.6399,
      "step": 84000
    },
    {
      "epoch": 1.96,
      "learning_rate": 4.3624817223034566e-05,
      "loss": 0.6351,
      "step": 84500
    },
    {
      "epoch": 1.98,
      "learning_rate": 4.358592850698442e-05,
      "loss": 0.631,
      "step": 85000
    },
    {
      "epoch": 1.99,
      "learning_rate": 4.354703979093426e-05,
      "loss": 0.6298,
      "step": 85500
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.350822885231621e-05,
      "loss": 0.6349,
      "step": 86000
    },
    {
      "epoch": 2.0,
      "eval_bleu": 58.202,
      "eval_gen_len": 15.9466,
      "eval_loss": 0.6546894311904907,
      "eval_runtime": 6844.8665,
      "eval_samples_per_second": 12.552,
      "eval_steps_per_second": 1.569,
      "step": 86048
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.346934013626606e-05,
      "loss": 0.5093,
      "step": 86500
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.3430451420215914e-05,
      "loss": 0.5066,
      "step": 87000
    },
    {
      "epoch": 2.03,
      "learning_rate": 4.339156270416576e-05,
      "loss": 0.4956,
      "step": 87500
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.335267398811561e-05,
      "loss": 0.5,
      "step": 88000
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.331378527206546e-05,
      "loss": 0.5034,
      "step": 88500
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.327489655601531e-05,
      "loss": 0.505,
      "step": 89000
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.323600783996516e-05,
      "loss": 0.4995,
      "step": 89500
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.319719690134711e-05,
      "loss": 0.5064,
      "step": 90000
    },
    {
      "epoch": 2.1,
      "learning_rate": 4.315830818529696e-05,
      "loss": 0.5065,
      "step": 90500
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.311949724667891e-05,
      "loss": 0.5031,
      "step": 91000
    },
    {
      "epoch": 2.13,
      "learning_rate": 4.3080608530628754e-05,
      "loss": 0.4967,
      "step": 91500
    },
    {
      "epoch": 2.14,
      "learning_rate": 4.30417198145786e-05,
      "loss": 0.5112,
      "step": 92000
    },
    {
      "epoch": 2.15,
      "learning_rate": 4.300290887596055e-05,
      "loss": 0.5069,
      "step": 92500
    },
    {
      "epoch": 2.16,
      "learning_rate": 4.29640201599104e-05,
      "loss": 0.5106,
      "step": 93000
    },
    {
      "epoch": 2.17,
      "learning_rate": 4.292513144386025e-05,
      "loss": 0.5077,
      "step": 93500
    },
    {
      "epoch": 2.18,
      "learning_rate": 4.28862427278101e-05,
      "loss": 0.5179,
      "step": 94000
    },
    {
      "epoch": 2.2,
      "learning_rate": 4.284735401175995e-05,
      "loss": 0.5192,
      "step": 94500
    },
    {
      "epoch": 2.21,
      "learning_rate": 4.28084652957098e-05,
      "loss": 0.5103,
      "step": 95000
    },
    {
      "epoch": 2.22,
      "learning_rate": 4.276957657965965e-05,
      "loss": 0.5203,
      "step": 95500
    },
    {
      "epoch": 2.23,
      "learning_rate": 4.2730687863609496e-05,
      "loss": 0.5179,
      "step": 96000
    },
    {
      "epoch": 2.24,
      "learning_rate": 4.269179914755935e-05,
      "loss": 0.515,
      "step": 96500
    },
    {
      "epoch": 2.25,
      "learning_rate": 4.26529104315092e-05,
      "loss": 0.5133,
      "step": 97000
    },
    {
      "epoch": 2.27,
      "learning_rate": 4.2614021715459043e-05,
      "loss": 0.5147,
      "step": 97500
    },
    {
      "epoch": 2.28,
      "learning_rate": 4.2575132999408895e-05,
      "loss": 0.5308,
      "step": 98000
    },
    {
      "epoch": 2.29,
      "learning_rate": 4.2536322060790845e-05,
      "loss": 0.5207,
      "step": 98500
    },
    {
      "epoch": 2.3,
      "learning_rate": 4.2497511122172794e-05,
      "loss": 0.5228,
      "step": 99000
    },
    {
      "epoch": 2.31,
      "learning_rate": 4.245862240612264e-05,
      "loss": 0.5207,
      "step": 99500
    },
    {
      "epoch": 2.32,
      "learning_rate": 4.241973369007249e-05,
      "loss": 0.5211,
      "step": 100000
    },
    {
      "epoch": 2.34,
      "learning_rate": 4.2380844974022335e-05,
      "loss": 0.5169,
      "step": 100500
    },
    {
      "epoch": 2.35,
      "learning_rate": 4.234195625797219e-05,
      "loss": 0.5302,
      "step": 101000
    },
    {
      "epoch": 2.36,
      "learning_rate": 4.230306754192204e-05,
      "loss": 0.5155,
      "step": 101500
    },
    {
      "epoch": 2.37,
      "learning_rate": 4.226417882587188e-05,
      "loss": 0.5193,
      "step": 102000
    },
    {
      "epoch": 2.38,
      "learning_rate": 4.2225290109821735e-05,
      "loss": 0.518,
      "step": 102500
    },
    {
      "epoch": 2.39,
      "learning_rate": 4.2186401393771586e-05,
      "loss": 0.5171,
      "step": 103000
    },
    {
      "epoch": 2.41,
      "learning_rate": 4.214751267772144e-05,
      "loss": 0.5208,
      "step": 103500
    },
    {
      "epoch": 2.42,
      "learning_rate": 4.210862396167128e-05,
      "loss": 0.5231,
      "step": 104000
    },
    {
      "epoch": 2.43,
      "learning_rate": 4.2069735245621134e-05,
      "loss": 0.5267,
      "step": 104500
    },
    {
      "epoch": 2.44,
      "learning_rate": 4.2030846529570985e-05,
      "loss": 0.5196,
      "step": 105000
    },
    {
      "epoch": 2.45,
      "learning_rate": 4.199195781352083e-05,
      "loss": 0.5234,
      "step": 105500
    },
    {
      "epoch": 2.46,
      "learning_rate": 4.1953224652334885e-05,
      "loss": 0.526,
      "step": 106000
    },
    {
      "epoch": 2.48,
      "learning_rate": 4.191449149114893e-05,
      "loss": 0.5312,
      "step": 106500
    },
    {
      "epoch": 2.49,
      "learning_rate": 4.187560277509878e-05,
      "loss": 0.5286,
      "step": 107000
    },
    {
      "epoch": 2.5,
      "learning_rate": 4.183671405904863e-05,
      "loss": 0.5225,
      "step": 107500
    },
    {
      "epoch": 2.51,
      "learning_rate": 4.1797825342998474e-05,
      "loss": 0.5207,
      "step": 108000
    },
    {
      "epoch": 2.52,
      "learning_rate": 4.1758936626948325e-05,
      "loss": 0.5263,
      "step": 108500
    },
    {
      "epoch": 2.53,
      "learning_rate": 4.172004791089818e-05,
      "loss": 0.5243,
      "step": 109000
    },
    {
      "epoch": 2.55,
      "learning_rate": 4.1681236972280126e-05,
      "loss": 0.5332,
      "step": 109500
    },
    {
      "epoch": 2.56,
      "learning_rate": 4.164234825622997e-05,
      "loss": 0.5297,
      "step": 110000
    },
    {
      "epoch": 2.57,
      "learning_rate": 4.160345954017982e-05,
      "loss": 0.5274,
      "step": 110500
    },
    {
      "epoch": 2.58,
      "learning_rate": 4.1564570824129674e-05,
      "loss": 0.5241,
      "step": 111000
    },
    {
      "epoch": 2.59,
      "learning_rate": 4.152568210807952e-05,
      "loss": 0.528,
      "step": 111500
    },
    {
      "epoch": 2.6,
      "learning_rate": 4.148679339202937e-05,
      "loss": 0.53,
      "step": 112000
    },
    {
      "epoch": 2.61,
      "learning_rate": 4.144790467597922e-05,
      "loss": 0.5291,
      "step": 112500
    },
    {
      "epoch": 2.63,
      "learning_rate": 4.1409015959929073e-05,
      "loss": 0.5431,
      "step": 113000
    },
    {
      "epoch": 2.64,
      "learning_rate": 4.137012724387892e-05,
      "loss": 0.5336,
      "step": 113500
    },
    {
      "epoch": 2.65,
      "learning_rate": 4.133123852782877e-05,
      "loss": 0.531,
      "step": 114000
    },
    {
      "epoch": 2.66,
      "learning_rate": 4.129234981177862e-05,
      "loss": 0.5278,
      "step": 114500
    },
    {
      "epoch": 2.67,
      "learning_rate": 4.1253461095728466e-05,
      "loss": 0.5214,
      "step": 115000
    },
    {
      "epoch": 2.68,
      "learning_rate": 4.121457237967831e-05,
      "loss": 0.5249,
      "step": 115500
    },
    {
      "epoch": 2.7,
      "learning_rate": 4.117568366362816e-05,
      "loss": 0.5351,
      "step": 116000
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.1136794947578014e-05,
      "loss": 0.5275,
      "step": 116500
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.109790623152786e-05,
      "loss": 0.5309,
      "step": 117000
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.105901751547771e-05,
      "loss": 0.5291,
      "step": 117500
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.102012879942756e-05,
      "loss": 0.5274,
      "step": 118000
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.098131786080951e-05,
      "loss": 0.532,
      "step": 118500
    },
    {
      "epoch": 2.77,
      "learning_rate": 4.0942429144759356e-05,
      "loss": 0.5278,
      "step": 119000
    },
    {
      "epoch": 2.78,
      "learning_rate": 4.090354042870921e-05,
      "loss": 0.5338,
      "step": 119500
    },
    {
      "epoch": 2.79,
      "learning_rate": 4.086465171265906e-05,
      "loss": 0.5289,
      "step": 120000
    },
    {
      "epoch": 2.8,
      "learning_rate": 4.082576299660891e-05,
      "loss": 0.5388,
      "step": 120500
    },
    {
      "epoch": 2.81,
      "learning_rate": 4.0786874280558755e-05,
      "loss": 0.5337,
      "step": 121000
    },
    {
      "epoch": 2.82,
      "learning_rate": 4.074798556450861e-05,
      "loss": 0.5308,
      "step": 121500
    },
    {
      "epoch": 2.84,
      "learning_rate": 4.0709174625890556e-05,
      "loss": 0.5294,
      "step": 122000
    },
    {
      "epoch": 2.85,
      "learning_rate": 4.067028590984041e-05,
      "loss": 0.5385,
      "step": 122500
    },
    {
      "epoch": 2.86,
      "learning_rate": 4.063139719379025e-05,
      "loss": 0.5379,
      "step": 123000
    },
    {
      "epoch": 2.87,
      "learning_rate": 4.05925084777401e-05,
      "loss": 0.5485,
      "step": 123500
    },
    {
      "epoch": 2.88,
      "learning_rate": 4.055361976168995e-05,
      "loss": 0.5391,
      "step": 124000
    },
    {
      "epoch": 2.89,
      "learning_rate": 4.05147310456398e-05,
      "loss": 0.5414,
      "step": 124500
    },
    {
      "epoch": 2.91,
      "learning_rate": 4.0475842329589645e-05,
      "loss": 0.5237,
      "step": 125000
    },
    {
      "epoch": 2.92,
      "learning_rate": 4.04369536135395e-05,
      "loss": 0.5363,
      "step": 125500
    },
    {
      "epoch": 2.93,
      "learning_rate": 4.039806489748935e-05,
      "loss": 0.543,
      "step": 126000
    },
    {
      "epoch": 2.94,
      "learning_rate": 4.0359331736303396e-05,
      "loss": 0.5265,
      "step": 126500
    },
    {
      "epoch": 2.95,
      "learning_rate": 4.032044302025325e-05,
      "loss": 0.5379,
      "step": 127000
    },
    {
      "epoch": 2.96,
      "learning_rate": 4.028155430420309e-05,
      "loss": 0.5311,
      "step": 127500
    },
    {
      "epoch": 2.98,
      "learning_rate": 4.0242665588152944e-05,
      "loss": 0.5361,
      "step": 128000
    },
    {
      "epoch": 2.99,
      "learning_rate": 4.0203776872102795e-05,
      "loss": 0.5405,
      "step": 128500
    },
    {
      "epoch": 3.0,
      "learning_rate": 4.016488815605265e-05,
      "loss": 0.537,
      "step": 129000
    },
    {
      "epoch": 3.0,
      "eval_bleu": 59.1835,
      "eval_gen_len": 15.7226,
      "eval_loss": 0.6417234539985657,
      "eval_runtime": 6645.2011,
      "eval_samples_per_second": 12.929,
      "eval_steps_per_second": 1.616,
      "step": 129072
    }
  ],
  "logging_steps": 500,
  "max_steps": 645360,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 15,
  "save_steps": 500,
  "total_flos": 4.4754523745125663e+18,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}