{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.5945861803404937,
  "eval_steps": 1000000,
  "global_step": 291948,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 5.000000000000001e-07,
      "loss": 9.9794,
      "step": 500
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 9.2638,
      "step": 1000
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.5e-06,
      "loss": 8.8815,
      "step": 1500
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 8.4333,
      "step": 2000
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.5e-06,
      "loss": 7.9493,
      "step": 2500
    },
    {
      "epoch": 0.02,
      "learning_rate": 3e-06,
      "loss": 7.5361,
      "step": 3000
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.5e-06,
      "loss": 7.2585,
      "step": 3500
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.000000000000001e-06,
      "loss": 7.0981,
      "step": 4000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.5e-06,
      "loss": 7.0153,
      "step": 4500
    },
    {
      "epoch": 0.03,
      "learning_rate": 5e-06,
      "loss": 6.9373,
      "step": 5000
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.500000000000001e-06,
      "loss": 6.8771,
      "step": 5500
    },
    {
      "epoch": 0.03,
      "learning_rate": 6e-06,
      "loss": 6.8348,
      "step": 6000
    },
    {
      "epoch": 0.04,
      "learning_rate": 6.5000000000000004e-06,
      "loss": 6.7952,
      "step": 6500
    },
    {
      "epoch": 0.04,
      "learning_rate": 7e-06,
      "loss": 6.7703,
      "step": 7000
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.500000000000001e-06,
      "loss": 6.734,
      "step": 7500
    },
    {
      "epoch": 0.04,
      "learning_rate": 8.000000000000001e-06,
      "loss": 6.7049,
      "step": 8000
    },
    {
      "epoch": 0.05,
      "learning_rate": 8.5e-06,
      "loss": 6.691,
      "step": 8500
    },
    {
      "epoch": 0.05,
      "learning_rate": 9e-06,
      "loss": 6.6627,
      "step": 9000
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.5e-06,
      "loss": 6.6363,
      "step": 9500
    },
    {
      "epoch": 0.05,
      "learning_rate": 1e-05,
      "loss": 6.6223,
      "step": 10000
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.999726756545549e-06,
      "loss": 6.5951,
      "step": 10500
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.999453513091095e-06,
      "loss": 6.5832,
      "step": 11000
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.999180269636641e-06,
      "loss": 6.5621,
      "step": 11500
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.998907026182189e-06,
      "loss": 6.5372,
      "step": 12000
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.998633782727735e-06,
      "loss": 6.5226,
      "step": 12500
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.998360539273283e-06,
      "loss": 6.5154,
      "step": 13000
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.99808729581883e-06,
      "loss": 6.4952,
      "step": 13500
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.997814052364376e-06,
      "loss": 6.4889,
      "step": 14000
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.997540808909924e-06,
      "loss": 6.4722,
      "step": 14500
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.99726756545547e-06,
      "loss": 6.476,
      "step": 15000
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.996994322001016e-06,
      "loss": 6.464,
      "step": 15500
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.996721078546564e-06,
      "loss": 6.4428,
      "step": 16000
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.996447835092112e-06,
      "loss": 6.4316,
      "step": 16500
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.996174591637658e-06,
      "loss": 6.4209,
      "step": 17000
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.995901348183204e-06,
      "loss": 6.4246,
      "step": 17500
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.995628104728752e-06,
      "loss": 6.4139,
      "step": 18000
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.995354861274299e-06,
      "loss": 6.4054,
      "step": 18500
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.995081617819847e-06,
      "loss": 6.3984,
      "step": 19000
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.994808374365393e-06,
      "loss": 6.3899,
      "step": 19500
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.994535130910939e-06,
      "loss": 6.3791,
      "step": 20000
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.994261887456487e-06,
      "loss": 6.385,
      "step": 20500
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.993988644002033e-06,
      "loss": 6.3711,
      "step": 21000
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.993715400547581e-06,
      "loss": 6.3632,
      "step": 21500
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.993442157093127e-06,
      "loss": 6.3601,
      "step": 22000
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.993168913638675e-06,
      "loss": 6.3542,
      "step": 22500
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.992895670184221e-06,
      "loss": 6.3442,
      "step": 23000
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.99262242672977e-06,
      "loss": 6.3412,
      "step": 23500
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.992349183275316e-06,
      "loss": 6.3376,
      "step": 24000
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.992075939820862e-06,
      "loss": 6.3354,
      "step": 24500
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.99180269636641e-06,
      "loss": 6.3251,
      "step": 25000
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.991529452911956e-06,
      "loss": 6.3238,
      "step": 25500
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.991256209457502e-06,
      "loss": 6.3216,
      "step": 26000
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.99098296600305e-06,
      "loss": 6.3202,
      "step": 26500
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.990709722548596e-06,
      "loss": 6.3066,
      "step": 27000
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.990436479094144e-06,
      "loss": 6.303,
      "step": 27500
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.990163235639692e-06,
      "loss": 6.2955,
      "step": 28000
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.989889992185238e-06,
      "loss": 6.2966,
      "step": 28500
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.989616748730785e-06,
      "loss": 6.2946,
      "step": 29000
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.989343505276333e-06,
      "loss": 6.2911,
      "step": 29500
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.989070261821879e-06,
      "loss": 6.2779,
      "step": 30000
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.988797018367425e-06,
      "loss": 6.2811,
      "step": 30500
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.988523774912973e-06,
      "loss": 6.2695,
      "step": 31000
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.98825053145852e-06,
      "loss": 6.2794,
      "step": 31500
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.987977288004067e-06,
      "loss": 6.2579,
      "step": 32000
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.987704044549613e-06,
      "loss": 6.2648,
      "step": 32500
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.98743080109516e-06,
      "loss": 6.2635,
      "step": 33000
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.987157557640708e-06,
      "loss": 6.2534,
      "step": 33500
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.986884314186255e-06,
      "loss": 6.2586,
      "step": 34000
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.986611070731802e-06,
      "loss": 6.2542,
      "step": 34500
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.986337827277348e-06,
      "loss": 6.2519,
      "step": 35000
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.986064583822896e-06,
      "loss": 6.2478,
      "step": 35500
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.985791340368442e-06,
      "loss": 6.2481,
      "step": 36000
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.98551809691399e-06,
      "loss": 6.2311,
      "step": 36500
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.985244853459536e-06,
      "loss": 6.2388,
      "step": 37000
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.984971610005082e-06,
      "loss": 6.2361,
      "step": 37500
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.98469836655063e-06,
      "loss": 6.227,
      "step": 38000
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.984425123096178e-06,
      "loss": 6.2394,
      "step": 38500
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.984151879641723e-06,
      "loss": 6.2325,
      "step": 39000
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.98387863618727e-06,
      "loss": 6.23,
      "step": 39500
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.983605392732819e-06,
      "loss": 6.2181,
      "step": 40000
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.983332149278365e-06,
      "loss": 6.2112,
      "step": 40500
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.983058905823911e-06,
      "loss": 6.2167,
      "step": 41000
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.982785662369459e-06,
      "loss": 6.2126,
      "step": 41500
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.982512418915005e-06,
      "loss": 6.2134,
      "step": 42000
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.982239175460553e-06,
      "loss": 6.2119,
      "step": 42500
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.9819659320061e-06,
      "loss": 6.206,
      "step": 43000
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.981692688551646e-06,
      "loss": 6.2065,
      "step": 43500
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.981419445097194e-06,
      "loss": 6.2027,
      "step": 44000
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.981146201642742e-06,
      "loss": 6.2028,
      "step": 44500
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.980872958188288e-06,
      "loss": 6.2003,
      "step": 45000
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.980599714733834e-06,
      "loss": 6.1951,
      "step": 45500
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.980326471279382e-06,
      "loss": 6.1924,
      "step": 46000
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.980053227824928e-06,
      "loss": 6.1826,
      "step": 46500
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.979779984370476e-06,
      "loss": 6.1876,
      "step": 47000
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.979506740916022e-06,
      "loss": 6.1793,
      "step": 47500
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.979233497461569e-06,
      "loss": 6.1782,
      "step": 48000
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.978960254007116e-06,
      "loss": 6.1666,
      "step": 48500
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.978687010552663e-06,
      "loss": 6.148,
      "step": 49000
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.978413767098209e-06,
      "loss": 6.1284,
      "step": 49500
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.978140523643757e-06,
      "loss": 6.1104,
      "step": 50000
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.977867280189305e-06,
      "loss": 6.0938,
      "step": 50500
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.977594036734851e-06,
      "loss": 6.0888,
      "step": 51000
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.977320793280397e-06,
      "loss": 6.0751,
      "step": 51500
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.977047549825945e-06,
      "loss": 6.0658,
      "step": 52000
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.976774306371491e-06,
      "loss": 6.0523,
      "step": 52500
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.97650106291704e-06,
      "loss": 6.0572,
      "step": 53000
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.976227819462586e-06,
      "loss": 6.0354,
      "step": 53500
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.975954576008132e-06,
      "loss": 6.0309,
      "step": 54000
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.97568133255368e-06,
      "loss": 6.0157,
      "step": 54500
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.975408089099226e-06,
      "loss": 6.0187,
      "step": 55000
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.975134845644774e-06,
      "loss": 6.0067,
      "step": 55500
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.97486160219032e-06,
      "loss": 6.0087,
      "step": 56000
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.974588358735868e-06,
      "loss": 5.987,
      "step": 56500
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.974315115281414e-06,
      "loss": 5.9804,
      "step": 57000
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.974041871826962e-06,
      "loss": 5.9881,
      "step": 57500
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.973768628372508e-06,
      "loss": 5.9743,
      "step": 58000
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.973495384918055e-06,
      "loss": 5.9702,
      "step": 58500
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.973222141463603e-06,
      "loss": 5.9545,
      "step": 59000
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.972948898009149e-06,
      "loss": 5.9549,
      "step": 59500
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.972675654554695e-06,
      "loss": 5.9456,
      "step": 60000
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.972402411100243e-06,
      "loss": 5.9337,
      "step": 60500
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.97212916764579e-06,
      "loss": 5.9409,
      "step": 61000
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.971855924191337e-06,
      "loss": 5.9309,
      "step": 61500
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.971582680736885e-06,
      "loss": 5.9318,
      "step": 62000
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.97130943728243e-06,
      "loss": 5.9146,
      "step": 62500
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.971036193827977e-06,
      "loss": 5.9062,
      "step": 63000
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.970762950373525e-06,
      "loss": 5.9043,
      "step": 63500
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.970489706919072e-06,
      "loss": 5.9019,
      "step": 64000
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.970216463464618e-06,
      "loss": 5.8974,
      "step": 64500
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.969943220010166e-06,
      "loss": 5.8888,
      "step": 65000
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.969669976555712e-06,
      "loss": 5.8834,
      "step": 65500
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.96939673310126e-06,
      "loss": 5.88,
      "step": 66000
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.969123489646806e-06,
      "loss": 5.8647,
      "step": 66500
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.968850246192352e-06,
      "loss": 5.8582,
      "step": 67000
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.9685770027379e-06,
      "loss": 5.8564,
      "step": 67500
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.968303759283448e-06,
      "loss": 5.8396,
      "step": 68000
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.968030515828993e-06,
      "loss": 5.8455,
      "step": 68500
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.96775727237454e-06,
      "loss": 5.8295,
      "step": 69000
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.967484028920089e-06,
      "loss": 5.8198,
      "step": 69500
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.967210785465635e-06,
      "loss": 5.8244,
      "step": 70000
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.966937542011183e-06,
      "loss": 5.8146,
      "step": 70500
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.966664298556729e-06,
      "loss": 5.8154,
      "step": 71000
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.966391055102275e-06,
      "loss": 5.8054,
      "step": 71500
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.966117811647823e-06,
      "loss": 5.7896,
      "step": 72000
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.96584456819337e-06,
      "loss": 5.7773,
      "step": 72500
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.965571324738916e-06,
      "loss": 5.7697,
      "step": 73000
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.965298081284464e-06,
      "loss": 5.7627,
      "step": 73500
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.965024837830011e-06,
      "loss": 5.7638,
      "step": 74000
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.964751594375558e-06,
      "loss": 5.7512,
      "step": 74500
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.964478350921104e-06,
      "loss": 5.7316,
      "step": 75000
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.964205107466652e-06,
      "loss": 5.7246,
      "step": 75500
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.963931864012198e-06,
      "loss": 5.7091,
      "step": 76000
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.963658620557746e-06,
      "loss": 5.6898,
      "step": 76500
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.963385377103292e-06,
      "loss": 5.6588,
      "step": 77000
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.963112133648838e-06,
      "loss": 5.6267,
      "step": 77500
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.962838890194386e-06,
      "loss": 5.59,
      "step": 78000
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.962565646739933e-06,
      "loss": 5.5757,
      "step": 78500
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.96229240328548e-06,
      "loss": 5.5375,
      "step": 79000
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.962019159831027e-06,
      "loss": 5.5121,
      "step": 79500
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.961745916376575e-06,
      "loss": 5.4834,
      "step": 80000
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.961472672922121e-06,
      "loss": 5.4608,
      "step": 80500
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.961199429467669e-06,
      "loss": 5.4326,
      "step": 81000
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.960926186013215e-06,
      "loss": 5.3968,
      "step": 81500
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.960652942558761e-06,
      "loss": 5.3704,
      "step": 82000
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.96037969910431e-06,
      "loss": 5.3503,
      "step": 82500
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.960106455649855e-06,
      "loss": 5.3166,
      "step": 83000
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.959833212195402e-06,
      "loss": 5.2936,
      "step": 83500
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.95955996874095e-06,
      "loss": 5.2733,
      "step": 84000
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.959286725286496e-06,
      "loss": 5.2458,
      "step": 84500
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.959013481832044e-06,
      "loss": 5.2177,
      "step": 85000
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.95874023837759e-06,
      "loss": 5.1938,
      "step": 85500
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.958466994923138e-06,
      "loss": 5.18,
      "step": 86000
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.958193751468684e-06,
      "loss": 5.1442,
      "step": 86500
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.957920508014232e-06,
      "loss": 5.1153,
      "step": 87000
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.957647264559778e-06,
      "loss": 5.1022,
      "step": 87500
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.957374021105325e-06,
      "loss": 5.0894,
      "step": 88000
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.957100777650872e-06,
      "loss": 5.0666,
      "step": 88500
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.956827534196419e-06,
      "loss": 5.0436,
      "step": 89000
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.956554290741967e-06,
      "loss": 5.0226,
      "step": 89500
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.956281047287513e-06,
      "loss": 5.0071,
      "step": 90000
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.956007803833059e-06,
      "loss": 4.9917,
      "step": 90500
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.955734560378607e-06,
      "loss": 4.9743,
      "step": 91000
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.955461316924155e-06,
      "loss": 4.9533,
      "step": 91500
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.955188073469701e-06,
      "loss": 4.9361,
      "step": 92000
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.954914830015247e-06,
      "loss": 4.9181,
      "step": 92500
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.954641586560795e-06,
      "loss": 4.9031,
      "step": 93000
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.954368343106342e-06,
      "loss": 4.8802,
      "step": 93500
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.954095099651888e-06,
      "loss": 4.8676,
      "step": 94000
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.953821856197436e-06,
      "loss": 4.8537,
      "step": 94500
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.953548612742982e-06,
      "loss": 4.8418,
      "step": 95000
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.95327536928853e-06,
      "loss": 4.8144,
      "step": 95500
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.953002125834078e-06,
      "loss": 4.8046,
      "step": 96000
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.952728882379622e-06,
      "loss": 4.7935,
      "step": 96500
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.95245563892517e-06,
      "loss": 4.7738,
      "step": 97000
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.952182395470718e-06,
      "loss": 4.7698,
      "step": 97500
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.951909152016264e-06,
      "loss": 4.7546,
      "step": 98000
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.95163590856181e-06,
      "loss": 4.7379,
      "step": 98500
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.951362665107359e-06,
      "loss": 4.7187,
      "step": 99000
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.951089421652905e-06,
      "loss": 4.7022,
      "step": 99500
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.950816178198453e-06,
      "loss": 4.701,
      "step": 100000
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.950542934743999e-06,
      "loss": 4.6832,
      "step": 100500
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.950269691289545e-06,
      "loss": 4.6588,
      "step": 101000
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.949996447835093e-06,
      "loss": 4.6484,
      "step": 101500
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.949723204380641e-06,
      "loss": 4.6276,
      "step": 102000
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.949449960926186e-06,
      "loss": 4.6143,
      "step": 102500
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.949176717471734e-06,
      "loss": 4.6045,
      "step": 103000
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.948903474017281e-06,
      "loss": 4.5886,
      "step": 103500
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.948630230562828e-06,
      "loss": 4.5653,
      "step": 104000
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.948356987108376e-06,
      "loss": 4.5474,
      "step": 104500
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.948083743653922e-06,
      "loss": 4.541,
      "step": 105000
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.947810500199468e-06,
      "loss": 4.5204,
      "step": 105500
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.947537256745016e-06,
      "loss": 4.4836,
      "step": 106000
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.947264013290562e-06,
      "loss": 4.451,
      "step": 106500
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.946990769836108e-06,
      "loss": 4.411,
      "step": 107000
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.946717526381656e-06,
      "loss": 4.3949,
      "step": 107500
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.946444282927203e-06,
      "loss": 4.3825,
      "step": 108000
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.94617103947275e-06,
      "loss": 4.3509,
      "step": 108500
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.945897796018297e-06,
      "loss": 4.3206,
      "step": 109000
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.945624552563845e-06,
      "loss": 4.3133,
      "step": 109500
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.945351309109391e-06,
      "loss": 4.2757,
      "step": 110000
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.945078065654939e-06,
      "loss": 4.2678,
      "step": 110500
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.944804822200485e-06,
      "loss": 4.2408,
      "step": 111000
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.944531578746031e-06,
      "loss": 4.2204,
      "step": 111500
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.94425833529158e-06,
      "loss": 4.2066,
      "step": 112000
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.943985091837125e-06,
      "loss": 4.182,
      "step": 112500
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.943711848382673e-06,
      "loss": 4.167,
      "step": 113000
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.94343860492822e-06,
      "loss": 4.1427,
      "step": 113500
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.943165361473766e-06,
      "loss": 4.1328,
      "step": 114000
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.942892118019314e-06,
      "loss": 4.1197,
      "step": 114500
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.942618874564862e-06,
      "loss": 4.09,
      "step": 115000
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.942345631110408e-06,
      "loss": 4.0778,
      "step": 115500
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.942072387655954e-06,
      "loss": 4.0566,
      "step": 116000
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.941799144201502e-06,
      "loss": 4.0456,
      "step": 116500
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.941525900747048e-06,
      "loss": 4.0192,
      "step": 117000
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.941252657292595e-06,
      "loss": 4.0168,
      "step": 117500
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.940979413838142e-06,
      "loss": 3.9962,
      "step": 118000
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.940706170383689e-06,
      "loss": 3.9763,
      "step": 118500
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.940432926929237e-06,
      "loss": 3.9619,
      "step": 119000
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.940159683474783e-06,
      "loss": 3.9564,
      "step": 119500
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.939886440020329e-06,
      "loss": 3.9373,
      "step": 120000
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.939613196565877e-06,
      "loss": 3.9195,
      "step": 120500
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.939339953111425e-06,
      "loss": 3.909,
      "step": 121000
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.939066709656971e-06,
      "loss": 3.8979,
      "step": 121500
    },
    {
      "epoch": 0.67,
      "learning_rate": 9.938793466202517e-06,
      "loss": 3.8775,
      "step": 122000
    },
    {
      "epoch": 0.67,
      "learning_rate": 9.938520222748065e-06,
      "loss": 3.8646,
      "step": 122500
    },
    {
      "epoch": 0.67,
      "learning_rate": 9.938246979293612e-06,
      "loss": 3.8639,
      "step": 123000
    },
    {
      "epoch": 0.67,
      "learning_rate": 9.93797373583916e-06,
      "loss": 3.8447,
      "step": 123500
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.937700492384706e-06,
      "loss": 3.8194,
      "step": 124000
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.937427248930252e-06,
      "loss": 3.8177,
      "step": 124500
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.9371540054758e-06,
      "loss": 3.8089,
      "step": 125000
    },
    {
      "epoch": 0.69,
      "learning_rate": 9.936880762021348e-06,
      "loss": 3.7943,
      "step": 125500
    },
    {
      "epoch": 0.69,
      "learning_rate": 9.936607518566892e-06,
      "loss": 3.7813,
      "step": 126000
    },
    {
      "epoch": 0.69,
      "learning_rate": 9.93633427511244e-06,
      "loss": 3.7618,
      "step": 126500
    },
    {
      "epoch": 0.69,
      "learning_rate": 9.936061031657988e-06,
      "loss": 3.7512,
      "step": 127000
    },
    {
      "epoch": 0.7,
      "learning_rate": 9.935787788203534e-06,
      "loss": 3.7484,
      "step": 127500
    },
    {
      "epoch": 0.7,
      "learning_rate": 9.93551454474908e-06,
      "loss": 3.7371,
      "step": 128000
    },
    {
      "epoch": 0.7,
      "learning_rate": 9.935241301294629e-06,
      "loss": 3.7187,
      "step": 128500
    },
    {
      "epoch": 0.7,
      "learning_rate": 9.934968057840175e-06,
      "loss": 3.7204,
      "step": 129000
    },
    {
      "epoch": 0.71,
      "learning_rate": 9.934694814385723e-06,
      "loss": 3.6997,
      "step": 129500
    },
    {
      "epoch": 0.71,
      "learning_rate": 9.934421570931269e-06,
      "loss": 3.6815,
      "step": 130000
    },
    {
      "epoch": 0.71,
      "learning_rate": 9.934148327476815e-06,
      "loss": 3.6848,
      "step": 130500
    },
    {
      "epoch": 0.72,
      "learning_rate": 9.933875084022363e-06,
      "loss": 3.6648,
      "step": 131000
    },
    {
      "epoch": 0.72,
      "learning_rate": 9.933601840567911e-06,
      "loss": 3.6632,
      "step": 131500
    },
    {
      "epoch": 0.72,
      "learning_rate": 9.933328597113457e-06,
      "loss": 3.6442,
      "step": 132000
    },
    {
      "epoch": 0.72,
      "learning_rate": 9.933055353659003e-06,
      "loss": 3.6456,
      "step": 132500
    },
    {
      "epoch": 0.73,
      "learning_rate": 9.932782110204551e-06,
      "loss": 3.6306,
      "step": 133000
    },
    {
      "epoch": 0.73,
      "learning_rate": 9.932508866750098e-06,
      "loss": 3.6161,
      "step": 133500
    },
    {
      "epoch": 0.73,
      "learning_rate": 9.932235623295646e-06,
      "loss": 3.61,
      "step": 134000
    },
    {
      "epoch": 0.73,
      "learning_rate": 9.931962379841192e-06,
      "loss": 3.5987,
      "step": 134500
    },
    {
      "epoch": 0.74,
      "learning_rate": 9.931689136386738e-06,
      "loss": 3.5921,
      "step": 135000
    },
    {
      "epoch": 0.74,
      "learning_rate": 9.931415892932286e-06,
      "loss": 3.5877,
      "step": 135500
    },
    {
      "epoch": 0.74,
      "learning_rate": 9.931142649477832e-06,
      "loss": 3.5808,
      "step": 136000
    },
    {
      "epoch": 0.75,
      "learning_rate": 9.930869406023378e-06,
      "loss": 3.5727,
      "step": 136500
    },
    {
      "epoch": 0.75,
      "learning_rate": 9.930596162568926e-06,
      "loss": 3.5528,
      "step": 137000
    },
    {
      "epoch": 0.75,
      "learning_rate": 9.930322919114474e-06,
      "loss": 3.5536,
      "step": 137500
    },
    {
      "epoch": 0.75,
      "learning_rate": 9.93004967566002e-06,
      "loss": 3.538,
      "step": 138000
    },
    {
      "epoch": 0.76,
      "learning_rate": 9.929776432205568e-06,
      "loss": 3.531,
      "step": 138500
    },
    {
      "epoch": 0.76,
      "learning_rate": 9.929503188751115e-06,
      "loss": 3.5204,
      "step": 139000
    },
    {
      "epoch": 0.76,
      "learning_rate": 9.92922994529666e-06,
      "loss": 3.5167,
      "step": 139500
    },
    {
      "epoch": 0.76,
      "learning_rate": 9.928956701842209e-06,
      "loss": 3.513,
      "step": 140000
    },
    {
      "epoch": 0.77,
      "learning_rate": 9.928683458387755e-06,
      "loss": 3.5084,
      "step": 140500
    },
    {
      "epoch": 0.77,
      "learning_rate": 9.928410214933301e-06,
      "loss": 3.4931,
      "step": 141000
    },
    {
      "epoch": 0.77,
      "learning_rate": 9.928136971478849e-06,
      "loss": 3.4863,
      "step": 141500
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.927863728024395e-06,
      "loss": 3.4663,
      "step": 142000
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.927590484569943e-06,
      "loss": 3.4624,
      "step": 142500
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.92731724111549e-06,
      "loss": 3.4671,
      "step": 143000
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.927043997661037e-06,
      "loss": 3.4598,
      "step": 143500
    },
    {
      "epoch": 0.79,
      "learning_rate": 9.926770754206584e-06,
      "loss": 3.4451,
      "step": 144000
    },
    {
      "epoch": 0.79,
      "learning_rate": 9.926497510752132e-06,
      "loss": 3.4398,
      "step": 144500
    },
    {
      "epoch": 0.79,
      "learning_rate": 9.926224267297678e-06,
      "loss": 3.4342,
      "step": 145000
    },
    {
      "epoch": 0.79,
      "learning_rate": 9.925951023843224e-06,
      "loss": 3.4244,
      "step": 145500
    },
    {
      "epoch": 0.8,
      "learning_rate": 9.925677780388772e-06,
      "loss": 3.4177,
      "step": 146000
    },
    {
      "epoch": 0.8,
      "learning_rate": 9.925404536934318e-06,
      "loss": 3.4139,
      "step": 146500
    },
    {
      "epoch": 0.8,
      "learning_rate": 9.925131293479866e-06,
      "loss": 3.4151,
      "step": 147000
    },
    {
      "epoch": 0.81,
      "learning_rate": 9.924858050025412e-06,
      "loss": 3.4056,
      "step": 147500
    },
    {
      "epoch": 0.81,
      "learning_rate": 9.924584806570959e-06,
      "loss": 3.4019,
      "step": 148000
    },
    {
      "epoch": 0.81,
      "learning_rate": 9.924311563116507e-06,
      "loss": 3.3941,
      "step": 148500
    },
    {
      "epoch": 0.81,
      "learning_rate": 9.924038319662054e-06,
      "loss": 3.3892,
      "step": 149000
    },
    {
      "epoch": 0.82,
      "learning_rate": 9.9237650762076e-06,
      "loss": 3.3853,
      "step": 149500
    },
    {
      "epoch": 0.82,
      "learning_rate": 9.923491832753147e-06,
      "loss": 3.3778,
      "step": 150000
    },
    {
      "epoch": 0.82,
      "learning_rate": 9.923218589298695e-06,
      "loss": 3.3689,
      "step": 150500
    },
    {
      "epoch": 0.82,
      "learning_rate": 9.922945345844241e-06,
      "loss": 3.361,
      "step": 151000
    },
    {
      "epoch": 0.83,
      "learning_rate": 9.922672102389787e-06,
      "loss": 3.3556,
      "step": 151500
    },
    {
      "epoch": 0.83,
      "learning_rate": 9.922398858935335e-06,
      "loss": 3.3438,
      "step": 152000
    },
    {
      "epoch": 0.83,
      "learning_rate": 9.922125615480881e-06,
      "loss": 3.347,
      "step": 152500
    },
    {
      "epoch": 0.84,
      "learning_rate": 9.92185237202643e-06,
      "loss": 3.3402,
      "step": 153000
    },
    {
      "epoch": 0.84,
      "learning_rate": 9.921579128571976e-06,
      "loss": 3.3374,
      "step": 153500
    },
    {
      "epoch": 0.84,
      "learning_rate": 9.921305885117522e-06,
      "loss": 3.3372,
      "step": 154000
    },
    {
      "epoch": 0.84,
      "learning_rate": 9.92103264166307e-06,
      "loss": 3.3291,
      "step": 154500
    },
    {
      "epoch": 0.85,
      "learning_rate": 9.920759398208618e-06,
      "loss": 3.3249,
      "step": 155000
    },
    {
      "epoch": 0.85,
      "learning_rate": 9.920486154754164e-06,
      "loss": 3.3149,
      "step": 155500
    },
    {
      "epoch": 0.85,
      "learning_rate": 9.92021291129971e-06,
      "loss": 3.3108,
      "step": 156000
    },
    {
      "epoch": 0.85,
      "learning_rate": 9.919939667845258e-06,
      "loss": 3.298,
      "step": 156500
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.919666424390804e-06,
      "loss": 3.2832,
      "step": 157000
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.919393180936352e-06,
      "loss": 3.29,
      "step": 157500
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.919119937481898e-06,
      "loss": 3.2901,
      "step": 158000
    },
    {
      "epoch": 0.87,
      "learning_rate": 9.918846694027445e-06,
      "loss": 3.2896,
      "step": 158500
    },
    {
      "epoch": 0.87,
      "learning_rate": 9.918573450572993e-06,
      "loss": 3.282,
      "step": 159000
    },
    {
      "epoch": 0.87,
      "learning_rate": 9.918300207118539e-06,
      "loss": 3.267,
      "step": 159500
    },
    {
      "epoch": 0.87,
      "learning_rate": 9.918026963664085e-06,
      "loss": 3.2681,
      "step": 160000
    },
    {
      "epoch": 0.88,
      "learning_rate": 9.917753720209633e-06,
      "loss": 3.2678,
      "step": 160500
    },
    {
      "epoch": 0.88,
      "learning_rate": 9.917480476755181e-06,
      "loss": 3.2638,
      "step": 161000
    },
    {
      "epoch": 0.88,
      "learning_rate": 9.917207233300727e-06,
      "loss": 3.2517,
      "step": 161500
    },
    {
      "epoch": 0.88,
      "learning_rate": 9.916933989846273e-06,
      "loss": 3.244,
      "step": 162000
    },
    {
      "epoch": 0.89,
      "learning_rate": 9.916660746391821e-06,
      "loss": 3.248,
      "step": 162500
    },
    {
      "epoch": 0.89,
      "learning_rate": 9.916387502937368e-06,
      "loss": 3.242,
      "step": 163000
    },
    {
      "epoch": 0.89,
      "learning_rate": 9.916114259482915e-06,
      "loss": 3.2385,
      "step": 163500
    },
    {
      "epoch": 0.9,
      "learning_rate": 9.915841016028462e-06,
      "loss": 3.229,
      "step": 164000
    },
    {
      "epoch": 0.9,
      "learning_rate": 9.915567772574008e-06,
      "loss": 3.2348,
      "step": 164500
    },
    {
      "epoch": 0.9,
      "learning_rate": 9.915294529119556e-06,
      "loss": 3.2281,
      "step": 165000
    },
    {
      "epoch": 0.9,
      "learning_rate": 9.915021285665102e-06,
      "loss": 3.2257,
      "step": 165500
    },
    {
      "epoch": 0.91,
      "learning_rate": 9.91474804221065e-06,
      "loss": 3.2176,
      "step": 166000
    },
    {
      "epoch": 0.91,
      "learning_rate": 9.914474798756196e-06,
      "loss": 3.2039,
      "step": 166500
    },
    {
      "epoch": 0.91,
      "learning_rate": 9.914201555301744e-06,
      "loss": 3.2046,
      "step": 167000
    },
    {
      "epoch": 0.91,
      "learning_rate": 9.91392831184729e-06,
      "loss": 3.2085,
      "step": 167500
    },
    {
      "epoch": 0.92,
      "learning_rate": 9.913655068392838e-06,
      "loss": 3.2028,
      "step": 168000
    },
    {
      "epoch": 0.92,
      "learning_rate": 9.913381824938385e-06,
      "loss": 3.1918,
      "step": 168500
    },
    {
      "epoch": 0.92,
      "learning_rate": 9.91310858148393e-06,
      "loss": 3.1765,
      "step": 169000
    },
    {
      "epoch": 0.93,
      "learning_rate": 9.912835338029479e-06,
      "loss": 3.1908,
      "step": 169500
    },
    {
      "epoch": 0.93,
      "learning_rate": 9.912562094575025e-06,
      "loss": 3.1877,
      "step": 170000
    },
    {
      "epoch": 0.93,
      "learning_rate": 9.912288851120571e-06,
      "loss": 3.1704,
      "step": 170500
    },
    {
      "epoch": 0.93,
      "learning_rate": 9.912015607666119e-06,
      "loss": 3.1809,
      "step": 171000
    },
    {
      "epoch": 0.94,
      "learning_rate": 9.911742364211665e-06,
      "loss": 3.169,
      "step": 171500
    },
    {
      "epoch": 0.94,
      "learning_rate": 9.911469120757213e-06,
      "loss": 3.1723,
      "step": 172000
    },
    {
      "epoch": 0.94,
      "learning_rate": 9.911195877302761e-06,
      "loss": 3.1625,
      "step": 172500
    },
    {
      "epoch": 0.94,
      "learning_rate": 9.910922633848307e-06,
      "loss": 3.1542,
      "step": 173000
    },
    {
      "epoch": 0.95,
      "learning_rate": 9.910649390393854e-06,
      "loss": 3.1579,
      "step": 173500
    },
    {
      "epoch": 0.95,
      "learning_rate": 9.910376146939402e-06,
      "loss": 3.1533,
      "step": 174000
    },
    {
      "epoch": 0.95,
      "learning_rate": 9.910102903484948e-06,
      "loss": 3.1527,
      "step": 174500
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.909829660030494e-06,
      "loss": 3.1454,
      "step": 175000
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.909556416576042e-06,
      "loss": 3.1412,
      "step": 175500
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.909283173121588e-06,
      "loss": 3.1366,
      "step": 176000
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.909009929667136e-06,
      "loss": 3.1403,
      "step": 176500
    },
    {
      "epoch": 0.97,
      "learning_rate": 9.908736686212682e-06,
      "loss": 3.1291,
      "step": 177000
    },
    {
      "epoch": 0.97,
      "learning_rate": 9.908463442758229e-06,
      "loss": 3.1344,
      "step": 177500
    },
    {
      "epoch": 0.97,
      "learning_rate": 9.908190199303776e-06,
      "loss": 3.1226,
      "step": 178000
    },
    {
      "epoch": 0.97,
      "learning_rate": 9.907916955849324e-06,
      "loss": 3.1289,
      "step": 178500
    },
    {
      "epoch": 0.98,
      "learning_rate": 9.90764371239487e-06,
      "loss": 3.1174,
      "step": 179000
    },
    {
      "epoch": 0.98,
      "learning_rate": 9.907370468940417e-06,
      "loss": 3.11,
      "step": 179500
    },
    {
      "epoch": 0.98,
      "learning_rate": 9.907097225485965e-06,
      "loss": 3.1201,
      "step": 180000
    },
    {
      "epoch": 0.99,
      "learning_rate": 9.906823982031511e-06,
      "loss": 3.0998,
      "step": 180500
    },
    {
      "epoch": 0.99,
      "learning_rate": 9.906550738577059e-06,
      "loss": 3.1008,
      "step": 181000
    },
    {
      "epoch": 0.99,
      "learning_rate": 9.906277495122605e-06,
      "loss": 3.1008,
      "step": 181500
    },
    {
      "epoch": 0.99,
      "learning_rate": 9.906004251668151e-06,
      "loss": 3.0861,
      "step": 182000
    },
    {
      "epoch": 1.0,
      "learning_rate": 9.9057310082137e-06,
      "loss": 3.099,
      "step": 182500
    },
    {
      "epoch": 1.0,
      "learning_rate": 9.905457764759247e-06,
      "loss": 3.0926,
      "step": 183000
    },
    {
      "epoch": 1.0,
      "learning_rate": 9.905184521304792e-06,
      "loss": 3.0864,
      "step": 183500
    },
    {
      "epoch": 1.0,
      "learning_rate": 9.90491127785034e-06,
      "loss": 3.0768,
      "step": 184000
    },
    {
      "epoch": 1.01,
      "learning_rate": 9.904638034395888e-06,
      "loss": 3.068,
      "step": 184500
    },
    {
      "epoch": 1.01,
      "learning_rate": 9.904364790941434e-06,
      "loss": 3.0731,
      "step": 185000
    },
    {
      "epoch": 1.01,
      "learning_rate": 9.90409154748698e-06,
      "loss": 3.0759,
      "step": 185500
    },
    {
      "epoch": 1.02,
      "learning_rate": 9.903818304032528e-06,
      "loss": 3.0576,
      "step": 186000
    },
    {
      "epoch": 1.02,
      "learning_rate": 9.903545060578074e-06,
      "loss": 3.0694,
      "step": 186500
    },
    {
      "epoch": 1.02,
      "learning_rate": 9.903271817123622e-06,
      "loss": 3.0627,
      "step": 187000
    },
    {
      "epoch": 1.02,
      "learning_rate": 9.902998573669168e-06,
      "loss": 3.0589,
      "step": 187500
    },
    {
      "epoch": 1.03,
      "learning_rate": 9.902725330214715e-06,
      "loss": 3.0554,
      "step": 188000
    },
    {
      "epoch": 1.03,
      "learning_rate": 9.902452086760263e-06,
      "loss": 3.0633,
      "step": 188500
    },
    {
      "epoch": 1.03,
      "learning_rate": 9.90217884330581e-06,
      "loss": 3.047,
      "step": 189000
    },
    {
      "epoch": 1.04,
      "learning_rate": 9.901905599851357e-06,
      "loss": 3.0462,
      "step": 189500
    },
    {
      "epoch": 1.04,
      "learning_rate": 9.901632356396903e-06,
      "loss": 3.0404,
      "step": 190000
    },
    {
      "epoch": 1.04,
      "learning_rate": 9.901359112942451e-06,
      "loss": 3.0442,
      "step": 190500
    },
    {
      "epoch": 1.04,
      "learning_rate": 9.901085869487997e-06,
      "loss": 3.0326,
      "step": 191000
    },
    {
      "epoch": 1.05,
      "learning_rate": 9.900812626033545e-06,
      "loss": 3.035,
      "step": 191500
    },
    {
      "epoch": 1.05,
      "learning_rate": 9.900539382579091e-06,
      "loss": 3.0382,
      "step": 192000
    },
    {
      "epoch": 1.05,
      "learning_rate": 9.900266139124637e-06,
      "loss": 3.0211,
      "step": 192500
    },
    {
      "epoch": 1.05,
      "learning_rate": 9.899992895670185e-06,
      "loss": 3.022,
      "step": 193000
    },
    {
      "epoch": 1.06,
      "learning_rate": 9.899719652215732e-06,
      "loss": 3.0279,
      "step": 193500
    },
    {
      "epoch": 1.06,
      "learning_rate": 9.899446408761278e-06,
      "loss": 3.0228,
      "step": 194000
    },
    {
      "epoch": 1.06,
      "learning_rate": 9.899173165306826e-06,
      "loss": 3.02,
      "step": 194500
    },
    {
      "epoch": 1.07,
      "learning_rate": 9.898899921852374e-06,
      "loss": 3.0146,
      "step": 195000
    },
    {
      "epoch": 1.07,
      "learning_rate": 9.89862667839792e-06,
      "loss": 3.0106,
      "step": 195500
    },
    {
      "epoch": 1.07,
      "learning_rate": 9.898353434943466e-06,
      "loss": 2.9964,
      "step": 196000
    },
    {
      "epoch": 1.07,
      "learning_rate": 9.898080191489014e-06,
      "loss": 3.0088,
      "step": 196500
    },
    {
      "epoch": 1.08,
      "learning_rate": 9.89780694803456e-06,
      "loss": 3.0076,
      "step": 197000
    },
    {
      "epoch": 1.08,
      "learning_rate": 9.897533704580108e-06,
      "loss": 2.9977,
      "step": 197500
    },
    {
      "epoch": 1.08,
      "learning_rate": 9.897260461125655e-06,
      "loss": 2.992,
      "step": 198000
    },
    {
      "epoch": 1.08,
      "learning_rate": 9.8969872176712e-06,
      "loss": 2.9985,
      "step": 198500
    },
    {
      "epoch": 1.09,
      "learning_rate": 9.896713974216749e-06,
      "loss": 2.9903,
      "step": 199000
    },
    {
      "epoch": 1.09,
      "learning_rate": 9.896440730762295e-06,
      "loss": 2.9923,
      "step": 199500
    },
    {
      "epoch": 1.09,
      "learning_rate": 9.896167487307843e-06,
      "loss": 2.9811,
      "step": 200000
    },
    {
      "epoch": 1.1,
      "learning_rate": 9.895894243853389e-06,
      "loss": 2.9821,
      "step": 200500
    },
    {
      "epoch": 1.1,
      "learning_rate": 9.895621000398937e-06,
      "loss": 2.9876,
      "step": 201000
    },
    {
      "epoch": 1.1,
      "learning_rate": 9.895347756944483e-06,
      "loss": 2.9726,
      "step": 201500
    },
    {
      "epoch": 1.1,
      "learning_rate": 9.895074513490031e-06,
      "loss": 2.9793,
      "step": 202000
    },
    {
      "epoch": 1.11,
      "learning_rate": 9.894801270035577e-06,
      "loss": 2.9729,
      "step": 202500
    },
    {
      "epoch": 1.11,
      "learning_rate": 9.894528026581124e-06,
      "loss": 2.9731,
      "step": 203000
    },
    {
      "epoch": 1.11,
      "learning_rate": 9.894254783126672e-06,
      "loss": 2.9761,
      "step": 203500
    },
    {
      "epoch": 1.11,
      "learning_rate": 9.893981539672218e-06,
      "loss": 2.9623,
      "step": 204000
    },
    {
      "epoch": 1.12,
      "learning_rate": 9.893708296217764e-06,
      "loss": 2.9668,
      "step": 204500
    },
    {
      "epoch": 1.12,
      "learning_rate": 9.893435052763312e-06,
      "loss": 2.9679,
      "step": 205000
    },
    {
      "epoch": 1.12,
      "learning_rate": 9.893161809308858e-06,
      "loss": 2.9583,
      "step": 205500
    },
    {
      "epoch": 1.13,
      "learning_rate": 9.892888565854406e-06,
      "loss": 2.9563,
      "step": 206000
    },
    {
      "epoch": 1.13,
      "learning_rate": 9.892615322399954e-06,
      "loss": 2.9499,
      "step": 206500
    },
    {
      "epoch": 1.13,
      "learning_rate": 9.8923420789455e-06,
      "loss": 2.9456,
      "step": 207000
    },
    {
      "epoch": 1.13,
      "learning_rate": 9.892068835491046e-06,
      "loss": 2.9508,
      "step": 207500
    },
    {
      "epoch": 1.14,
      "learning_rate": 9.891795592036594e-06,
      "loss": 2.9509,
      "step": 208000
    },
    {
      "epoch": 1.14,
      "learning_rate": 9.89152234858214e-06,
      "loss": 2.9471,
      "step": 208500
    },
    {
      "epoch": 1.14,
      "learning_rate": 9.891249105127687e-06,
      "loss": 2.9403,
      "step": 209000
    },
    {
      "epoch": 1.14,
      "learning_rate": 9.890975861673235e-06,
      "loss": 2.9365,
      "step": 209500
    },
    {
      "epoch": 1.15,
      "learning_rate": 9.890702618218781e-06,
      "loss": 2.9358,
      "step": 210000
    },
    {
      "epoch": 1.15,
      "learning_rate": 9.890429374764329e-06,
      "loss": 2.9409,
      "step": 210500
    },
    {
      "epoch": 1.15,
      "learning_rate": 9.890156131309875e-06,
      "loss": 2.938,
      "step": 211000
    },
    {
      "epoch": 1.16,
      "learning_rate": 9.889882887855421e-06,
      "loss": 2.9338,
      "step": 211500
    },
    {
      "epoch": 1.16,
      "learning_rate": 9.88960964440097e-06,
      "loss": 2.9184,
      "step": 212000
    },
    {
      "epoch": 1.16,
      "learning_rate": 9.889336400946517e-06,
      "loss": 2.9212,
      "step": 212500
    },
    {
      "epoch": 1.16,
      "learning_rate": 9.889063157492062e-06,
      "loss": 2.9309,
      "step": 213000
    },
    {
      "epoch": 1.17,
      "learning_rate": 9.88878991403761e-06,
      "loss": 2.9195,
      "step": 213500
    },
    {
      "epoch": 1.17,
      "learning_rate": 9.888516670583158e-06,
      "loss": 2.9194,
      "step": 214000
    },
    {
      "epoch": 1.17,
      "learning_rate": 9.888243427128704e-06,
      "loss": 2.9158,
      "step": 214500
    },
    {
      "epoch": 1.17,
      "learning_rate": 9.887970183674252e-06,
      "loss": 2.9166,
      "step": 215000
    },
    {
      "epoch": 1.18,
      "learning_rate": 9.887696940219798e-06,
      "loss": 2.9161,
      "step": 215500
    },
    {
      "epoch": 1.18,
      "learning_rate": 9.887423696765344e-06,
      "loss": 2.9108,
      "step": 216000
    },
    {
      "epoch": 1.18,
      "learning_rate": 9.887150453310892e-06,
      "loss": 2.9073,
      "step": 216500
    },
    {
      "epoch": 1.19,
      "learning_rate": 9.886877209856438e-06,
      "loss": 2.8961,
      "step": 217000
    },
    {
      "epoch": 1.19,
      "learning_rate": 9.886603966401985e-06,
      "loss": 2.9083,
      "step": 217500
    },
    {
      "epoch": 1.19,
      "learning_rate": 9.886330722947533e-06,
      "loss": 2.9036,
      "step": 218000
    },
    {
      "epoch": 1.19,
      "learning_rate": 9.88605747949308e-06,
      "loss": 2.9115,
      "step": 218500
    },
    {
      "epoch": 1.2,
      "learning_rate": 9.885784236038627e-06,
      "loss": 2.8945,
      "step": 219000
    },
    {
      "epoch": 1.2,
      "learning_rate": 9.885510992584173e-06,
      "loss": 2.8848,
      "step": 219500
    },
    {
      "epoch": 1.2,
      "learning_rate": 9.88523774912972e-06,
      "loss": 2.8962,
      "step": 220000
    },
    {
      "epoch": 1.2,
      "learning_rate": 9.884964505675267e-06,
      "loss": 2.8834,
      "step": 220500
    },
    {
      "epoch": 1.21,
      "learning_rate": 9.884691262220815e-06,
      "loss": 2.8847,
      "step": 221000
    },
    {
      "epoch": 1.21,
      "learning_rate": 9.884418018766361e-06,
      "loss": 2.8826,
      "step": 221500
    },
    {
      "epoch": 1.21,
      "learning_rate": 9.884144775311907e-06,
      "loss": 2.8852,
      "step": 222000
    },
    {
      "epoch": 1.22,
      "learning_rate": 9.883871531857455e-06,
      "loss": 2.8759,
      "step": 222500
    },
    {
      "epoch": 1.22,
      "learning_rate": 9.883598288403002e-06,
      "loss": 2.8836,
      "step": 223000
    },
    {
      "epoch": 1.22,
      "learning_rate": 9.88332504494855e-06,
      "loss": 2.8705,
      "step": 223500
    },
    {
      "epoch": 1.22,
      "learning_rate": 9.883051801494096e-06,
      "loss": 2.8723,
      "step": 224000
    },
    {
      "epoch": 1.23,
      "learning_rate": 9.882778558039644e-06,
      "loss": 2.8717,
      "step": 224500
    },
    {
      "epoch": 1.23,
      "learning_rate": 9.88250531458519e-06,
      "loss": 2.8776,
      "step": 225000
    },
    {
      "epoch": 1.23,
      "learning_rate": 9.882232071130738e-06,
      "loss": 2.8581,
      "step": 225500
    },
    {
      "epoch": 1.23,
      "learning_rate": 9.881958827676284e-06,
      "loss": 2.8573,
      "step": 226000
    },
    {
      "epoch": 1.24,
      "learning_rate": 9.88168558422183e-06,
      "loss": 2.8536,
      "step": 226500
    },
    {
      "epoch": 1.24,
      "learning_rate": 9.881412340767378e-06,
      "loss": 2.8717,
      "step": 227000
    },
    {
      "epoch": 1.24,
      "learning_rate": 9.881139097312924e-06,
      "loss": 2.8704,
      "step": 227500
    },
    {
      "epoch": 1.25,
      "learning_rate": 9.88086585385847e-06,
      "loss": 2.8519,
      "step": 228000
    },
    {
      "epoch": 1.25,
      "learning_rate": 9.880592610404019e-06,
      "loss": 2.859,
      "step": 228500
    },
    {
      "epoch": 1.25,
      "learning_rate": 9.880319366949565e-06,
      "loss": 2.8534,
      "step": 229000
    },
    {
      "epoch": 1.25,
      "learning_rate": 9.880046123495113e-06,
      "loss": 2.8597,
      "step": 229500
    },
    {
      "epoch": 1.26,
      "learning_rate": 9.879772880040659e-06,
      "loss": 2.8416,
      "step": 230000
    },
    {
      "epoch": 1.26,
      "learning_rate": 9.879499636586207e-06,
      "loss": 2.8479,
      "step": 230500
    },
    {
      "epoch": 1.26,
      "learning_rate": 9.879226393131753e-06,
      "loss": 2.8523,
      "step": 231000
    },
    {
      "epoch": 1.26,
      "learning_rate": 9.878953149677301e-06,
      "loss": 2.8461,
      "step": 231500
    },
    {
      "epoch": 1.27,
      "learning_rate": 9.878679906222847e-06,
      "loss": 2.8488,
      "step": 232000
    },
    {
      "epoch": 1.27,
      "learning_rate": 9.878406662768394e-06,
      "loss": 2.8337,
      "step": 232500
    },
    {
      "epoch": 1.27,
      "learning_rate": 9.878133419313941e-06,
      "loss": 2.836,
      "step": 233000
    },
    {
      "epoch": 1.28,
      "learning_rate": 9.877860175859488e-06,
      "loss": 2.8352,
      "step": 233500
    },
    {
      "epoch": 1.28,
      "learning_rate": 9.877586932405036e-06,
      "loss": 2.8422,
      "step": 234000
    },
    {
      "epoch": 1.28,
      "learning_rate": 9.877313688950582e-06,
      "loss": 2.8429,
      "step": 234500
    },
    {
      "epoch": 1.28,
      "learning_rate": 9.877040445496128e-06,
      "loss": 2.8367,
      "step": 235000
    },
    {
      "epoch": 1.29,
      "learning_rate": 9.876767202041676e-06,
      "loss": 2.8288,
      "step": 235500
    },
    {
      "epoch": 1.29,
      "learning_rate": 9.876493958587224e-06,
      "loss": 2.8235,
      "step": 236000
    },
    {
      "epoch": 1.29,
      "learning_rate": 9.87622071513277e-06,
      "loss": 2.8287,
      "step": 236500
    },
    {
      "epoch": 1.29,
      "learning_rate": 9.875947471678316e-06,
      "loss": 2.8247,
      "step": 237000
    },
    {
      "epoch": 1.3,
      "learning_rate": 9.875674228223864e-06,
      "loss": 2.8184,
      "step": 237500
    },
    {
      "epoch": 1.3,
      "learning_rate": 9.87540098476941e-06,
      "loss": 2.8098,
      "step": 238000
    },
    {
      "epoch": 1.3,
      "learning_rate": 9.875127741314957e-06,
      "loss": 2.8192,
      "step": 238500
    },
    {
      "epoch": 1.31,
      "learning_rate": 9.874854497860505e-06,
      "loss": 2.8212,
      "step": 239000
    },
    {
      "epoch": 1.31,
      "learning_rate": 9.874581254406051e-06,
      "loss": 2.8122,
      "step": 239500
    },
    {
      "epoch": 1.31,
      "learning_rate": 9.874308010951599e-06,
      "loss": 2.8226,
      "step": 240000
    },
    {
      "epoch": 1.31,
      "learning_rate": 9.874034767497147e-06,
      "loss": 2.8088,
      "step": 240500
    },
    {
      "epoch": 1.32,
      "learning_rate": 9.873761524042691e-06,
      "loss": 2.8141,
      "step": 241000
    },
    {
      "epoch": 1.32,
      "learning_rate": 9.87348828058824e-06,
      "loss": 2.8021,
      "step": 241500
    },
    {
      "epoch": 1.32,
      "learning_rate": 9.873215037133787e-06,
      "loss": 2.7997,
      "step": 242000
    },
    {
      "epoch": 1.32,
      "learning_rate": 9.872941793679333e-06,
      "loss": 2.7989,
      "step": 242500
    },
    {
      "epoch": 1.33,
      "learning_rate": 9.87266855022488e-06,
      "loss": 2.8033,
      "step": 243000
    },
    {
      "epoch": 1.33,
      "learning_rate": 9.872395306770428e-06,
      "loss": 2.8018,
      "step": 243500
    },
    {
      "epoch": 1.33,
      "learning_rate": 9.872122063315974e-06,
      "loss": 2.8071,
      "step": 244000
    },
    {
      "epoch": 1.34,
      "learning_rate": 9.871848819861522e-06,
      "loss": 2.7986,
      "step": 244500
    },
    {
      "epoch": 1.34,
      "learning_rate": 9.871575576407068e-06,
      "loss": 2.7878,
      "step": 245000
    },
    {
      "epoch": 1.34,
      "learning_rate": 9.871302332952614e-06,
      "loss": 2.7945,
      "step": 245500
    },
    {
      "epoch": 1.34,
      "learning_rate": 9.871029089498162e-06,
      "loss": 2.795,
      "step": 246000
    },
    {
      "epoch": 1.35,
      "learning_rate": 9.87075584604371e-06,
      "loss": 2.7882,
      "step": 246500
    },
    {
      "epoch": 1.35,
      "learning_rate": 9.870482602589255e-06,
      "loss": 2.7896,
      "step": 247000
    },
    {
      "epoch": 1.35,
      "learning_rate": 9.870209359134802e-06,
      "loss": 2.7849,
      "step": 247500
    },
    {
      "epoch": 1.35,
      "learning_rate": 9.86993611568035e-06,
      "loss": 2.7878,
      "step": 248000
    },
    {
      "epoch": 1.36,
      "learning_rate": 9.869662872225897e-06,
      "loss": 2.7791,
      "step": 248500
    },
    {
      "epoch": 1.36,
      "learning_rate": 9.869389628771445e-06,
      "loss": 2.7728,
      "step": 249000
    },
    {
      "epoch": 1.36,
      "learning_rate": 9.86911638531699e-06,
      "loss": 2.7844,
      "step": 249500
    },
    {
      "epoch": 1.37,
      "learning_rate": 9.868843141862537e-06,
      "loss": 2.7719,
      "step": 250000
    },
    {
      "epoch": 1.37,
      "learning_rate": 9.868569898408085e-06,
      "loss": 2.7813,
      "step": 250500
    },
    {
      "epoch": 1.37,
      "learning_rate": 9.868296654953631e-06,
      "loss": 2.7711,
      "step": 251000
    },
    {
      "epoch": 1.37,
      "learning_rate": 9.868023411499177e-06,
      "loss": 2.7753,
      "step": 251500
    },
    {
      "epoch": 1.38,
      "learning_rate": 9.867750168044725e-06,
      "loss": 2.7808,
      "step": 252000
    },
    {
      "epoch": 1.38,
      "learning_rate": 9.867476924590273e-06,
      "loss": 2.7645,
      "step": 252500
    },
    {
      "epoch": 1.38,
      "learning_rate": 9.86720368113582e-06,
      "loss": 2.7709,
      "step": 253000
    },
    {
      "epoch": 1.38,
      "learning_rate": 9.866930437681366e-06,
      "loss": 2.7677,
      "step": 253500
    },
    {
      "epoch": 1.39,
      "learning_rate": 9.866657194226914e-06,
      "loss": 2.7677,
      "step": 254000
    },
    {
      "epoch": 1.39,
      "learning_rate": 9.86638395077246e-06,
      "loss": 2.7655,
      "step": 254500
    },
    {
      "epoch": 1.39,
      "learning_rate": 9.866110707318008e-06,
      "loss": 2.7718,
      "step": 255000
    },
    {
      "epoch": 1.4,
      "learning_rate": 9.865837463863554e-06,
      "loss": 2.7524,
      "step": 255500
    },
    {
      "epoch": 1.4,
      "learning_rate": 9.8655642204091e-06,
      "loss": 2.7602,
      "step": 256000
    },
    {
      "epoch": 1.4,
      "learning_rate": 9.865290976954648e-06,
      "loss": 2.7603,
      "step": 256500
    },
    {
      "epoch": 1.4,
      "learning_rate": 9.865017733500194e-06,
      "loss": 2.7592,
      "step": 257000
    },
    {
      "epoch": 1.41,
      "learning_rate": 9.864744490045742e-06,
      "loss": 2.7458,
      "step": 257500
    },
    {
      "epoch": 1.41,
      "learning_rate": 9.864471246591289e-06,
      "loss": 2.7494,
      "step": 258000
    },
    {
      "epoch": 1.41,
      "learning_rate": 9.864198003136835e-06,
      "loss": 2.7464,
      "step": 258500
    },
    {
      "epoch": 1.41,
      "learning_rate": 9.863924759682383e-06,
      "loss": 2.7482,
      "step": 259000
    },
    {
      "epoch": 1.42,
      "learning_rate": 9.86365151622793e-06,
      "loss": 2.7485,
      "step": 259500
    },
    {
      "epoch": 1.42,
      "learning_rate": 9.863378272773477e-06,
      "loss": 2.7445,
      "step": 260000
    },
    {
      "epoch": 1.42,
      "learning_rate": 9.863105029319023e-06,
      "loss": 2.747,
      "step": 260500
    },
    {
      "epoch": 1.43,
      "learning_rate": 9.862831785864571e-06,
      "loss": 2.7409,
      "step": 261000
    },
    {
      "epoch": 1.43,
      "learning_rate": 9.862558542410117e-06,
      "loss": 2.7415,
      "step": 261500
    },
    {
      "epoch": 1.43,
      "learning_rate": 9.862285298955663e-06,
      "loss": 2.742,
      "step": 262000
    },
    {
      "epoch": 1.43,
      "learning_rate": 9.862012055501211e-06,
      "loss": 2.7361,
      "step": 262500
    },
    {
      "epoch": 1.44,
      "learning_rate": 9.861738812046758e-06,
      "loss": 2.7427,
      "step": 263000
    },
    {
      "epoch": 1.44,
      "learning_rate": 9.861465568592306e-06,
      "loss": 2.7305,
      "step": 263500
    },
    {
      "epoch": 1.44,
      "learning_rate": 9.861192325137852e-06,
      "loss": 2.7291,
      "step": 264000
    },
    {
      "epoch": 1.44,
      "learning_rate": 9.860919081683398e-06,
      "loss": 2.7295,
      "step": 264500
    },
    {
      "epoch": 1.45,
      "learning_rate": 9.860645838228946e-06,
      "loss": 2.7306,
      "step": 265000
    },
    {
      "epoch": 1.45,
      "learning_rate": 9.860372594774494e-06,
      "loss": 2.7323,
      "step": 265500
    },
    {
      "epoch": 1.45,
      "learning_rate": 9.86009935132004e-06,
      "loss": 2.7336,
      "step": 266000
    },
    {
      "epoch": 1.46,
      "learning_rate": 9.859826107865586e-06,
      "loss": 2.721,
      "step": 266500
    },
    {
      "epoch": 1.46,
      "learning_rate": 9.859552864411134e-06,
      "loss": 2.7328,
      "step": 267000
    },
    {
      "epoch": 1.46,
      "learning_rate": 9.85927962095668e-06,
      "loss": 2.7221,
      "step": 267500
    },
    {
      "epoch": 1.46,
      "learning_rate": 9.859006377502228e-06,
      "loss": 2.7195,
      "step": 268000
    },
    {
      "epoch": 1.47,
      "learning_rate": 9.858733134047775e-06,
      "loss": 2.7208,
      "step": 268500
    },
    {
      "epoch": 1.47,
      "learning_rate": 9.858459890593321e-06,
      "loss": 2.7248,
      "step": 269000
    },
    {
      "epoch": 1.47,
      "learning_rate": 9.858186647138869e-06,
      "loss": 2.7114,
      "step": 269500
    },
    {
      "epoch": 1.47,
      "learning_rate": 9.857913403684417e-06,
      "loss": 2.7171,
      "step": 270000
    },
    {
      "epoch": 1.48,
      "learning_rate": 9.857640160229961e-06,
      "loss": 2.7128,
      "step": 270500
    },
    {
      "epoch": 1.48,
      "learning_rate": 9.85736691677551e-06,
      "loss": 2.7156,
      "step": 271000
    },
    {
      "epoch": 1.48,
      "learning_rate": 9.857093673321057e-06,
      "loss": 2.7127,
      "step": 271500
    },
    {
      "epoch": 1.49,
      "learning_rate": 9.856820429866603e-06,
      "loss": 2.7144,
      "step": 272000
    },
    {
      "epoch": 1.49,
      "learning_rate": 9.85654718641215e-06,
      "loss": 2.7049,
      "step": 272500
    },
    {
      "epoch": 1.49,
      "learning_rate": 9.856273942957697e-06,
      "loss": 2.7132,
      "step": 273000
    },
    {
      "epoch": 1.49,
      "learning_rate": 9.856000699503244e-06,
      "loss": 2.7009,
      "step": 273500
    },
    {
      "epoch": 1.5,
      "learning_rate": 9.855727456048792e-06,
      "loss": 2.692,
      "step": 274000
    },
    {
      "epoch": 1.5,
      "learning_rate": 9.855454212594338e-06,
      "loss": 2.7116,
      "step": 274500
    },
    {
      "epoch": 1.5,
      "learning_rate": 9.855180969139884e-06,
      "loss": 2.7019,
      "step": 275000
    },
    {
      "epoch": 1.5,
      "learning_rate": 9.854907725685432e-06,
      "loss": 2.7019,
      "step": 275500
    },
    {
      "epoch": 1.51,
      "learning_rate": 9.85463448223098e-06,
      "loss": 2.6916,
      "step": 276000
    },
    {
      "epoch": 1.51,
      "learning_rate": 9.854361238776526e-06,
      "loss": 2.6941,
      "step": 276500
    },
    {
      "epoch": 1.51,
      "learning_rate": 9.854087995322072e-06,
      "loss": 2.7009,
      "step": 277000
    },
    {
      "epoch": 1.52,
      "learning_rate": 9.85381475186762e-06,
      "loss": 2.7022,
      "step": 277500
    },
    {
      "epoch": 1.52,
      "learning_rate": 9.853541508413167e-06,
      "loss": 2.6958,
      "step": 278000
    },
    {
      "epoch": 1.52,
      "learning_rate": 9.853268264958714e-06,
      "loss": 2.6851,
      "step": 278500
    },
    {
      "epoch": 1.52,
      "learning_rate": 9.85299502150426e-06,
      "loss": 2.6867,
      "step": 279000
    },
    {
      "epoch": 1.53,
      "learning_rate": 9.852721778049807e-06,
      "loss": 2.6856,
      "step": 279500
    },
    {
      "epoch": 1.53,
      "learning_rate": 9.852448534595355e-06,
      "loss": 2.6922,
      "step": 280000
    },
    {
      "epoch": 1.53,
      "learning_rate": 9.852175291140901e-06,
      "loss": 2.6843,
      "step": 280500
    },
    {
      "epoch": 1.53,
      "learning_rate": 9.851902047686447e-06,
      "loss": 2.691,
      "step": 281000
    },
    {
      "epoch": 1.54,
      "learning_rate": 9.851628804231995e-06,
      "loss": 2.683,
      "step": 281500
    },
    {
      "epoch": 1.54,
      "learning_rate": 9.851355560777543e-06,
      "loss": 2.6879,
      "step": 282000
    },
    {
      "epoch": 1.54,
      "learning_rate": 9.85108231732309e-06,
      "loss": 2.6801,
      "step": 282500
    },
    {
      "epoch": 1.55,
      "learning_rate": 9.850809073868637e-06,
      "loss": 2.6786,
      "step": 283000
    },
    {
      "epoch": 1.55,
      "learning_rate": 9.850535830414184e-06,
      "loss": 2.6792,
      "step": 283500
    },
    {
      "epoch": 1.55,
      "learning_rate": 9.85026258695973e-06,
      "loss": 2.6696,
      "step": 284000
    },
    {
      "epoch": 1.55,
      "learning_rate": 9.849989343505278e-06,
      "loss": 2.6815,
      "step": 284500
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.849716100050824e-06,
      "loss": 2.6803,
      "step": 285000
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.84944285659637e-06,
      "loss": 2.6839,
      "step": 285500
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.849169613141918e-06,
      "loss": 2.6759,
      "step": 286000
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.848896369687464e-06,
      "loss": 2.6577,
      "step": 286500
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.848623126233012e-06,
      "loss": 2.6649,
      "step": 287000
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.848349882778558e-06,
      "loss": 2.6682,
      "step": 287500
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.848076639324106e-06,
      "loss": 2.6699,
      "step": 288000
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.847803395869653e-06,
      "loss": 2.6639,
      "step": 288500
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.8475301524152e-06,
      "loss": 2.6767,
      "step": 289000
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.847256908960747e-06,
      "loss": 2.6677,
      "step": 289500
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.846983665506293e-06,
      "loss": 2.6525,
      "step": 290000
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.846710422051841e-06,
      "loss": 2.6733,
      "step": 290500
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.846437178597387e-06,
      "loss": 2.6571,
      "step": 291000
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.846163935142935e-06,
      "loss": 2.6668,
      "step": 291500
    },
    {
      "epoch": 1.59,
      "step": 291948,
      "total_flos": 3.1748015285904343e+18,
      "train_loss": 4.151520017065994,
      "train_runtime": 197999.1247,
      "train_samples_per_second": 2958.982,
      "train_steps_per_second": 92.469
    }
  ],
  "logging_steps": 500,
  "max_steps": 18308700,
  "num_train_epochs": 100,
  "save_steps": 1000000,
  "total_flos": 3.1748015285904343e+18,
  "trial_name": null,
  "trial_params": null
}