{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.2067952933391237,
  "global_step": 700000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.05,
      "learning_rate": 6.893176444637453e-06,
      "loss": 11.9733,
      "step": 10000
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.3786352889274907e-05,
      "loss": 1.1347,
      "step": 20000
    },
    {
      "epoch": 0.05,
      "learning_rate": 2.067952933391236e-05,
      "loss": 0.9519,
      "step": 30000
    },
    {
      "epoch": 0.1,
      "learning_rate": 2.7572705778549813e-05,
      "loss": 0.8738,
      "step": 40000
    },
    {
      "epoch": 0.16,
      "learning_rate": 3.446588222318727e-05,
      "loss": 0.8298,
      "step": 50000
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.135905866782472e-05,
      "loss": 0.7905,
      "step": 60000
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.825223511246217e-05,
      "loss": 0.6585,
      "step": 70000
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.5145411557099626e-05,
      "loss": 0.5349,
      "step": 80000
    },
    {
      "epoch": 0.1,
      "learning_rate": 6.203858800173708e-05,
      "loss": 0.4895,
      "step": 90000
    },
    {
      "epoch": 0.16,
      "learning_rate": 6.893176444637454e-05,
      "loss": 0.4541,
      "step": 100000
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.582494089101199e-05,
      "loss": 0.4319,
      "step": 110000
    },
    {
      "epoch": 0.26,
      "learning_rate": 8.271811733564945e-05,
      "loss": 0.4165,
      "step": 120000
    },
    {
      "epoch": 0.05,
      "learning_rate": 8.96112937802869e-05,
      "loss": 0.4059,
      "step": 130000
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.650447022492434e-05,
      "loss": 0.3957,
      "step": 140000
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.982117649107569e-05,
      "loss": 0.388,
      "step": 150000
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.945837773083162e-05,
      "loss": 0.3787,
      "step": 160000
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.909557897058755e-05,
      "loss": 0.3705,
      "step": 170000
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.873278021034346e-05,
      "loss": 0.3678,
      "step": 180000
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.836998145009939e-05,
      "loss": 0.3595,
      "step": 190000
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.800718268985532e-05,
      "loss": 0.3543,
      "step": 200000
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.764438392961124e-05,
      "loss": 0.3495,
      "step": 210000
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.728158516936716e-05,
      "loss": 0.3452,
      "step": 220000
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.691878640912308e-05,
      "loss": 0.3419,
      "step": 230000
    },
    {
      "epoch": 1.05,
      "learning_rate": 9.655598764887901e-05,
      "loss": 0.3365,
      "step": 240000
    },
    {
      "epoch": 1.1,
      "learning_rate": 9.619318888863494e-05,
      "loss": 0.3336,
      "step": 250000
    },
    {
      "epoch": 1.05,
      "learning_rate": 9.583039012839085e-05,
      "loss": 0.331,
      "step": 260000
    },
    {
      "epoch": 1.1,
      "learning_rate": 9.546759136814678e-05,
      "loss": 0.3282,
      "step": 270000
    },
    {
      "epoch": 1.16,
      "learning_rate": 9.510479260790271e-05,
      "loss": 0.3263,
      "step": 280000
    },
    {
      "epoch": 1.21,
      "learning_rate": 9.474199384765862e-05,
      "loss": 0.3234,
      "step": 290000
    },
    {
      "epoch": 1.26,
      "learning_rate": 9.437919508741455e-05,
      "loss": 0.3213,
      "step": 300000
    },
    {
      "epoch": 1.05,
      "learning_rate": 9.401639632717048e-05,
      "loss": 0.3173,
      "step": 310000
    },
    {
      "epoch": 1.1,
      "learning_rate": 9.36535975669264e-05,
      "loss": 0.3166,
      "step": 320000
    },
    {
      "epoch": 1.16,
      "learning_rate": 9.329079880668232e-05,
      "loss": 0.3137,
      "step": 330000
    },
    {
      "epoch": 1.21,
      "learning_rate": 9.292800004643825e-05,
      "loss": 0.3129,
      "step": 340000
    },
    {
      "epoch": 1.26,
      "learning_rate": 9.256520128619417e-05,
      "loss": 0.3125,
      "step": 350000
    },
    {
      "epoch": 1.05,
      "learning_rate": 9.22024025259501e-05,
      "loss": 0.3079,
      "step": 360000
    },
    {
      "epoch": 1.1,
      "learning_rate": 9.183960376570602e-05,
      "loss": 0.3074,
      "step": 370000
    },
    {
      "epoch": 1.16,
      "learning_rate": 9.147680500546194e-05,
      "loss": 0.3062,
      "step": 380000
    },
    {
      "epoch": 1.21,
      "learning_rate": 9.111400624521787e-05,
      "loss": 0.3052,
      "step": 390000
    },
    {
      "epoch": 1.26,
      "learning_rate": 9.07512074849738e-05,
      "loss": 0.3037,
      "step": 400000
    },
    {
      "epoch": 2.05,
      "learning_rate": 9.038840872472971e-05,
      "loss": 0.3003,
      "step": 410000
    },
    {
      "epoch": 2.1,
      "learning_rate": 9.002560996448564e-05,
      "loss": 0.3007,
      "step": 420000
    },
    {
      "epoch": 2.16,
      "learning_rate": 8.966281120424155e-05,
      "loss": 0.2991,
      "step": 430000
    },
    {
      "epoch": 2.21,
      "learning_rate": 8.930001244399748e-05,
      "loss": 0.2972,
      "step": 440000
    },
    {
      "epoch": 2.26,
      "learning_rate": 8.893721368375341e-05,
      "loss": 0.2974,
      "step": 450000
    },
    {
      "epoch": 2.05,
      "learning_rate": 8.857441492350932e-05,
      "loss": 0.2938,
      "step": 460000
    },
    {
      "epoch": 2.1,
      "learning_rate": 8.821161616326525e-05,
      "loss": 0.2921,
      "step": 470000
    },
    {
      "epoch": 2.16,
      "learning_rate": 8.784881740302118e-05,
      "loss": 0.2932,
      "step": 480000
    },
    {
      "epoch": 2.21,
      "learning_rate": 8.74860186427771e-05,
      "loss": 0.2914,
      "step": 490000
    },
    {
      "epoch": 2.26,
      "learning_rate": 8.712321988253302e-05,
      "loss": 0.292,
      "step": 500000
    },
    {
      "epoch": 2.05,
      "learning_rate": 8.676042112228895e-05,
      "loss": 0.2871,
      "step": 510000
    },
    {
      "epoch": 2.1,
      "learning_rate": 8.639762236204487e-05,
      "loss": 0.2884,
      "step": 520000
    },
    {
      "epoch": 2.16,
      "learning_rate": 8.60348236018008e-05,
      "loss": 0.2874,
      "step": 530000
    },
    {
      "epoch": 2.21,
      "learning_rate": 8.567202484155672e-05,
      "loss": 0.2859,
      "step": 540000
    },
    {
      "epoch": 2.26,
      "learning_rate": 8.530922608131264e-05,
      "loss": 0.2867,
      "step": 550000
    },
    {
      "epoch": 2.05,
      "learning_rate": 8.494642732106857e-05,
      "loss": 0.2828,
      "step": 560000
    },
    {
      "epoch": 2.05,
      "learning_rate": 8.45836285608245e-05,
      "loss": 0.2829,
      "step": 570000
    },
    {
      "epoch": 2.1,
      "learning_rate": 8.422082980058041e-05,
      "loss": 0.2817,
      "step": 580000
    },
    {
      "epoch": 2.16,
      "learning_rate": 8.385803104033634e-05,
      "loss": 0.281,
      "step": 590000
    },
    {
      "epoch": 2.21,
      "learning_rate": 8.349523228009227e-05,
      "loss": 0.2819,
      "step": 600000
    },
    {
      "epoch": 2.26,
      "learning_rate": 8.313243351984818e-05,
      "loss": 0.2803,
      "step": 610000
    },
    {
      "epoch": 3.05,
      "learning_rate": 8.276963475960411e-05,
      "loss": 0.2782,
      "step": 620000
    },
    {
      "epoch": 3.1,
      "learning_rate": 8.240683599936003e-05,
      "loss": 0.2779,
      "step": 630000
    },
    {
      "epoch": 3.16,
      "learning_rate": 8.204403723911594e-05,
      "loss": 0.2793,
      "step": 640000
    },
    {
      "epoch": 3.21,
      "learning_rate": 8.168123847887187e-05,
      "loss": 0.2774,
      "step": 650000
    },
    {
      "epoch": 3.26,
      "learning_rate": 8.13184397186278e-05,
      "loss": 0.2765,
      "step": 660000
    },
    {
      "epoch": 3.05,
      "learning_rate": 8.095564095838371e-05,
      "loss": 0.2738,
      "step": 670000
    },
    {
      "epoch": 3.1,
      "learning_rate": 8.059284219813964e-05,
      "loss": 0.2742,
      "step": 680000
    },
    {
      "epoch": 3.16,
      "learning_rate": 8.023004343789557e-05,
      "loss": 0.2746,
      "step": 690000
    },
    {
      "epoch": 3.21,
      "learning_rate": 7.986724467765148e-05,
      "loss": 0.274,
      "step": 700000
    }
  ],
  "max_steps": 2901420,
  "num_train_epochs": 15,
  "total_flos": 1.6750927872e+18,
  "trial_name": null,
  "trial_params": null
}