{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9961987154279722,
  "global_step": 38000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 1.973784244330843e-05,
      "loss": 0.8507,
      "step": 500
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.947568488661686e-05,
      "loss": 0.8336,
      "step": 1000
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9213527329925285e-05,
      "loss": 0.8291,
      "step": 1500
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.8951369773233716e-05,
      "loss": 0.8236,
      "step": 2000
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.8689212216542143e-05,
      "loss": 0.8227,
      "step": 2500
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.842705465985057e-05,
      "loss": 0.8166,
      "step": 3000
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.8164897103159e-05,
      "loss": 0.809,
      "step": 3500
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.790273954646743e-05,
      "loss": 0.8051,
      "step": 4000
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.7640581989775854e-05,
      "loss": 0.8054,
      "step": 4500
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.7378424433084285e-05,
      "loss": 0.7966,
      "step": 5000
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.7116266876392713e-05,
      "loss": 0.8015,
      "step": 5500
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.685410931970114e-05,
      "loss": 0.7975,
      "step": 6000
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.659195176300957e-05,
      "loss": 0.7953,
      "step": 6500
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.6329794206318e-05,
      "loss": 0.7941,
      "step": 7000
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.6067636649626427e-05,
      "loss": 0.7957,
      "step": 7500
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.5805479092934855e-05,
      "loss": 0.7934,
      "step": 8000
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.5543321536243282e-05,
      "loss": 0.7941,
      "step": 8500
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.5281163979551713e-05,
      "loss": 0.7866,
      "step": 9000
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.5019006422860141e-05,
      "loss": 0.7834,
      "step": 9500
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.4756848866168569e-05,
      "loss": 0.778,
      "step": 10000
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.4494691309476996e-05,
      "loss": 0.7863,
      "step": 10500
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.4232533752785424e-05,
      "loss": 0.7808,
      "step": 11000
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.3970376196093854e-05,
      "loss": 0.788,
      "step": 11500
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.3708218639402281e-05,
      "loss": 0.7829,
      "step": 12000
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.344606108271071e-05,
      "loss": 0.776,
      "step": 12500
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.318390352601914e-05,
      "loss": 0.7761,
      "step": 13000
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.2921745969327566e-05,
      "loss": 0.7812,
      "step": 13500
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.2659588412635995e-05,
      "loss": 0.7776,
      "step": 14000
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.2397430855944423e-05,
      "loss": 0.7727,
      "step": 14500
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.2135273299252852e-05,
      "loss": 0.7734,
      "step": 15000
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.187311574256128e-05,
      "loss": 0.7733,
      "step": 15500
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.161095818586971e-05,
      "loss": 0.7683,
      "step": 16000
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.1348800629178135e-05,
      "loss": 0.7743,
      "step": 16500
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1086643072486565e-05,
      "loss": 0.7718,
      "step": 17000
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.0824485515794994e-05,
      "loss": 0.7553,
      "step": 17500
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.0562327959103422e-05,
      "loss": 0.7674,
      "step": 18000
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0300170402411851e-05,
      "loss": 0.7684,
      "step": 18500
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0038012845720279e-05,
      "loss": 0.7634,
      "step": 19000
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.775855289028707e-06,
      "loss": 0.7624,
      "step": 19500
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.513697732337134e-06,
      "loss": 0.7656,
      "step": 20000
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.251540175645564e-06,
      "loss": 0.761,
      "step": 20500
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.989382618953993e-06,
      "loss": 0.768,
      "step": 21000
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.72722506226242e-06,
      "loss": 0.7666,
      "step": 21500
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.465067505570848e-06,
      "loss": 0.7604,
      "step": 22000
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.202909948879278e-06,
      "loss": 0.7543,
      "step": 22500
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.940752392187705e-06,
      "loss": 0.7604,
      "step": 23000
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.678594835496133e-06,
      "loss": 0.7613,
      "step": 23500
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.4164372788045625e-06,
      "loss": 0.7617,
      "step": 24000
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.15427972211299e-06,
      "loss": 0.7637,
      "step": 24500
    },
    {
      "epoch": 0.66,
      "learning_rate": 6.892122165421419e-06,
      "loss": 0.764,
      "step": 25000
    },
    {
      "epoch": 0.67,
      "learning_rate": 6.629964608729847e-06,
      "loss": 0.7506,
      "step": 25500
    },
    {
      "epoch": 0.68,
      "learning_rate": 6.367807052038275e-06,
      "loss": 0.7515,
      "step": 26000
    },
    {
      "epoch": 0.69,
      "learning_rate": 6.105649495346703e-06,
      "loss": 0.7501,
      "step": 26500
    },
    {
      "epoch": 0.71,
      "learning_rate": 5.843491938655133e-06,
      "loss": 0.7624,
      "step": 27000
    },
    {
      "epoch": 0.72,
      "learning_rate": 5.5813343819635605e-06,
      "loss": 0.7529,
      "step": 27500
    },
    {
      "epoch": 0.73,
      "learning_rate": 5.319176825271989e-06,
      "loss": 0.7521,
      "step": 28000
    },
    {
      "epoch": 0.75,
      "learning_rate": 5.0570192685804175e-06,
      "loss": 0.7527,
      "step": 28500
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.794861711888846e-06,
      "loss": 0.7535,
      "step": 29000
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.532704155197274e-06,
      "loss": 0.7479,
      "step": 29500
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.270546598505702e-06,
      "loss": 0.7495,
      "step": 30000
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.008389041814131e-06,
      "loss": 0.7477,
      "step": 30500
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.746231485122559e-06,
      "loss": 0.751,
      "step": 31000
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.484073928430987e-06,
      "loss": 0.7441,
      "step": 31500
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.221916371739416e-06,
      "loss": 0.7467,
      "step": 32000
    },
    {
      "epoch": 0.85,
      "learning_rate": 2.959758815047844e-06,
      "loss": 0.746,
      "step": 32500
    },
    {
      "epoch": 0.87,
      "learning_rate": 2.697601258356272e-06,
      "loss": 0.7512,
      "step": 33000
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.4354437016647007e-06,
      "loss": 0.7387,
      "step": 33500
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.173286144973129e-06,
      "loss": 0.7449,
      "step": 34000
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.9111285882815573e-06,
      "loss": 0.7443,
      "step": 34500
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.6489710315899856e-06,
      "loss": 0.7498,
      "step": 35000
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.3868134748984141e-06,
      "loss": 0.7377,
      "step": 35500
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.1246559182068425e-06,
      "loss": 0.7444,
      "step": 36000
    },
    {
      "epoch": 0.96,
      "learning_rate": 8.624983615152708e-07,
      "loss": 0.741,
      "step": 36500
    },
    {
      "epoch": 0.97,
      "learning_rate": 6.003408048236991e-07,
      "loss": 0.759,
      "step": 37000
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.3818324813212746e-07,
      "loss": 0.7429,
      "step": 37500
    },
    {
      "epoch": 1.0,
      "learning_rate": 7.602569144055578e-08,
      "loss": 0.7439,
      "step": 38000
    }
  ],
  "max_steps": 38145,
  "num_train_epochs": 1,
  "total_flos": 1.0054839784439808e+16,
  "trial_name": null,
  "trial_params": null
}