{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.15416238437821173,
  "eval_steps": 500,
  "global_step": 300,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 4.9999979150098865e-05,
      "loss": 4.0418,
      "step": 5
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999989444743508e-05,
      "loss": 3.8361,
      "step": 10
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999974458911041e-05,
      "loss": 3.8737,
      "step": 15
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9999529575515425e-05,
      "loss": 3.806,
      "step": 20
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9999249407210515e-05,
      "loss": 3.4125,
      "step": 25
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9998904084925845e-05,
      "loss": 3.6489,
      "step": 30
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.999849360956141e-05,
      "loss": 3.4719,
      "step": 35
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.999811831976564e-05,
      "loss": 3.5265,
      "step": 40
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.999759057166764e-05,
      "loss": 3.6889,
      "step": 45
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9996997673913204e-05,
      "loss": 3.5104,
      "step": 50
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.999633962804756e-05,
      "loss": 3.5778,
      "step": 55
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.999561643578575e-05,
      "loss": 3.3353,
      "step": 60
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.999482809901257e-05,
      "loss": 3.3511,
      "step": 65
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9993974619782625e-05,
      "loss": 3.6355,
      "step": 70
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.999305600032027e-05,
      "loss": 3.3915,
      "step": 75
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.999207224301965e-05,
      "loss": 3.4926,
      "step": 80
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.999102335044467e-05,
      "loss": 3.4918,
      "step": 85
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9989909325328996e-05,
      "loss": 3.3984,
      "step": 90
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.998873017057605e-05,
      "loss": 3.4989,
      "step": 95
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.998748588925897e-05,
      "loss": 3.6368,
      "step": 100
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9986176484620665e-05,
      "loss": 3.4412,
      "step": 105
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.998480196007375e-05,
      "loss": 3.4708,
      "step": 110
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9983362319200554e-05,
      "loss": 3.37,
      "step": 115
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.998185756575313e-05,
      "loss": 3.4524,
      "step": 120
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9980287703653225e-05,
      "loss": 3.3863,
      "step": 125
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.997865273699227e-05,
      "loss": 3.4414,
      "step": 130
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.997729789123732e-05,
      "loss": 3.3335,
      "step": 135
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.997554574721849e-05,
      "loss": 3.231,
      "step": 140
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.997372851099727e-05,
      "loss": 3.5962,
      "step": 145
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.997184618730983e-05,
      "loss": 3.2795,
      "step": 150
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.996989878106192e-05,
      "loss": 3.5384,
      "step": 155
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.996788629732897e-05,
      "loss": 3.3274,
      "step": 160
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.996580874135597e-05,
      "loss": 3.2292,
      "step": 165
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.996366611855753e-05,
      "loss": 3.4829,
      "step": 170
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.996145843451785e-05,
      "loss": 3.1998,
      "step": 175
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9959185694990655e-05,
      "loss": 3.3757,
      "step": 180
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.995684790589926e-05,
      "loss": 3.1001,
      "step": 185
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.99544450733365e-05,
      "loss": 3.2301,
      "step": 190
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.995197720356471e-05,
      "loss": 3.3187,
      "step": 195
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.994944430301576e-05,
      "loss": 3.527,
      "step": 200
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.994684637829098e-05,
      "loss": 3.3808,
      "step": 205
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.994418343616119e-05,
      "loss": 3.2406,
      "step": 210
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.994145548356664e-05,
      "loss": 3.3463,
      "step": 215
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.993866252761702e-05,
      "loss": 3.2972,
      "step": 220
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.993580457559145e-05,
      "loss": 3.4135,
      "step": 225
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.9932881634938424e-05,
      "loss": 3.4025,
      "step": 230
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.992989371327581e-05,
      "loss": 3.662,
      "step": 235
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.992684081839086e-05,
      "loss": 3.4279,
      "step": 240
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.9923722958240124e-05,
      "loss": 3.3075,
      "step": 245
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.992054014094951e-05,
      "loss": 3.2826,
      "step": 250
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.991729237481417e-05,
      "loss": 3.4992,
      "step": 255
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.991397966829856e-05,
      "loss": 3.2421,
      "step": 260
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.991060203003639e-05,
      "loss": 3.2615,
      "step": 265
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.9907159468830585e-05,
      "loss": 3.3299,
      "step": 270
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.990365199365327e-05,
      "loss": 3.3117,
      "step": 275
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.9900079613645754e-05,
      "loss": 3.4729,
      "step": 280
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.9896442338118524e-05,
      "loss": 3.2521,
      "step": 285
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.989274017655117e-05,
      "loss": 3.261,
      "step": 290
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.9888973138592414e-05,
      "loss": 3.4403,
      "step": 295
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.988514123406004e-05,
      "loss": 3.3408,
      "step": 300
    }
  ],
  "logging_steps": 5,
  "max_steps": 9730,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 300,
  "total_flos": 3.55665860235264e+16,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}