{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.970873786407767,
  "global_step": 75,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.04,
      "learning_rate": 9.999999999999999e-06,
      "loss": 1.144,
      "step": 1
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9999999999999998e-05,
      "loss": 1.3398,
      "step": 2
    },
    {
      "epoch": 0.12,
      "learning_rate": 3e-05,
      "loss": 0.7985,
      "step": 3
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.998572332372787e-05,
      "loss": 0.7269,
      "step": 4
    },
    {
      "epoch": 0.19,
      "learning_rate": 2.9942920471376185e-05,
      "loss": 0.6506,
      "step": 5
    },
    {
      "epoch": 0.23,
      "learning_rate": 2.9871672920607158e-05,
      "loss": 0.6182,
      "step": 6
    },
    {
      "epoch": 0.27,
      "learning_rate": 2.977211629518312e-05,
      "loss": 0.5439,
      "step": 7
    },
    {
      "epoch": 0.31,
      "learning_rate": 2.9644440106799003e-05,
      "loss": 0.5743,
      "step": 8
    },
    {
      "epoch": 0.35,
      "learning_rate": 2.9488887394336025e-05,
      "loss": 0.5574,
      "step": 9
    },
    {
      "epoch": 0.39,
      "learning_rate": 2.9305754261223403e-05,
      "loss": 0.5959,
      "step": 10
    },
    {
      "epoch": 0.43,
      "learning_rate": 2.9095389311788626e-05,
      "loss": 0.5722,
      "step": 11
    },
    {
      "epoch": 0.47,
      "learning_rate": 2.8858192987669303e-05,
      "loss": 0.6372,
      "step": 12
    },
    {
      "epoch": 0.5,
      "learning_rate": 2.8594616805549752e-05,
      "loss": 0.5282,
      "step": 13
    },
    {
      "epoch": 0.54,
      "learning_rate": 2.8305162497673325e-05,
      "loss": 0.5699,
      "step": 14
    },
    {
      "epoch": 0.58,
      "learning_rate": 2.7990381056766583e-05,
      "loss": 0.5106,
      "step": 15
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.7650871687193287e-05,
      "loss": 0.4858,
      "step": 16
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.728728066433488e-05,
      "loss": 0.5103,
      "step": 17
    },
    {
      "epoch": 0.7,
      "learning_rate": 2.6900300104368527e-05,
      "loss": 0.4594,
      "step": 18
    },
    {
      "epoch": 0.74,
      "learning_rate": 2.649066664678467e-05,
      "loss": 0.5136,
      "step": 19
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.605916005215186e-05,
      "loss": 0.443,
      "step": 20
    },
    {
      "epoch": 0.82,
      "learning_rate": 2.5606601717798212e-05,
      "loss": 0.4905,
      "step": 21
    },
    {
      "epoch": 0.85,
      "learning_rate": 2.5133853114234907e-05,
      "loss": 0.5694,
      "step": 22
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.464181414529809e-05,
      "loss": 0.4642,
      "step": 23
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.413142143513081e-05,
      "loss": 0.4738,
      "step": 24
    },
    {
      "epoch": 0.97,
      "learning_rate": 2.360364654526569e-05,
      "loss": 0.522,
      "step": 25
    },
    {
      "epoch": 1.04,
      "learning_rate": 2.305949412520236e-05,
      "loss": 0.7219,
      "step": 26
    },
    {
      "epoch": 1.08,
      "learning_rate": 2.25e-05,
      "loss": 0.4399,
      "step": 27
    },
    {
      "epoch": 1.12,
      "learning_rate": 2.192622919852551e-05,
      "loss": 0.4449,
      "step": 28
    },
    {
      "epoch": 1.16,
      "learning_rate": 2.1339273926110493e-05,
      "loss": 0.3547,
      "step": 29
    },
    {
      "epoch": 1.19,
      "learning_rate": 2.074025148547635e-05,
      "loss": 0.4001,
      "step": 30
    },
    {
      "epoch": 1.23,
      "learning_rate": 2.0130302149885033e-05,
      "loss": 0.386,
      "step": 31
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.9510586992564095e-05,
      "loss": 0.4169,
      "step": 32
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.888228567653781e-05,
      "loss": 0.3796,
      "step": 33
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.8246594209071543e-05,
      "loss": 0.3544,
      "step": 34
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.760472266500396e-05,
      "loss": 0.3437,
      "step": 35
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.6957892883300778e-05,
      "loss": 0.3457,
      "step": 36
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.6307336141214877e-05,
      "loss": 0.32,
      "step": 37
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.5654290810480043e-05,
      "loss": 0.342,
      "step": 38
    },
    {
      "epoch": 1.54,
      "learning_rate": 1.5e-05,
      "loss": 0.2976,
      "step": 39
    },
    {
      "epoch": 1.58,
      "learning_rate": 1.4345709189519963e-05,
      "loss": 0.2522,
      "step": 40
    },
    {
      "epoch": 1.62,
      "learning_rate": 1.3692663858785127e-05,
      "loss": 0.2336,
      "step": 41
    },
    {
      "epoch": 1.66,
      "learning_rate": 1.304210711669923e-05,
      "loss": 0.2732,
      "step": 42
    },
    {
      "epoch": 1.7,
      "learning_rate": 1.2395277334996045e-05,
      "loss": 0.2377,
      "step": 43
    },
    {
      "epoch": 1.74,
      "learning_rate": 1.1753405790928458e-05,
      "loss": 0.3129,
      "step": 44
    },
    {
      "epoch": 1.78,
      "learning_rate": 1.1117714323462188e-05,
      "loss": 0.2584,
      "step": 45
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.0489413007435905e-05,
      "loss": 0.3123,
      "step": 46
    },
    {
      "epoch": 1.85,
      "learning_rate": 9.86969785011497e-06,
      "loss": 0.2232,
      "step": 47
    },
    {
      "epoch": 1.89,
      "learning_rate": 9.259748514523654e-06,
      "loss": 0.1568,
      "step": 48
    },
    {
      "epoch": 1.93,
      "learning_rate": 8.66072607388951e-06,
      "loss": 0.2494,
      "step": 49
    },
    {
      "epoch": 1.97,
      "learning_rate": 8.073770801474494e-06,
      "loss": 0.3024,
      "step": 50
    },
    {
      "epoch": 2.04,
      "learning_rate": 7.500000000000004e-06,
      "loss": 0.551,
      "step": 51
    },
    {
      "epoch": 2.08,
      "learning_rate": 6.9405058747976406e-06,
      "loss": 0.2052,
      "step": 52
    },
    {
      "epoch": 2.12,
      "learning_rate": 6.3963534547343126e-06,
      "loss": 0.1905,
      "step": 53
    },
    {
      "epoch": 2.16,
      "learning_rate": 5.86857856486919e-06,
      "loss": 0.2084,
      "step": 54
    },
    {
      "epoch": 2.19,
      "learning_rate": 5.3581858547019095e-06,
      "loss": 0.1803,
      "step": 55
    },
    {
      "epoch": 2.23,
      "learning_rate": 4.8661468857650965e-06,
      "loss": 0.2179,
      "step": 56
    },
    {
      "epoch": 2.27,
      "learning_rate": 4.393398282201788e-06,
      "loss": 0.2408,
      "step": 57
    },
    {
      "epoch": 2.31,
      "learning_rate": 3.940839947848141e-06,
      "loss": 0.1756,
      "step": 58
    },
    {
      "epoch": 2.35,
      "learning_rate": 3.5093333532153316e-06,
      "loss": 0.1596,
      "step": 59
    },
    {
      "epoch": 2.39,
      "learning_rate": 3.0996998956314744e-06,
      "loss": 0.2365,
      "step": 60
    },
    {
      "epoch": 2.43,
      "learning_rate": 2.7127193356651214e-06,
      "loss": 0.172,
      "step": 61
    },
    {
      "epoch": 2.47,
      "learning_rate": 2.3491283128067174e-06,
      "loss": 0.1681,
      "step": 62
    },
    {
      "epoch": 2.5,
      "learning_rate": 2.0096189432334194e-06,
      "loss": 0.21,
      "step": 63
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.6948375023266743e-06,
      "loss": 0.2293,
      "step": 64
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.4053831944502509e-06,
      "loss": 0.1471,
      "step": 65
    },
    {
      "epoch": 2.62,
      "learning_rate": 1.1418070123306989e-06,
      "loss": 0.2285,
      "step": 66
    },
    {
      "epoch": 2.66,
      "learning_rate": 9.046106882113753e-07,
      "loss": 0.2122,
      "step": 67
    },
    {
      "epoch": 2.7,
      "learning_rate": 6.942457387765977e-07,
      "loss": 0.1508,
      "step": 68
    },
    {
      "epoch": 2.74,
      "learning_rate": 5.11112605663977e-07,
      "loss": 0.2877,
      "step": 69
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.5555989320099957e-07,
      "loss": 0.2275,
      "step": 70
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.278837048168797e-07,
      "loss": 0.2798,
      "step": 71
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.2832707939284428e-07,
      "loss": 0.2241,
      "step": 72
    },
    {
      "epoch": 2.89,
      "learning_rate": 5.7079528623816824e-08,
      "loss": 0.1843,
      "step": 73
    },
    {
      "epoch": 2.93,
      "learning_rate": 1.4276676272133027e-08,
      "loss": 0.1802,
      "step": 74
    },
    {
      "epoch": 2.97,
      "learning_rate": 0.0,
      "loss": 0.2303,
      "step": 75
    },
    {
      "epoch": 2.97,
      "step": 75,
      "total_flos": 2.4061184768475136e+16,
      "train_loss": 0.3887578751643499,
      "train_runtime": 5064.5824,
      "train_samples_per_second": 0.97,
      "train_steps_per_second": 0.015
    },
    {
      "epoch": 2.97,
      "eval_loss": 0.23003457486629486,
      "eval_runtime": 84.5662,
      "eval_samples_per_second": 4.848,
      "eval_steps_per_second": 0.307,
      "step": 75
    }
  ],
  "max_steps": 75,
  "num_train_epochs": 3,
  "total_flos": 2.4061184768475136e+16,
  "trial_name": null,
  "trial_params": null
}