{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "global_step": 32735,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.08,
      "learning_rate": 1.9697571406751184e-05,
      "loss": 0.4892,
      "step": 500
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9392087979227128e-05,
      "loss": 0.4021,
      "step": 1000
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.908660455170307e-05,
      "loss": 0.3816,
      "step": 1500
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.8781121124179017e-05,
      "loss": 0.3486,
      "step": 2000
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.8475637696654957e-05,
      "loss": 0.335,
      "step": 2500
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.81701542691309e-05,
      "loss": 0.3268,
      "step": 3000
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.786528180846189e-05,
      "loss": 0.3362,
      "step": 3500
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.7559798380937837e-05,
      "loss": 0.3123,
      "step": 4000
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.7254314953413777e-05,
      "loss": 0.3114,
      "step": 4500
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.694883152588972e-05,
      "loss": 0.2938,
      "step": 5000
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.6643348098365667e-05,
      "loss": 0.297,
      "step": 5500
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.6337864670841607e-05,
      "loss": 0.2818,
      "step": 6000
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.60329922101726e-05,
      "loss": 0.2854,
      "step": 6500
    },
    {
      "epoch": 1.07,
      "learning_rate": 1.5727508782648543e-05,
      "loss": 0.2273,
      "step": 7000
    },
    {
      "epoch": 1.15,
      "learning_rate": 1.5422025355124486e-05,
      "loss": 0.2193,
      "step": 7500
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.511654192760043e-05,
      "loss": 0.2276,
      "step": 8000
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.4811058500076371e-05,
      "loss": 0.2151,
      "step": 8500
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.4505575072552316e-05,
      "loss": 0.2193,
      "step": 9000
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.4200702611883306e-05,
      "loss": 0.2363,
      "step": 9500
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.3897052084924394e-05,
      "loss": 0.2273,
      "step": 10000
    },
    {
      "epoch": 1.6,
      "learning_rate": 1.3591568657400339e-05,
      "loss": 0.2275,
      "step": 10500
    },
    {
      "epoch": 1.68,
      "learning_rate": 1.328608522987628e-05,
      "loss": 0.2247,
      "step": 11000
    },
    {
      "epoch": 1.76,
      "learning_rate": 1.2980601802352224e-05,
      "loss": 0.2133,
      "step": 11500
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.2675118374828165e-05,
      "loss": 0.2222,
      "step": 12000
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.236963494730411e-05,
      "loss": 0.2136,
      "step": 12500
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.2064151519780052e-05,
      "loss": 0.2185,
      "step": 13000
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.1758668092255997e-05,
      "loss": 0.1531,
      "step": 13500
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.145318466473194e-05,
      "loss": 0.1496,
      "step": 14000
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.1147701237207882e-05,
      "loss": 0.1467,
      "step": 14500
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.0842217809683826e-05,
      "loss": 0.1674,
      "step": 15000
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0536734382159768e-05,
      "loss": 0.1505,
      "step": 15500
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.0231250954635711e-05,
      "loss": 0.1447,
      "step": 16000
    },
    {
      "epoch": 2.52,
      "learning_rate": 9.925767527111655e-06,
      "loss": 0.1407,
      "step": 16500
    },
    {
      "epoch": 2.6,
      "learning_rate": 9.620284099587598e-06,
      "loss": 0.153,
      "step": 17000
    },
    {
      "epoch": 2.67,
      "learning_rate": 9.314800672063541e-06,
      "loss": 0.1534,
      "step": 17500
    },
    {
      "epoch": 2.75,
      "learning_rate": 9.009317244539484e-06,
      "loss": 0.1499,
      "step": 18000
    },
    {
      "epoch": 2.83,
      "learning_rate": 8.703833817015428e-06,
      "loss": 0.1605,
      "step": 18500
    },
    {
      "epoch": 2.9,
      "learning_rate": 8.398350389491371e-06,
      "loss": 0.1674,
      "step": 19000
    },
    {
      "epoch": 2.98,
      "learning_rate": 8.093477928822362e-06,
      "loss": 0.1524,
      "step": 19500
    },
    {
      "epoch": 3.05,
      "learning_rate": 7.788605468153352e-06,
      "loss": 0.0993,
      "step": 20000
    },
    {
      "epoch": 3.13,
      "learning_rate": 7.483122040629296e-06,
      "loss": 0.081,
      "step": 20500
    },
    {
      "epoch": 3.21,
      "learning_rate": 7.17763861310524e-06,
      "loss": 0.0811,
      "step": 21000
    },
    {
      "epoch": 3.28,
      "learning_rate": 6.872155185581183e-06,
      "loss": 0.0825,
      "step": 21500
    },
    {
      "epoch": 3.36,
      "learning_rate": 6.566671758057126e-06,
      "loss": 0.0823,
      "step": 22000
    },
    {
      "epoch": 3.44,
      "learning_rate": 6.261188330533069e-06,
      "loss": 0.0808,
      "step": 22500
    },
    {
      "epoch": 3.51,
      "learning_rate": 5.955704903009012e-06,
      "loss": 0.078,
      "step": 23000
    },
    {
      "epoch": 3.59,
      "learning_rate": 5.650221475484955e-06,
      "loss": 0.0764,
      "step": 23500
    },
    {
      "epoch": 3.67,
      "learning_rate": 5.344738047960899e-06,
      "loss": 0.0769,
      "step": 24000
    },
    {
      "epoch": 3.74,
      "learning_rate": 5.040476554146938e-06,
      "loss": 0.0861,
      "step": 24500
    },
    {
      "epoch": 3.82,
      "learning_rate": 4.734993126622881e-06,
      "loss": 0.0956,
      "step": 25000
    },
    {
      "epoch": 3.89,
      "learning_rate": 4.429509699098824e-06,
      "loss": 0.0827,
      "step": 25500
    },
    {
      "epoch": 3.97,
      "learning_rate": 4.124026271574768e-06,
      "loss": 0.0836,
      "step": 26000
    },
    {
      "epoch": 4.05,
      "learning_rate": 3.81854284405071e-06,
      "loss": 0.0529,
      "step": 26500
    },
    {
      "epoch": 4.12,
      "learning_rate": 3.5130594165266534e-06,
      "loss": 0.0371,
      "step": 27000
    },
    {
      "epoch": 4.2,
      "learning_rate": 3.208797922712693e-06,
      "loss": 0.0408,
      "step": 27500
    },
    {
      "epoch": 4.28,
      "learning_rate": 2.903314495188636e-06,
      "loss": 0.044,
      "step": 28000
    },
    {
      "epoch": 4.35,
      "learning_rate": 2.5978310676645796e-06,
      "loss": 0.034,
      "step": 28500
    },
    {
      "epoch": 4.43,
      "learning_rate": 2.2923476401405224e-06,
      "loss": 0.041,
      "step": 29000
    },
    {
      "epoch": 4.51,
      "learning_rate": 1.9868642126164657e-06,
      "loss": 0.0402,
      "step": 29500
    },
    {
      "epoch": 4.58,
      "learning_rate": 1.6813807850924087e-06,
      "loss": 0.0321,
      "step": 30000
    },
    {
      "epoch": 4.66,
      "learning_rate": 1.375897357568352e-06,
      "loss": 0.0349,
      "step": 30500
    },
    {
      "epoch": 4.73,
      "learning_rate": 1.0704139300442952e-06,
      "loss": 0.0422,
      "step": 31000
    },
    {
      "epoch": 4.81,
      "learning_rate": 7.649305025202383e-07,
      "loss": 0.041,
      "step": 31500
    },
    {
      "epoch": 4.89,
      "learning_rate": 4.594470749961815e-07,
      "loss": 0.0445,
      "step": 32000
    },
    {
      "epoch": 4.96,
      "learning_rate": 1.5396364747212466e-07,
      "loss": 0.0376,
      "step": 32500
    },
    {
      "epoch": 5.0,
      "step": 32735,
      "total_flos": 1.2201660848633088e+17,
      "train_loss": 0.16656628906298943,
      "train_runtime": 2957.7214,
      "train_samples_per_second": 177.067,
      "train_steps_per_second": 11.068
    }
  ],
  "max_steps": 32735,
  "num_train_epochs": 5,
  "total_flos": 1.2201660848633088e+17,
  "trial_name": null,
  "trial_params": null
}