trainer_state.json

{
  "best_metric": 0.9949109414758269,
  "best_model_checkpoint": "dit-base-finetuned-rvlcdip-finetuned-custom-first/checkpoint-237",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 237,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.13,
      "learning_rate": 2.0833333333333336e-05,
      "loss": 1.9058,
      "step": 10
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.166666666666667e-05,
      "loss": 1.6887,
      "step": 20
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.8591549295774653e-05,
      "loss": 1.4773,
      "step": 30
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.624413145539906e-05,
      "loss": 1.1597,
      "step": 40
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.389671361502348e-05,
      "loss": 0.792,
      "step": 50
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.154929577464789e-05,
      "loss": 0.5423,
      "step": 60
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.9201877934272305e-05,
      "loss": 0.3686,
      "step": 70
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.9745547073791349,
      "eval_loss": 0.2355559766292572,
      "eval_runtime": 120.4299,
      "eval_samples_per_second": 19.58,
      "eval_steps_per_second": 0.614,
      "step": 79
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.6854460093896714e-05,
      "loss": 0.2791,
      "step": 80
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.450704225352113e-05,
      "loss": 0.2155,
      "step": 90
    },
    {
      "epoch": 1.27,
      "learning_rate": 3.215962441314554e-05,
      "loss": 0.1837,
      "step": 100
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.9812206572769952e-05,
      "loss": 0.1594,
      "step": 110
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.746478873239437e-05,
      "loss": 0.1428,
      "step": 120
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.511737089201878e-05,
      "loss": 0.1049,
      "step": 130
    },
    {
      "epoch": 1.77,
      "learning_rate": 2.2769953051643194e-05,
      "loss": 0.0977,
      "step": 140
    },
    {
      "epoch": 1.9,
      "learning_rate": 2.0422535211267607e-05,
      "loss": 0.0891,
      "step": 150
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9936386768447837,
      "eval_loss": 0.07920637726783752,
      "eval_runtime": 116.9591,
      "eval_samples_per_second": 20.161,
      "eval_steps_per_second": 0.633,
      "step": 158
    },
    {
      "epoch": 2.03,
      "learning_rate": 1.807511737089202e-05,
      "loss": 0.0908,
      "step": 160
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.5727699530516433e-05,
      "loss": 0.0788,
      "step": 170
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.3380281690140845e-05,
      "loss": 0.0866,
      "step": 180
    },
    {
      "epoch": 2.41,
      "learning_rate": 1.1032863849765258e-05,
      "loss": 0.0754,
      "step": 190
    },
    {
      "epoch": 2.53,
      "learning_rate": 8.685446009389673e-06,
      "loss": 0.0684,
      "step": 200
    },
    {
      "epoch": 2.66,
      "learning_rate": 6.338028169014085e-06,
      "loss": 0.0585,
      "step": 210
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.990610328638498e-06,
      "loss": 0.0715,
      "step": 220
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.643192488262911e-06,
      "loss": 0.0652,
      "step": 230
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9949109414758269,
      "eval_loss": 0.05667030066251755,
      "eval_runtime": 116.045,
      "eval_samples_per_second": 20.32,
      "eval_steps_per_second": 0.638,
      "step": 237
    },
    {
      "epoch": 3.0,
      "step": 237,
      "total_flos": 2.349294757953454e+18,
      "train_loss": 0.41549516755317334,
      "train_runtime": 2328.1847,
      "train_samples_per_second": 13.02,
      "train_steps_per_second": 0.102
    }
  ],
  "logging_steps": 10,
  "max_steps": 237,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 2.349294757953454e+18,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}