checkpoint-2000/trainer_state.json

{
  "best_global_step": 2000,
  "best_metric": 2.37511335409849,
  "best_model_checkpoint": "./SALAMA_NEWMED6/checkpoint-2000",
  "epoch": 1.1267605633802817,
  "eval_steps": 2000,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.056338028169014086,
      "grad_norm": 2.0612876415252686,
      "learning_rate": 1.98e-06,
      "loss": 0.0564,
      "step": 100
    },
    {
      "epoch": 0.11267605633802817,
      "grad_norm": 1.467324137687683,
      "learning_rate": 3.980000000000001e-06,
      "loss": 0.0525,
      "step": 200
    },
    {
      "epoch": 0.16901408450704225,
      "grad_norm": 1.1306756734848022,
      "learning_rate": 5.98e-06,
      "loss": 0.0595,
      "step": 300
    },
    {
      "epoch": 0.22535211267605634,
      "grad_norm": 2.680368661880493,
      "learning_rate": 7.980000000000002e-06,
      "loss": 0.0561,
      "step": 400
    },
    {
      "epoch": 0.28169014084507044,
      "grad_norm": 4.0752339363098145,
      "learning_rate": 9.980000000000001e-06,
      "loss": 0.0614,
      "step": 500
    },
    {
      "epoch": 0.3380281690140845,
      "grad_norm": 2.2631568908691406,
      "learning_rate": 9.794818652849742e-06,
      "loss": 0.0635,
      "step": 600
    },
    {
      "epoch": 0.39436619718309857,
      "grad_norm": 3.81473445892334,
      "learning_rate": 9.587564766839379e-06,
      "loss": 0.0656,
      "step": 700
    },
    {
      "epoch": 0.4507042253521127,
      "grad_norm": 4.52515983581543,
      "learning_rate": 9.380310880829016e-06,
      "loss": 0.0674,
      "step": 800
    },
    {
      "epoch": 0.5070422535211268,
      "grad_norm": 5.0459513664245605,
      "learning_rate": 9.173056994818653e-06,
      "loss": 0.0746,
      "step": 900
    },
    {
      "epoch": 0.5633802816901409,
      "grad_norm": 2.449471950531006,
      "learning_rate": 8.96580310880829e-06,
      "loss": 0.0677,
      "step": 1000
    },
    {
      "epoch": 0.6197183098591549,
      "grad_norm": 2.445279121398926,
      "learning_rate": 8.758549222797929e-06,
      "loss": 0.0696,
      "step": 1100
    },
    {
      "epoch": 0.676056338028169,
      "grad_norm": 4.043335437774658,
      "learning_rate": 8.551295336787564e-06,
      "loss": 0.0685,
      "step": 1200
    },
    {
      "epoch": 0.7323943661971831,
      "grad_norm": 1.7198883295059204,
      "learning_rate": 8.344041450777203e-06,
      "loss": 0.0647,
      "step": 1300
    },
    {
      "epoch": 0.7887323943661971,
      "grad_norm": 3.9584922790527344,
      "learning_rate": 8.13678756476684e-06,
      "loss": 0.0754,
      "step": 1400
    },
    {
      "epoch": 0.8450704225352113,
      "grad_norm": 1.50466787815094,
      "learning_rate": 7.929533678756477e-06,
      "loss": 0.0658,
      "step": 1500
    },
    {
      "epoch": 0.9014084507042254,
      "grad_norm": 3.0848331451416016,
      "learning_rate": 7.722279792746114e-06,
      "loss": 0.0683,
      "step": 1600
    },
    {
      "epoch": 0.9577464788732394,
      "grad_norm": 2.9346296787261963,
      "learning_rate": 7.515025906735752e-06,
      "loss": 0.0664,
      "step": 1700
    },
    {
      "epoch": 1.0140845070422535,
      "grad_norm": 2.4397850036621094,
      "learning_rate": 7.307772020725389e-06,
      "loss": 0.0512,
      "step": 1800
    },
    {
      "epoch": 1.0704225352112675,
      "grad_norm": 1.455031156539917,
      "learning_rate": 7.100518134715026e-06,
      "loss": 0.0204,
      "step": 1900
    },
    {
      "epoch": 1.1267605633802817,
      "grad_norm": 1.2992125749588013,
      "learning_rate": 6.893264248704664e-06,
      "loss": 0.0225,
      "step": 2000
    },
    {
      "epoch": 1.1267605633802817,
      "eval_loss": 0.02766764909029007,
      "eval_runtime": 10449.8663,
      "eval_samples_per_second": 1.359,
      "eval_steps_per_second": 0.17,
      "eval_wer": 2.37511335409849,
      "step": 2000
    }
  ],
  "logging_steps": 100,
  "max_steps": 5325,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 2000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 6.531667287146496e+19,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}