graphcodebert-swa-from-epoch-1/checkpoint-1400/config_hyperparams.json

{
  "train_config": {
    "model_name": "/kaggle/input/models/dzung271828/microsoft-graphcodebert-base/transformers/default/1",
    "num_epochs": 4,
    "batch_size": 1024,
    "learning_rate": 1e-06,
    "max_length": 512,
    "num_labels": 2,
    "loss_type": "r-drop",
    "focal_alpha": 1.0,
    "focal_gamma": 2.0,
    "r_drop_alpha": 10.0,
    "infonce_temperature": 0.07,
    "infonce_weight": 0.5,
    "label_smoothing": 0.5,
    "adversarial_epsilon": 0.5,
    "use_swa": true,
    "swa_start_epoch": 1,
    "swa_lr": 1e-05,
    "data_augmentation": true,
    "aug_rename_prob": 0.8,
    "aug_format_prob": 0.8,
    "freeze_base": true,
    "seed": 42,
    "use_wandb": false,
    "mixup_alpha": 1.0,
    "low_pass_keep_ratio": 0.5,
    "freq_consistency_weight": 0.5
  },
  "training_arguments": {
    "output_dir": "graphcodebert-swa-from-epoch-1/",
    "num_train_epochs": 4,
    "per_device_train_batch_size": 1024,
    "per_device_eval_batch_size": 2048,
    "learning_rate": 1e-06,
    "warmup_steps": 195,
    "weight_decay": 0.1,
    "logging_steps": 5,
    "eval_steps": 100,
    "save_steps": 100,
    "metric_for_best_model": "macro_f1",
    "greater_is_better": true,
    "save_total_limit": 5,
    "fp16": true,
    "seed": 42
  },
  "training_state": {
    "global_step": 1400,
    "epoch": 2.8629856850715747,
    "best_metric": 0.6459359532496649,
    "best_model_checkpoint": "graphcodebert-swa-from-epoch-1/checkpoint-1400"
  }
}