training_config.json

{
    "model_name": "microsoft/deberta-v3-large",
    "num_labels": 2,
    "train_file": "train_windows.jsonl",
    "val_file": "val_windows.jsonl",
    "max_length": 512,
    "batch_size": 8,
    "gradient_accumulation_steps": 8,
    "num_epochs": 3,
    "learning_rate": 1e-06,
    "warmup_ratio": 0.1,
    "weight_decay": 0.01,
    "max_grad_norm": 1.0,
    "label_smoothing": 0.0,
    "device": "cuda",
    "num_workers": 0,
    "seed": 42,
    "bf16": true,
    "logging_steps": 1,
    "eval_steps": 5000,
    "save_steps": 10000,
    "output_dir": "./deberta_link_output",
    "wandb_project": "deberta-link-classification",
    "wandb_name": "deberta-v3-large-link-tokens",
    "patience": 2,
    "min_delta": 0.0001,
    "max_checkpoints": 5,
    "protect_latest_epoch_step": true
}