data/config_marker.json

{
    "mode": "marker",
    "data_path": "data/data_augmented_256.jsonl",
    "output_dir": "models/marker",
    "model_name": "distilbert-base-uncased",
    "max_len": 256,
    "num_train_epochs": 3,
    "per_device_train_batch_size": 32,
    "per_device_eval_batch_size": 64,
    "gradient_accumulation_steps": 1,
    "learning_rate": 1e-5,
    "warmup_ratio": 0.1,
    "weight_decay": 0.1,
    "val_split": 0.1,
    "test_split": 0.1,
    "early_stopping_patience": 2,
    "fp16": true,
    "seed": 42,
    "logging_steps": 50,
    "save_total_limit": 2
}