src/config.py

"""
CONFIGURATION FILE
Centralized configuration for the project
"""

import torch
from pathlib import Path

class Config:
    """Global configuration"""
    
    # Paths
    PROJECT_ROOT = Path(__file__).parent.parent
    DATA_DIR = PROJECT_ROOT / 'data'
    CHECKPOINT_DIR = PROJECT_ROOT / 'checkpoints'
    RESULTS_DIR = PROJECT_ROOT / 'results'
    LOGS_DIR = PROJECT_ROOT / 'logs'
    
    # Data
    PROCESSED_DATA_DIR = DATA_DIR / 'processed'
    VI_VOCAB_PATH = PROCESSED_DATA_DIR / 'vi_vocab.pkl'
    EN_VOCAB_PATH = PROCESSED_DATA_DIR / 'en_vocab.pkl'
    PROCESSED_DATA_PATH = PROCESSED_DATA_DIR / 'processed_data.pkl'
    
    # Model
    MODEL_SIZE = 'base'  # 'tiny', 'small', 'base', 'large'
    PAD_IDX = 0
    
    # Training
    NUM_EPOCHS = 20
    BATCH_SIZE = 32
    WARMUP_STEPS = 4000
    LABEL_SMOOTHING = 0.1
    
    # Device
    DEVICE = 'cuda' if torch.cuda.is_available() else 'cpu'