download_model.py

#!/usr/bin/env python3
"""
Helper script to download the int4 model files at build time for Hugging Face Spaces
"""

import os
import sys
import subprocess
import logging
from pathlib import Path

# Configure logging
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)

# Model configuration
MAIN_MODEL_ID = "Tonic/petite-elle-L-aime-3-sft"
INT4_MODEL_ID = "Tonic/petite-elle-L-aime-3-sft/int4"
LOCAL_MODEL_PATH = "./int4"

def download_model():
    """Download the int4 model files to local directory"""
    try:
        logger.info(f"Downloading int4 model from {INT4_MODEL_ID}")
        
        # Create local directory if it doesn't exist
        os.makedirs(LOCAL_MODEL_PATH, exist_ok=True)
        
        # Use huggingface_hub to download the model
        from huggingface_hub import snapshot_download
        
        # Download the int4 model files
        snapshot_download(
            repo_id=INT4_MODEL_ID,
            local_dir=LOCAL_MODEL_PATH,
            local_dir_use_symlinks=False,
            ignore_patterns=["*.md", "*.txt", "*.git*", "*.ipynb", "*.py"]
        )
        
        logger.info(f"Model downloaded successfully to {LOCAL_MODEL_PATH}")
        return True
        
    except Exception as e:
        logger.error(f"Error downloading model: {e}")
        return False

def check_model_files():
    """Check if required model files exist"""
    required_files = [
        "config.json",
        "pytorch_model.bin",
        "tokenizer.json",
        "tokenizer_config.json"
    ]
    
    missing_files = []
    for file in required_files:
        file_path = os.path.join(LOCAL_MODEL_PATH, file)
        if not os.path.exists(file_path):
            missing_files.append(file)
    
    if missing_files:
        logger.error(f"Missing model files: {missing_files}")
        return False
    
    logger.info("All required model files found")
    return True

def verify_model_integrity():
    """Verify that the downloaded model files are valid"""
    try:
        # Try to load the tokenizer to verify it's working
        from transformers import AutoTokenizer
        tokenizer = AutoTokenizer.from_pretrained(LOCAL_MODEL_PATH)
        logger.info("Tokenizer loaded successfully from local files")
        
        # Try to load the model config
        from transformers import AutoConfig
        config = AutoConfig.from_pretrained(LOCAL_MODEL_PATH)
        logger.info("Model config loaded successfully from local files")
        
        return True
        
    except Exception as e:
        logger.error(f"Error verifying model integrity: {e}")
        return False

def main():
    """Main function to download model at build time"""
    logger.info("Starting model download for Hugging Face Space...")
    
    # Check if model files already exist
    if check_model_files():
        logger.info("Model files already exist, verifying integrity...")
        if verify_model_integrity():
            logger.info("Model files verified successfully")
            return True
        else:
            logger.warning("Model files exist but failed integrity check, re-downloading...")
    
    # Download the model
    if download_model():
        logger.info("Model download completed successfully")
        
        # Verify the downloaded files
        if check_model_files() and verify_model_integrity():
            logger.info("Model download and verification completed successfully")
            return True
        else:
            logger.error("Model download completed but verification failed")
            return False
    else:
        logger.error("Model download failed")
        return False

if __name__ == "__main__":
    success = main()
    sys.exit(0 if success else 1)