download_original.py

#!/usr/bin/env python3
"""
Download original DeepSeek-V2-Lite model and copy to our directory
"""

import os
from huggingface_hub import snapshot_download
import shutil

# Download original model
print("Downloading original DeepSeek-V2-Lite model...")
original_path = snapshot_download(
    repo_id="deepseek-ai/DeepSeek-V2-Lite",
    cache_dir="/tmp/deepseek_download"
)

print(f"Downloaded to: {original_path}")

# Target directory
target_dir = "/mnt/weka/home/shibo.hao/feng/code/junxia/deepseekv2lite"

# Files to copy (only model weights, keep our custom files)
files_to_copy = [
    "model-00001-of-000004.safetensors",
    "model-00002-of-000004.safetensors", 
    "model-00003-of-000004.safetensors",
    "model-00004-of-000004.safetensors",
    "model.safetensors.index.json"
]

# Copy files
print("Copying model weight files...")
for filename in files_to_copy:
    src = os.path.join(original_path, filename)
    dst = os.path.join(target_dir, filename)
    
    if os.path.exists(src):
        print(f"Copying {filename}...")
        shutil.copy2(src, dst)
        # Check file size
        size = os.path.getsize(dst)
        print(f"  Size: {size / (1024**3):.2f} GB")
    else:
        print(f"Warning: {filename} not found in original model")

print("Done! Model weights copied successfully.")