deep-cache-cleaner 1.0.1

Comprehensive cache clearing utility for PyTorch, HuggingFace, and other ML libraries

File: README.md

Deep Cache Cleaner

A comprehensive cache clearing utility for PyTorch, HuggingFace, and other ML libraries.

Features

• 🧹 Clear PyTorch GPU memory cache
• 🤗 Clear HuggingFace transformers cache
• 📝 Clear sentence-transformers cache
• 🗑️ Clear Python pycache directories
• 📊 Check cache sizes and GPU memory usage
• 💣 Aggressive cleanup mode (deletes downloaded models)
• 🔧 Command-line interface
• 📦 Easy to use Python API

Installation

pip install deep-cache-cleaner

Optional Dependencies

For full functionality, install with optional dependencies:

# For HuggingFace transformers support
pip install deep-cache-cleaner[transformers]

# For sentence-transformers support  
pip install deep-cache-cleaner[sentence-transformers]

# For development
pip install deep-cache-cleaner[dev]

Usage

Command Line Interface

# Clear all caches
deep-cache-cleaner clear

# Check cache sizes
deep-cache-cleaner check  

# Aggressive cleanup (deletes downloaded models)
deep-cache-cleaner aggressive

# Auto-confirm aggressive cleanup
deep-cache-cleaner aggressive --yes

# Quiet mode
deep-cache-cleaner clear --quiet

Python API

import cache_cleaner

# Clear all caches
cache_cleaner.clear_all_caches()

# Check cache sizes
sizes = cache_cleaner.check_cache_sizes()
print(f"Total cache size: {sizes['Total']:.2f} GB")

# Aggressive cleanup
cache_cleaner.aggressive_cache_clear()

# Use the class interface for more control
cleaner = cache_cleaner.CacheCleaner(verbose=True)
cleaner.clear_all_caches()

What Gets Cleared

Standard Cleanup (clear)

• PyTorch GPU memory cache
• Python garbage collection
• In-memory HuggingFace model cache
• Python pycache directories
• Reports on cache sizes and temp files

Aggressive Cleanup (aggressive)

⚠️ Warning: This deletes downloaded models!

Everything from standard cleanup, plus:

• HuggingFace cache directory (~/.cache/huggingface)
• Sentence-transformers cache (~/.cache/sentence_transformers)
• Torch cache directory (~/.cache/torch)
• Cache directories from environment variables

Requirements

• Python 3.8+
• PyTorch (for GPU cache clearing)
• Optional: transformers, sentence-transformers

License

MIT License

Contributing

Contributions welcome! Please read our contributing guidelines and submit pull requests.