Find your next open-source project

Search by name, purpose, or keyword — or paste a GitHub repository URL
All skills

optimization-bitsandbytes

Quantizes LLMs to 8-bit or 4-bit to cut GPU memory use by 50-75% with little accuracy loss.