turboderp/exllamav2

Features

• Inference Engines - High-speed inference library for modern consumer-grade GPUs.
• Model Quantization - Fast inference library optimized for low-bitwidth model quantization.