Last released Oct 17, 2024
A high-throughput and memory-efficient inference and serving engine for LLMs
Last released Sep 5, 2024
Forward-only flash-attn
Supported by