Last released Jun 1, 2024
A high-throughput and memory-efficient inference and serving engine for LLMs
Last released May 31, 2024
Forward-only flash-attn
Supported by