1 resource · 0 articles
High-throughput open-source LLM serving library using PagedAttention for efficient inference.