Community maintained hardware plugin for vLLM on Ascend

A high-throughput and memory-efficient inference and serving engine for LLMs

vLLM’s reference system for K8S-native cluster-wide deployment with community-driven performance optimization

Transformers-compatible library for applying various compression algorithms to LLMs for optimized deployment with vLLM

Community maintained hardware plugin for vLLM on Spyre

Cost-efficient and pluggable Infrastructure components for GenAI inference

This repo hosts code for vLLM CI & Performance Benchmark infrastructure.