目录

vLLM-Omni MetaX

vLLM Omni MetaX Plugin

Docs

| About MetaX | Documentation | #sig-maca |

About

vLLM-Omni MetaX is an adapter plugin that enables vLLM-Omni to run on MetaX GPUs by leveraging the existing vllm-metax backend.

Unlike vllm-metax, which provides the low-level hardware platform integration, this plugin focuses on bridging the vLLM-Omni execution stack (multi-stage, multimodal pipelines) to the MetaX runtime.

Responsibilities

Component Responsibility
vllm-metax Provides MetaX platform, kernels, and runtime
vllm-omni Handles multi-stage multimodal inference
vllm-omni-metax Bridges Omni to MetaX backend

Prerequisites

  • MetaX GPU (C-series)
  • Linux
  • Python 3.10 – 3.12
  • vLLM (version aligned with vllm-metax)
  • vLLM-Omni
  • vllm-metax

Installation

Install vllm-omni-metax as a Python package:

pip install -e .

Compatibility

The current documentation and adaptation target the 0.20.0 stack:

  • vllm-omni-metax 0.20.0
  • vllm-omni 0.20.0
  • vllm-metax 0.20.0

See docs for the full setup flow and runtime patch notes.

关于
66.0 KB
邀请码
    Gitlink(确实开源)
  • 加入我们
  • 官网邮箱:gitlink@ccf.org.cn
  • QQ群
  • QQ群
  • 公众号
  • 公众号

版权所有:中国计算机学会技术支持:开源发展技术委员会
京ICP备13000930号-9 京公网安备 11010802047560号