Implementation:InternLM Lmdeploy Daily Ete Test H800

Knowledge Sources	InternLM_Lmdeploy
Domains	CI, Testing, GPU
Last Updated	2026-02-07 15:00 GMT

Overview

A GitHub Actions workflow that runs daily end-to-end tests for lmdeploy on NVIDIA H800 GPUs, targeting large-scale model testing with multi-GPU configurations.

Description

The daily_ete_test_h800.yml workflow is a hardware-specific variant designed for H800 GPUs. It runs on self-hosted runners labeled h800-r1 and uses the Docker image m.daocloud.io/docker.io/openmmlab/lmdeploy:latest-cu12.8 (pulled from a Chinese mirror registry).

The workflow includes:

• linux-build: Builds the lmdeploy wheel on ubuntu-latest with CUDA 12.8.
• download_pkgs: Downloads and stages build artifacts on the H800 runner. Mounts multiple NVMe volumes for model storage.
• test_tools: Matrix job covering backends x models x functions. Runs multi-GPU tests including 1, 2, 4, and 8 GPU configurations with the not other marker filter. This is the only daily test variant that routinely exercises 8-GPU parallelism.
• test_restful: Tests the RESTful API with Intern-S1 model at TP=8 (8 GPUs), validating large model serving with a 15-minute startup wait.
• get_coverage_report: Combines coverage data from all test jobs.

Key differences from other variants: no quantization test job, uses Chinese Docker registry mirror, tests large models requiring 8 GPUs (e.g., Intern-S1), and the default regression functions are limited to ['tools','restful'].

Usage

Triggered automatically on weekdays at 14:00 UTC or manually via GitHub Actions dispatch. Primarily used for validating large model support on high-end GPU hardware.

Code Reference

Source Location

• Repository: InternLM_Lmdeploy
• File: .github/workflows/daily_ete_test_h800.yml
• Lines: 1-355

Signature

name: daily_ete_test_h800

on:
  workflow_dispatch:
    inputs:
      repo_org: { type: string, default: 'InternLM/lmdeploy' }
      repo_ref: { type: string, default: 'main' }
      backend: { type: string, default: "['turbomind', 'pytorch']" }
      model: { type: string, default: "['llm','mllm']" }
      function: { type: string, default: '["pipeline", "restful", "chat"]' }
      offline_mode: { type: boolean, default: false }
      regression_func: { type: string, default: "['tools','restful']" }
  schedule:
    - cron: '00 14 * * 0-4'

jobs:
  linux-build: ...
  download_pkgs: ...
  test_tools: ...
  test_restful: ...
  get_coverage_report: ...

I/O Contract

Inputs

Name	Type	Required	Description
repo_org	string	No	Repository organization/name (default: InternLM/lmdeploy)
repo_ref	string	No	Branch, tag, or commit ID (default: main)
backend	string	Yes	JSON list of backends to test
model	string	Yes	JSON list of model types: llm, mllm
function	string	Yes	JSON list of test functions
offline_mode	boolean	Yes	Whether to use pre-prepared offline packages
regression_func	string	Yes	JSON list: tools, restful

Outputs

Name	Type	Description
Build artifacts	wheel file	lmdeploy wheel package for CUDA 12.8
Test reports	Allure reports	Test results on H800 hardware
Coverage report	XML/text	Combined code coverage report

Usage Examples

# Manual trigger to test large model restful API on H800:
# Go to Actions > daily_ete_test_h800 > Run workflow
# Set regression_func: "['restful']"
# This will test Intern-S1 at TP=8 on 8 H800 GPUs

Related Pages

• Environment:InternLM_Lmdeploy_CUDA_GPU_Runtime