Skip to content

Latest commit

 

History

History

InferenceX end-to-end benchmarks

English | 中文

This project contains model-serving benchmarks, configurations, launchers, Python tooling, documentation, and performance history. Start with the documentation index.

The Python manifest and lockfile belong to this project. From the repository root, install and run the local tooling with:

cd inferencex-e2e
uv sync --locked
uv run --locked python -m infx.matrix.generate test-config \
  --config-files configs/nvidia-master.yaml --config-keys <key>

Replace <key> with the selected config key. Workflow dispatch generator arguments also use paths relative to this directory. See the testing guide for dependency groups and checks.

The project-local Python version selects its interpreter. Repository policy and GitHub workflows remain at the repository root. The local license keeps attribution available when containers mount only this project directory.