2.1 KiB
MLPerf EDU Install Guide
This is the install contract for the runnable preview. The public command is
mlperf; the distribution package is mlperf-edu; the Python compatibility
package is mlperf_edu.
Recommended: uv From A Checkout
cd mlperf-edu
uv sync --extra dev
uv run mlperf doctor
uv run mlperf list profiles
uv run mlperf init --profile min
Use this path for development, classrooms, and artifact evaluation from a
checked-out repository. It installs the project in an isolated .venv and keeps
the mlperf command tied to the current source tree.
Install The CLI As A Tool
cd mlperf-edu
uv tool install .
mlperf doctor
mlperf run --profile min --dry-run
Use this path when the goal is to make mlperf available as a normal command.
The command uses the packaged workloads.yaml registry when it is not running
inside a source checkout.
Build A Wheel And Install It
cd mlperf-edu
uv build
uv tool install --force dist/mlperf_edu-0.1.0-py3-none-any.whl
mlperf doctor
mlperf audit
mlperf validate smoke --dry-run
The wheel contains src/mlperf_edu/workloads.yaml, a materialized copy of the
generated flat registry. Keep it synchronized with:
python3 tools/export_flat_registry.py --check
python3 tools/export_flat_registry.py
pip Fallback
cd mlperf-edu
python3 -m venv .venv
. .venv/bin/activate
python -m pip install --upgrade pip
python -m pip install -e ".[dev]"
mlperf doctor
Optional Extras
uv sync --extra audio
uv sync --extra dev --extra audio
audio installs torchaudio for full Speech Commands experiments. The default
suite can run without it because the local preview uses lightweight paths for
the tiny workloads.
Packaging Checks Before Release
python3 tools/export_registry_layout.py --check
python3 tools/export_flat_registry.py --check
python3 tools/generate_review_packets.py --check
uv build
python3 -m pytest -q
mlperf audit
mlperf validate smoke --dry-run
Long validation remains separate:
mlperf validate release --keep-going --skip-doctor