Skip to content
name: MLPerf inference bert (deepsparse, tf, onnxruntime, pytorch)
on:
pull_request:
branches: [ "main", "dev" ]
paths:
- '.github/workflows/test-mlperf-inference-bert-deepsparse-tf-onnxruntime-pytorch.yml'
- '**'
- '!**.md'
jobs:
build:
runs-on: ${{ matrix.os }}
strategy:
fail-fast: false
matrix:
# 3.12 didn't work on 20240305 - need to check
python-version: [ "3.11" ]
backend: [ "deepsparse", "tf", "onnxruntime", "pytorch" ]
precision: [ "int8", "fp32" ]
os: [ubuntu-latest, windows-latest, macos-latest]
exclude:
- backend: tf
- backend: pytorch
- backend: onnxruntime
- precision: fp32
- os: windows-latest
steps:
- uses: actions/checkout@v3
- name: Set up Python ${{ matrix.python-version }}
uses: actions/setup-python@v3
with:
python-version: ${{ matrix.python-version }}
- name: Install mlcflow
run: |
python -m pip install --upgrade pip
python -m pip install --ignore-installed --verbose pip setuptools
python -m pip install .
mlc pull repo mlcommons@mlperf-automations --branch=dev
- name: Test MLPerf Inference Bert ${{ matrix.backend }} on ${{ matrix.os }}
if: matrix.os == 'windows-latest'
run: |
mlcr --tags=run,mlperf,inference,generate-run-cmds,_submission,_short --submitter="MLCommons" --hw_name=gh_${{ matrix.os }} --model=bert-99 --backend=${{ matrix.backend }} --device=cpu --scenario=Offline --test_query_count=5 --adr.loadgen.tags=_from-pip --pip_loadgen=yes --precision=${{ matrix.precision }} --target_qps=1 -v --quiet
- name: Test MLPerf Inference Bert ${{ matrix.backend }} on ${{ matrix.os }}
if: matrix.os != 'windows-latest'
run: |
mlcr --tags=run,mlperf,inference,generate-run-cmds,_submission,_short --submitter="MLCommons" --hw_name=gh_${{ matrix.os }}_x86 --model=bert-99 --backend=${{ matrix.backend }} --device=cpu --scenario=Offline --test_query_count=5 --precision=${{ matrix.precision }} --target_qps=1 -v --quiet