hotpotqa-nightly #34
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| name: hotpotqa-nightly | |
| on: | |
| workflow_dispatch: | |
| inputs: | |
| sample_limit: | |
| description: "Optional sample limit. Use 0 for the full downloaded file." | |
| required: false | |
| default: "0" | |
| top_k: | |
| description: "Retriever top_k for each mode." | |
| required: false | |
| default: "10" | |
| modes: | |
| description: "Comma-separated retriever modes." | |
| required: false | |
| default: "lexical_baseline,embedding_baseline,weighted_graph,activation_spreading_v1" | |
| schedule: | |
| - cron: "0 2 * * *" | |
| jobs: | |
| hotpotqa-nightly: | |
| runs-on: ubuntu-latest | |
| timeout-minutes: 90 | |
| permissions: | |
| contents: read | |
| steps: | |
| - uses: actions/checkout@v5 | |
| - uses: actions/setup-python@v6 | |
| with: | |
| python-version: "3.11" | |
| cache: "pip" | |
| cache-dependency-path: "pyproject.toml" | |
| - name: Install | |
| run: python -m pip install --no-build-isolation -e . | |
| - name: Download HotpotQA dev distractor | |
| run: python scripts/download_hotpotqa.py --force | |
| - name: Run HotpotQA nightly benchmark | |
| run: > | |
| python scripts/run_hotpotqa_benchmark.py | |
| --dataset "benchmarks/external/hotpotqa/data/hotpot_dev_distractor_v1.json" | |
| --limit "${{ github.event.inputs.sample_limit || '0' }}" | |
| --top-k "${{ github.event.inputs.top_k || '10' }}" | |
| --modes "${{ github.event.inputs.modes || 'lexical_baseline,embedding_baseline,weighted_graph,activation_spreading_v1' }}" | |
| --output "benchmarks/external/hotpotqa/data/hotpotqa-nightly-report.json" | |
| --summary-output "benchmarks/external/hotpotqa/data/hotpotqa-nightly-summary.json" | |
| - name: Upload HotpotQA report artifact | |
| uses: actions/upload-artifact@v7 | |
| with: | |
| name: hotpotqa-nightly-report | |
| path: benchmarks/external/hotpotqa/data/hotpotqa-nightly-report.json | |
| if-no-files-found: error | |
| - name: Upload HotpotQA summary artifact | |
| uses: actions/upload-artifact@v7 | |
| with: | |
| name: hotpotqa-nightly-summary | |
| path: benchmarks/external/hotpotqa/data/hotpotqa-nightly-summary.json | |
| if-no-files-found: error |