Use Dockerfile with context #70

	name: validate-model

	on:
	workflow_dispatch:
	pull_request:
	branches: main


	jobs:
	cml:
	runs-on: ubuntu-latest

	steps:
	- uses: actions/checkout@v4

	- uses: iterative/setup-cml@v2

	- uses: astral-sh/setup-uv@v1

	- uses: actions/setup-python@v4
	with:
	python-version-file: ".python-version"

	- name: Run model validation
	env:
	repo_token: ${{ secrets.PERSONAL_ACCESS_TOKEN }}
	run: \|
	# Configure PAT to git clone the private repo: pg2-dataset
	git config --global credential.helper store
	echo "https://${{ secrets.USERNAME }}:${{ secrets.PERSONAL_ACCESS_TOKEN }}@github.com" > ~/.git-credentials
	uv sync

	# Run DVC repro
	echo "https://${{ secrets.USERNAME }}:${{ secrets.PERSONAL_ACCESS_TOKEN }}@github.com" > supervised/git-auth.txt
	echo "https://${{ secrets.USERNAME }}:${{ secrets.PERSONAL_ACCESS_TOKEN }}@github.com" > zero_shot/git-auth.txt
	uv run dvc repro supervised/dvc.yaml
	uv run dvc repro zero_shot/dvc.yaml

	# Create report with metrics
	# Pick one supervised model
	echo "✅ Supervised models have all passed validation." >> report.md
	echo '```csv' >> report.md
	cat ./supervised/metric/local/charge_ladder_pls.csv >> report.md
	echo '```' >> report.md

	# Pick one zero-shot model
	echo "✅ Zero-shot models have all passed validation." >> report.md
	echo '```csv' >> report.md
	cat ./zero_shot/metric/local/ranganathan_esm.csv >> report.md
	echo '```' >> report.md

	# Send comment to PR
	echo "Send comment to PR"
	cml comment create report.md

Provide feedback