HF_SPACE_DEPLOY.md

HF Space deploy runbook (Docker SDK)

1) Create Space

• Visibility: Public
• SDK: Docker
• Add tag: openenv

2) Push files

# ...existing code...
git add .
git commit -m "DataQualityEnv OpenEnv submission"
git push

3) Set Space secrets/variables

• API_BASE_URL=https://router.huggingface.co/v1
• MODEL_NAME=meta-llama/Llama-3.1-8B-Instruct
• HF_TOKEN=<your token>
• ENV_URL is no longer needed for the Space UI path.

Notes:

• Keep these Space secrets configured so the deployment stays healthy during the evaluation window.
• The Space app now runs in-process through space_app.py and Gradio.
• Judge runs are separate: they still inject API_BASE_URL, HF_TOKEN, and MODEL_NAME when executing python inference.py.
• Do not rely on hardcoded model credentials or ENV_URL in the Space path.

4) Verify endpoints

curl https://<your-space>.hf.space/health
curl -X POST https://<your-space>.hf.space/reset \
  -H 'content-type: application/json' \
  -d '{"task_id":1,"seed":42}'

5) Validate submission

./validate-submission.sh https://<your-space>.hf.space
python scripts/check_graders.py  # run locally against local server first

6) Final checks

• openenv validate passes
• /health returns {"status":"ok"}
• /reset and /step both return valid JSON
• Inference completes under 20 minutes