Agents Course Unit 4 — GAIA Evaluation Runner

Instructions

1. Duplicate this Space (or push these files to a new Space) and keep it public.
2. In Space Settings → Secrets, set at least HF_TOKEN (and optionally OPENAI_API_KEY / GROQ_API_KEY / XAI_API_KEY).
3. Log in with the Hugging Face button below (needed for submission username).
4. Click Run Evaluation & Submit All Answers. Expect several minutes for all 20 questions.
5. Target: ≥ 30% (6/20) exact-match on GAIA Level-1 validation questions.

Answers are submitted as bare strings (no FINAL ANSWER prefix). File-backed tasks (png / mp3 / py / xlsx) are downloaded automatically and passed to the agent as local paths.