🤖 GAIA Agent Evaluation

Test your Hugging Face Agent before submitting the full evaluation.

Recommended workflow

1. Login with Hugging Face.
2. Click Test One Question.
3. Check the generated answer.
4. If everything works, run the Full 20-Question Evaluation.

🧪 Step 1 — Test One Question

This is the fast test.

It downloads one random GAIA question, runs your agent, and shows the answer.

Nothing is submitted to the leaderboard.

Waiting for test...

🚀 Step 2 — Full Evaluation

Only click this after the one-question test works.

This downloads the official evaluation set and submits the agent's answers to the course evaluator.

Waiting for full evaluation...