test_gaia_questions.py

# test_gaia_questions.py

import requests
from langgraph_agents import graph

def test_with_real_gaia_questions():
    # Fetch questions directly from the benchmark API
    url = "https://agents-course-unit4-scoring.hf.space/questions"
    response = requests.get(url)
    questions = response.json()

    for q in questions[:5]:  # Limit to first 5 for testing
        question = q["question"]
        task_id = q["task_id"]
        state = {"question": question, "answer": ""}
        result = graph.invoke(state)
        print(f"[{task_id}] Q: {question}")
        print(f"→ {result['answer']}")
        print("-" * 60)

if __name__ == "__main__":
    test_with_real_gaia_questions()