|
|
|
|
|
|
|
|
import os |
|
|
import requests |
|
|
import pandas as pd |
|
|
import gradio as gr |
|
|
from agent import answer_question |
|
|
import asyncio |
|
|
|
|
|
DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space" |
|
|
|
|
|
class GAIALlamaAgent: |
|
|
def __init__(self): |
|
|
print("✅ LangChain/LlamaIndex Agent initialized.") |
|
|
|
|
|
def __call__(self, question: str) -> str: |
|
|
print(f"📨 Agent received: {question[:50]}...") |
|
|
try: |
|
|
return asyncio.run(answer_question(question)) |
|
|
except Exception as e: |
|
|
return f"[ERROR] {str(e)}" |
|
|
|
|
|
def run_and_submit_all(profile: gr.OAuthProfile | None): |
|
|
space_id = os.getenv("SPACE_ID") |
|
|
username = profile.username if profile else None |
|
|
if not username: |
|
|
return "Please log in to Hugging Face.", None |
|
|
|
|
|
print(f"👤 User: {username}") |
|
|
api_url = DEFAULT_API_URL |
|
|
questions_url = f"{api_url}/questions" |
|
|
submit_url = f"{api_url}/submit" |
|
|
agent_code = f"https://huggingface.co/spaces/{space_id}/tree/main" if space_id else "" |
|
|
|
|
|
try: |
|
|
response = requests.get(questions_url, timeout=15) |
|
|
response.raise_for_status() |
|
|
questions_data = response.json() |
|
|
print(f"📥 Fetched {len(questions_data)} questions") |
|
|
except Exception as e: |
|
|
return f"❌ Error fetching questions: {e}", None |
|
|
|
|
|
agent = GAIALlamaAgent() |
|
|
answers_payload = [] |
|
|
results_log = [] |
|
|
|
|
|
for item in questions_data: |
|
|
qid = item.get("task_id") |
|
|
question = item.get("question") |
|
|
if not qid or not question: |
|
|
continue |
|
|
try: |
|
|
answer = agent(question) |
|
|
except Exception as e: |
|
|
answer = f"[AGENT ERROR] {e}" |
|
|
answers_payload.append({"task_id": qid, "submitted_answer": answer}) |
|
|
results_log.append({"Task ID": qid, "Question": question, "Submitted Answer": answer}) |
|
|
|
|
|
if not answers_payload: |
|
|
return "No answers to submit.", pd.DataFrame(results_log) |
|
|
|
|
|
submission_data = { |
|
|
"username": username, |
|
|
"agent_code": agent_code, |
|
|
"answers": answers_payload |
|
|
} |
|
|
|
|
|
try: |
|
|
response = requests.post(submit_url, json=submission_data, timeout=60) |
|
|
response.raise_for_status() |
|
|
result_data = response.json() |
|
|
status = ( |
|
|
f"✅ Submission Successful!\n" |
|
|
f"User: {result_data.get('username')}\n" |
|
|
f"Score: {result_data.get('score')}%\n" |
|
|
f"Correct: {result_data.get('correct_count')}/{result_data.get('total_attempted')}\n" |
|
|
f"Message: {result_data.get('message')}" |
|
|
) |
|
|
return status, pd.DataFrame(results_log) |
|
|
except Exception as e: |
|
|
return f"❌ Submission failed: {e}", pd.DataFrame(results_log) |
|
|
|
|
|
|
|
|
with gr.Blocks() as demo: |
|
|
gr.Markdown(""" |
|
|
# 🧠 GAIA Agent Evaluation |
|
|
|
|
|
This app runs a LlamaIndex + LangChain powered agent through the GAIA benchmark. |
|
|
|
|
|
1. Login to Hugging Face below |
|
|
2. Click **Run Evaluation** to test all questions |
|
|
3. Answers will be submitted and scored |
|
|
""") |
|
|
gr.LoginButton() |
|
|
run_button = gr.Button("Run Evaluation & Submit All Answers") |
|
|
status_output = gr.Textbox(label="Status", lines=5, interactive=False) |
|
|
results_table = gr.DataFrame(label="Questions and Agent Answers", wrap=True) |
|
|
|
|
|
run_button.click(fn=run_and_submit_all, outputs=[status_output, results_table]) |
|
|
|
|
|
if __name__ == "__main__": |
|
|
print("\n🔁 App Starting Up...") |
|
|
if os.getenv("SPACE_ID"): |
|
|
print(f"🔗 Space: https://huggingface.co/spaces/{os.getenv('SPACE_ID')}") |
|
|
demo.launch(debug=True) |
