Update app.py

app.py

@@ -1,115 +1,81 @@
 import os
-import gradio as gr
 import requests
-import pandas as pd
-
-from agent import create_agent, fetch_random_question
-
-# --- Constants ---
-DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
-
-
-def run_and_submit_all(profile):
-    """
-    Fetch all questions, run the SmolAgent on them, submit all answers,
-    and display the results.
+from smolagents import CodeAgent, tool, OpenAIServerModel
+
+# ------------------------
+# Constants
+# ------------------------
+API_URL = "https://agents-course-unit4-scoring.hf.space"
+
+# ------------------------
+# Tool definitions
+# ------------------------
+@tool
+def fetch_questions() -> list:
     """
-    space_id = os.getenv("SPACE_ID")
+    Fetch the full list of GAIA evaluation questions.
 
-    if profile is None:
-        return "Please login to Hugging Face with the button.", None
-    username = profile.username
+    Returns:
+        list: A list of question dicts, each with 'task_id' and 'question'.
+    """
+    resp = requests.get(f"{API_URL}/questions", timeout=15)
+    resp.raise_for_status()
+    return resp.json()
 
-    # Instantiate the agent
-    try:
-        agent = create_agent()
-    except Exception as e:
-        return f"Error initializing agent: {e}", None
 
-    agent_code = f"https://huggingface.co/spaces/{space_id}/tree/main"
+@tool
+def fetch_random_question() -> dict:
+    """
+    Fetch a single random GAIA question.
 
-    # Fetch questions
-    try:
-        response = requests.get(f"{DEFAULT_API_URL}/questions", timeout=15)
-        response.raise_for_status()
-        questions = response.json()
-    except Exception as e:
-        return f"Error fetching questions: {e}", None
+    Returns:
+        dict: A dict with keys 'task_id' and 'question'.
+    """
+    resp = requests.get(f"{API_URL}/random-question", timeout=15)
+    resp.raise_for_status()
+    return resp.json()
 
-    if not questions:
-        return "No questions fetched.", None
 
-    # Run agent on each question
-    results = []
-    payload = []
-    for q in questions:
-        task_id = q.get("task_id")
-        question_text = q.get("question")
-        if not task_id or not question_text:
-            continue
-        try:
-            answer = agent.run(question=question_text)
-        except Exception as e:
-            answer = f"ERROR: {e}"
-        results.append({"Task ID": task_id, "Question": question_text, "Answer": answer})
-        payload.append({"task_id": task_id, "submitted_answer": answer})
+@tool
+def submit_answers(username: str, agent_code: str, answers: list) -> dict:
+    """
+    Submit the agent's answers to GAIA and get the scoring.
 
-    # Submit answers
-    submit_payload = {"username": username, "agent_code": agent_code, "answers": payload}
-    try:
-        resp = requests.post(f"{DEFAULT_API_URL}/submit", json=submit_payload, timeout=60)
-        resp.raise_for_status()
-        data = resp.json()
-        status = (
-            f"Submission Successful!\n"
-            f"User: {data.get('username')}\n"
-            f"Score: {data.get('score')}% ({data.get('correct_count')}/{data.get('total_attempted')})\n"
-            f"Message: {data.get('message')}"
-        )
-    except Exception as e:
-        status = f"Submission Failed: {e}"
+    Args:
+        username (str): The Hugging Face username identifying the submission.
+        agent_code (str): URL to your Space code repository for verification.
+        answers (list): A list of dicts, each with 'task_id' and 'submitted_answer'.
 
-    return status, pd.DataFrame(results)
+    Returns:
+        dict: A dict containing 'score', 'correct_count', 'total_attempted', 'message', etc.
+    """
+    payload = {
+        "username": username,
+        "agent_code": agent_code,
+        "answers": answers
+    }
+    resp = requests.post(f"{API_URL}/submit", json=payload, timeout=60)
+    resp.raise_for_status()
+    return resp.json()
 
 
-def test_random_question(profile):
-    """
-    Fetch a random GAIA question and return its answer by the agent.
+def create_agent() -> CodeAgent:
     """
-    if profile is None:
-        return "Please login to test.", ""
-    try:
-        q = fetch_random_question()
-        agent = create_agent()
-        ans = agent.run(question=q.get("question", ""))
-        return q.get("question", ""), ans
-    except Exception as e:
-        return f"Error: {e}", ""
-
+    Build and return a configured CodeAgent using OpenAI GPT-3.5 Turbo.
+    Requires OPENAI_API_KEY in the environment.
 
-# --- Gradio Interface ---
-with gr.Blocks() as demo:
-    gr.Markdown("# SmolAgent Evaluation Runner")
-    gr.Markdown(
-        """
-        **Instructions:**
-        1. Clone this space and define your agent logic in agent.py.
-        2. Log in with your Hugging Face account.
-        3. Use 'Run Evaluation & Submit All Answers' or 'Test Random Question'.
-        """
+    Returns:
+        CodeAgent: An instance of CodeAgent configured with the GAIA tools.
+    """
+    # Use 'model_id' to match the OpenAIServerModel signature
+    model = OpenAIServerModel(model_id="gpt-3.5-turbo")
+    agent = CodeAgent(
+        tools=[fetch_questions, fetch_random_question, submit_answers],
+        model=model,
+        prompt_template=(
+            "Here is a GAIA question:\n"
+            "{question}\n"
+            "Provide ONLY the exact answer (exact-match), with no extra text."
+        )
     )
-
-    login = gr.LoginButton()
-    run_all_btn = gr.Button("Run Evaluation & Submit All Answers")
-    test_btn = gr.Button("Test Random Question")
-
-    status_output = gr.Textbox(label="Status / Result", lines=5, interactive=False)
-    table = gr.DataFrame(label="Full Results Table", wrap=True)
-    question_box = gr.Textbox(label="Random Question", lines=3, interactive=False)
-    answer_box = gr.Textbox(label="Agent Answer", lines=3, interactive=False)
-
-    run_all_btn.click(fn=run_and_submit_all, inputs=[login], outputs=[status_output, table])
-    test_btn.click(fn=test_random_question, inputs=[login], outputs=[question_box, answer_box])
-
-if __name__ == "__main__":
-    demo.launch(debug=True, share=False)
+    return agent