Deploy GAIA agent

app.py

@@ -6,22 +6,19 @@ import requests
 import pandas as pd
 
 from smolagents import CodeAgent, tool
-from litellm import LiteLLMModel
+from smolagents.models import LiteLLMModel  # ✅ correct import
 
 # --- Constants ---
 DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
 
-# --- Tool: simple DuckDuckGo search via HTML scraping ---
+# --- Simple Web Search Tool ---
 @tool
 def simple_search(query: str) -> str:
     """
-    Perform a basic DuckDuckGo search.
-
     Args:
-        query (str): Search query.
-
+        query (str): Search query text.
     Returns:
-        str: Top 3 titles and URLs from search.
+        str: Top 3 DuckDuckGo search result titles & links.
     """
     try:
         resp = requests.get(
@@ -33,15 +30,14 @@ def simple_search(query: str) -> str:
         from bs4 import BeautifulSoup
         soup = BeautifulSoup(resp.text, "html.parser")
         items = soup.select("a.result__a")[:3]
-        results = [f"{a.get_text()}\n{a['href']}" for a in items]
-        return "\n\n".join(results) if results else "No results found."
+        return "\n\n".join(f"{a.get_text()}\n{a['href']}" for a in items) or "No results found."
     except Exception as e:
         return f"Search error: {e}"
 
-# --- Enhanced Agent using a public model ---
+# --- Enhanced Agent using Light Model ---
 class BasicAgent:
     def __init__(self):
-        print("BasicAgent initialized with Falcon 7B Instruct model.")
+        print("BasicAgent initialized with LiteLLMModel (falcon-7b-instruct).")
         self.model = LiteLLMModel(
             model_id="tiiuae/falcon-7b-instruct",
             max_tokens=512,
@@ -53,17 +49,17 @@ class BasicAgent:
         )
 
     def __call__(self, question: str) -> str:
-        print(f"Received question: {question[:50]}...")
+        print(f"Question: {question[:60]}...")
         try:
             return self.agent.run(question)
         except Exception as e:
             return f"Agent error: {e}"
 
 def run_and_submit_all(profile: gr.OAuthProfile | None):
-    space_id = os.getenv("SPACE_ID")
     if not profile:
-        return "Please log in to Hugging Face to submit.", None
+        return "Please log in to Hugging Face to submit answers.", None
     username = profile.username
+    space_id = os.getenv("SPACE_ID", "")
 
     questions_url = f"{DEFAULT_API_URL}/questions"
     submit_url = f"{DEFAULT_API_URL}/submit"
@@ -71,35 +67,35 @@ def run_and_submit_all(profile: gr.OAuthProfile | None):
     try:
         agent = BasicAgent()
     except Exception as e:
-        return f"Agent init failed: {e}", None
+        return f"Agent initialization failed: {e}", None
 
     agent_code = f"https://huggingface.co/spaces/{space_id}/tree/main"
 
     try:
-        resp = requests.get(questions_url, timeout=15)
-        resp.raise_for_status()
-        qs = resp.json()
+        r = requests.get(questions_url, timeout=15)
+        r.raise_for_status()
+        questions = r.json()
     except Exception as e:
-        return f"Failed to fetch questions: {e}", None
+        return f"Error fetching questions: {e}", None
 
-    logs, payload = [], []
-    for item in qs:
-        tid = item.get("task_id")
-        q = item.get("question")
-        if not tid or q is None:
+    logs, answers = [], []
+    for item in questions:
+        task_id = item.get("task_id")
+        question = item.get("question")
+        if not task_id or question is None:
             continue
-        ans = agent(q)
-        payload.append({"task_id": tid, "submitted_answer": ans})
-        logs.append({"Task ID": tid, "Question": q, "Submitted Answer": ans})
+        ans = agent(question)
+        answers.append({"task_id": task_id, "submitted_answer": ans})
+        logs.append({"Task ID": task_id, "Question": question, "Submitted Answer": ans})
 
-    if not payload:
-        return "No answers generated.", pd.DataFrame(logs)
+    if not answers:
+        return "Agent produced no answers.", pd.DataFrame(logs)
 
-    submit = {"username": username, "agent_code": agent_code, "answers": payload}
+    payload = {"username": username, "agent_code": agent_code, "answers": answers}
     try:
-        r2 = requests.post(submit_url, json=submit, timeout=60)
-        r2.raise_for_status()
-        data = r2.json()
+        resp = requests.post(submit_url, json=payload, timeout=60)
+        resp.raise_for_status()
+        data = resp.json()
         status = (
             f"✅ Submission Successful!\n"
             f"Score: {data.get('score','N/A')}% "
@@ -108,17 +104,17 @@ def run_and_submit_all(profile: gr.OAuthProfile | None):
         )
         return status, pd.DataFrame(logs)
     except Exception as e:
-        return f"Submit failed: {e}", pd.DataFrame(logs)
+        return f"Submission failed: {e}", pd.DataFrame(logs)
 
-# --- Gradio UI ---
+# --- Gradio Interface ---
 with gr.Blocks() as demo:
     gr.Markdown("# GAIA Agent Evaluation Runner")
     gr.LoginButton()
-    run_btn = gr.Button("Run Evaluation & Submit All Answers")
-    status = gr.Textbox(label="Status / Submission Result", lines=5, interactive=False)
-    results = gr.DataFrame(label="Questions & Submitted Answers", wrap=True)
+    run_button = gr.Button("Run Evaluation & Submit All Answers")
+    status_box = gr.Textbox(label="Status / Submission Result", lines=5, interactive=False)
+    result_table = gr.DataFrame(label="Questions & Agent Answers", wrap=True)
 
-    run_btn.click(run_and_submit_all, outputs=[status, results])
+    run_button.click(run_and_submit_all, outputs=[status_box, result_table])
 
 if __name__ == "__main__":
     print("Launching Gradio app...")