Update app.py

app.py

@@ -1,142 +1,171 @@
 import os
 import gradio as gr
 import requests
+import inspect
 import pandas as pd
 import asyncio
-import json
+from smolagents import ToolCallingAgent, InferenceClientModel, HfApiModel
+from smolagents import DuckDuckGoSearchTool, Tool, CodeAgent
 from huggingface_hub import login
-from smolagents import CodeAgent, InferenceClientModel, DuckDuckGoSearchTool
 
-# --- Constants ---
 DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
-QUESTIONS_URL = f"{DEFAULT_API_URL}/questions"
-SUBMIT_URL = f"{DEFAULT_API_URL}/submit"
 
-# --- Hugging Face Login ---
 login(token=os.environ["HUGGINGFACEHUB_API_TOKEN"])
 
-# --- Define Tools ---
 search_tool = DuckDuckGoSearchTool()
 
-# --- Main Async Function with Progress Logs ---
 async def run_and_submit_all(profile: gr.OAuthProfile | None):
     log_output = ""
+
     try:
         agent = CodeAgent(
             tools=[search_tool],
             model=InferenceClientModel(model="mistralai/Magistral-Small-2506"),
             max_steps=5,
-            verbosity_level=2
+            verbosity_level=2        
         )
     except Exception as e:
-        yield f"❌ Agent Initialization Error: {e}", None, log_output
+        yield f"Error initializing agent: {e}", None, log_output
         return
 
-    space_id = os.getenv("SPACE_ID", "unknown")
+    space_id = os.getenv("SPACE_ID")
     agent_code = f"https://huggingface.co/spaces/{space_id}/tree/main"
 
+    questions_url = f"{DEFAULT_API_URL}/questions"
     try:
-        response = requests.get(QUESTIONS_URL, timeout=15)
+        response = requests.get(questions_url, timeout=15)
         response.raise_for_status()
-        questions = response.json()
-        if not questions:
-            yield "⚠️ No questions received.", None, log_output
+        questions_data = response.json()
+        if not questions_data:
+            yield "Fetched questions list is empty or invalid format.", None, log_output
             return
     except Exception as e:
-        yield f"❌ Failed to fetch questions: {e}", None, log_output
+        yield f"Error fetching questions: {e}", None, log_output
         return
 
-    answers = []
-    logs = []
-    loop = asyncio.get_running_loop()
+    results_log = []
+    answers_payload = []
+    loop = asyncio.get_event_loop()
 
-    for item in questions:
+    for item in questions_data:
         task_id = item.get("task_id")
-        question = item.get("question")
-        if not task_id or not question:
+        question_text = item.get("question")
+        if not task_id or question_text is None:
             continue
 
         log_output += f"🔍 Solving Task ID: {task_id}...\n"
-        yield None, None, log_output  # Live update
-
-        system_prompt = (
-            "You are a general AI assistant. I will ask you a question. "
-            "Report your thoughts, and finish your answer with the following template: "
-            "FINAL ANSWER: [YOUR FINAL ANSWER]. YOUR FINAL ANSWER should be a number OR as few words as possible OR a comma separated list of numbers and/or strings.\n\n"
-        )
-        full_prompt = system_prompt + f"Question: {question.strip()}"
+        yield None, None, log_output
 
         try:
-            result = await loop.run_in_executor(None, lambda: agent(full_prompt))
-
-            if isinstance(result, dict) and "final_answer" in result:
-                final_answer = str(result["final_answer"]).strip()
-            elif isinstance(result, str):
-                if "FINAL ANSWER:" in result:
-                    final_answer = result.split("FINAL ANSWER:")[-1].strip()
+            system_prompt = (
+                "You are a general AI assistant. I will ask you a question. "
+                "Report your thoughts, and finish your answer with the following template: "
+                "FINAL ANSWER: [YOUR FINAL ANSWER]. YOUR FINAL ANSWER should be a number OR as few words as possible OR a comma separated list of numbers and/or strings. "
+                "If you are asked for a number, don't use comma to write your number neither use units such as $ or percent sign unless specified otherwise. "
+                "If you are asked for a string, don't use articles, neither abbreviations (e.g. for cities), and write the digits in plain text unless specified otherwise. "
+                "If you are asked for a comma separated list, apply the above rules depending of whether the element to be put in the list is a number or a string.\n\n"
+            )
+            full_prompt = system_prompt + f"Question: {question_text.strip()}"
+
+            agent_result = await loop.run_in_executor(None, agent, full_prompt)
+
+            # Extract final answer cleanly
+            if isinstance(agent_result, dict) and "final_answer" in agent_result:
+                final_answer = str(agent_result["final_answer"]).strip()
+            elif isinstance(agent_result, str):
+                response_text = agent_result.strip()
+
+                # Remove known boilerplate
+                if "Here is the final answer from your managed agent" in response_text:
+                    response_text = response_text.split(":", 1)[-1].strip()
+
+                if "FINAL ANSWER:" in response_text:
+                    _, final_answer = response_text.rsplit("FINAL ANSWER:", 1)
+                    final_answer = final_answer.strip()
                 else:
-                    final_answer = result.strip()
+                    final_answer = response_text
             else:
-                final_answer = str(result).strip()
+                final_answer = str(agent_result).strip()
 
-        except Exception as e:
-            final_answer = f"AGENT ERROR: {e}"
-            print(f"[ERROR] Task {task_id} failed: {e}")
+            answers_payload.append({
+                "task_id": task_id,
+                "model_answer": final_answer
+            })
 
-        answers.append({"task_id": task_id, "model_answer": final_answer})
-        logs.append({"Task ID": task_id, "Question": question, "Submitted Answer": final_answer})
-        log_output += f"✅ Done: {task_id} — Answer: {final_answer[:60]}\n"
-        yield None, None, log_output  # Live update
+            results_log.append({
+                "Task ID": task_id,
+                "Question": question_text,
+                "Submitted Answer": final_answer
+            })
 
-    valid_answers = [a for a in answers if isinstance(a["task_id"], str) and isinstance(a["model_answer"], str)]
+            log_output += f"✅ Done: {task_id} — Answer: {final_answer[:60]}\n"
+            yield None, None, log_output
 
-    if not valid_answers:
-        yield "❌ Agent produced no valid answers.", pd.DataFrame(logs), log_output
+        except Exception as e:
+            print(f"Error running agent on task {task_id}: {e}")
+            results_log.append({
+                "Task ID": task_id,
+                "Question": question_text,
+                "Submitted Answer": f"AGENT ERROR: {e}"
+            })
+            log_output += f"⛔️ Error: {task_id} — {e}\n"
+            yield None, None, log_output
+
+    if not answers_payload:
+        yield "Agent did not produce any answers to submit.", pd.DataFrame(results_log), log_output
         return
 
-    submission = {
-        "username": profile.username if profile else "unknown",
-        "agent_code": agent_code,
-        "answers": valid_answers
-    }
-
-    print("[DEBUG] Submitting:\n", json.dumps(submission, indent=2))
-
+    username = profile.username if profile else "unknown"
+    submit_url = f"{DEFAULT_API_URL}/submit"
+    submission_data = {"username": username.strip(), "agent_code": agent_code, "answers": answers_payload}
     try:
-        resp = requests.post(SUBMIT_URL, json=submission, timeout=60)
-        resp.raise_for_status()
-        result_data = resp.json()
-
-        summary = (
-            f"✅ Submission Successful\n"
+        response = requests.post(submit_url, json=submission_data, timeout=60)
+        response.raise_for_status()
+        result_data = response.json()
+        final_status = (
+            f"Submission Successful!\n"
             f"User: {result_data.get('username')}\n"
-            f"Score: {result_data.get('score', 'N/A')}% "
-            f"({result_data.get('correct_count')}/{result_data.get('total_attempted')})\n"
-            f"Message: {result_data.get('message', 'No message.')}"
+            f"Overall Score: {result_data.get('score', 'N/A')}% "
+            f"({result_data.get('correct_count', '?')}/{result_data.get('total_attempted', '?')} correct)\n"
+            f"Message: {result_data.get('message', 'No message received.')}"
         )
-        yield summary, pd.DataFrame(logs), log_output
-
+        results_df = pd.DataFrame(results_log)
+        yield final_status, results_df, log_output
     except Exception as e:
-        yield f"❌ Submission failed: {e}", pd.DataFrame(logs), log_output
+        status_message = f"Submission Failed: {e}"
+        results_df = pd.DataFrame(results_log)
+        yield status_message, results_df, log_output
 
-# --- Gradio UI ---
 with gr.Blocks() as demo:
-    gr.Markdown("# 🧠 GAIA Agent Evaluation Interface")
+    gr.Markdown("# Basic Agent Evaluation Runner")
     gr.Markdown("""
-    - Log in with your Hugging Face account.
-    - Click the button below to run the agent and submit the answers.
-    - Watch the log to see which question is being solved in real-time.
+    **Instructions:**
+    1. Clone this space and define your agent logic.
+    2. Log in to your Hugging Face account.
+    3. Click 'Run Evaluation & Submit All Answers'.
+    ---
+    **Note:**
+    The run may take time. Async is now used to improve responsiveness.
     """)
 
     gr.LoginButton()
-    run_button = gr.Button("🚀 Run Evaluation & Submit")
-    status = gr.Textbox(label="Final Status", lines=6)
-    table = gr.DataFrame(label="Answer Log")
-    progress_log = gr.Textbox(label="Live Progress Log", lines=10, interactive=False)
 
-    run_button.click(fn=run_and_submit_all, outputs=[status, table, progress_log])
+    run_button = gr.Button("Run Evaluation & Submit All Answers")
+    status_output = gr.Textbox(label="Run Status / Submission Result", lines=5, interactive=False)
+    results_table = gr.DataFrame(label="Questions and Agent Answers", wrap=True)
+    progress_log = gr.Textbox(label="Progress Log", lines=10, interactive=False)
+
+    run_button.click(fn=run_and_submit_all, outputs=[status_output, results_table, progress_log])
 
-# --- Launch ---
 if __name__ == "__main__":
-    print("Launching Agent Space...")
-    demo.launch(debug=True)
+    print("\n" + "-"*30 + " App Starting " + "-"*30)
+    space_host_startup = os.getenv("SPACE_HOST")
+    space_id_startup = os.getenv("SPACE_ID")
+
+    if space_host_startup:
+        print(f"✅ SPACE_HOST: https://{space_host_startup}.hf.space")
+    if space_id_startup:
+        print(f"✅ SPACE_ID: https://huggingface.co/spaces/{space_id_startup}")
+
+    print("Launching Gradio Interface...")
+    demo.launch(debug=True, share=False)