Spaces:
Sleeping
Sleeping
| import os | |
| import gradio as gr | |
| import requests | |
| import pandas as pd | |
| from smolagents import CodeAgent, DuckDuckGoSearchTool, LiteLLMModel, VisitWebpageTool | |
| # --- Constants --- | |
| DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space" | |
| # --- Agent Definition --- | |
| class AgentArchitect: | |
| def __init__(self): | |
| # SECURE: Fetches the OpenAI API key from your Space Secrets | |
| openai_api_key = os.getenv("OPENAI_API_KEY") | |
| if not openai_api_key: | |
| print("CRITICAL: OPENAI_API_KEY is missing. Please add it to your Space Secrets!") | |
| # Bypassing Hugging Face billing completely. | |
| # We use gpt-4o-mini because it is highly capable at coding and very cost-effective. | |
| self.model = LiteLLMModel( | |
| model_id="gpt-4o-mini", | |
| api_key=openai_api_key | |
| ) | |
| self.tools = [DuckDuckGoSearchTool(), VisitWebpageTool()] | |
| self.agent = CodeAgent( | |
| tools=self.tools, | |
| model=self.model, | |
| add_base_tools=True | |
| ) | |
| def __call__(self, question: str) -> str: | |
| try: | |
| # Enforce Exact Match scoring formatting | |
| prompt = ( | |
| f"{question}\n\n" | |
| f"Instructions: Think step-by-step. Solve the problem using your tools. " | |
| f"Provide ONLY the final, concise answer." | |
| ) | |
| result = self.agent.run(prompt) | |
| return str(result) | |
| except Exception as e: | |
| return f"Error: {e}" | |
| def run_and_submit_all(profile: gr.OAuthProfile | None): | |
| space_id = os.getenv("SPACE_ID") | |
| if not profile: | |
| return "Please Login to Hugging Face with the button.", None | |
| username = profile.username | |
| api_url = DEFAULT_API_URL | |
| questions_url = f"{api_url}/questions" | |
| submit_url = f"{api_url}/submit" | |
| try: | |
| agent_instance = AgentArchitect() | |
| response = requests.get(questions_url, timeout=15) | |
| response.raise_for_status() | |
| questions_data = response.json() | |
| results_log = [] | |
| answers_payload = [] | |
| for item in questions_data: | |
| task_id = item.get("task_id") | |
| question_text = item.get("question") | |
| # Agent logic with OpenAI's brain | |
| submitted_answer = agent_instance(question_text) | |
| answers_payload.append({"task_id": task_id, "submitted_answer": submitted_answer}) | |
| results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": submitted_answer}) | |
| agent_code_link = f"https://huggingface.co/spaces/{space_id}/tree/main" | |
| submission_data = { | |
| "username": username.strip(), | |
| "agent_code": agent_code_link, | |
| "answers": answers_payload | |
| } | |
| submit_response = requests.post(submit_url, json=submission_data, timeout=60) | |
| submit_response.raise_for_status() | |
| result_data = submit_response.json() | |
| final_status = ( | |
| f"Submission Successful!\n" | |
| f"User: {result_data.get('username')}\n" | |
| f"Final Score: {result_data.get('score')}% \n" | |
| f"Message: {result_data.get('message')}" | |
| ) | |
| return final_status, pd.DataFrame(results_log) | |
| except Exception as e: | |
| return f"Submission Failed: {e}", None | |
| # --- Gradio UI --- | |
| with gr.Blocks(theme=gr.themes.Soft()) as demo: | |
| gr.Markdown("# 🚀 Professional Agent Evaluator (OpenAI Edition)") | |
| gr.LoginButton() | |
| run_button = gr.Button("Run Evaluation & Submit All Answers", variant="primary") | |
| status_output = gr.Textbox(label="Leaderboard Status", lines=4) | |
| results_table = gr.DataFrame(label="Agent Reasoning Trace", wrap=True) | |
| run_button.click(fn=run_and_submit_all, outputs=[status_output, results_table]) | |
| if __name__ == "__main__": | |
| demo.launch() |