| |
| |
| |
|
|
| import subprocess |
| import sys |
|
|
| subprocess.check_call([ |
| sys.executable, "-m", "pip", "install", "-q", |
| "smolagents[toolkit]", |
| "gradio", |
| "requests", |
| "pandas" |
| ]) |
|
|
|
|
| |
| |
| |
|
|
| import os |
| import gradio as gr |
| import requests |
| import pandas as pd |
|
|
| from smolagents import ( |
| CodeAgent, |
| InferenceClientModel, |
| DuckDuckGoSearchTool, |
| PythonInterpreterTool |
| ) |
|
|
|
|
| |
| |
| |
|
|
| DEFAULT_API_URL = "https://huggingface.co/Shrutipanchal086/structrural_AI_agent" |
|
|
|
|
| |
| |
| |
|
|
| class BasicAgent: |
|
|
| def __init__(self): |
|
|
| print("Initializing StructuralGPT GAIA Agent...") |
|
|
| |
| |
| |
| |
|
|
| hf_token = os.getenv("hf_token") |
|
|
| if not hf_token: |
| raise ValueError( |
| "HF_TOKEN not found. " |
| "Please add HF_TOKEN in Space Settings -> Secrets." |
| ) |
|
|
| |
| |
| |
|
|
| self.model = InferenceClientModel( |
| model_id="Qwen/Qwen3-Next-80B-A3B-Thinking", |
| token=hf_token, |
| max_tokens=3000 |
| ) |
|
|
| |
| |
| |
|
|
| self.search_tool = DuckDuckGoSearchTool( |
| max_results=8 |
| ) |
|
|
| |
| |
| |
|
|
| self.python_tool = PythonInterpreterTool( |
| authorized_imports=[ |
| "math", |
| "statistics", |
| "datetime", |
| "json", |
| "re" |
| ] |
| ) |
|
|
| |
| |
| |
|
|
| self.agent = CodeAgent( |
| model=self.model, |
| tools=[ |
| self.search_tool, |
| self.python_tool |
| ], |
| max_steps=10 |
| ) |
|
|
| print("StructuralGPT GAIA Agent initialized successfully.") |
|
|
|
|
| def __call__(self, question: str) -> str: |
|
|
| print("\n" + "=" * 60) |
| print("QUESTION:") |
| print(question) |
| print("=" * 60) |
|
|
| prompt = f""" |
| You are a powerful general-purpose AI agent participating |
| in the GAIA benchmark. |
| |
| Your job is to solve the user's question accurately. |
| |
| IMPORTANT RULES: |
| |
| 1. Understand the question completely before answering. |
| |
| 2. If the question requires current, factual, or external |
| information, use the web search tool. |
| |
| 3. If calculations are required, use the Python tool. |
| Do not rely on mental arithmetic for complicated calculations. |
| |
| 4. Break difficult problems into smaller steps. |
| |
| 5. Verify important calculations and facts before producing |
| the final answer. |
| |
| 6. If multiple pieces of information are required, collect |
| all necessary information before answering. |
| |
| 7. Do not invent facts, sources, numbers, or results. |
| |
| 8. Give ONLY the final answer required by the question. |
| Do not unnecessarily explain your internal reasoning. |
| |
| 9. Pay very close attention to: |
| - units |
| - dates |
| - names |
| - numerical values |
| - percentages |
| - requested formats |
| |
| 10. If the question asks for a specific format, follow that |
| format exactly. |
| |
| You are also knowledgeable in civil and structural engineering, |
| including RCC design, steel design, structural analysis, |
| foundation engineering, transportation engineering, |
| water resources engineering, and construction management. |
| |
| For engineering questions, use Indian Standards when relevant, |
| including IS 456, IS 875, IS 1893, IS 800 and IS 13920. |
| |
| USER QUESTION: |
| {question} |
| """ |
|
|
| try: |
|
|
| result = self.agent.run(prompt) |
|
|
| answer = str(result).strip() |
|
|
| print("\nFINAL ANSWER:") |
| print(answer) |
|
|
| return answer |
|
|
| except Exception as e: |
|
|
| print("Agent error:", e) |
|
|
| return f"Unable to solve the question because of an agent error: {e}" |
|
|
|
|
| |
| |
| |
|
|
| def run_and_submit_all(profile: gr.OAuthProfile | None): |
|
|
| """ |
| Fetch all GAIA questions, |
| run the agent, |
| submit answers, |
| and display results. |
| """ |
|
|
| |
| |
| |
|
|
| if profile: |
|
|
| username = profile.username |
|
|
| print(f"User logged in: {username}") |
|
|
| else: |
|
|
| print("User not logged in.") |
|
|
| return ( |
| "Please login to Hugging Face using the Login button.", |
| None |
| ) |
|
|
|
|
| |
| |
| |
|
|
| api_url = DEFAULT_API_URL |
|
|
| questions_url = f"{api_url}/questions" |
|
|
| submit_url = f"{api_url}/submit" |
|
|
|
|
| |
| |
| |
|
|
| space_id = os.getenv("SPACE_ID") |
|
|
| if space_id: |
|
|
| agent_code = ( |
| f"https://huggingface.co/spaces/" |
| f"{space_id}/tree/main" |
| ) |
|
|
| else: |
|
|
| agent_code = "Local/Unknown-Space" |
|
|
|
|
| print("Agent code URL:") |
| print(agent_code) |
|
|
|
|
| |
| |
| |
|
|
| try: |
|
|
| agent = BasicAgent() |
|
|
| except Exception as e: |
|
|
| print("Error creating agent:", e) |
|
|
| return ( |
| f"Error initializing agent: {e}", |
| None |
| ) |
|
|
|
|
| |
| |
| |
|
|
| print("\nFetching GAIA questions...") |
|
|
| try: |
|
|
| response = requests.get( |
| questions_url, |
| timeout=30 |
| ) |
|
|
| response.raise_for_status() |
|
|
| questions_data = response.json() |
|
|
| if not questions_data: |
|
|
| return ( |
| "No questions were received.", |
| None |
| ) |
|
|
| print( |
| f"Fetched {len(questions_data)} questions." |
| ) |
|
|
| except Exception as e: |
|
|
| print("Error fetching questions:", e) |
|
|
| return ( |
| f"Error fetching questions: {e}", |
| None |
| ) |
|
|
|
|
| |
| |
| |
|
|
| results_log = [] |
|
|
| answers_payload = [] |
|
|
| print("\nRunning agent...") |
|
|
| for number, item in enumerate( |
| questions_data, |
| start=1 |
| ): |
|
|
| task_id = item.get("task_id") |
|
|
| question_text = item.get("question") |
|
|
| if not task_id or question_text is None: |
|
|
| print( |
| "Skipping invalid question:", |
| item |
| ) |
|
|
| continue |
|
|
|
|
| print( |
| f"\nProcessing question " |
| f"{number}/{len(questions_data)}" |
| ) |
|
|
|
|
| try: |
|
|
| submitted_answer = agent( |
| 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 |
| } |
| ) |
|
|
| except Exception as e: |
|
|
| print( |
| f"Agent error on task {task_id}: {e}" |
| ) |
|
|
| results_log.append( |
| { |
| "Task ID": task_id, |
| "Question": question_text, |
| "Submitted Answer": |
| f"AGENT ERROR: {e}" |
| } |
| ) |
|
|
|
|
| |
| |
| |
|
|
| if not answers_payload: |
|
|
| return ( |
| "Agent did not produce any answers.", |
| pd.DataFrame(results_log) |
| ) |
|
|
|
|
| |
| |
| |
|
|
| submission_data = { |
|
|
| "username": |
| username.strip(), |
|
|
| "agent_code": |
| agent_code, |
|
|
| "answers": |
| answers_payload |
| } |
|
|
|
|
| status_update = ( |
| f"Agent finished. " |
| f"Submitting {len(answers_payload)} answers..." |
| ) |
|
|
| print(status_update) |
|
|
|
|
| |
| |
| |
|
|
| try: |
|
|
| response = requests.post( |
| submit_url, |
| json=submission_data, |
| timeout=120 |
| ) |
|
|
| response.raise_for_status() |
|
|
| result_data = response.json() |
|
|
|
|
| final_status = ( |
|
|
| "Submission Successful!\n\n" |
|
|
| f"User: " |
| f"{result_data.get('username')}\n" |
|
|
| f"Overall Score: " |
| f"{result_data.get('score', 'N/A')}%\n" |
|
|
| f"Correct: " |
| f"{result_data.get('correct_count', '?')}/" |
| f"{result_data.get('total_attempted', '?')}\n\n" |
|
|
| f"Message: " |
| f"{result_data.get('message', '')}" |
| ) |
|
|
|
|
| results_df = pd.DataFrame( |
| results_log |
| ) |
|
|
| return ( |
| final_status, |
| results_df |
| ) |
|
|
|
|
| except requests.exceptions.HTTPError as e: |
|
|
| error_detail = ( |
| f"Server responded with " |
| f"status {e.response.status_code}." |
| ) |
|
|
| try: |
|
|
| error_json = e.response.json() |
|
|
| error_detail += ( |
| f" Detail: " |
| f"{error_json.get('detail', '')}" |
| ) |
|
|
| except Exception: |
|
|
| error_detail += ( |
| f" Response: " |
| f"{e.response.text[:500]}" |
| ) |
|
|
|
|
| return ( |
| f"Submission Failed: {error_detail}", |
| pd.DataFrame(results_log) |
| ) |
|
|
|
|
| except requests.exceptions.Timeout: |
|
|
| return ( |
| "Submission Failed: Request timed out.", |
| pd.DataFrame(results_log) |
| ) |
|
|
|
|
| except requests.exceptions.RequestException as e: |
|
|
| return ( |
| f"Submission Failed: Network error - {e}", |
| pd.DataFrame(results_log) |
| ) |
|
|
|
|
| except Exception as e: |
|
|
| return ( |
| f"Unexpected submission error: {e}", |
| pd.DataFrame(results_log) |
| ) |
|
|
|
|
| |
| |
| |
|
|
| with gr.Blocks() as demo: |
|
|
| gr.Markdown( |
| "# 🚀 StructuralGPT - GAIA Agent" |
| ) |
|
|
| gr.Markdown( |
| """ |
| ### Instructions |
| |
| 1. Login to Hugging Face. |
| 2. The agent uses Qwen through Hugging Face. |
| 3. The agent can search the web. |
| 4. The agent can perform calculations using Python. |
| 5. Click **Run Evaluation & Submit All Answers**. |
| """ |
| ) |
|
|
| gr.LoginButton() |
|
|
|
|
| run_button = gr.Button( |
| "Run Evaluation & Submit All Answers" |
| ) |
|
|
|
|
| status_output = gr.Textbox( |
| label="Run Status / Submission Result", |
| lines=8, |
| 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" + |
| "-" * 30 + |
| " App Starting " + |
| "-" * 30 |
| ) |
|
|
|
|
| space_host = os.getenv( |
| "SPACE_HOST" |
| ) |
|
|
| space_id = os.getenv( |
| "SPACE_ID" |
| ) |
|
|
|
|
| if space_host: |
|
|
| print( |
| f"SPACE_HOST: {space_host}" |
| ) |
|
|
| if space_id: |
|
|
| print( |
| f"SPACE_ID: {space_id}" |
|
|
| ) |
|
|
| print( |
| "Repository:" |
| ) |
|
|
| print( |
| f"https://huggingface.co/spaces/{space_id}" |
| ) |
|
|
|
|
| print( |
| "\nLaunching StructuralGPT..." |
| ) |
|
|
|
|
| demo.launch( |
| debug=True, |
| share=False |
| ) |
