Spaces:
Sleeping
Sleeping
| import os | |
| import gradio as gr | |
| import requests | |
| import pandas as pd | |
| import logging | |
| # Configure logging for more detailed output | |
| logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s') | |
| logger = logging.getLogger(__name__) | |
| # --- LangChain Imports --- | |
| from langchain_community.tools import DuckDuckGoSearchRun | |
| from langchain.agents import AgentExecutor, create_react_agent | |
| from langchain_huggingface import ChatHuggingFace, HuggingFaceEndpoint | |
| from langchain_core.prompts import PromptTemplate | |
| from langchain_core.tools import Tool | |
| from langchain_core.messages import HumanMessage | |
| # --- Constants --- | |
| DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space" | |
| # --- Basic Agent Definition --- | |
| class BasicAgent: | |
| def __init__(self): | |
| logger.info("Initializing LangChain-based Agent...") | |
| hf_token = os.getenv("HUGGINGFACEHUB_API_TOKEN") | |
| if not hf_token: | |
| logger.error("HUGGINGFACEHUB_API_TOKEN not found in environment variables.") | |
| logger.error("Please add it as a secret in your Hugging Face Space settings.") | |
| self.llm = None | |
| else: | |
| try: | |
| # Using Mistral-7B-Instruct-v0.2 with text-generation task, | |
| # which is a good general-purpose model and task type. | |
| logger.info(f"Attempting to initialize ChatHuggingFace with repo_id: mistralai/Mistral-7B-Instruct-v0.2. Token starts with: {hf_token[:5]}*****") | |
| self.llm = ChatHuggingFace( | |
| llm=HuggingFaceEndpoint( | |
| repo_id="mistralai/Mistral-7B-Instruct-v0.2", | |
| temperature=0.1, | |
| huggingfacehub_api_token=hf_token, | |
| task="text-generation", # Mistral-Instruct often works well with this task | |
| max_new_tokens=512, | |
| ) | |
| ) | |
| logger.info(f"ChatHuggingFace LLM instance created successfully for repo_id: mistralai/Mistral-7B-Instruct-v0.2") | |
| # Perform a small test of the LLM instance to ensure it's functional | |
| logger.info("Performing a small test inference with the initialized ChatHuggingFace LLM.") | |
| try: | |
| test_message = HumanMessage(content="What is the capital of France?") | |
| response_from_llm_test = self.llm.invoke([test_message]) | |
| if response_from_llm_test and response_from_llm_test.content: | |
| logger.info(f"LLM test inference successful. Response snippet: '{response_from_llm_test.content[:50]}...'") | |
| else: | |
| logger.error("LLM test inference returned an empty or invalid response.") | |
| self.llm = None | |
| raise ValueError("LLM test inference failed to return content.") | |
| except Exception as test_e: | |
| logger.error(f"LLM POST-INITIALIZATION INFERENCE TEST FAILED: {test_e}") | |
| self.llm = None | |
| raise | |
| logger.info("ChatHuggingFace LLM fully initialized and tested.") | |
| except Exception as e: | |
| logger.exception(f"CRITICAL ERROR during LLM (ChatHuggingFace/HuggingFaceEndpoint) initialization: {e}") | |
| self.llm = None | |
| # --- Define Tools --- | |
| self.tools = [ | |
| DuckDuckGoSearchRun( | |
| name="web_search", | |
| description="A useful tool for searching the internet for information, especially for current events or factual queries. Use this when you need to find specific data or verify facts." | |
| ) | |
| ] | |
| # --- Create Agent Prompt --- | |
| self.agent_prompt = PromptTemplate.from_template( | |
| """Answer the following question as concisely and directly as possible. | |
| If the question requires factual information, use your tools to find it. | |
| If the question requires a numerical answer, provide only the number. | |
| If the question requires a specific string, provide only that string. | |
| Do not include any conversational filler, explanations, or extra text unless explicitly asked. | |
| You have access to the following tools: | |
| {tools} | |
| Use the following format: | |
| Question: the input question you must answer | |
| Thought: you should always think about what to do | |
| Action: the action to take, should be one of [{tool_names}] | |
| Action Input: the input to the action | |
| Observation: the result of the action | |
| ... (this Thought/Action/Action Input/Observation can repeat N times) | |
| Thought: I now know the final answer | |
| Final Answer: the final answer to the original input question. Provide only the final answer without any additional text or explanations. | |
| Question: {input} | |
| Thought:{agent_scratchpad} | |
| """ | |
| ) | |
| # --- Initialize the Agent Executor --- | |
| if self.llm: | |
| try: | |
| self.agent = create_react_agent(self.llm, self.tools, self.agent_prompt) | |
| self.agent_executor = AgentExecutor( | |
| agent=self.agent, | |
| tools=self.tools, | |
| verbose=True, | |
| handle_parsing_errors=True, | |
| max_iterations=7, | |
| # early_stopping_method="generate" # <-- REMOVED THIS LINE | |
| ) | |
| logger.info("LangChain Agent Executor initialized successfully.") | |
| except Exception as e: | |
| logger.exception(f"ERROR creating agent or agent executor: {e}") | |
| self.agent_executor = None | |
| else: | |
| self.agent_executor = None | |
| logger.warning("Agent Executor not initialized because LLM failed to initialize.") | |
| def __call__(self, question: str) -> str: | |
| logger.info(f"\n--- Agent received question: {question[:150]}...") | |
| if not self.agent_executor: | |
| logger.error("Agent not configured. Returning placeholder answer.") | |
| return "AGENT_INIT_ERROR: Agent not configured due to prior LLM/agent setup failure. Check Space logs." | |
| try: | |
| response = self.agent_executor.invoke({"input": question}) | |
| final_answer = response.get('output', "No answer generated by agent.") | |
| final_answer = str(final_answer).strip() | |
| logger.info(f"--- Agent returning answer: '{final_answer[:150]}'") | |
| return final_answer | |
| except Exception as e: | |
| logger.exception(f"ERROR during agent execution for question: {question[:50]}... Error: {e}") | |
| return f"AGENT_EXECUTION_ERROR: {e}" | |
| def run_and_submit_all(profile: gr.OAuthProfile | None): | |
| space_id = os.getenv("SPACE_ID") | |
| if profile: | |
| username= f"{profile.username}" | |
| logger.info(f"User logged in: {username}") | |
| else: | |
| logger.warning("User not logged in.") | |
| return "Please Login to Hugging Face with the button.", None | |
| api_url = DEFAULT_API_URL | |
| questions_url = f"{api_url}/questions" | |
| submit_url = f"{api_url}/submit" | |
| # 1. Instantiate Agent | |
| try: | |
| agent = BasicAgent() | |
| if not agent.llm or not agent.agent_executor: | |
| return "Agent could not be initialized. HUGGINGFACEHUB_API_TOKEN might be missing or invalid, or there's an issue with the LLM/Agent setup. Check Space logs for details.", None | |
| except Exception as e: | |
| logger.exception(f"Critical error during agent instantiation: {e}") | |
| return f"Error initializing agent: {e}", None | |
| agent_code = f"https://huggingface.co/spaces/{space_id}/tree/main" | |
| logger.info(agent_code) | |
| # 2. Fetch Questions | |
| logger.info(f"Fetching questions from: {questions_url}") | |
| try: | |
| response = requests.get(questions_url, timeout=30) | |
| response.raise_for_status() | |
| questions_data = response.json() | |
| if not questions_data: | |
| logger.warning("Fetched questions list is empty.") | |
| return "Fetched questions list is empty or invalid format.", None | |
| logger.info(f"Fetched {len(questions_data)} questions.") | |
| except requests.exceptions.RequestException as e: | |
| logger.error(f"Error fetching questions: {e}") | |
| return f"Error fetching questions: {e}", None | |
| except requests.exceptions.JSONDecodeError as e: | |
| logger.error(f"Error decoding JSON response from questions endpoint: {e}") | |
| logger.error(f"Response text: {response.text[:500]}") | |
| return f"Error decoding server response for questions: {e}", None | |
| except Exception as e: | |
| logger.exception(f"An unexpected error occurred fetching questions: {e}") | |
| return f"An unexpected error occurred fetching questions: {e}", None | |
| # 3. Run your Agent | |
| results_log = [] | |
| answers_payload = [] | |
| logger.info(f"Running agent on {len(questions_data)} questions...") | |
| for i, item in enumerate(questions_data): | |
| task_id = item.get("task_id") | |
| question_text = item.get("question") | |
| logger.info(f"\nProcessing question {i+1}/{len(questions_data)}: Task ID {task_id}") | |
| if not task_id or question_text is None: | |
| logger.warning(f"Skipping item with missing task_id or question: {item}") | |
| continue | |
| 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: | |
| logger.exception(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}"}) | |
| if not answers_payload: | |
| logger.warning("Agent did not produce any answers to submit.") | |
| return "Agent did not produce any answers to submit.", pd.DataFrame(results_log) | |
| # 4. Prepare Submission | |
| submission_data = {"username": username.strip(), "agent_code": agent_code, "answers": answers_payload} | |
| status_update = f"Agent finished. Submitting {len(answers_payload)} answers for user '{username}'..." | |
| logger.info(status_update) | |
| # 5. Submit | |
| logger.info(f"Submitting {len(answers_payload)} answers to: {submit_url}") | |
| try: | |
| response = requests.post(submit_url, json=submission_data, timeout=120) | |
| response.raise_for_status() | |
| result_data = response.json() | |
| final_status = ( | |
| f"Submission Successful!\n" | |
| f"User: {result_data.get('username')}\n" | |
| 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.')}" | |
| ) | |
| logger.info("Submission successful.") | |
| results_df = pd.DataFrame(results_log) | |
| return final_status, results_df | |
| except requests.exceptions.HTTPError as e: | |
| error_detail = f"Server responded with status {e.response.status_code}." | |
| try: | |
| error_json = e.response.json() | |
| error_detail += f" Detail: {error_json.get('detail', e.response.text)}" | |
| except requests.exceptions.JSONDecodeError: | |
| error_detail += f" Response: {e.response.text[:500]}" | |
| status_message = f"Submission Failed: {error_detail}" | |
| logger.error(status_message) | |
| results_df = pd.DataFrame(results_log) | |
| return status_message, results_df | |
| except requests.exceptions.Timeout: | |
| status_message = "Submission Failed: The request timed out." | |
| logger.error(status_message) | |
| results_df = pd.DataFrame(results_log) | |
| return status_message, results_df | |
| except requests.exceptions.RequestException as e: | |
| status_message = f"Submission Failed: Network error - {e}" | |
| logger.error(status_message) | |
| results_df = pd.DataFrame(results_log) | |
| return status_message, results_df | |
| except Exception as e: | |
| status_message = f"An unexpected error occurred during submission: {e}" | |
| logger.exception(status_message) | |
| results_df = pd.DataFrame(results_log) | |
| return status_message, results_df | |
| # --- Build Gradio Interface using Blocks --- | |
| with gr.Blocks() as demo: | |
| gr.Markdown("# Advanced Agent Evaluation Runner for GAIA Benchmark") | |
| gr.Markdown( | |
| """ | |
| **Instructions:** | |
| 1. **Clone this space** to your account. | |
| 2. **Add `HUGGINGFACEHUB_API_TOKEN` as a Space Secret** under the "Settings" tab of your duplicated Space. This is crucial for the agent to use the Hugging Face Inference API. | |
| 3. Modify the `BasicAgent` class in `app.py` to refine its logic, add more tools (like a calculator or file reader), and optimize its prompting to improve GAIA scores. | |
| 4. Log in to your Hugging Face account using the button below. This uses your HF username for submission. | |
| 5. Click 'Run Evaluation & Submit All Answers' to fetch questions, run your agent, submit answers, and see your score. | |
| --- | |
| **Disclaimers:** | |
| The submission process can take time as the agent processes multiple questions. This setup provides a foundational framework; continuous improvement in agent logic and tool integration is key to higher GAIA scores. Consider strategies like answer caching or asynchronous processing for larger benchmarks. | |
| """ | |
| ) | |
| gr.LoginButton() | |
| 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) | |
| run_button.click( | |
| fn=run_and_submit_all, | |
| outputs=[status_output, results_table] | |
| ) | |
| if __name__ == "__main__": | |
| logger.info("\n" + "-"*30 + " App Starting " + "-"*30) | |
| space_host_startup = os.getenv("SPACE_HOST") | |
| space_id_startup = os.getenv("SPACE_ID") | |
| if space_host_startup: | |
| logger.info(f"✅ SPACE_HOST found: {space_host_startup}") | |
| logger.info(f" Runtime URL should be: https://{space_host_startup}.hf.space") | |
| else: | |
| logger.info("ℹ️ SPACE_HOST environment variable not found (running locally?).") | |
| if space_id_startup: | |
| logger.info(f"✅ SPACE_ID found: {space_id_startup}") | |
| logger.info(f" Repo URL: https://huggingface.co/spaces/{space_id_startup}") | |
| logger.info(f" Repo Tree URL: https://huggingface.co/spaces/{space_id_startup}/tree/main") | |
| else: | |
| logger.info("ℹ️ SPACE_ID environment variable not found (running locally?). Repo URL cannot be determined.") | |
| logger.info("-"*(60 + len(" App Starting ")) + "\n") | |
| logger.info("Launching Gradio Interface for Basic Agent Evaluation...") | |
| demo.launch(debug=True, share=False) |