drAbreu's picture
Improved the interface
cb2e2ec
raw
history blame
19 kB
import os
import gradio as gr
import requests
import inspect
import pandas as pd
import json
import asyncio
from pathlib import Path
from datetime import datetime
from typing import List, Dict, Any, Optional
from tqdm.asyncio import tqdm as async_tqdm
from agents.llama_index_agent import GaiaAgent
# --- Constants ---
DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
CACHE_DIR = "cache"
CACHE_FILE = os.path.join(CACHE_DIR, "agent_cache.json")
MAX_CONCURRENT_REQUESTS = 3 # Limit concurrent API calls
# Model configurations
CLAUDE = {
"model_provider": "anthropic",
"model_name": "claude-3-7-sonnet-latest"
}
OPENAI = {
"model_provider": "openai",
"model_name": "gpt-4o"
}
# --- Optimized Agent Implementation ---
class OptimizedGaiaAgent:
"""
Enhanced GAIA agent with caching and asynchronous processing capabilities.
"""
def __init__(
self,
model_config=CLAUDE,
use_cache=True,
cache_file=CACHE_FILE,
max_concurrent=MAX_CONCURRENT_REQUESTS
):
"""
Initialize the optimized agent.
Args:
model_config: Dictionary with model_provider and model_name
use_cache: Whether to use caching
cache_file: Path to the cache file
max_concurrent: Maximum number of concurrent requests
"""
self.agent = GaiaAgent(**model_config)
self.model_config = model_config
self.use_cache = use_cache
self.cache_file = cache_file
self.cache = self._load_cache() if use_cache else {}
self.semaphore = asyncio.Semaphore(max_concurrent)
print(f"OptimizedGaiaAgent initialized with {model_config['model_provider']} {model_config['model_name']}")
if use_cache:
print(f"Cache loaded with {len(self.cache)} answers")
def _load_cache(self) -> Dict[str, str]:
"""Load cached answers from file"""
# Create cache directory if it doesn't exist
os.makedirs(os.path.dirname(self.cache_file), exist_ok=True)
cache_path = Path(self.cache_file)
if cache_path.exists():
try:
with open(cache_path, 'r') as f:
return json.load(f)
except Exception as e:
print(f"Error loading cache: {e}")
return {}
return {}
def _save_cache(self) -> None:
"""Save cached answers to file"""
try:
with open(self.cache_file, 'w') as f:
json.dump(self.cache, f, indent=2)
except Exception as e:
print(f"Error saving cache: {e}")
def _get_cache_key(self, question: str) -> str:
"""Generate a consistent key for the cache"""
# Strip whitespace and normalize
return question.strip()
async def process_question(self, task_id: str, question: str) -> Dict[str, Any]:
"""
Process a single question, using cache if available.
Args:
task_id: ID of the task/question
question: The question text
Returns:
Dictionary with task_id, question, answer, and metadata
"""
cache_key = self._get_cache_key(question)
# Check cache first
if self.use_cache and cache_key in self.cache:
print(f"🔄 Cache hit for task {task_id[:8]}...")
return {
"task_id": task_id,
"question": question,
"submitted_answer": self.cache[cache_key],
"cached": True,
"error": False
}
# Process the question (with semaphore to limit concurrent requests)
async with self.semaphore:
print(f"⚙️ Processing task {task_id[:8]}...")
try:
response = await self.agent.run(question)
answer = response.response.blocks[-1].text
# Save to cache
if self.use_cache:
self.cache[cache_key] = answer
# Use asyncio.to_thread for file I/O to avoid blocking
await asyncio.to_thread(self._save_cache)
return {
"task_id": task_id,
"question": question,
"submitted_answer": answer,
"cached": False,
"error": False
}
except Exception as e:
error_message = f"ERROR: {str(e)}"
print(f"❌ Error processing task {task_id[:8]}: {error_message}")
return {
"task_id": task_id,
"question": question,
"submitted_answer": error_message,
"cached": False,
"error": True
}
async def process_all(
self,
questions_data: List[Dict[str, Any]],
progress_callback=None
) -> List[Dict[str, Any]]:
"""
Process all questions, with progress reporting.
Args:
questions_data: List of question dictionaries
progress_callback: Function to call with progress updates
Returns:
List of results with answers and metadata
"""
# Filter out invalid questions
valid_questions = [
item for item in questions_data
if item.get("task_id") and item.get("question") is not None
]
if not valid_questions:
print("No valid questions to process.")
return []
total = len(valid_questions)
print(f"Processing {total} questions with {MAX_CONCURRENT_REQUESTS} concurrent tasks...")
# Process questions and collect results
results = []
# Create async tasks
tasks = [
self.process_question(item["task_id"], item["question"])
for item in valid_questions
]
# Process with progress tracking
if progress_callback:
progress_callback(0, desc="Starting processing...")
# Process tasks one by one with progress updates
for i, task in enumerate(asyncio.as_completed(tasks)):
result = await task
results.append(result)
# Update progress
if progress_callback:
progress_callback((i + 1) / total, desc=f"Processed {i + 1}/{total} questions")
# Sort results to match original order
id_to_result = {result["task_id"]: result for result in results}
ordered_results = [
id_to_result.get(
item["task_id"],
{"task_id": item["task_id"], "question": item.get("question"), "submitted_answer": "ERROR: Processing failed", "error": True}
)
for item in valid_questions
]
return ordered_results
# --- Main Application Class ---
class BasicAgent:
"""
Optimized agent wrapper for the GAIA benchmark.
"""
def __init__(
self,
model_provider="anthropic",
model_name="claude-3-7-sonnet-latest",
api_key=None,
use_cache=True,
max_concurrent=MAX_CONCURRENT_REQUESTS
):
"""
Initialize the BasicAgent with caching and async capabilities.
Args:
model_provider: LLM provider to use
model_name: Specific model to use
api_key: Optional API key
use_cache: Whether to use caching
max_concurrent: Maximum concurrent requests
"""
model_config = {
"model_provider": model_provider,
"model_name": model_name,
"api_key": api_key
}
self.agent = OptimizedGaiaAgent(
model_config=model_config,
use_cache=use_cache,
max_concurrent=max_concurrent
)
print(f"BasicAgent initialized with {model_provider} {model_name}.")
async def process_async(self, questions_data, progress_callback=None):
"""Process questions asynchronously with progress reporting"""
return await self.agent.process_all(questions_data, progress_callback)
def __call__(self, question: str) -> str:
"""
Process a single question (for compatibility with the original interface).
This method is synchronous for backward compatibility.
"""
print(f"Agent received question (first 50 chars): {question[:50]}...")
async def agentic_main():
result = await self.agent.process_question("single", question)
return result["submitted_answer"]
final_answer = asyncio.run(agentic_main())
print(f"Agent returning answer: {final_answer}")
return final_answer
# --- Async Run and Submit Function ---
async def async_run_and_submit_all(
profile: gr.OAuthProfile | None,
progress=gr.Progress()
) -> tuple:
"""
Asynchronous version of run_and_submit_all.
Fetches questions, processes them concurrently, and submits answers.
"""
# --- Determine HF Space Runtime URL and Repo URL ---
space_id = os.getenv("SPACE_ID") # Get the SPACE_ID for sending link to the code
if not profile:
print("User not logged in.")
return "Please Login to Hugging Face with the button.", None
username = f"{profile.username}"
print(f"User logged in: {username}")
api_url = DEFAULT_API_URL
questions_url = f"{api_url}/questions"
submit_url = f"{api_url}/submit"
# 1. Instantiate Agent
try:
progress(0, desc="Initializing agent...")
agent = BasicAgent()
except Exception as e:
print(f"Error instantiating agent: {e}")
return f"Error initializing agent: {e}", None
# In the case of an app running as a Hugging Face space, this link points toward your codebase
agent_code = f"https://huggingface.co/spaces/{space_id}/tree/main"
print(agent_code)
# 2. Fetch Questions
print(f"Fetching questions from: {questions_url}")
progress(0.1, desc="Fetching questions...")
try:
# Use asyncio for the HTTP request
async def fetch_questions():
loop = asyncio.get_event_loop()
return await loop.run_in_executor(
None,
lambda: requests.get(questions_url, timeout=15)
)
response = await fetch_questions()
response.raise_for_status()
questions_data = response.json()
if not questions_data:
print("Fetched questions list is empty.")
return "Fetched questions list is empty or invalid format.", None
print(f"Fetched {len(questions_data)} questions.")
progress(0.2, desc=f"Successfully fetched {len(questions_data)} questions.")
except requests.exceptions.RequestException as e:
print(f"Error fetching questions: {e}")
return f"Error fetching questions: {e}", None
except requests.exceptions.JSONDecodeError as e:
print(f"Error decoding JSON response from questions endpoint: {e}")
print(f"Response text: {response.text[:500]}")
return f"Error decoding server response for questions: {e}", None
except Exception as e:
print(f"An unexpected error occurred fetching questions: {e}")
return f"An unexpected error occurred fetching questions: {e}", None
# 3. Process Questions Asynchronously
print(f"Processing {len(questions_data)} questions...")
try:
# Define progress update function
def update_progress(value, desc=""):
# Scale progress from 0.2-0.8 for the processing phase
progress(0.2 + (value * 0.6), desc=desc)
results = await agent.process_async(questions_data, update_progress)
# Convert results to the expected format
answers_payload = [
{"task_id": result["task_id"], "submitted_answer": result["submitted_answer"]}
for result in results
]
# Format for display
results_log = [
{"Task ID": result["task_id"], "Question": result["question"], "Submitted Answer": result["submitted_answer"]}
for result in results
]
progress(0.8, desc=f"Processed all {len(results)} questions. Preparing submission...")
except Exception as e:
print(f"Error during question processing: {e}")
return f"Error during question processing: {e}", None
if not answers_payload:
print("Agent did not produce any answers to submit.")
return "Agent did not produce any answers to submit.", pd.DataFrame([])
# 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}'..."
print(status_update)
progress(0.9, desc="Submitting answers...")
# 5. Submit
print(f"Submitting {len(answers_payload)} answers to: {submit_url}")
try:
async def submit_answers():
loop = asyncio.get_event_loop()
return await loop.run_in_executor(
None,
lambda: requests.post(submit_url, json=submission_data, timeout=60)
)
response = await submit_answers()
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.')}"
)
print("Submission successful.")
progress(1.0, desc="Complete!")
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}"
print(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."
print(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}"
print(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}"
print(status_message)
results_df = pd.DataFrame(results_log)
return status_message, results_df
# Synchronous wrapper for the async function (for Gradio compatibility)
def run_and_submit_all(profile: gr.OAuthProfile | None, progress=gr.Progress()):
"""Synchronous wrapper for the async function"""
return asyncio.run(async_run_and_submit_all(profile, progress))
# --- Build Gradio Interface using Blocks ---
with gr.Blocks() as demo:
gr.Markdown("# Optimized GAIA Agent Evaluation Runner")
gr.Markdown(
"""
**Instructions:**
1. Please clone this space, then modify the code to define your agent's logic, the tools, and necessary packages.
2. Log in to your Hugging Face account using the button below. This uses your HF username for submission.
3. Click 'Run Evaluation & Submit All Answers' to fetch questions, process them, and see your score.
This implementation features:
- **Caching**: Answers are saved to avoid reprocessing the same questions
- **Asynchronous Processing**: Questions are processed concurrently for better performance
- **Progress Tracking**: See real-time progress as questions are processed
"""
)
with gr.Row():
gr.LoginButton()
clear_cache_button = gr.Button("Clear Cache")
run_button = gr.Button("Run Evaluation & Submit All Answers", variant="primary")
status_output = gr.Textbox(label="Run Status / Submission Result", lines=5, interactive=False)
results_table = gr.DataFrame(label="Questions and Agent Answers", wrap=True)
# Define clear cache function
def clear_cache():
if os.path.exists(CACHE_FILE):
try:
os.remove(CACHE_FILE)
return f"Cache cleared successfully! ({CACHE_FILE})"
except Exception as e:
return f"Error clearing cache: {e}"
return "No cache file found."
# Connect the components
clear_cache_button.click(
fn=clear_cache,
outputs=status_output
)
run_button.click(
fn=run_and_submit_all,
inputs=[gr.OAuthProfile()],
outputs=[status_output, results_table]
)
# --- App Entry Point ---
if __name__ == "__main__":
print("\n" + "-"*30 + " App Starting " + "-"*30)
# Check for SPACE_HOST and SPACE_ID at startup for information
space_host_startup = os.getenv("SPACE_HOST")
space_id_startup = os.getenv("SPACE_ID") # Get SPACE_ID at startup
if space_host_startup:
print(f"✅ SPACE_HOST found: {space_host_startup}")
print(f" Runtime URL should be: https://{space_host_startup}.hf.space")
else:
print("ℹ️ SPACE_HOST environment variable not found (running locally?).")
if space_id_startup: # Print repo URLs if SPACE_ID is found
print(f"✅ SPACE_ID found: {space_id_startup}")
print(f" Repo URL: https://huggingface.co/spaces/{space_id_startup}")
print(f" Repo Tree URL: https://huggingface.co/spaces/{space_id_startup}/tree/main")
else:
print("ℹ️ SPACE_ID environment variable not found (running locally?). Repo URL cannot be determined.")
print("-"*(60 + len(" App Starting ")) + "\n")
print("Launching Gradio Interface for Optimized Agent Evaluation...")
demo.launch(debug=True, share=False)