HF_Agents_Final_Project / docs /LOCAL_TESTING.md
Yago Bolivar
reorder
073b7fb

A newer version of the Gradio SDK is available: 5.43.1

Upgrade

Local Testing Guide for GAIA Agent

This document outlines how to test the GAIA agent locally during development.

Setup

  1. Install dependencies:

    pip install -r requirements.txt
    
  2. If you want to use the OAuth features locally:

    huggingface-cli login
    

    Or set the HF_TOKEN environment variable with your token from HF Settings.

Running the Application

Option 1: Simplified Local Testing (Recommended for Development)

Use app_local.py which has a mock agent and doesn't require OAuth:

python app_local.py

Or use the helper script:

bash run_local.sh

This will:

  • Install required dependencies
  • Run the local version of the app
  • Use a mock agent that returns test responses
  • Use local sample questions without making API calls
  • Not submit any answers to the actual API

Option 2: Full Application with Test Username

If you want to test the full application but without requiring login:

python app2.py

When the application loads:

  1. Enter a test username in the "Or enter test username for local development" field
  2. Click "Run Evaluation & Submit All Answers"

Option 3: Full Application with OAuth

To test the complete application with OAuth authentication:

  1. Make sure you're logged in to Hugging Face CLI: huggingface-cli login
  2. Run: python app.py or python app2.py
  3. Click the "Login" button in the interface
  4. After logging in, click "Run Evaluation & Submit All Answers"

Debugging

If you encounter OAuth-related errors:

  1. Check if you're logged in with huggingface-cli whoami
  2. Try setting your Hugging Face token as an environment variable:
    export HF_TOKEN=your_token_here
    
  3. Use the local testing version (app_local.py) which avoids OAuth entirely

Next Steps

  1. Replace the mock agent in app_local.py with your real agent implementation
  2. Test with a small set of sample questions before scaling up
  3. Gradually add and test tools (web search, file reader, etc.)
  4. When ready, deploy to Hugging Face Spaces for full evaluation