Skip to content

SETUP_GUIDE.md

Latest commit

 

History

History
80 lines (60 loc) · 1.8 KB

SETUP_GUIDE.md

File metadata and controls

80 lines (60 loc) · 1.8 KB

Setup Guide

Quick Setup

This framework supports multiple LLM providers through a flexible configuration system.

Option 1: Environment Variables (Recommended)

For Standard OpenAI:

export OPENAI_API_KEY='sk-your-openai-key-here'
python simuleval_core.py

For Custom LLM Proxy:

export OTHER_API_KEY='your-proxy-api-key'
python simuleval_core.py

Option 2: Configuration File

  1. Copy the example configuration:
cp config/private_config.example.py config/private_config.py
  1. Edit config/private_config.py with your credentials:
# For OpenAI
OPENAI_CONFIG = {
    "api_key": "sk-your-actual-key-here",
    "model": "gpt-4o-mini",
    "provider": "openai"
}

# OR for custom proxy (organization-specific)
OTHER_CONFIG = {
    "api_key": "your-actual-proxy-key",
    "model": "gpt-4o-mini", 
    "base_url": "https://your-internal-proxy.company.com/v1",
    "provider": "openai"
}
  1. Run the simulation:
python simuleval_core.py

Testing Your Setup

Before running the full simulation, test your connection:

python test_llm_connection.py

This will verify your API configuration is working correctly.

Configuration Options

API Configuration

  • OPENAI_CONFIG: Uses standard OpenAI API with gpt-4o-mini
  • OTHER_CONFIG: Uses custom proxy endpoint with gpt-4o-mini
  • Environment variables: Override file-based config

Priority Order

The system checks for configuration in this order:

  1. OTHER_API_KEY environment variable
  2. OPENAI_API_KEY environment variable
  3. OTHER_CONFIG in private_config.py
  4. OPENAI_CONFIG in private_config.py

Security Notes

  • The config/private_config.py file is automatically git-ignored
  • Never commit API keys to version control
  • Use environment variables in production environments