Llama.Ipsum

A lightweight mock server that simulates the OpenAI API for development and testing purposes.

Features

• Implements key OpenAI API endpoints:
  • /v1/chat/completions
  • /v1/completions
  • /v1/models
• Supports API key authentication
• Logs requests and responses
• Supports streaming responses
  • Configurable throughput (tokens per second)
• Handles tool calls

Installation

1. Clone the repository:

   git clone https://github.com/taha-yassine/llama-ipsum.git
   cd llama-ipsum

2. Run the server:

   just run

Template Customization

The mock server uses Jinja2 templates to generate responses. You can customize these templates to fit your specific testing needs.

Using Custom Templates

1. Create a directory for your custom templates:

   mkdir -p my_templates

2. Copy the templates you want to customize:

   # Example: customize chat completion response
   cp app/templates/chat/completion.json.jinja my_templates/chat/

3. Edit the templates according to your needs.

4. Start the server with your custom templates:

   uv run -m app.main --template-dir /path/to/my_templates

License

This project is licensed under the MIT License.