eval-track Documentation

eval-track is a library that provides observability and tracking services for LLM-ML. This documentation explains the usage and functionality of the main components.

Main Components

Client

Client class for retrieving and storing trace data.

Tracer

Decorator for capturing function and coroutine responses.

Router

API endpoints for trace data.

Installation

uv pip install "git+https://github.com/tied-inc/eval-track/tracker"

For detailed usage instructions, please refer to each component's documentation.

Examples

The following examples demonstrate different ways to integrate and use eval-track in your applications:

Advanced FastAPI Usage

Demonstrates advanced patterns including async processing, error handling, and Pydantic model integration.

Rust Actix Usage

Demonstrates capturing function responses in an Actix Web application using procedural macros.

API Orchestration

Illustrates microservice orchestration with parallel request processing and comprehensive trace aggregation.

Each example includes detailed setup instructions, implementation details, and best practices for using eval-track in different scenarios.