Skip to content

jalpp/neuralnetapi

Repository files navigation

Chess ONNX Policy Nets API

A high-performance inference API endpoint for chess neural network policy evaluation, designed to integrate seamlessly with ChessAgine's MCP (Model Context Protocol) net tool. Overview This repository provides a REST API for running inference on chess position evaluation using ONNX-optimized neural networks. The API accepts chess positions in FEN notation and returns policy predictions for move evaluation, enabling real-time chess analysis and AI-powered decision making. Features

Features

  • Fast Inference: Leverages ONNX Runtime for optimized neural network inference
  • Chess Position Analysis: Accepts positions in FEN format for universal compatibility
  • Policy Network Evaluation: Returns move probabilities and policy predictions
  • MCP Integration: Built specifically to work with ChessAgine's Model Context Protocol tools
  • RESTful API: Simple HTTP endpoints for easy integration

License:

GPL

Credits

  • Hollowleaf for leela pre-proccessing code
  • Maia devs for Maia class
  • Lichess API

Author

@jalpp

About

A high-performance inference API endpoint for chess neural network policy evaluation

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published