https://github.com/pierric/smart-chess-rust

https://github.com/pierric/smart-chess-rust

Last synced: 16 days ago
JSON representation

• Host: GitHub
• URL: https://github.com/pierric/smart-chess-rust
• Owner: pierric
• License: bsd-3-clause
• Created: 2023-11-01T14:52:14.000Z (about 1 year ago)
• Default Branch: main
• Last Pushed: 2024-12-30T21:54:01.000Z (16 days ago)
• Last Synced: 2024-12-30T22:31:16.327Z (16 days ago)
• Language: Rust
• Size: 3.01 MB
• Stars: 0
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
[![Rust](https://github.com/pierric/smart-chess-rust/actions/workflows/rust.yml/badge.svg)](https://github.com/pierric/smart-chess-rust/actions/workflows/rust.yml)

# Build

## Python part

```

poetry install

```

## Rust part

```

poetry shell

export LIBTORCH_USE_PYTORCH=1

export LIBTORCH_BYPASS_VERSION_CHECK=1

cargo build

```

Tch is always built against the torch with patch-version being 0. But very safe to be run with the other patch versions. It is necessary to export the environment variable `LIBTORCH_BYPASS_VERSION_CHECK=1`.

# Quantization

Experiments with quantized model (int8) is also tried. No convicing results are seen, neither faster nor better performance. If interested, installing pytorch_quantization as described here: https://github.com/NVIDIA/TensorRT/tree/release/9.3/tools/pytorch-quantization/pytorch_quantization