hasktorch/hasktorch

Tensors and neural networks in Haskell

What it solves

Hasktorch provides a way to build tensors and neural networks using the Haskell programming language, allowing developers to leverage the power of PyTorch's core C++ libraries within a strongly-typed functional environment.

How it works

It functions as a Haskell library that leverages the core C++ libraries shared by PyTorch (libtorch). It uses a low-level Foreign Function Interface (FFI) and code generation to bind PyTorch's C++ functionality to Haskell. The library automatically handles the downloading and caching of official libtorch binaries for various platforms, including support for CPU, CUDA, and Apple's MPS backend.

Who it’s for

Haskell developers and researchers who want to implement deep learning models and tensor operations using a functional programming language.

Highlights

  • PyTorch Integration: Built on top of the same C++ core used by PyTorch.
  • Cross-Platform Support: Works on Linux (CPU/CUDA), macOS Apple Silicon (CPU/MPS), and macOS Intel (CPU).
  • Automated Setup: Automatically downloads and caches required libtorch binaries during the build process.
  • Flexible Build Tools: Compatible with Cabal, Stack, and Nix.

Related

  • Project
    LaurentMazare/tch-rs

    Rust bindings for the PyTorch C++ API (libtorch), enabling the use of PyTorch's tensor operations and deep learning tools in Rust.

  • Project
    neuralhydrology/neuralhydrology

    A PyTorch-based Python library designed for deep learning research in hydrology, allowing users to train neural networks via configuration files.

  • Project
    skorch-dev/skorch

    A scikit-learn compatible neural network library that wraps PyTorch, enabling the use of scikit-learn's workflow tools with PyTorch models.

  • Project
    nschaetti/EchoTorch

    A PyTorch-based module for implementing and testing Echo State Networks and Reservoir Computing, designed for research purposes and integration into deep architectures.

  • Project
    Ascend/pytorch

    A PyTorch adapter plugin that enables the PyTorch framework to run on Ascend AI processors, providing hardware acceleration for AI model training and inference.