openxla/xla

A machine learning compiler for GPUs, CPUs, and ML accelerators

What it solves

XLA addresses the challenge of running machine learning models efficiently across diverse hardware. It prevents the need to write hardware-specific code for every model, allowing high-performance execution on GPUs, CPUs, and specialized ML accelerators.

How it works

XLA acts as a compiler that takes models from popular frameworks like PyTorch, TensorFlow, and JAX and optimizes them specifically for the target hardware platform to ensure high-performance execution.

Who it’s for

This project is primarily for ML framework developers, XLA contributors who want to develop the compiler, and integrators who need to debug or add support for new ML frontends or hardware backends.

Highlights

  • Support for major frameworks including PyTorch, TensorFlow, and JAX.
  • Cross-platform optimization for CPUs, GPUs, and ML accelerators.
  • Open-source machine learning compiler.

Related

  • Project
    jax-ml/jax

    A Python library for accelerator-oriented array computation and program transformation, designed for high-performance numerical computing and large-scale machine learning.

  • Project
    tenstorrent/tt-forge

    An open-source AI compiler stack for Tenstorrent hardware that enables the execution and training of models from PyTorch, JAX, and ONNX.

  • Project
    microsoft/onnxruntime

    A cross-platform machine learning accelerator that optimizes inference and training for models from various frameworks like PyTorch and TensorFlow.

  • Project
    openxla/stablehlo

    A portability layer and operation set for ML models that enables interoperability between different ML frameworks and ML compilers.

  • Dispatch
    Faster Text Generation with TensorFlow and XLA

    Hugging Face has enabled XLA compilation for text generation in the transformers library using TensorFlow, resulting in speedups of up to 100x and performance that often exceeds PyTorch.