Changelog

This page contains the release history of PyLO.

Version 0.2.0

Release Date: November 2025

New Features

• Added CUDA-accelerated implementation of VeLO optimizer (pylo.optim.VeLO_CUDA)

  • Provides significant speedup for training on CUDA-enabled GPUs

  • Fully compatible with the existing VeLO optimizer API

  • Requires CUDA toolkit for installation

Improvements

• Enhanced documentation with API reference

• Improved installation instructions

Version 0.1.0

Release Date: May 2025

• Initial release of PyLO

• Core learned optimizer implementations

• Integration with PyTorch optim library

• Hugging Face Hub support for sharing models

• Basic benchmarking utilities