ReleaseNVIDIANVIDIApublished Sep 6, 2024seen 15h

NVIDIA/TorchFort v0.2.0

NVIDIA/TorchFort

Open original ↗

Captured source

source ↗
GH
GitHub/github.com/NVIDIA/TorchFort
NVIDIA/TorchFort v0.2.0
Source ↗
published Sep 6, 2024seen 15hcaptured 15hhttp 200method plain

v0.2.0

Repository: NVIDIA/TorchFort

Tag: v0.2.0

Published: 2024-09-06T17:48:03Z

Prerelease: yes

Release notes:

What's Changed

This release includes several major updates to TorchFort, including:

  • Enabling compilation of library with alternative compilers to NVHPC (e.g., GNU)
  • Enabling model/RL system training and inference on CPU
  • Enabling CPU-only builds without CUDA/NCCL
  • New reinforcement learning features, including PPO algorithms and on-policy algorithms
  • Improvements to build scripts

Breaking Changes

#14 enables placing and running models/RL systems on CPU. To enable this, an additional device argument was added to the model/system creation APIs (e.g., torchfort_model_create). Please refer to the documentation for more details.

PRs included in this release

  • Enable support for complex gradient reduction in distributed cases. (https://github.com/NVIDIA/TorchFort/pull/2)
  • remove extraneous mpi call from cmake (https://github.com/NVIDIA/TorchFort/pull/4)
  • generalize cmake to build for different cuda archs (https://github.com/NVIDIA/TorchFort/pull/7)
  • remove hardcoded yaml-cpp path from CMakeLists.txt (https://github.com/NVIDIA/TorchFort/pull/5)
  • Build updates and improvements (https://github.com/NVIDIA/TorchFort/pull/10)
  • Update setup.cpp (https://github.com/NVIDIA/TorchFort/pull/11)
  • merging rl changes ( https://github.com/NVIDIA/TorchFort/pull/13)
  • Enable model training/inference on CPU or GPU devices. Enabling usage of alternative compilers to NVHPC, (https://github.com/NVIDIA/TorchFort/pull/14)
  • Tkurth/rl ppo (https://github.com/NVIDIA/TorchFort/pull/12)
  • Tkurth/rl tests (https://github.com/NVIDIA/TorchFort/pull/15)
  • Add train and inference functions for 5d Fortran arrays. (https://github.com/NVIDIA/TorchFort/pull/17)
  • Fix interface issues in Fortran module with gfortran. (https://github.com/NVIDIA/TorchFort/pull/18)
  • Enable builds without CUDA/GPU support (https://github.com/NVIDIA/TorchFort/pull/19)
  • Fixing up documentation. (https://github.com/NVIDIA/TorchFort/pull/20)
  • 0.2.0 release (https://github.com/NVIDIA/TorchFort/pull/21)

Full Changelog: https://github.com/NVIDIA/TorchFort/compare/v0.1.0...v0.2.0