TorchOpt is an efficient library for differentiable optimization built upon PyTorch.
optree
compatibility for multi-tree-map with None
values by @XuehaiPan in #195.Full Changelog: https://github.com/metaopt/torchopt/compare/v0.7.2...v0.7.3
Adadelta
, RAdam
, Adamax
optimizer by @JieRen98 and @Benjamin-eecs in #171.Full Changelog: https://github.com/metaopt/torchopt/compare/v0.7.1...v0.7.2
ruff
and flake8
plugins integration by @XuehaiPan in #138 and #139.extract_state_dict
by @StefanoWoerner in #162.zip(*nested)
in transformations by @XuehaiPan in #145.Full Changelog: https://github.com/metaopt/torchopt/compare/v0.7.0...v0.7.1
Full Changelog: https://github.com/metaopt/torchopt/compare/v0.6.0...v0.7.0
nan_to_num
hook and gradient transformation by @XuehaiPan in #119.functorch
integration by @vmoens and @Benjamin-eecs and @XuehaiPan in #6.None
in module containers by @XuehaiPan.sqrt_
and add_
by @Benjamin-eecs and @JieRen98 and @XuehaiPan.shape=(1,)
) can change the shape of the scalar updates (shape=()
) by @XuehaiPan in #71.Full Changelog: https://github.com/metaopt/torchopt/compare/v0.5.0...v0.6.0
params
to update function in gradient transformations by @XuehaiPan in #65.weight_decay
option to optimizers by @XuehaiPan in #65.maximize
option to optimizers by @XuehaiPan in #64.pytest.mark.parametrize
and enabling parallel testing by @XuehaiPan and @Benjamin-eecs in #55.nvcc
requirement by @XuehaiPan in #51.cibuildwheel
to build wheels by @XuehaiPan in #45.Full Changelog: https://github.com/metaopt/TorchOpt/compare/v0.4.2...v0.4.3 Pre-built wheels: https://pypi.org/project/torchopt/0.4.3/#files
CMakeLists.txt
by @XuehaiPan in #18.pyproject.toml
for packaging and update GitHub Action workflows by @XuehaiPan in #31.TorchOpt
to torchopt
by @XuehaiPan in #20.conda
environment recipe by @XuehaiPan in #24.Full Changelog: https://github.com/metaopt/TorchOpt/compare/v0.4.1...v0.4.2
N/A