Skip to content

Releases: f-dangel/backpack

Version 1.6.0

12 Jul 18:19
1ebfb40
Compare
Choose a tag to compare

See the changelog for details

Version 1.5.2

19 Dec 12:59
ed362e4
Compare
Choose a tag to compare

For details, refer to the changelog.

Version 1.5.1

03 Nov 09:37
6414462
Compare
Choose a tag to compare

For details, refer to the changelog.

Version 1.5.0

15 Feb 11:42
0ab9421
Compare
Choose a tag to compare

For details see the changelog.

Version 1.4.0

11 Oct 20:03
1da7e53
Compare
Choose a tag to compare

See the changelog for details.

Version 1.3.0

16 Jun 14:16
1f72b00
Compare
Choose a tag to compare

See the changelog for details.

1.2.0

02 Nov 20:20
6a1ac37
Compare
Choose a tag to compare

Added

  • Deprecated python3.5, tested compatibility with PyTorch 1.6.0 [PR]
  • Support first-order extensions for Conv1d, Conv3d,
    ConvTranspose1d, ConvTranspose2d, ConvTranspose3d
    • extensions.BatchGrad [PR]
    • extensions.BatchL2Grad [PR]
    • extensions.SumGradSquared and extensions.Variance [PR]
    • Raise exceptions for unsupported exotic hyperparameters [PR1, PR2]
  • New example: Backpropagating through BackPACK quantities [commit]
  • New extensions in API: Block-diagonal curvature products
    • Exposed via extensions.HMP, extensions.GGNMP, extensions.PCHMP [PR]
    • Examples: Hutchinson trace estimation [PR] and Hessian-free optimization with CG [PR]

Fixed

  • Add missing zero_grad in the diagonal GGN second-order optimization example [PR]

Internal

  • Increased test coverage
    • New test suite for backpack.extensions [PR]
    • New test suite for backpack.core [PR]
  • Implemented derivatives of the following operations in backpack.core
    • More activation functions [PR]
    • Conv1d, Conv3d [PR]
    • ConvTranspose1d, ConvTranspose2d, ConvTranspose3d [PR]
  • Refactor firstorder extensions to share more code [PR1, PR2]
  • Removed detachs to support differentiating through quantities [PR]

1.1.1

29 Apr 18:54
ad8899e
Compare
Choose a tag to compare

Added

  • Improved documentation, moved to ReadTheDocs [PR1, PR2, PR3]
  • Tested compatibility with PyTorch 1.5.0.
  • Support 2nd-order backprop for vectors in MSELoss [PR]
  • Sanity checks to raise warnings if the following are used.
    inplace modification [PR],
    unsupported loss parameters [PR],
    custom losses in 2nd-order backpropagation [PR]

Fixed

  • Removed opt_einsum dependency [PR]
  • Missing implementations and wrong backpropagation of KFRA for Conv2d, MaxPool2d, and AvgPool2d [PR]
  • Remove try_view and use reshape to use PyTorch 1.4.0 improvements [PR]

Internal

  • Docstring style [PR]

Release 1.1.0

11 Feb 19:27
3122de0
Compare
Choose a tag to compare
Merge pull request #45 from f-dangel/release

Release BackPACK1.1.0

1.0.1

23 Jan 19:38
bce5207
Compare
Choose a tag to compare

This is not a new release at the time of writing. It is a tag to track v1.0.1 on PyPI