Releases: chr5tphr/zennit
Releases · chr5tphr/zennit
v0.3.3
v0.3.2
v0.3.1
What's Changed
- adding software paper reference by @sebastian-lapuschkin in #34
- Implemented custom module backward hook by @chr5tphr in #35
- Fixed a typo in README.md by @chr5tphr in #36
- Allow empty classes in feed_forward.py by @chr5tphr in #38
New Contributors
- @sebastian-lapuschkin made their first contribution in #34
Full Changelog: 0.3.0...0.3.1
v0.3.0
What's Changed
- Add Attributor to compute gradient by @chr5tphr in #28
- SmoothGrad by @chr5tphr in #29
- Added IntegratedGradients by @chr5tphr in #30
- Occlusion Attributor by @chr5tphr in #31
- Backprop Approaches by @chr5tphr in #32
- Added a few other example composites to README by @chr5tphr in #33
Full Changelog: 0.2.1...0.3.0
v0.2.1
v0.2.0
What's Changed
- Add ResNet support by @chr5tphr in #1
- Readme indent by @chr5tphr in #2
- added simple zennit logo by @chr5tphr in #3
- Alpha beta rule change by @rachtibat in #4
- zplus bug fix by @rachtibat in #5
- added better text for alpha beta value errors, cleaned up some style problems by @chr5tphr in #7
- Added notes to the zplus rule, referring to the original DTD paper, w… by @chr5tphr in #8
- Make the example runnable from reading the README alone by @chr5tphr in #10
- Be more careful with tensor references in Hook by @chr5tphr in #13
- Set parameter to not require gradients in example by @chr5tphr in #14
- Readme Install by @chr5tphr in #15
- Add parameter control to mod_params and LinearHook by @chr5tphr in #16
- Basic hook by @chr5tphr in #17
- Changed SSH to HTTPS for cloning in README.md by @chr5tphr in #18
- Style fixes: unused import and binary op linebreak by @chr5tphr in #20
- Added palette_swap.py utility script by @chr5tphr in #21
- Added palette_fit.py, a script to fit palettes by @chr5tphr in #22
- Use float for dist computations in palette_fit by @chr5tphr in #23
- License by @chr5tphr in #25
New Contributors
- @rachtibat made their first contribution in #4
Full Changelog: 0.1.0...0.2.0