Releases: xai4space/meteors
Releases · xai4space/meteors
v0.2.0
🛎️ Chores
- bump up the package version (#158)
🩺 Bug Fixes
- Consistency of terminology for Attributes (#157)
- Set the keep_gradient argument to False in methods that can store the gradient in the final results (#151)
🔨 Features
- standardize relative imports across codebase (#155)
- Added band specific spatial attribution visualisation (#146)
- Added internal batch size to Integrated Gradients (#148)
- added option for title in the charts (#138)
- Updating Rye and preparing for shifting to uv (#145)
📚 Documentation
v0.1.2
🛎️ Chores
- update the package version (#142)
🩺 Bug Fixes
- documentation version alias update fixed (#139)
📚 Documentation
- tutorials fixes (#129)
- fixed documentation versioning (#135)
- Documentation template update (#134)
- fix reference docs for visualizer modules (#133)
- fix navbar for new tutorials (#132)
- update changelog.md for v0.1.1 [skip ci] (#131)
🔨 Features
- Added Hyperlinks to the PR in changelog (#136)
v0.1.1
🛎️ Chores
- update package version (#130)
📚 Documentation
- refine the tutorial for lime and add tutorial for attributions methods (#128)
- refine the example for segmentation problem type (#124)
- update changelog.md for 0.1.0 [skip ci] (#113)
🩺 Bug Fixes
- Moved the postprocessing to the ExplainableModel (#123)
- Corrected visualisers (#118)
- Corrected Attributes Functionalities (#117)
- changelog.md and release notes do not contain all commits due to the commit processor not parsing multi-line commit messages. (#116)
🔨 Features
- Updated release and docs GitHub actions to tigger when GitHub tag is pushed (#119)
v0.1.0
🛎️ Chores
🔨 Features
- 96 feat segmentation support for attribution methods (#103)
- refactored the package structure (#111)
- Custom errors (#101)
🩺 Bug Fixes
- corrected visualizations (#106)
- Updated occlusion (#93)
- 91 docs the lime illustration image is missing in the docs (#92)
📚 Documentation
- update changelog.md for 0.0.4 [skip ci] (#90)
v0.0.4
v0.0.3
LIME method
meteors
version 0.0.1 is the pre-release version implementing the very basics of LIME attribution method.