v3.0.0

RelevancePropagation v3.0.0

Diff since v2.0.2

• Update XAIBase interface to v4. This adds a field to the Explanation return type and removes the add_batch_dim keyword argument. Refer to the XAIBase.jl changelog for more information (#19)
• Add option to skip normalization of output layer relevance (#22)

Merged pull requests:

• Update CI workflows (#18) (@adrhill)
• Update XAIBase dependency to v4, drop support for Julia <1.10 (#19) (@adrhill)
• Test benchmarks with PkgJogger (#20) (@adrhill)
• Add ExplicitImports.jl tests (#21) (@adrhill)
• Add option to skip normalization of output layer relevance (#22) (@adrhill)

v2.0.2

RelevancePropagation v2.0.2

Diff since v2.0.1

Merged pull requests:

• Modularize package tests (#17) (@adrhill)

v2.0.1

RelevancePropagation v2.0.1

Diff since v2.0.0

Merged pull requests:

• canonize SkipConnection and fix splitting of parallel layers (#14) (@Maximilian-Stefan-Ernst)

v2.0.0

RelevancePropagation v2.0.0

Diff since v1.1.0

This release removes the automatic reexport of heatmapping functionality.
Users are now required to manually load VisionHeatmaps.jl and/or TextHeatmaps.jl.

This reduces the maintenance burden for new heatmapping features and the amount of dependencies for users who don't require heatmapping functionality.

• Remove reexport of heatmapping functionality by updating XAIBase dependency to v3.0.0 (#13).
• Add support for LayerNorm and Scale layers (#9)
• Add LayerNormRule (#9)
• Add LRP rule overview to docs (#12)

Some internal improvements were made as well:

• Update canonize mechanism to include model splitting pass canonize_split (#9)
• Improve modify_layer by introducing get_weight and get_bias abstractions to handle varying field names (#9)
• Update LayerMap to use ModelIndex (#10)
• Make chainzip more robust (#11)

Merged pull requests:

• Add LayerNorm support, add model splitting to canonize (#9) (@Maximilian-Stefan-Ernst)
• Update LayerMap to use ModelIndex (#10) (@adrhill)
• Make chainzip more robust (#11) (@adrhill)
• Add LRP rule overview to docs (#12) (@adrhill)
• Update XAIBase to v3.0.0 (#13) (@adrhill)

v1.1.0

RelevancePropagation v1.1.0

Diff since v1.0.1

• Support SkipConnection layers (#8)
• Document LRP rule notation in API reference (e11c234)

Merged pull requests:

• Support SkipConnection layers (#8) (@adrhill)

v1.0.1

RelevancePropagation v1.0.1

Diff since v1.0.0

Merged pull requests:

• Reorganize documentation (#7) (@adrhill)

v1.0.0

RelevancePropagation v1.0.0

Merged pull requests:

• Initial functionality (#1) (@adrhill)
• Remove Tullio package extension (#2) (@adrhill)
• Rename package to RelevancePropagation.jl (#3) (@adrhill)
• Add LRP documentation from ExplainableAI.jl (#4) (@adrhill)
• Refactor CRP to use XAIBase feature selectors (#5) (@adrhill)