Docs: Clarification about requirements Modules/ Canonizers

chr5tphr / zennit

Zennit is a high-level framework in Python using PyTorch for explaining/exploring neural networks using attribution methods like LRP.

Other

190 stars 32 forks source link

Closed chr5tphr closed 2 years ago

chr5tphr commented 2 years ago

added Note to SequentialBatchNorm to clarify the assumptions made to detect adjacent modules
- clarify the requirement of models to be implemented using pytorch modules in getting-started.rst and README.md
- clarify the requirement of ordered children in sequential models for SequentialMergeBatchNorm in getting-started.rst

Closes #63