braindatalab / gecobench

NLP Benchmark for XAI methods
BSD 3-Clause "New" or "Revised" License
3 stars 1 forks source link

Add normalized attribution plot and fix spacing for sentence wise plot #44

Closed venondev closed 7 months ago

venondev commented 7 months ago

Gender:

Raw accumulated:

most_common_xai_attributions

Normalized: most_common_xai_attributions_normalized

Sentencewise sentence_wise_attribution_diff

Sentiment:

Raw accumulated:

most_common_xai_attributions

Normalized:

most_common_xai_attributions_normalized

Sentencewise:

sentence_wise_attribution_diff

rickwg commented 7 months ago

Looks good. We were also talking about a more nuanced normalization scheme accounting for the issue that mainly names appear with the current normalization strategy. But we can do that in a second step. I will create an issue. Please resolve the conflicts in the sentiment config file and then go ahead and merge this branch.