Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Update howto.md #21

Open
wants to merge 1 commit into
base: master
Choose a base branch
from
Open

Update howto.md #21

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion howto.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -57,7 +57,7 @@ Here are some guidelines for selecting baseline depending on the input type:

**Text**: Consider using the all-zero baseline (i.e. using all-zero embedding vector). This may seem unintuitive, but nevertheless works. The reason is that the training process causes the input embeddings of unimportant stop words to have a small norm; these are precisely words that tend to "nothing". Some models constrain embeddings to have unit norm. In these cases, a padding term is an adequate baseline. The other alternative is to use "stop" words.

**Binary and continuous features**: Some times binary and continuous features are binned and then embedded. In these cases, follow the approach for 'text features above. The other alternative is to use several training data inputs as a baseline and to average the attributions over these.
**Binary and continuous features**: Sometimes binary and continuous features are binned and then embedded. In these cases, follow the approach for 'text features above. The other alternative is to use several training data inputs as a baseline and to average the attributions over these.

If you have trouble selecting a baseline, reach out to [integrated-gradients@gmail.com](mailto:integrated-gradients@gmail.com).

Expand Down