jettjaniak / teren

Linking activation space features to model behavior
Apache License 2.0
0 stars 1 forks source link

split `get_pert_loss` function s.t. we can access perturbed activations if we want to #13

Closed jettjaniak closed 4 months ago

jettjaniak commented 4 months ago

this can still be a single function, but it should call a function that computes the perturbed activations