search

Trustworthy-ML-Lab / Label-free-CBM

A new framework to transform any neural networks into an interpretable concept-bottleneck-model (CBM) without needing labeled concept data
73 stars 13 forks source link

[test acc -1.0000] #5

Closed renos closed 9 months ago

renos commented 9 months ago

When I run cifar 10 / cifar 100 I get [test acc -1.0000] when I run python train_cbm.py --concept_set data/concept_sets/cifar10_filtered.txt

ChoiDae1 commented 6 months ago

Hi! I also face this issue. Is it okay?

tuomaso commented 6 months ago

This is not an issue it just means test data was not provided to glm-saga. Val acc is the accuracy of the method on the unseen data.