Research code for evaluating category knowledge acquired by neural networks.
Usage
Classification Performance
A measure how well pairs of learned representations can be correctly judged to belong to the same category
Divergence-from-Prototype
A measure of how abstract learned representations are.
Category-Spread
A measure of spread between learned representations that belong to the same category
Silhouette Score
A measure of how well the similarity structure of learned representations captures the gold category structure
Compatibility
Developed on Ubuntu 18.04 and Python 3.7