Sign in
Does CLIP's Generalization Performance Mainly Stem from High Train-Test Similarity?
Prasanna Mayilvahanan
Thaddäus Wiedemer
Evgenia Rusak
Matthias Bethge
Wieland Brendel
Published in:
CoRR (2023)
Keyphrases
</>
similarity measure
wide range
structural similarity
euclidean distance
high dimensional
edit distance
similarity metric
highly similar
real world
data mining
social networks
e learning
multiscale
test cases
test data
high precision