Cosine distance between sentence embedding variations (averaged tokens, CLS token, random token) does not predict their relative performance on FlashHolmes linguistic tasks across BERT-family models.
Title resolution pending
1 Pith paper cite this work, alongside 89 external citations. Polarity classification is still indexing.
1
Pith paper citing it
89
external citations · OpenAlex
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Testing the assumptions about the geometry of sentence embedding spaces: the cosine measure need not apply
Cosine distance between sentence embedding variations (averaged tokens, CLS token, random token) does not predict their relative performance on FlashHolmes linguistic tasks across BERT-family models.