Soft-ZCA whitening, an eigenvalue-regularized ZCA transform, is proposed and shown to improve semantic code search MRR on CodeSearchNet and a low-resource R dataset, although test-set whitening and test-set-tuned epsilon weaken the evidence.
Isotropic repre- sentation can improve zero-shot cross-lingual transfer on multilingual lan- guage models
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2024 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Isotropy Matters: Soft-ZCA Whitening of Embeddings for Semantic Code Search
Soft-ZCA whitening, an eigenvalue-regularized ZCA transform, is proposed and shown to improve semantic code search MRR on CodeSearchNet and a low-resource R dataset, although test-set whitening and test-set-tuned epsilon weaken the evidence.