Injected learned Gaussian noise, tuned with a variational loss, improves CLIP's few-shot classification accuracy slightly over CoOp and CLIP-Adapter.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2024 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Enhance Vision-Language Alignment with Noise
Injected learned Gaussian noise, tuned with a variational loss, improves CLIP's few-shot classification accuracy slightly over CoOp and CLIP-Adapter.