PromptMargin adapts CLIP to few-shot classification under distribution shift using selective augmentations and a multimodal margin regularizer, beating MaPLe on most of fifteen datasets.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Prompt Tuning Vision Language Models with Margin Regularizer for Few-Shot Learning under Distribution Shifts
PromptMargin adapts CLIP to few-shot classification under distribution shift using selective augmentations and a multimodal margin regularizer, beating MaPLe on most of fifteen datasets.