Using class-name text embeddings, rather than image features, to condition prompts reduces the base-new accuracy tradeoff in vision-language prompt tuning.
A closer look at few-shot classification again
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
-
A Closer Look at Conditional Prompt Tuning for Vision-Language Models
Using class-name text embeddings, rather than image features, to condition prompts reduces the base-new accuracy tradeoff in vision-language prompt tuning.