Papers/2609.30434
🧪 Test?View on arXiv

ProCAP: Probabilistic Cross-Attentive Prompt Learning for Vision-Language Models

Not provided in the content

multimodalprompt learningfew-shot learningtransfer learning
2609.30434
Builder Relevance
80%
1h ago

Abstract

ProCAP enhances cross-modal interaction and training stability in vision-language models without updating the backbone.

Reality Card

Core Claim

ProCAP achieves strong base-to-novel generalization and competitive transfer performance across multiple datasets while keeping the CLIP backbone unchanged.

Method / Result

ProCAP improves few-shot learning performance across 11 datasets.

Limitations

The method's reliance on specific parameterization and regularization techniques may limit its generalizability to other architectures.

Paper to code

Verified implementation resources so builders can test the paper’s claims instead of stopping at the abstract.

No verified implementation link has been attached yet. AIBuzzHub will keep this panel separate from unverified search results.
← Back to all papers