CATP: Cross-Attention Token Pruning for Accuracy Preserved Multimodal Model Inference
This work addresses efficiency challenges for users of multimodal models, though it appears incremental as it builds on existing pruning techniques.
The paper tackles the trade-off between computational efficiency and model precision in large multimodal models by introducing Cross-Attention Token Pruning (CATP), achieving up to 12.1x higher accuracy compared to existing token pruning methods.
In response to the rising interest in large multimodal models, we introduce Cross-Attention Token Pruning (CATP), a precision-focused token pruning method. Our approach leverages cross-attention layers in multimodal models, exemplified by BLIP-2, to extract valuable information for token importance determination. CATP employs a refined voting strategy across model heads and layers. In evaluations, CATP achieves up to 12.1X higher accuracy compared to existing token pruning methods, addressing the trade-off between computational efficiency and model precision.