Efficient LLaMA-3.2-Vision by Trimming Cross-attended Visual Features
read the original abstract
Visual token reduction lowers inference costs caused by extensive image features in large vision-language models (LVLMs). Unlike relevant studies that prune tokens in self-attention-only LVLMs, our work uniquely addresses cross-attention-based models, which achieve superior performance. We identify that the key-value (KV) cache size for image tokens in cross-attention layers significantly exceeds that of text tokens in self-attention layers, posing a major compute bottleneck. To mitigate this issue, we exploit the sparse nature in cross-attention maps to selectively prune redundant visual features. Our Trimmed Llama effectively reduces KV cache demands without requiring additional training. By benefiting from 50%-reduced visual features, our model can reduce inference latency and memory usage while achieving benchmark parity.
This paper has not been read by Pith yet.
Forward citations
Cited by 3 Pith papers
-
ProjLens: Unveiling the Role of Projectors in Multimodal Model Safety
ProjLens shows that backdoor parameters in MLLMs are encoded in low-rank subspaces of the projector and that embeddings shift toward the target direction with magnitude linear in input norm, activating only on poisone...
-
HotComment: A Benchmark for Evaluating Popularity of Online Comments
HotComment is a new multimodal benchmark that quantifies online comment popularity via content quality assessment, interaction-based prediction, and agent-simulated user engagement, accompanied by the StyleCmt stylist...
-
C-CoT: Counterfactual Chain-of-Thought with Vision-Language Models for Safe Autonomous Driving
C-CoT applies VLMs to autonomous driving via five-stage reasoning with a meta-action tree for counterfactuals, yielding 81.9% risk recall, 3.52% collision rate, and 1.98 m L2 error on a new dataset.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.