Enhancing Instruction-Following Capability of Visual-Language Models by Reducing Image Redundancy
This work addresses a key bottleneck in MLLMs for improving their usability in tasks requiring precise human command interpretation, though it is incremental as it builds on existing methods to optimize token processing.
The paper tackles the problem of multimodal large language models (MLLMs) having inferior instruction-following capability compared to large language models (LLMs) by proposing Visual-Modality Token Compression (VMTC) and Cross-Modality Attention Inhibition (CMAI) strategies to reduce image redundancy, which significantly enhances instruction-following while preserving multimodal understanding across benchmarks like VQA-V2 and MMBench.
Large Language Models (LLMs) have strong instruction-following capability to interpret and execute tasks as directed by human commands. Multimodal Large Language Models (MLLMs) have inferior instruction-following ability compared to LLMs. However, there is a significant gap in the instruction-following capabilities between the MLLMs and LLMs. In this study, we conduct a pilot experiment, which demonstrates that spatially down-sampling visual tokens significantly enhances the instruction-following capability of MLLMs. This is attributed to the substantial redundancy in visual modality. However, this intuitive method severely impairs the MLLM's multimodal understanding capability. In this paper, we propose Visual-Modality Token Compression (VMTC) and Cross-Modality Attention Inhibition (CMAI) strategies to alleviate this gap between MLLMs and LLMs by inhibiting the influence of irrelevant visual tokens during content generation, increasing the instruction-following ability of the MLLMs while retaining their multimodal understanding capacity. In VMTC module, the primary tokens are retained and the redundant tokens are condensed by token clustering and merging. In CMAI process, we aggregate text-to-image attentions by text-to-text attentions to obtain a text-to-image focus score. Attention inhibition is performed on the text-image token pairs with low scores. Our comprehensive experiments over instruction-following capabilities and VQA-V2, GQA, TextVQA, MME and MMBench five benchmarks, demonstrate that proposed strategy significantly enhances the instruction following capability of MLLMs while preserving the ability to understand and process multimodal inputs.