这篇论文告诉你如何在不牺牲性能的前提下大幅降低多模态大模型的计算量。他们在Qwen3-VL上砍了33.7%的算力，性能只掉了0.5%，方法很巧妙。
论文提出一种操作级视觉令牌跳过框架，将Transformer层分解为注意力与FFN操作，选择性跳过冗余计算。实验覆盖3种MLLM架构（含Qwen3-VL）和10个VQA基准，实现精度-效率权衡：在Qwen3-VL上减少33.7% TFLOPs，保留99.5%的原始性能。关键发现是晚期视觉令牌更新对答案表示影响很小，且有效计算具有操作主导性和层依赖性。
Attend, Transform, or Silence: Operator-Level Visual Skipping for Efficient Multimodal LLM Inference
Multimodal large language models (MLLMs) increasingly process long visual-token sequences, increasing the overall inference computation. Existing acceleration methods usually remove visual tokens or skip visual-token updates in entire layers, but these coarse strategies may discard fine-grained evidence or suppress useful operators together with redundant ones. In this paper, we study visual-token computation from an answer-observable perspective and find that late visual-token updates can remain large while having little effect on answer-token representations. Motivated by this answer-silent redundancy, we decompose each Transformer layer into attention and FFN operators and show that useful visual computation is often operator-dominant and layer-dependent. We propose an operator-level visual-token skipping framework that preserves the full visual-token sequence while selectively bypassing redundant attention, FFN, or both. Experiments across three MLLM architectures and 10 VQA benchmarks show that our method achieves strong efficiency-accuracy trade-offs, reducing \textbf{33.7\%} TFLOPs on Qwen3-VL while retaining \textbf{99.5\%} of the vanilla model performance.