Hi-Lo Prune：在剪枝之前先看清将失去什么的分层词元选择方法
Hi-Lo Prune: Look at What You'll Lose before Pruning with Hierarchical Token Selection
对做多模态大模型视觉侧推理加速的人，这篇给出了一个可直接复用的三段式流程：先用由粗到细的分层选择挑出冗余词元与待剪候选，再在 LLM 前几层用注意力把候选词元的信息显式迁移到保留词元上，最后在指定层一次性移除，从而支持激进的高比例早期剪枝而无需微调。
原文依据 a training-free pruning strategy that is built on a simple principle: look at what you will lose before you remove it
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论多模态大语言模型（MLLMs）在视觉语言理解方面取得了显著进展，但处理长视觉词元序列的计算开销仍然很高。现有方法通过在视觉编码器之后丢弃图像词元，或在 LLM 的早期 Transformer 层中对其进行剪枝来降低这一开销。这些策略虽然提升了效率，却不可避免地丢弃了有信息量的视觉内容，并可能损害下游推理性能。为应对这一挑战，本文提出 Hi-Lo Prune，一种免训练的剪枝策略，其建立在一个简单的原则之上：在移除之前先看清将会失去什么。Hi-Lo Prune 不直接丢弃词元，而是先通过由粗到细的选择过程确定哪些词元可以安全剪枝，再在剪枝发生之前促使模型吸收这些词元的信息。具体而言，该框架包含三个阶段：（1）分层剪枝词元选择。在视觉编码之后，应用由粗到细的过程确定需要保留的词元，并从冗余词元中选出一组关键的剪枝候选。（2）注意力引导的候选词元合并。在移除所选词元之前，在 LLM 的早期层施加注意力机制，将来自这些候选词元的信息显式迁移到保留词元上。（3）低信息量候选词元移除。在指定的 Transformer 层移除被剪枝的词元，从而降低后续所有层的计算量。该设计使得在早期层进行激进剪枝的同时仍能保留关键视觉线索。在 Qwen2-VL、Qwen2.5-VL 和 Qwen3-VL 上的实验表明，Hi-Lo Prune 在多个基准上持续优于现有剪枝方法，即使在高剪枝比例下也能在无需任何微调的情况下取得强劲性能。
英文原文摘要 1874 字符
Multimodal Large Language Models (MLLMs) have achieved remarkable progress in vision-language understanding, yet processing long visual token sequences remains computationally expensive. Existing approaches mitigate this cost by reducing image tokens, either by discarding them after the visual encoder or by pruning them in the early Transformer layers of the LLM. While these strategies improve efficiency, they inevitably discard informative visual content and risk degrading downstream reasoning performance. To address this challenge, we introduce Hi-Lo Prune, a training-free pruning strategy that is built on a simple principle: look at what you will lose before you remove it. Instead of directly dropping tokens, Hi-Lo Prune first identifies which tokens are safe to prune through a coarse-to-fine selection process, and then encourages the model to absorb their information before pruning occurs. Specifically, the framework consists of three stages: (1) Hierarchical Pruning Token Selection. After visual encoding, we apply a coarse-to-fine process that identifies tokens to retain and selects a critical set of pruning candidates from redundant ones. (2) Attention-Guided Candidate Token Merge. Before removing selected tokens, an attention mechanism is applied to the early LLM layers, which explicitly transfers information from these candidates to the retained tokens. (3) Low-informative Candidate Token Removal. At a designated Transformer layer, the pruned tokens are removed, reducing computation for all subsequent layers. This design enables aggressive early-layer pruning while preserving critical visual cues. Experiments on Qwen2-VL, Qwen2.5-VL, and Qwen3-VL demonstrate that Hi-Lo Prune consistently outperforms existing pruning methods across multiple benchmarks, achieving strong performance even under high pruning ratios without any fine-tuning.