GLASS：用于LLM推理时稀疏化的全局-局部聚合
GLASS: Global-Local Aggregation for Inference-time Sparsification of LLMs
AI总结：
GLASS通过融合全局和局部信号提升LLM推理时的稀疏化效果，尤其在短提示和长生成场景中表现优异，显著降低困惑度和KL散度，同时加快设备解码速度。
摘要：
推理时的稀疏化是将大语言模型部署到资源受限设备的有希望路径，但现有无训练方法通常仅从输入提示估计前馈网络（FFN）神经元的重要性。我们发现仅基于提示的信号往往不可靠，尤其是在短提示和长形式解码中，导致不准确的掩码和降质的生成保真度。我们提出GLASS，一种即插即用、无训练的框架，通过聚合两个互补的神经元关键性视角——局部提示特定的激活和全局模型内在先验，稳定动态FFN剪枝。GLASS通过秩聚合融合全局和局部信号，即使在提示较短时也能获得稳健的神经元选择。我们将GLASS解释为在基于排列的概率模型下的最大后验共识排名，为其加权秩聚合规则提供了原理基础。我们将其应用于多种开源LLM，并证明其在挑战性的短提示、长生成场景中优于现有无训练基线，实现困惑度降低达45.10%，KL散度降低25.73%，同时提供显著的设备解码加速。
英文摘要：
Inference-time sparsification is a promising path to deploy large language models (LLMs) on resource-constrained devices, yet existing training-free methods typically estimate feedforward network (FFN) neuron importance from the input prompt alone. We show this prompt-only signal is often unreliable, especially for short prompts and long-form decoding, leading to inaccurate masks and degraded generation fidelity. We propose GLASS, a plug-and-play, training-free framework that stabilizes dynamic FFN pruning by aggregating two complementary views of neuron criticality: local prompt-specific activations and a global model-intrinsic prior. GLASS fuses global and local signals via rank aggregation, yielding robust critical-neuron selection even when the prompt is short. We interpret GLASS as the maximum-a-posteriori consensus ranking under a permutation-based probabilistic model, providing a principled foundation for its weighted rank-aggregation rule. We apply GLASS to a diverse set of open-source LLMs, and show that it yields substantial improvements over prior training-free baselines in the challenging short-prompt, long-generation scenarios, achieving up to 45.10% lower perplexity and 25.73% lower KL divergence, while delivering significant on-device decoding speedup.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
