最新研究发现：大模型层剪枝并非无损，反而会破坏测试时扩展能力的链式推理，值得关注。
arXiv:2510.22228v2 Announce Type: replace-cross Abstract: Layer pruning has emerged as a widely adopted technique for improving the efficiency of larg…
研究修剪注意力层如何影响LLM的解释忠实性与置信度校准，揭示模型优化新视角。
arXiv:2606.24970v1 Announce Type: new Abstract: Pruning Large Language Models (LLMs) reduces memory and inference costs by removing parts of the netwo…
提出拓扑感知层剪枝方法，有效降低大视觉语言模型的计算与内存成本，助力高效部署。
arXiv:2604.16502v2 Announce Type: replace Abstract: Large Language Models (LLMs) have demonstrated strong capabilities in natural language understandi…
提出Ghosted Layers，无需训练即可恢复层剪枝后LLM的性能，通过激活对齐解决隐藏状态不匹配问题。
arXiv:2605.15491v1 Announce Type: cross Abstract: Layer pruning removes entire Transformer decoder blocks from large language models, but introduces a…