Papers › A Sliding Layer Merging Method for Efficient Depth-Wise Pruning in LLMs

A Sliding Layer Merging Method for Efficient Depth-Wise Pruning in LLMs

26 Feb 2025arXiv:2502.19159archive 2025-07-28

Xuan Ding, Rui Sun, Yunjian Zhang, Xiu Yan, Yueqi Zhou, Kaihao Huang, Suzhong Fu, Angelica I Aviles-Rivero, Chuanlong Xie, Yao Zhu

Compared to width-wise pruning, depth-wise pruning can significantly accelerate inference in resource-constrained scenarios. However, treating the entire Transformer layer as the minimum pruning unit may degrade model performance by indiscriminately discarding the entire information of the layer. This paper reveals the ``Patch-like'' feature relationship between layers in large language models by analyzing the correlation of the outputs of different layers in the reproducing kernel Hilbert space. Building on this observation, we propose a sliding layer merging method that dynamically selects and fuses consecutive layers from top to bottom according to a pre-defined similarity threshold, thereby simplifying the model structure while maintaining its performance. Extensive experiments on LLMs with various architectures and different parameter scales show that our method outperforms existing pruning techniques in both zero-shot inference performance and retraining recovery quality after pruning. In particular, in the experiment with 35% pruning on the Vicuna-7B model, our method achieved a 1.654% improvement in average performance on zero-shot tasks compared to the existing method. Moreover, we further reveal the potential of combining depth pruning with width pruning to enhance the pruning effect. Our codes are available at https://github.com/920927/SLM-a-sliding-layer-merging-method.

PaperPDFCode

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

920927/slm-a-sliding-layer-merging-method officialmentioned in paperpytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Methods

Absolute Position EncodingsAdamAttentionBPEDense ConnectionsDropoutLabel SmoothingLayer NormalizationLinear LayerMulti-Head AttentionPosition-Wise Feed-Forward LayerPruningResidual ConnectionSoftmaxTransformer

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections