Adaptive Pruning of Pretrained Transformer via Differential Inclusions
This work addresses the computational inefficiency of existing pruning methods for transformers, offering a more flexible and customizable approach for model compression in AI applications.
The paper tackles the problem of compressing large transformers to reduce inference costs by introducing SPP, a method that prunes pretrained transformers at any desired compression ratio in a single stage, achieving competitive performance across various backbones.
Large transformers have demonstrated remarkable success, making it necessary to compress these models to reduce inference costs while preserving their perfor-mance. Current compression algorithms prune transformers at fixed compression ratios, requiring a unique pruning process for each ratio, which results in high computational costs. In contrast, we propose pruning of pretrained transformers at any desired ratio within a single pruning stage, based on a differential inclusion for a mask parameter. This dynamic can generate the whole regularization solution path of the mask parameter, whose support set identifies the network structure. Therefore, the solution path identifies a Transformer weight family with various sparsity levels, offering greater flexibility and customization. In this paper, we introduce such an effective pruning method, termed SPP (Solution Path Pruning). To achieve effective pruning, we segment the transformers into paired modules, including query-key pairs, value-projection pairs, and sequential linear layers, and apply low-rank compression to these pairs, maintaining the output structure while enabling structural compression within the inner states. Extensive experiments conducted on various well-known transformer backbones have demonstrated the efficacy of SPP.