Saved in:
Bibliographic Details
Main Authors: Xiong, Yizhe, Huang, Wei, Ye, Xin, Chen, Hui, Lin, Zijia, Lian, Haoran, Su, Zhenpeng, Han, Jungong, Ding, Guiguang
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2502.00439
Tags: Add Tag
No Tags, Be the first to tag this record!
Table of Contents:
  • Post-training is essential for adapting Large Language Models (LLMs) to real-world applications. Deploying post-trained models faces significant challenges due to substantial memory overhead and noticeable inference latency. Existing work has identified significant redundancies in LLMs and proposed efficient architectures, namely intra-layer KV sharing and cross-layer KV sharing. However, these methods still result in high inference time overhead, remaining suboptimal for post-training pre-trained LLMs. In this paper, we identify that the \texttt{Softmax} operation is a primary bottleneck for LLM inference and discover that it is actually highly redundant during post-training. We propose Softmax \textbf{Uni}fication in \textbf{Att}e\textbf{n}tion (\textbf{UniAttn}), a novel post-training method that unifies Softmax activations across transformer blocks to reduce LLM inference costs. Additionally, UniAttn adopts a linear projection to compensate for the errors induced by Softmax unification. Experiments show that UniAttn matches the performance of standard post-training while significantly reducing inference costs, outperforming existing efficient architectures during post-training.