arXiv · 2512.03870
Reconstructing KV Caches with Cross-layer Fusion For Enhanced Transformers
Abstract
Transformer decoders have achieved strong results across tasks, but the memory required for the KV cache becomes prohibitive at long sequence lengths. Although Cross-layer KV Cache sharing (e.g., YOCO, CLA) offers a path to mitigate KV Cache bottleneck, it typically underperforms within-layer methods like GQA. To understand the root cause, we investigate the information flow of keys and values of the top-layers. Our preliminary reveals a clear distribution: values are predominantly derived from the bottom layer, while keys draw more information from both bottom and middle layers. Building upon this, we propose FusedKV, whose top-layer KV caches are a learnable fusion of the most informative ones from the bottom and middle layers. This fusion operates directly on post-RoPE keys, preserving relative positional information without the computational cost of re-applying rotary embeddings. To further improve efficiency, we propose FusedKV-Lite, an cross-layer sharing approach, where top-layer KV caches are directly derived from the bottom-layer values and the middle-layer keys. Compared to FusedKV, FusedKV-Lite reduces I/O overhead at the cost of a slight increase in perplexity. In experiments on LLMs ranging from 332M to 4B parameters, our proposed method reduce 50\% cache memory while achieving lower validation perplexity than the standard Transformer decoder, establishing it as a memory-efficient, high-performance architectural alternative.
Explore related subjects
Keep this discovery
Explore connections, maps & timelines
Hongzhan Lin, Zhiqi Bai, Xinmiao Zhang, Sen Yang, Xiang Li, Siran Yang, Yunlong Xu, Jiaheng Liu, Yongchi Zhao, Jiamang Wang, Yuchi Xu, Wenbo Su, Bo Zheng. 2026-02-19. Reconstructing KV Caches with Cross-layer Fusion For Enhanced Transformers. https://arxiv.org/abs/2512.03870
Cite the original work for its findings. Save a collection to share your selection of sources.