cs.LGNov 16, 2026

GaugeRoPE: RoPE-Compatible Executable Latent Attention

Authors: Yuanhang Gao, Xiangrui Yang, Lingyuan Meng, Yuanfeng Chen, Dongsheng Li

Organizations: National University of Defense Technology

Abstract

Latent KV cache compression reduces memory footprint by mapping historical keys and values into compact representations. However, generic compressed key states require reconstruction to native dimension before applying RoPE and computing attention scores, adding sequence-scaling overhead to the decode path. We characterize when a latent KV representation can serve directly as the execution state of pretrained RoPE. For RoPE-based transformers, exact static linear execution of the rotary key path requires the latent-to-native map to preserve the pretrained RoPE action through an intertwining relation, restricting exact mixing to coordinates that share the same RoPE frequency. Guided by this characterization, we build GaugeRoPE, a training-free executable latent KV representation. It combines a RoPE-compatible carrier that preserves the rotary action with an approximate position-independent residual key--value (RKV) correction for complementary key and value information, both evaluated without per-token dense historical K/V reconstruction. On Mistral-7B, GaugeRoPE achieves 5.85×5.85\times KV compression while maintaining quality close to the full-cache baseline. Additional MHA results validate the construction beyond GQA. GaugeRoPE achieves up to 2.48×2.48\times the decode throughput of the vanilla uncompressed baseline through direct latent execution.

Figures & tables

Appendix figures & tables7 assets

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

CardsList
  1. RACE: RoPE-Aligned Attention Compression Based on Query-Key Activation Energies

    Oct 20, 2026Irene Tenison, Abdulrahman Alabdulkareem, Daniel Karl I. Weidele +2KV-Cache CompressionTransformer Attention

  2. Block-GTQ: RoPE-Block Rate Allocation for KV-Cache Quantization

    Jun 23, 2026Fengfeng Liang, Yuechen Zhang, Jiaya JiaKV CachingLLM Inference Acceleration