Searcharxiv⌕ Search

arXiv subjects

Yuanhang Gao

Publications and source records attributed to Yuanhang Gao.

3 recordsLinked to original sources

GaugeRoPE: RoPE-Compatible Executable Latent Attention

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 $5.85\times$ KV compression while maintaining quality close to the full-cache baseline. Additional MHA results validate the construction beyond GQA. GaugeRoPE achieves up to $2.48\times$ the decode throughput of the vanilla uncompressed baseline through direct latent execution.

cs.LG↗

vToken: Token-Level Virtualization for Reclaimable KV Caches

Large language model serving faces a critical memory bottleneck: the KV cache grows with sequence length and batch size. PagedAttention uses fixed-size memory blocks to reduce allocator-level fragmentation, but recent KV eviction algorithms operate at a token granularity finer than block-level management. This mismatch causes intra-block fragmentation, leaving a large fraction of allocated KV memory unreclaimable. We present vToken, a lightweight token-level virtualization layer that decouples logical token liveness from physical block placement. vToken maintains a stable logical token view through token-table indirection and realizes physical reclamation by repacking live tokens asynchronously. The design preserves PagedAttention kernels and CUDA Graph compatibility. We implement vToken in vLLM and evaluate it with H2O, Random, and Scissorhands across models. Compared with a paired Naive-Evict baseline, vToken reduces retained KV blocks per request by 27.2\%--72.3\% and improves SLA-constrained throughput by up to 1.37$\times$. Under a constrained active-KV budget, it extends the maximum feasible concurrency by up to 2$\times$, while reducing the per-policy integration footprint from 500+ lines to under 50.

cs.AI↗

ALL-E: Aesthetics-guided Low-light Image Enhancement

Evaluating the performance of low-light image enhancement (LLE) is highly subjective, thus making integrating human preferences into image enhancement a necessity. Existing methods fail to consider this and present a series of potentially valid heuristic criteria for training enhancement models. In this paper, we propose a new paradigm, i.e., aesthetics-guided low-light image enhancement (ALL-E), which introduces aesthetic preferences to LLE and motivates training in a reinforcement learning framework with an aesthetic reward. Each pixel, functioning as an agent, refines itself by recursive actions, i.e., its corresponding adjustment curve is estimated sequentially. Extensive experiments show that integrating aesthetic assessment improves both subjective experience and objective evaluation. Our results on various benchmarks demonstrate the superiority of ALL-E over state-of-the-art methods.

cs.CV↗