arXiv subjects

Zhaokai Luo

Publications and source records attributed to Zhaokai Luo.

8 recordsLinked to original sources

arXiv2026

RedKnot-MLA: Multi-Head Offline-Online Reuse for DeepSeek-V4 Long-Context Serving

Multi-head latent attention (MLA) exposes many logical query heads through one packed latent KV stream. This representation is memory efficient, but it removes the physical per-head cache boundary assumed by conventional head-wise reuse. We present our system, a DeepSeek-V4 realization of RedKnot's head-aware reuse principle. Each immutable document is processed offline at canonical position zero; certified Local-head contributions are retained as MLA-Off. At serving time, query-side RoPE relocation restores the document's request position, a small Global-head set and protected Local token rows are recomputed as MLA-Online, and the two paths are merged before a single shared output projection. The packed MLA latent is never split. DeepSeek-V4-Flash uses 37 reusable layers and a 56/8 Local/Global partition, giving a 75.29% analytic logical head-row ceiling; the Pro-0813 profile uses 55 layers and 112/16 heads, giving 78.89%. Frozen Flash operating points show hot-artifact TTFT speedups of 2.02-3.84x. At 256K, the archived three-dataset study reports an aggregate F1 change of +3.24 percentage points, an EM change of +4.16 points, and a 78.7-79.5% analytic major-operator arithmetic saving, while one dataset decreases by 2.81 F1 points. A separate author-reported 256K hot-artifact QPS measurement is approximately 2.0x; because its raw concurrency trace is not included in this bundle, we mark it as preliminary rather than archived evidence. We describe the factorization, position repair, token-row closure, sparse-MoE support, TP8 integration, and the measurement boundaries needed to interpret these results.

cs.AI↗

arXiv2026

AtomRec: Evolving Atomic Memory for Agentic Recommendation

Agentic recommender systems use large language models to maintain semantic memory and support evidence-aware recommendation. However, existing memory mechanisms often compress user and item information into coarse summaries and connect them with scalar collaborative links, making it difficult to preserve fine-grained preference stages or retrieve interpretable evidence as user interests evolve. We propose \textsc{AtomRec}, an agentic recommender with evolving atomic collaborative memory. \textsc{AtomRec} represents user and item memories as structured atomic units, builds semantic links across related memories, and evolves related historical fields when new interactions arrive. During recommendation, it retrieves linked memories as multi-hop evidence paths rather than isolated neighbor summaries, allowing collaborative signals to support grounded ranking. Experiments on four public benchmarks show that \textsc{AtomRec} consistently outperforms state-of-the-art agentic and memory-augmented baselines, with around 8.5\% average relative improvement across metrics.

cs.IR↗

arXiv2026

PILOT in the Loop: Live Self-Improvement for Long-Horizon Agents

Long-horizon agent runs generate experience that can improve both the current run and future work. Most self-improvement methods process this experience only after execution ends, so they cannot redirect the active run or immediately apply and validate lessons learned from it. We argue that self-improvement should instead be live, using emerging experience both to redirect the active run and to update the persistent harness. Existing agent architectures do not fully support this goal. Single-agent self-correction combines task execution and trajectory assessment within one context, while subagent delegation separates execution but typically cannot redirect an active subagent. We present PILOT, a supervisor-worker harness for live self-improvement through two coupled mechanisms: (1) live steering lets a separate supervisor redirect or abort the active worker during execution; and (2) live self-evolution distils procedures and failure modes revealed during execution into reusable skills and memory. Across two frozen backbones and three benchmarks, PILOT ranks first in five of six configurations. On Terminal-Bench 2.0, PILOT outperforms counterpart harnesses by up to 9.8 percentage points. In the self-improvement setting, PILOT gains 14.6 points with GLM-5.1 and 12.4 points with Kimi-K2.6. Mean output tokens fall by 42.9% and 47.4%, while successful evaluations per million output tokens rise by 110.3% and 134.0%, respectively.

cs.AI↗

arXiv2026

OneModel: A Unified Foundation for Platform-Scale Multi-Scenario Ranking

Platform-scale recommender systems often span multiple business streams such as organic recommendation, advertising, and merchant services, where user behaviors form a continuous cross-stream trajectory. Maintaining separate ranking systems fragments user representations and increases engineering cost. We propose \textbf{OneModel}, a unified framework for multi-stream final ranking. OneModel maps heterogeneous behaviors into shared event sequences, learns long-context user representations with an action-oriented backbone, and introduces \emph{Scenario-aware Information Modulation} to balance cross-stream transfer and stream-specific specialization. For production deployment, OneModel further adopts stratified user representation, multi-objective training, and optimized online serving with feature decomposition, user feature prefetching, shared user-tower computation, and graph-level inference optimization. We deploy OneModel in production at \emph{Xiaohongshu}, where it delivers consistent offline gains over strong baselines and scales favorably with context length and model capacity. Online A/B tests improve Time Spent by \textbf{+0.33\%} and Engagement by \textbf{+1.25\%} in Explore Feed, lift advertising value by \textbf{+3.43\%} and CTR by \textbf{+8.18\%} in Feed Advertising, and raise DGMV by \textbf{+1.1867\%} and GPM by \textbf{+2.1585\%} in Merchant Recommendation, validating unified multi-stream ranking as an effective production foundation.

cs.IR↗

arXiv2026

Hierarchical Latent Reasoning for LLM-based Recommendation

Large Language Models (LLMs) have shown strong potential for recommendation by leveraging their semantic understanding and contextual modeling capabilities. Recent studies further introduce reasoning mechanisms to improve user preference modeling. However, explicit natural-language reasoning incurs substantial inference overhead, whereas existing latent reasoning methods mainly focus on generating or verifying intermediate states, leaving their layer-wise preference roles and contributions insufficiently characterized. We propose HiLaR, a Hierarchical Latent Reasoning framework with layer-aware reinforcement optimization for LLM-based recommendation. HiLaR constructs temporal-guided hierarchical user preference representations, aligns them with multiple LLM latent reasoning states, and organizes the reasoning process from broad preferences to fine-grained current intents. To further optimize the reasoning trajectory, HiLaR combines final recommendation feedback with layer-aware process rewards derived from the marginal target-likelihood gain of each state. Experiments on four Amazon benchmark datasets show that HiLaR generally outperforms strong sequential, generative, and LLM-based recommendation baselines. Ablation and sensitivity analyses further verify the contribution of hierarchical representation learning, latent alignment, and process-level optimization. Our code is available in https://github.com/hupeiyu21/HiLaR.

cs.IR↗

arXiv2026

FlowBlock: Wavefront-Parallel Decoding for Self-Correcting Diffusion Language Models

Block-wise diffusion large language models (dLLMs) decode sequentially at the block level, enabling effective KV-cache reuse across blocks but making inter-block decoding strictly serial. Prior work has attempted to unlock inter-block parallelism through post-training methods, but achieves only modest speedups and often degrades accuracy. We observe that self-correcting dLLMs offer a training-free alternative: token-to-token (T2T) editing can repair tokens drafted with a slightly stale upstream context, so a downstream block requires only an informative draft rather than a finalized predecessor. This turns block finality from a hard dependency into a scheduling resource. We propose \textbf{\flowblock{}}, a training-free parallel decoding framework built on two mechanisms. (i) \emph{Gated Wavefront Decoding} admits blocks into a bounded wavefront only when a readiness gate is satisfied, jointly refines active blocks via T2T editing, and commits blocks in order under a windowed block-causal mask that preserves exact frozen-prefix KV caches reuse. (ii) \emph{Heterogeneous Wavefront Packing} assigns each request an independent wavefront while packing asynchronous windows into dense, shape-stable batched forwards. Across different benchmarks, \flowblock{} improves tokens per second (TPS) over LLaDA-2.1 and LLaDA-2.0, two serial block-wise dLLMs, by up to 2.95$\times$ and 4.01$\times$, while reducing latency by up to 53.6\% and 77.1\%, respectively. It also improves average accuracy by 1.3 points. Compared with D2F, a training-based inter-block-parallel baseline, \flowblock{} achieves higher accuracy and up to 16$\times$ higher batched serving throughput.

cs.AI↗

arXiv2026

Akashic: A Low-Overhead LLM Inference Service with MemAttention

Recent LLM-based agent systems continuously accumulate context across multi-turn interactions, tool invocations, and cross-session workflows. Replaying the full history for every request quickly becomes impractical: long contexts increase prefill cost, may exceed context limits, and often bury task-relevant evidence in irrelevant content, degrading both serving efficiency and output quality. We propose Akashic, a low-overhead memory system built around MemAttention, which organizes context into bounded chunks and models semantic relationships across chunks, preserving cross-chunk evidence without repeatedly rewriting the full history. Akashic further applies hardware-software co-designed memory placement to co-locate likely co-retrieved chunks, reducing retrieval fragmentation and I/O overhead. Across four representative workloads and three model sizes, Akashic improves task accuracy by up to 10.2 points, throughput by up to 1.21x, and sustainable request rate by up to 1.88x over strong prior memory baselines.

cs.AI↗

arXiv2026

RedKnot: Efficient Long-Context LLM Serving with Head-Aware KV Reuse and SegPagedAttention

As the input length of large language model (LLM) serving continues to grow, the KV cache has become a dominant bottleneck in AI infrastructure. It limits GPU memory capacity, serving concurrency, cache reuse, and distributed scalability. Multiple important problems, including position-independent KV cache, prefix KV cache compression, hot/cold KV cache separation, and distributed KV cache management, all depend on how the KV cache is represented and managed. However, existing serving systems largely rely on a monolithic KV cache abstraction, where the KV cache is treated as a homogeneous sequence of token-level memory blocks and managed with similar policies across attention heads and serving scenarios. We observe that KV cache utility is highly structured across KV heads: different heads exhibit different functional roles, attention distances, and runtime importance. Therefore, a full KV cache is not always necessary for every head, token range, or serving scenario. We present RedKnot, a head-aware KV cache management system for LLM serving. RedKnot breaks the conventional monolithic KV cache abstraction by decomposing the KV cache along KV heads, whose importance and effective attention ranges vary significantly across serving scenarios. This head-level decomposition turns the KV cache from a monolithic tensor abstraction into a structured memory object, enabling RedKnot to uniformly support position-independent KV reuse, prefix KV compression, hot/cold KV separation, and distributed KV placement while preserving output fidelity and improving resource efficiency, without requiring model retraining or fine-tuning. RedKnot establishes a new foundation for AI infrastructure by transforming the KV cache from a monolithic, passive runtime artifact into a dynamic, model-aware runtime substrate for scalable LLM serving.

cs.AI↗