arXiv subjects

Xinpeng Zhao

Publications and source records attributed to Xinpeng Zhao.

8 recordsLinked to original sources

arXiv2026

Decoupling Disaggregated Memory Optimizations from Indexing: A Compiler-Runtime Approach

Disaggregated memory (DM) decouples compute and memory into independently scalable pools, connected over a slower interconnect rather than a local bus. This decoupling is exactly what makes DM attractive--but it also means that every index must now reason explicitly about remote-memory access and its associated optimizations. State-of-the-art index designs respond to this by embedding remote-memory logic and optimizations directly into their core data structures and concurrency control mechanisms. Consequently, an optimization tuned for one index cannot be lifted and reused in another, and even the same index cannot be ported to a different DM architecture without a fresh round of redesign. This escalating, per-index, per-platform engineering burden is unsustainable as hardware and index requirements evolve. In this paper, we present Nox, a compiler-runtime framework that breaks this coupling by taking an unmodified, concurrent index as input and automatically generating its disaggregated-memory counterpart, without touching the original index logic. A compiler layer rewrites the index's LLVM IR to expose allocation, address, and pointer-dependency information that a centralized runtime uses to drive caching and address translation. Empirically, Nox-generated B+-trees, hash tables, and skip lists scale robustly on real RDMA and CXL hardware across every workload tested. They can also outperform some specialized, hand-crafted indexes and match others, especially on workloads that are closer to real-world ones. These results show that today's fastest disaggregated-memory optimizations need not stay locked inside monolithic, hand-crafted code--a compiler-runtime stack can generalize them while preserving the scalability of proven index implementations, without sacrificing it for portability.

cs.DB↗

arXiv2026

Token-Level Credit Assignment Optimization for Generative Document Retrieval

Generative retrieval models perform document retrieval by autoregressively generating document identifiers (DocIDs). This process naturally forms a sequential decision problem, i.e., the model makes a sequence of token-level decisions, selecting a DocID token at each decoding step, with the resulting complete sequence identifying the retrieved document. However, relevance feedback is available only after the complete DocID has been generated and mapped to a document, resulting in a granularity mismatch between token-level generation decisions and document-level retrieval supervision. Consequently, existing reinforcement learning methods for generative retrieval rely on sequence-level rewards, assigning the same document-level relevance signal to every decoding step. Such uniform credit assignment obscures the contribution of individual token decisions, making it difficult to identify which decisions contribute to retrieval success or failure. In this paper, we propose Token-Level Credit Assignment for Generative Retrieval (TCA), a fine-grained reinforcement learning framework that aligns the granularity of credit assignment with that of autoregressive DocID generation. Unlike assigning a single reward to an entire generated DocID, TCA derives fine-grained rewards by comparing the hidden-state trajectory of each generated DocID with the gold DocID trajectory obtained from a frozen reference model. These trajectory-based rewards provide differentiated feedback across decoding steps, allowing the policy to reinforce generation paths that remain aligned with the target DocID. Moreover, TCA decouples token-level credit assignment from policy optimization and can be instantiated with both GRPO and PPO. Experiments on benchmarks show that our method consistently outperforms baselines, demonstrating the effectiveness of fine-grained supervision for aligning DocID generation.

cs.IR↗

arXiv2026

Surviving Partial Rank Failures in Wide Expert-Parallel MoE Inference

Mixture-of-Experts (MoE) serving relies on wide expert parallelism (EP) to aggregate the memory capacity and bandwidth of many GPUs within one inference instance. This efficiency comes with a systems cost: every decoding step depends on token dispatch and combination across all active EP ranks, so even one rank failure can disrupt the entire service. Existing EP stacks handle such failures poorly because they treat membership as a fixed configuration established at initialization. The same rank set determines communicator state, expert placement, and the routing metadata baked into CUDA execution graphs, leaving the system with no way to shrink around a failure while keeping the instance valid. This paper argues that partial-failure tolerance should instead be formulated as a live EP validity problem. We present EEP, a communication and runtime substrate that represents membership as explicit, mutable runtime state. EEP repairs the specific state invalidated by a fault: it restores peer reachability without rebuilding the communication substrate, repairs lost expert coverage through a bandwidth-aware hierarchy, and reintegrates repaired ranks without forcing healthy ranks to recapture their CUDA graphs. We implement EEP in an EP serving stack integrated with SGLang and evaluate it under steady-state serving, failure recovery, and rank reintegration. The results show that explicit mutable membership preserves the steady-state fast path, staying within 4.4% of a fixed-membership DeepEP baseline under static serving, while turning a local rank fault from whole-instance downtime into two bounded interruptions. On a single-rank failure workload, EEP incurs an 11s recovery pause and an 8s reintegration pause, and restores throughput to within 95% of the pre-fault level within 52s, whereas a fixed-membership full-restart baseline remains unavailable until 348s.

cs.DC↗

arXiv2026

Pooling Engram Conditional Memory in Large Language Models using CXL

Engram conditional memory has emerged as a promising component for LLMs by decoupling static knowledge lookup from dynamic computation. Since Engram exhibits sparse access patterns and supports prefetching, its massive embedding tables are well-suited for offloading to lower-tier memory. In this paper, we propose using Compute Express Link (CXL) memory pool for Engram storage. Compared to RDMA, CXL provides fine-grained and low-latency access required by minimal and discrete retrieval patterns of Engram. We integrate the CXL-based Engram pool into SGLang, achieving near-DRAM end-to-end performance. This provides a scalable and cost-efficient storage solution for future Engram-integrated LLMs without compromising inference performance.

cs.AR↗

arXiv2026

DiffuGR: Generative Document Retrieval with Diffusion Language Models

Generative retrieval (GR) reframes document retrieval as an end-to-end task of generating sequential document identifiers (DocIDs). Existing GR methods predominantly rely on left-to-right auto-regressive decoding, which suffers from two fundamental limitations: (i) a \emph{mismatch between DocID generation and natural language generation}, whereby an incorrect DocID token generated at an early step can lead to entirely erroneous retrieval; and (ii) an \emph{inability to dynamically balance the trade-off between retrieval efficiency and accuracy}, which is crucial for practical applications. To tackle these challenges, we propose generative document retrieval with diffusion language models, termed \emph{DiffuGR}. DiffuGR formulates DocID generation as a discrete diffusion process. During training, DocIDs are corrupted through a stochastic masking process, and a diffusion language model is trained to recover them under a retrieval-aware objective. For inference, DiffuGR generates DocID tokens in parallel and refines them through a controllable number of denoising steps. Unlike auto-regressive decoding, DiffuGR introduce \emph{a novel mechanism to first generate plenty of confident DocID tokens and then refine the generation through diffusion-based denoising}. Moreover, DiffuGR also offers \emph{explicit runtime control over the quality-latency tradeoff}. Extensive experiments on widely-applied retrieval benchmarks show that DiffuGR outperforms strong auto-regressive generative retrievers. Additionally, we verify that DiffuGR achieves flexible control over the quality-latency trade-off via variable denoising budgets.

cs.IR↗

arXiv2025

CPCL: Cross-Modal Prototypical Contrastive Learning for Weakly Supervised Text-based Person Retrieval

Weakly supervised text-based person retrieval seeks to retrieve images of a target person using textual descriptions, without relying on identity annotations and is more challenging and practical. The primary challenge is the intra-class differences, encompassing intra-modal feature variations and cross-modal semantic gaps. Prior works have focused on instance-level samples and ignored prototypical features of each person which are intrinsic and invariant. Toward this, we propose a Cross-Modal Prototypical Contrastive Learning (CPCL) method. In practice, the CPCL introduces the CLIP model to weakly supervised text-based person retrieval to map visual and textual instances into a shared latent space. Subsequently, the proposed Prototypical Multi-modal Memory (PMM) module captures associations between heterogeneous modalities of image-text pairs belonging to the same person through the Hybrid Cross-modal Matching (HCM) module in a many-to-many mapping fashion. Moreover, the Outlier Pseudo Label Mining (OPLM) module further distinguishes valuable outlier samples from each modality, enhancing the creation of more reliable clusters by mining implicit relationships between image-text pairs. We conduct extensive experiments on popular benchmarks of weakly supervised text-based person retrieval, which validate the effectiveness, generalizability of CPCL.

cs.CV↗

arXiv2024

Temporal-Spatial Object Relations Modeling for Vision-and-Language Navigation

Vision-and-Language Navigation (VLN) is a challenging task where an agent is required to navigate to a natural language described location via vision observations. The navigation abilities of the agent can be enhanced by the relations between objects, which are usually learned using internal objects or external datasets. The relationships between internal objects are modeled employing graph convolutional network (GCN) in traditional studies. However, GCN tends to be shallow, limiting its modeling ability. To address this issue, we utilize a cross attention mechanism to learn the connections between objects over a trajectory, which takes temporal continuity into account, termed as Temporal Object Relations (TOR). The external datasets have a gap with the navigation environment, leading to inaccurate modeling of relations. To avoid this problem, we construct object connections based on observations from all viewpoints in the navigational environment, which ensures complete spatial coverage and eliminates the gap, called Spatial Object Relations (SOR). Additionally, we observe that agents may repeatedly visit the same location during navigation, significantly hindering their performance. For resolving this matter, we introduce the Turning Back Penalty (TBP) loss function, which penalizes the agent's repetitive visiting behavior, substantially reducing the navigational distance. Experimental results on the REVERIE, SOON, and R2R datasets demonstrate the effectiveness of the proposed method.

cs.CV↗

arXiv2020

Tunable Anisotropic Thermal Transport in Super-Aligned Carbon Nanotube Films

Super-aligned carbon nanotube (CNT) films have intriguing anisotropic thermal transport properties due to the anisotropic nature of individual nanotubes and the important role of nanotube alignment. However, the relationship between the alignment and the anisotropic thermal conductivities was not well understood due to the challenges in both the preparation of high-quality super-aligned CNT film samples and the thermal characterization of such highly anisotropic and porous thin films. Here, super-aligned CNT films with different alignment configurations are designed and their anisotropic thermal conductivities are measured using time-domain thermoreflectance (TDTR) with an elliptical-beam approach. The results suggest that the alignment configuration could tune the cross-plane thermal conductivity k_z from 6.4 to 1.5 W/mK and the in-plane anisotropic ratio from 1.2 to 13.5. This work confirms the important role of CNT alignment in tuning the thermal transport properties of super-aligned CNT films and provides an efficient way to design thermally anisotropic films for thermal management.

physics.app-ph↗