arXiv subjects

Zhen Li

Publications and source records attributed to Zhen Li.

At least 55 records · Page 3Linked to original sources

arXiv2026

CodeHID: Learning an Addressable Hierarchical Code Index for Generative Code Retrieval

Code retrieval models have predominantly relied on a flat matching paradigm that treats code snippets as independent candidates, making them less capable of distinguishing similar code candidates. Generative retrieval offers a solution by constructing a learnable index over the code corpus, guiding the retriever to better understand how code candidates are semantically organized and addressed. However, naively applying generative retrieval in the code retrieval task may result in operating over an identifier space whose prefixes do not correspond to meaningful code-semantic regions. In this paper, we propose CodeHID, a generative code retrieval framework that reformulates the code retrieval task from flat candidate matching to coarse-to-fine semantic address generation. CodeHID relies on two core components. First, Pseudo-Neighbor Guided DocID Learning constructs a globally static hierarchical index by applying multi-level residual quantization and $k$-nearest-neighbor pseudo-supervision, ensuring that semantically related code snippets share prefixes while preserving target-level separability. Second, Dual-Phase DocID Generation Guidance reliably navigates this fixed index by combining training-side ranking enhancements, using hard negatives and rank distillation, with inference-side candidate constraints and prefix-aware decoding. Extensive experiments on CoSQA and ProCQA benchmarks demonstrate that CodeHID outperforms existing sparse retrieval, pre-trained code models, dense code retrieval, and generative retrieval baselines by a large margin in most cases, achieving particularly strong improvements in rank-one retrieval metrics.

cs.IR↗

arXiv2026

Observational Signatures of Thin Accretion Disks around Rotating Black Holes Embedded in Dark Matter Halos

Black holes embedded in a dark matter background represent an important candidate model that extends the standard vacuum case, characterizing the strong gravitational environments at galactic centers and the gravitational influence of dark matter. This study focuses on dark matter halo-modified Kerr black holes and systematically investigates the observational characteristics of their surrounding thin accretion disks. The dark matter halo is modeled by a general double power-law density profile, whose integrated mass $M_D(r)$ modifies the radial mass function of Kerr metric to $m(r) = M + M_D(r)$. Based on this spacetime, we analyze the observational properties of their thin accretion disk, including the radiative flux, temperature, differential luminosity, and spectral luminosity as a function of radius, with respect to the black hole spin and dark matter halo parameters. Furthermore, by combining ray-tracing methods, we simulate the bolometric image of the thin accretion disk under different black hole spins, dark matter halo parameters, and viewing inclinations. The dark matter halo enhances the gravitational potential, shifting the innermost stable circular orbit (ISCO) of the accretion disk outward, suppressing the peak radiative flux, and producing a fainter image, thereby generating signatures that are possibly observational distinguishable from standard vacuum Kerr black holes.

gr-qc↗

arXiv2026

Broadband phonon-velocity suppression and a finite anisotropic crossover in twisted bilayer SnSe

Moiré superlattices reshape lattice dynamics without altering chemical composition, yet how crystal anisotropy modifies this control remains unclear. We combine density-functional-theory (DFT)-calibrated lattice-dynamical calculations with angle-matched untwisted controls to study puckered bilayer SnSe across seven commensurate twist angles ($3.18^\circ$--$8.77^\circ$). At 300 K, twisting suppresses the band-path heat-capacity-weighted mean-square group velocity to 2.6--8.4\% of the control values; the suppression spans a broad frequency range rather than a few soft branches. The velocity response crosses over between $4.78^\circ$ and $3.82^\circ$ into a regime where the relaxed stacking textures and frequency-resolved velocity profiles become self-similar, with the normalized mean-square velocity ratio spanning only 11.1\% of its mean across the three smallest angles---a finite anisotropic crossover, not a singular-angle condition. Direct DFT--MACE force-constant agreement ($r=0.996$), uniform $4\times4\times1$ stability scans, and acoustic-sum-rule and path-density tests support the trend. The equilibrium trend is defined by six structures after excluding one relaxation-sensitive case. These results extend phonon twistronics to low-symmetry layered materials and identify crystal anisotropy as a key determinant of finite-angle phonon crossover behavior.

cond-mat.mtrl-sci↗

arXiv2026

InfiniVerse: Occupancy Guided Unbounded Scene Generation for Autonomous Driving

Generating realistic, controllable, and temporally coherent urban environments is a critical yet unresolved challenge in the autonomous driving community. In this paper, we introduce InfiniVerse, a unified pipeline for long-range, 2D-3D-aligned, and controllable synthesis of dynamic urban scenes from a single frame. In practice, our approach first reconstructs a 3D occupancy representation from the input multi-view frame. This representation serves as a foundation for autoregressive scene extension along arbitrary trajectories. Subsequently, a video diffusion model translates the coarse occupancy grid into realistic, spatiotemporally consistent video sequences. Moreover, we propose a hierarchical sketch-and-refine paradigm, in which the generated videos are re-projected as image-conditioned feedback to enhance the 3D occupancy representation, establishing cross-modal alignment and mutual enhancement between the visual and spatial domains. Extensive evaluations on the Waymo Open Dataset and nuScenes demonstrate that InfiniVerse achieves state-of-the-art performance, with a FID of 6.4 and FVD of 67.97, significantly outperforming existing benchmarks in both duration and stability.

cs.CV↗

arXiv2026

Marionette: Predicting World States, Rendering Geometry, Painting Appearance

Interactive game world models typically autoregress visual observations directly in pixel or latent space, forcing structured properties such as pose, geometry, and occlusion to be implicitly maintained by the same generative sequence. Over long horizons, errors in these latent world properties accumulate, making consistency and controllability fragile. We explicitly model the evolving world state, delegate exact geometric computation to a fixed, zero-parameter renderer, and leave the neural model to synthesize appearance. We instantiate this idea as Marionette, a world model for interactive games with articulated characters. First, a two-stage autoregressive dynamics model predicts an explicit and interpretable 276-dimensional 3D world state comprising multi-entity articulated skeletons, metric root trajectories, and rotations. Second, a zero-parameter graphics bridge converts the predicted state into pose-control videos, computing world-space geometry and occlusion in closed form. Third, a control-conditioned video-diffusion observation model synthesizes photorealistic RGB observations from the resulting structured controls. Our experiments establish two properties of Marionette. First, the predicted world state is directly controllable. Forcing a mismatched action stream changes root-aligned joint error by 31% across 48 held-out segments. Second, long-horizon behaviour is determined in the state, and can be repaired there. Left free, the two generated characters drift to 21.2 m apart (recorded sessions stay near 5 m) and a third of frames show ground penetration. Two rules imposed on the explicit state, a terrain collider and a separation cap, cut penetration by 66% and keep the pair engaged, with no change to the observation model. Routing appearance through the predicted state costs no fidelity we can detect, at an FVD of 831 against 799 for recorded pose.

cs.CV↗

arXiv2026

AlayaWorld: Interactive Long-Horizon World Modeling - Full Technical Report (v1.1)

This report presents an improved version of AlayaWorld. While the backbone architecture, chunk-wise autoregressive generation scheme, and training data remain unchanged from the previous release, we substantially revise how conditioning signals are represented and integrated into the model. The new design is guided by a simple principle: conditioning signals should match the generated content as closely as possible in both latent representation and temporal structure. To this end, we make two major changes. First, we replace the previous depth-warping-based spatial memory with a streaming 3D point-cache renderer. Second, we redesign the conditioning pipeline so that visual conditions are encoded in the same causal-VAE latent space, with temporal statistics consistent with those of the generated video. Concretely, the new version introduces six modifications: (1) replacing static-frame image conditioning with motion-aware latent conditioning; (2) causally encoding re-rendered spatial memory as a continuous sequence; (3) aligning the temporal-memory window in pixel space; (4) adopting hard memory dropout that removes memory tokens rather than zeroing them; (5) unifying the VAE encoding and decoding protocol across training and inference; and (6) removing the camera AdaLN branch, such that viewpoint control is provided entirely through the re-rendered spatial condition.

cs.AI↗

arXiv2026

Quantum-Centric Geometry Optimization with Wave-Function-Based Embedding

The EWF-(FCI,SQD) method, a wave-function-based embedding approach combining full configuration interaction (FCI) and sample-based quantum diagonalization (SQD), is a promising new tool for the simulation of molecular systems. However, applications of EWF-(FCI,SQD) have so far been limited to single-point calculations, whereas the study of complex chemical processes requires the ability to explore potential energy surfaces. In this work, we demonstrate geometry optimization with EWF-(FCI,SQD), scaling our simulations to molecules as large as menthone and benzidine within the STO-3G basis set. Without fragmentation, these systems comprise 73 and 82 molecular orbitals respectively, presenting an intractable Hilbert space for conventional exact or high-level subspace solvers and establishing a clear necessity for fragmentation-based methodologies. The underlying fragment SQD simulations in the EWF-(FCI,SQD) geometry optimizations use up to 70 qubits. The resulting geometries show exceptional accuracy relative to the classical reference, with deviations below 4 picometers.

quant-ph↗

arXiv2026

Combating Knowledge Corruption in Agent Systems: A Byzantine-Tolerant Secure Collaborative RAG Framework

While retrieval-augmented generation systems partially address the hallucination issues in large language models, it also introduces new vulnerabilities to knowledge corruption attacks. Adversaries exploit these vulnerabilities by poisoning documents provided by RAG system to manipulate LLM outputs. To counter this threat, we propose SecureCollaRAG, a Byzantine-tolerant collaborative RAG framework leveraging Multi-source Knowledge Validation Mechanism. Our approach enables agent system to securely verify document provenance through dynamic GNN-based credibility scoring, effectively preventing stealthy knowledge corruption attacks while preserving essential domain knowledge integrity. Through extensive evaluations and formal analysis, we demonstrate that SecureCollaRAG maintains robustness against attackers under non-IID data distributions.

cs.CR↗

arXiv2026

AST: Adaptive, Seamless, and Training-Free Precise Speech Editing

Text-based speech editing aims to modify specific segments while preserving speaker identity and acoustic context. Current approaches generally involve either expensive task-specific training or adapting pre-trained Text-to-Speech (TTS) models. However, both paradigms face challenges: task-specific methods often degrade fidelity in unedited regions, whereas TTS adaptations struggle with a trade-off between editing naturalness and temporal fidelity. To address these issues, we propose AST, an Adaptive, Seamless, and Training-free speech editing framework. Built upon pre-trained TTS, AST leverages Latent Recomposition to stitch preserved source segments with synthesized targets, guaranteeing fidelity in unedited regions. To break the quality-controllability trade-off, we introduce Adaptive Weak Fact Guidance (AWFG), which modulates a mel-space signal to ensure seamless boundary transitions without disrupting the generative manifold. Furthermore, to address evaluation gaps in temporal fidelity, we propose a new benchmark suite: the LibriSpeech-Edit dataset and a novel metric, Word-level Dynamic Time Warping (WDTW). Extensive experiments demonstrate that AST consistently outperforms existing task-specific and fine-tuned speech editing methods across content accuracy, perceptual quality, speaker preservation, and temporal fidelity. Remarkably, AST achieves state-of-the-art zero-shot speech editing performance without any task-specific training or paired editing data, validating the effectiveness of latent recomposition and AWFG in bridging the quality-controllability trade-off.

cs.SD↗

arXiv2026

A Comprehensive FP8 Training Recipe for Reasoning-Enhanced Language Models

The immense computational cost of training Large Language Models (LLMs) presents a major barrier to innovation. While FP8 training offers a promising solution with significant theoretical efficiency gains, its widespread adoption has been hindered by the lack of a comprehensive, open-source training recipe. To bridge this gap, we introduce an end-to-end FP8 training recipe that seamlessly integrates continual pre-training and supervised fine-tuning. Our methodology employs a fine-grained, hybrid-granularity quantization strategy to maintain numerical fidelity while maximizing computational efficiency. Through extensive experiments, including the continue pre-training of models on a 160B-token corpus, we demonstrate that our recipe is not only remarkably stable but also essentially lossless, achieving performance on par with the BF16 baseline across a suite of reasoning benchmarks. Crucially, this is achieved with substantial efficiency improvements, including up to a 22% reduction in training time, a 14% decrease in peak memory usage, and a 19% increase in throughput. Our results establish FP8 as a practical and robust alternative to BF16, and we will release the accompanying code to further democratize large-scale model training.

cs.CL↗

arXiv2026

Quantum Computing Enabled ab initio Molecular Dynamics Simulations

We demonstrate a quantum-classical workflow for ab initio molecular dynamics (AIMD) in which quantum measurements from a chemistry-inspired LUCJ ansatz are post-processed using Sample-based Quantum Diagonalization (SQD) to recover determinant subspaces and deliver energies and analytical nuclear gradients for dynamics. As an exact benchmark, we use full configuration interaction (FCI) in the STO-3G basis, enabling a direct assessment of the accuracy of SQD. In gas-phase benchmarks, SQD reproduces FCI energies and gradients to within 1 kcal mol$^{-1}$ of the FCI reference and yields stable AIMD trajectories. In explicit-solvent QM/MM simulations, SQD retains this agreement, matching FCI energy fluctuations and RMS gradient profiles and reproducing solute-solvent structure as quantified by radial distribution functions. Overall, these benchmarks establish LUCJ+SQD as a practical route for integrating current quantum hardware into QM/MM molecular dynamics and provide an early demonstration of condensed-phase QM/MM dynamics driven by a quantum electronic-structure engine.

physics.chem-ph↗

arXiv2026

Reasoning to Regulate: Chain-of-Thought for Traffic Rule Understanding

Understanding and complying with traffic regulations is a safety-critical requirement for autonomous driving, yet remains challenging due to the diversity and context dependence of traffic signage. Importantly, regulation understanding is not a simple recognition task, but a reasoning problem: whether a rule applies depends on interpreting the sign in relation to the spatial layout of lanes and scene context. To support such reasoning, MapDR provide fine-grained annotations that link each traffic sign's regulatory rules to the specific lanes they govern. Existing methods, however, largely treat this as direct sequence prediction, ignoring the underlying reasoning that connects sign semantics and map structure. To address this limitation, we explicitly incorporate reasoning into this task and propose a framework that equips vision-language models (VLMs) with chain-of-thought (CoT) capabilities. We first design a scalable CoT curation pipeline that bootstraps rationales from a strong LLM through a two-round strategy and employs a VLM-based verifier to filter out incorrect cases, yielding a high-quality set of (CoT, answer) pairs. Building on this foundation, we adopt a two-stage training scheme: supervised fine-tuning (SFT) to teach rationale-to-answer generation, followed by GRPO reinforcement learning with answer-grounded, fine-grained rewards to further improve final answer accuracy. Extensive experiments on MapDR show that our approach significantly improves both interpretability and accuracy, establishing the first reasoning-based framework for regulation-aware autonomous driving.

cs.CV↗

arXiv2026

OxygenREC-v2: Internalizing Discrimination into Generative Recommendation

Generative recommendation unifies retrieval and ranking within a single model by autoregressively decoding semantic identifier (SID) sequences. Yet reliably incorporating behavior signals from clicks, cart additions, and orders remains challenging. Existing approaches either jointly optimize generative and discriminative objectives, requiring delicate trade-offs, or use a separate ranker as a post-hoc reinforcement-learning reward, risking out-of-distribution scoring and reward misalignment. We propose OxygenREC-v2, a generative recommender that Internalizes Discrimination into Generative Recommendation (IDGR). Rather than adding a separate discriminative objective, OxygenREC-v2 uses logged behavior to condition generation and supervise training. During pre-training, a behavior instruction conditions generation on the target behavior. During post-training, future interaction behaviors are exploited as privileged knowledge in our entropy-aware trajectory optimization self-distillation framework, enabling reward-model-free policy optimization. Throughout both training stages, OxygenREC-v2 maintains a single unified backbone. We implement OxygenREC-v2 as a 3B-parameter, 1B-activated MoE and deploy it on JD.com's large-scale e-commerce platform. Across multiple online A/B tests, OxygenREC-v2 improves user click-through conversion rate (UCTCVR) by 1.6--4.4% and GMV by 2.8--6.8% over OxygenREC-v1.

cs.IR↗

arXiv2026

Geometry-Guided Representations for Coherent Lane and Traffic Topology Reasoning in Driving Scenes

Road topology reasoning is fundamental for autonomous driving, requiring both accurate perception of road elements and understanding of their complex connectivity, including lane connectivity (Lane-to-Lane, L2L) and traffic regulation (Lane-to-Traffic signs, L2T). However, existing methods typically treat perception and topology reasoning as fragmented tasks, ignoring their potential for mutual enhancement. Crucially, while topology is inherently relational, prior works often overlook geometric relationships during feature extraction, relying instead on brittle post-processing or coordinate-based heuristics applied only at inference time. To bridge this gap, we propose CoPo (Coherent Perception and toPology), a unified framework that integrates geometry-guided relational modeling across three levels: 1) Perception-level: We introduce a relation-aware lane detector that utilizes geometry-biased self-attention and curve-guided cross-attention to enrich lane representations with structural priors; 2) Reasoning-level: We design relation-enhanced topology heads, including a geometry-enhanced L2L head and a cross-view L2T head, which effectively align features to infer connectivity; and 3) Supervision-level: We implement a contrastive InfoNCE strategy to regularize relational embeddings, pulling connected pairs closer in the latent space. This coherent multi-level design enables end-to-end joint optimization of perception and reasoning. Extensive experiments on OpenLane-V2 demonstrate that CoPo significantly outperforms existing methods, achieving gains of {\textbf{+3.1}} in DET$_l$, {\textbf{+5.3}} in TOP$_{ll}$, {\textbf{+4.9}} in TOP$_{lt}$, and {\textbf{+4.4}} overall in OLS, setting a new state-of-the-art.

cs.CV↗

arXiv2026

OmniCustom: Sync Audio-Video Customization Via Joint Audio-Video Generation Model

Existing mainstream video customization methods focus on generating identity-consistent videos based on given reference images and textual prompts. Benefiting from the rapid advancement of joint audio-video generation, this paper proposes a more compelling new task: sync audio-video customization, which aims to synchronously customize both video identity and audio timbre. Specifically, given a reference image $I^{r}$ and a reference audio $A^{r}$, this novel task requires generating videos that maintain the identity of the reference image while imitating the timbre of the reference audio, with spoken content freely specifiable through user-provided textual prompts. To this end, we propose OmniCustom, a powerful DiT-based audio-video customization framework that can synthesize a video following reference image identity, audio timbre, and text prompts all at once in a zero-shot manner. Our framework is built on three key contributions. First, identity and audio timbre control are achieved through separate reference identity and audio LoRA modules that operate through self-attention layers within the base audio-video generation model. Second, we introduce a contrastive learning objective alongside the standard flow matching objective. It uses predicted flows conditioned on reference inputs as positive examples and those without reference conditions as negative examples, thereby enhancing the model ability to preserve identity and timbre. Third, we train OmniCustom on our constructed large-scale, high-quality audio-visual human dataset. Extensive experiments demonstrate that OmniCustom outperforms existing methods in generating audio-video content with consistent identity and timbre fidelity. Project page: https://omnicustom-project.github.io/page/.

cs.SD↗

arXiv2026

BitNet Text Embeddings

LLM-based text embedders have substantially improved retrieval and semantic representation quality, but their deployment remains costly: large backbone models slow down embedding inference, while high-dimensional full-precision embeddings impose substantial storage and bandwidth overhead on large-scale indexes. In this paper, we present BITEMBED, an extreme low-bit framework for LLM-based text embedding that jointly targets encoding efficiency and vector storage. BITEMBED converts pretrained LLM backbones into BitNet-style embedding encoders with ternary weights, quantized activations, and lightweight normalization refinement. The converted model is adapted to representation learning through continual contrastive pre-training, followed by supervised contrastive fine-tuning with both similarity-distribution distillation and attention-relation distillation from a full-precision teacher. Beyond quantizing the backbone, BITEMBED further trains output embeddings to support multiple storage precisions meeting different storage needs in various scenarios. Experiments on MMTEB (eng, v2) with Qwen3-0.6B and Gemma3-270M show that BITEMBED is largely comparable to full precision teacher embedders. Moreover, BITEMBED flexibly obtains text embeddings of various precisions, achieving a trade-off between performance and storage cost.

cs.CL↗

arXiv2026

GPE: Evaluating Robust Evidence Aggregation for Fact Verification under Controllable GEO-Style Poisoning

Large language models increasingly use search tools to retrieve up-to-date information, introducing a new attack surface in which retrieved documents can be manipulated. This risk is amplified by the development of generative engine optimization, which can make selected content more likely to be retrieved, cited, and adopted by models. Existing fact-verification benchmarks and evaluation frameworks do not provide the controlled evidence environments needed to assess robustness against GEO poisoning. We therefore propose GPE, which consists of a multi-domain fact-verification benchmark and an evaluation framework for controlling evidence sources and poisoning ratios. Experiments across multiple verification methods and poisoning attacks demonstrate that GPE exposes robustness degradation and efficiency trade-offs that cannot be observed through clean evaluation alone, confirming the need to evaluate fact verification under adversarial evidence environments.

cs.CR↗

arXiv2026

AlayaWorld: Interactive Long-Horizon World Modeling -- Full Technical Report

Unlike conventional video game development, which relies on labor-intensive pipelines for asset production, animation, physics, and programming, video world models generate interactive environments from user inputs instantly. It enable us to create customized, explorable, and continuously evolving virtual world from text, an image, or video. Realizing this vision requires four tightly coupled capabilities: interaction, persistent spatiotemporal consistency, stable long-horizon generation, and efficient response. We present AlayaWorld, an interactive long-horizon video world model that generates 24-fps video at 540p and 720p. Built on a 15B video diffusion transformer, AlayaWorld generates short latent chunks autoregressively under camera trajectories and switchable text prompts. Its bounded visual context combines a persistent sink frame, compressed temporal history, geometry-aligned spatial memory, and recent-frame conditioning. To reduce long-term drift, the model is trained with corrupted histories and prediction residuals collected from its own roll-outs. We further introduce a discrete autoregressive distillation formulation that combines distribution-matching distillation, self-forcing++, and consistency distillation, reducing inference from approximately 30 sampling steps to four steps per chunk. On iWorld-Bench, AlayaWorld achieves the best performance over long-horizon generation. Conceived as a full-stack, open-source, and long-term project, AlayaWorld is intended to provide an extensible foundation for future research on interactive video world models.

cs.AI↗