Search arXiv⌕ Search

arXiv subjects

Tong Zhang

Publications and source records attributed to Tong Zhang.

At least 55 records · Page 3Linked to original sources
arXiv2026

REACH: Controller-Managed Long-Span ECC for HBM AI Inference

High-Bandwidth Memory (HBM) cost motivates stronger controller protection that can support a wider range of device error rates. Long-span error-correcting codes provide stronger protection at a comparable code rate, but a direct implementation couples small accesses to span-wide state and requires costly decoding at HBM bandwidth. Read-dominated LLM decode offers a favorable setting: sequential reads support span aggregation, while sparse writes limit parity-update traffic. This paper presents REACH, a controller microarchitecture that uses established inner codes to correct common errors and identify unresolved chunks, reserving a long outer code for known-erasure repair. Differential parity bounds write traffic, and a co-designed endpoint preserves 32\,B transactions without an extra data burst. Ramulator2 sustains 1.88\,TB/s of application traffic at the highest error stress, while separate full-interface sizing supports a 2.69\,TB/s application target using ASAP7-synthesized kernels. At this analytical target, REACH's nominal composition uses 55.8\% less controller area and 57.7\% less modeled power than the evaluated mean-work direct-long design, showing the benefit of reserving long-span recovery for exceptional requests.

cs.AR↗
arXiv2026

KDFlow: A User-Friendly and Efficient Knowledge Distillation Framework for Large Language Models

Knowledge distillation (KD) is widely used to compress and post-train large language models (LLMs), yet many existing frameworks execute teacher inference with the same training-oriented backend as student optimization, leading to suboptimal efficiency. In this paper, we propose KDFlow, a novel framework for LLM distillation that features a decoupled architecture and employs SGLang for teacher inference. KDFlow combines SGLang for teacher inference with PyTorch FSDP2 for student optimization, allowing each model to run on a backend tailored to its workload. To enable efficient full-vocabulary distillation in this decoupled architecture, KDFlow transfers the teacher's final hidden states via Ray's object store and recomputes teacher logits on each student worker using a frozen copy of the teacher's output head. Furthermore, our framework supports both off-policy and on-policy distillation and incorporates cross-tokenizer algorithms through highly extensible and user-friendly APIs. Experiments show that KDFlow achieves a 1.44$\times$ to 6.36$\times$ speedup over MS-SWIFT in off-policy distillation and a 1.43$\times$ to 1.75$\times$ speedup over verl in on-policy distillation. KDFlow further scales to 64 GPUs, achieving 3.68$\times$ and 2.52$\times$ strong-scaling speedups in two representative model configurations. The code and documentation are publicly available.

cs.CL↗
arXiv2026

CIVA: Critic-Induced Value-Subspace Attacks on Visual World-Model Agents

Visual world-model agents such as DreamerV3 act through a recurrent latent state rather than a single observation, which weakens frame-wise observation attacks and makes their perturbations vary sharply over time under a strict per-frame perturbation constraint. We study white-box, causal, online attacks on such agents and propose Critic-Induced Value-Subspace Attacks (\textbf{CIVA}). Our key observation is that, along a rollout, critic-guided perturbations concentrate in a low-dimensional subspace induced by the victim's own critic. Based on this observation, CIVA first probes the frozen victim offline with critic-guided PGD and extracts a low-rank value-subspace by SVD. At test time, it optimizes only the subspace coefficients, smooths them with an exponential moving average (EMA), and maps them back to pixels. This design attacks value-sensitive recurrent dynamics while keeping the online optimization cheap and temporally coherent. Extensive experiments on DMC walker walk, Atari Pong, and Crafter show that CIVA consistently outperforms five recent methods; on DMC walker walk, it achieves the largest reward drop of 26.07\% while keeping temporal variation low, with TempAbs of 0.646.

cs.CV↗
arXiv2026

A multicenter benchmark and clinically structured metric for coronary CTA report generation

Reliable evaluation of automated coronary computed tomography angiography (CCTA) report generation requires standardized multicentre benchmarks and clinically structured metrics. We established a four-centre benchmark comprising 3,021 CCTA series from 818 patient-report pairs to evaluate seven open-source three-dimensional vision-language models. We developed CSM$_{\text{CCTA}}$, a clinically structured metric for CCTA report evaluation, with patient-, vessel-, and segment-level variables defined according to clinical guidelines. Report pairs are compared at the finest shared anatomical level, and the contributions of different clinical components are weighted based on expert assessments. We estimated these weights using 70 expert-scored cases and evaluated clinical alignment in a non-overlapping set of 30 cases. CSM$_{\text{CCTA}}$ showed a strong correlation with radiologist scores (Pearson's $r=0.97$, $p<0.001$), exceeding the next-best metric, FORTE ($r=0.70$), by 0.27, and agreed with expert preferences in 115 of 160 pairwise comparisons (71.9\%). Under controlled perturbations, CSM$_{\text{CCTA}}$ remained stable to clinically equivalent wording and decreased monotonically with progressive information omission. In the multicenter benchmark, the CCTA-trained C2RG model achieved the highest CSM$_{\text{CCTA}}$ scores across all four hospitals, although its performance remained far from optimal. In contrast, CCTA-irrelevant reports accounted for up to 98.7\% of the outputs from generalist models. Together, the benchmark provides a standardized setting for model comparison, while CSM$_{\text{CCTA}}$ enables clinically structured evaluation of finding agreement and anatomical specificity. These results support a more clinically aligned and anatomically resolved approach to evaluating CCTA report generation. Code is available at https://openi.pcl.ac.cn/OpenMedIA/CSM_CCTA.

cs.CV↗
arXiv2026

One-step group factor analysis via penalized least squares

In this article, we revisit the problem of group factor analysis and propose a one-step penalized least squares method to estimate the factor loadings and factors in large-dimensional group factor models, offering a distinct alternative to the conventional two-step principal component approach. Our procedure originates from the equivalence between the group factor structure and the carefully tailored identification conditions. Leveraging this insight, we develop a tricky Lagrange multiplier formulation for a penalized least square loss function. This one-step optimization framework, combined with the fine-tuned penalty parameters, facilitates the direct derivation of central limit theorems for factor loadings, factor scores, common and local components, as well as the convergence rates for them. Our theory demonstrates that our one-step approach achieves the same rate as the two-step aggregated principal-component method and even the same limiting standard error in the balanced group panel case, but smaller limiting standard error than the two-step canonical correlation procedure. Extensive simulation studies justify the theory. Applications to U.S. house prices and CSI300 weekly returns confirm that our method identifies global factors and heterogeneous local patterns.

stat.ME↗
arXiv2026

Understanding Deep Learning via Entropy Space Theory

Deep learning is often criticized for its theoretical research lagging behind practice. To make deep learning easier to understand, the entropy space theory is first introduced here. The entropy space can cover all the possibilities of any deep learning model by topological structure. It is independent of network parameters. Through the designed fundamental operations and norm, entropy space is proven to be a normed space within the formal axiomatic framework. Based on the theory, a unified coordinate system is proposed. It can coordinatize every state of a model and rank them by compression of the maximal value of information entropy. The theory offers a novel priori framework for mathematical fundamentals of deep learning.

cs.AI↗
arXiv2026

Low-Altitude Fluid Antenna Network with Multi-Agent Reinforcement Learning

Low-altitude wireless networks (LAWNs) integrate terrestrial and aerial platforms to provide ubiquitous communication, sensing, and localization services for unmanned aerial vehicles (UAVs) and electric vertical takeoff and landing (eVTOL) aircraft. However, dynamic air-ground and air-air channels, abrupt blockages, and heterogeneous interference hinder the realization of this goal. Nevertheless, fluid antenna (FA), a cutting-edge multiple-input multiple-output (MIMO) technique, overcomes these challenges by reconfiguring antenna positions to unlock additional spatial degrees-of-freedom. In this paper, towards bringing low-altitude FA networks into reality, we study the fast and high-performance FA reconfiguration for low-altitude FA networks with multi-agent reinforcement learning (MARL). Specifically, we present an electromagnetic digital twin (EM-DT)-assisted MARL framework. To fill the sim-to-real gap, we introduce a two-stage transfer learning framework. Our case study shows that joint FA positions and beamforming optimization can enhance the system sum-rate by 118.5%, compared to the fixed position baseline. This gain comes from the dynamic millisecond timescale reconfiguration of FA arrays and the adaptive steering of beams toward aerial users with mobility.

cs.IT↗
arXiv2026

Mitigating Rubric Interference in LLM Judges via On-Policy Self-Distillation

LLM judges increasingly evaluate responses against fine-grained rubric checklists. When a sample requires multiple rubrics, current methods typically assess each in a separate inference call. Evaluating all rubrics in a single pass is a natural alternative with greater efficiency, but we find that it introduces rubric interference: the verdict on one rubric shifts depending on which other rubrics are co-present. In a preliminary study, only one-third of samples receive fully consistent verdicts when evaluated under rubric sets of varying composition. We develop a measurement framework that probes interference through four controlled operations: rubric set expansion, subsetting, reordering, and noise injection. To mitigate interference without external supervision, we propose Self-Anchored Rubric Alignment (SARA). SARA uses a model's own single-rubric judgments as stable anchors and aligns multi-rubric reasoning with these anchors through on-policy self-distillation. We validate SARA on three datasets (HealthBench, FLASK, ResearchQA) and two model families (Qwen3, Llama-3.1). SARA consistently improves evaluation consistency while maintaining agreement with both base models and GPT-4.1 as a reference judge. Furthermore, the learned consistency transfers across datasets, confirming that SARA teaches a general capability rather than fitting dataset-specific patterns.

cs.LG↗
arXiv2026

StoSignSGD: Unbiased Structural Stochasticity Fixes SignSGD for Training Large Language Models

Sign-based optimization algorithms, such as SignSGD, have garnered attention for their performance in distributed learning and training large foundation models. Despite their empirical superiority, SignSGD is known to diverge on non-smooth objectives, which are ubiquitous due to ReLUs, max-pools, and mixture-of-experts. To overcome this limitation, we propose StoSignSGD, an algorithm that injects structural stochasticity into the sign operator while maintaining an unbiased update step. In the regime of (online) convex optimization, StoSignSGD rigorously resolves the non-convergence issues of SignSGD, achieving a sharp convergence rate matching the lower bound. For the more challenging non-convex non-smooth optimization, we introduce generalized stationary measures that encompass prior definitions, proving that StoSignSGD improves upon the best-known complexity bounds by dimensional factors. Empirically, StoSignSGD is stable and efficient across diverse large language model (LLM) training regimes. In aggressive low-precision pretraining, which spans both FP8 and the far more demanding FP4 regime where AdamW fails catastrophically, StoSignSGD stays stable and consistently performs the best. It attains a 1.44x to 2.14x speedup over established baselines under FP8. Under 4-bit precision, it improves downstream accuracy on the largest OLMo2-370M model by 1.13 points over the strongest stable baseline, and this advantage grows as both the model size and the data scale up. When fine-tuning 7B LLMs on mathematical reasoning tasks, StoSignSGD also delivers clear gains over both AdamW and SignSGD. Finally, to explain why it works, we develop a sign conversion framework that turns any general optimizer into its unbiased, sign-based counterpart. Using this framework, we decompose the core components of StoSignSGD and run a comprehensive ablation study to validate our design choices.

cs.LG↗
arXiv2026

XPolicyLab: A Unified Standard and Open Ecosystem for Robot Policy Evaluation and Deployment

Robot policy evaluation and deployment remain fragmented by model-specific software dependencies, data representations, and runtime interfaces, so that connecting N policies to M evaluation environments requires O(NM) separate integrations. We present XPolicyLab, a unified standard and open ecosystem that reduces this cost to O(N+M). XPolicyLab specifies common observation, action, and trajectory schemas together with a minimal adapter interface for observation updates, action prediction, batched execution, and episode reset, while a dependency-isolated client/server architecture separates policy inference from environment execution, so that each side retains its native software stack and may run locally or remotely. The ecosystem integrates 42 robot policies and standardizes their installation, debugging, serving, and evaluation workflows. Across these adapters, model-specific code varies by an order of magnitude while the environment-facing loop stays within a few lines of a fixed reference, confirming that the contract confines heterogeneity to the policy side. In a controlled study, conforming to the standard reduces the integration effort of a representative policy from over five hours to two hours, and packaged agent skills reduce it further to thirty minutes. The same adapters serve RoboTwin, RoboDojo simulation, and standardized real-robot evaluation through one interface. XPolicyLab is released as shared infrastructure for reproducible policy comparison and standardized deployment across simulation and physical platforms. Project website: https://xpolicylab.github.io/.

cs.RO↗
arXiv2026

DiSCO: Defending text-to-image generation through distribution-guided contrastive prompt optimization

As text-to-image generative models advance, they raise critical safety concerns, particularly the generation of Not-Safe-For-Work (NSFW) content such as violence and nudity, further exacerbated by red-teaming adversarial attacks. Existing defenses predominantly operate under white-box assumptions, relying on text encoder optimization, weight editing, or inference-time intervention, and fundamentally cannot scale to proprietary models. Black-box alternatives based on LLM prompt rewriting offer broader applicability, yet fail in a critical regime we identify as the \textit{benign adversarial} problem: prompts that are linguistically safe but still trigger harmful generation due to the model's learned data distribution. We propose DiSCO, a zero-shot, strictly black-box defense that operates entirely at the prompt level as a plug-and-play module, requiring no model retraining, fine-tuning, or access to model internals. DiSCO performs distribution-guided suffix expansion via beam search, optimized through contrastive scoring over safe and unsafe image pools generated by the target model itself, with iterative adaptive feedback until safe content is produced. We demonstrate that DiSCO consistently enhances the safety of both undefended and defended models on the I2P benchmark under multiple red-teaming attacks, achieving 37.7% and 25.13% ASR reduction, respectively, while maintaining semantic fidelity and improving image coherence. As a black-box, architecture-agnostic module, DiSCO can be readily applied to any text-to-image system without necessitating any changes to the model itself.

cs.AI↗
arXiv2026

DualStabSleepNet: A Dual-Domain Diffusion Stabilization Network for Robust Sleep Staging

Existing deep learning approaches for automatic sleep staging suffer from limited robustness under heterogeneous recording conditions, where non-stationary noise, inter-subject differences and cross-dataset distribution shifts cause unstable features and poor generalization. This work proposes DualStabSleepNet (DSSNet), a dual-domain diffusion stabilization network for robust sleep staging, which improves robustness in both data and feature domains. After preprocessing multi-channel polysomnography (PSG), a continuous-scale diffusion-based stabilization module suppresses noise while preserving physiological signal structures. Stabilized signals are converted to time-frequency representations and fed into a Vision Transformer backbone. A teacher-student guided diffusion feature stabilization module further mitigates feature drift and enforces multi-level feature consistency. Evaluated on four public PSG datasets SleepEDF-20, SleepEDF-78, SHHS and ISRUC-S3, DSSNet achieves state-of-the-art accuracy of 89.2%, 88.0%, 89.7%, 86.7% with improved macro-F1 and Cohen's kappa. It obtains notable improvements on hard transitional stages (e.g., 12.5% gain for N1 on SHHS) and boosts N2/REM recognition. Under cross-dataset settings, DSSNet is robust to distribution shift and performs on par with or superior to target-dataset trained baselines, demonstrating its practical potential for real-world sleep staging across heterogeneous cohorts.

cs.CV↗
arXiv2026

ToolHazard: Scaling Adversarial Environments for Security Evaluation and Alignment of LLM-based Agents

Large language model (LLM) agents integrated with external tools are vulnerable to indirect prompt injections embedded in environmental states. However, existing studies largely rely on manually implemented or reused environments, stochastic LLM-based tool simulation, and predefined injection locations, limiting scalable security research across broader domains. To bridge this gap, we propose **ToolHazard**, a scalable adversarial environment synthesis framework that reduces human engineering and supports expansion with additional seed domains and compute. Through an Environment Simulator, an Attacker Agent, and a User Simulator, ToolHazard synthesizes executable stateful environments, discovers viable injection points and generates environment-specific payloads, and constructs state-grounded long-horizon tasks. Based on ToolHazard, we build **ToolHazard-Bench** for stress-testing agents under complex workflows and diverse environmental attacks. Experiments reveal substantial agent vulnerabilities and show that injection timing and placement affect attack effectiveness. Moreover, ToolHazard-generated alignment data improves security on both ToolHazard-Bench and AgentDojo while preserving benign task utility.

cs.CR↗
arXiv2026

Dimensional crossover and local strain induced deflection of the spin spiral state in multiferroic NiI2

Low-dimensional multiferroics hold great promise for integrated magnetoelectric devices. Spin spiral state has recently been shown to induce ferroelectricity in single-layer van der Waals (vdW) material NiI2. However, how this state evolves and can be tuned towards the two-dimensional limit remain unclear. Here, we combine spin-polarized scanning tunneling microscopy, layer-by-layer film growth, and multi-scale theoretical modeling to investigate the spin spirals in NiI2 thin films. As the film thickness increases from 1 to 7 monolayers, we observed a continuous increase of spin-spiral wavelength and a rotation of wavevector from near [110] to [1-10] direction, which evidences a dimensional crossover primarily driven by enhanced interlayer exchange energy. Moreover, we find that the film wrinkles can cause deflection of the spin spiral wavevector, which is caused by local curvature induced modification of exchange interactions. Our findings establish thickness and local strain as two tuning methods for engineering non-collinear helical magnetism and accompanied electric polarization in vdW multiferroics.

cond-mat.mes-hall↗
arXiv2026

LGFNet: A CTC-Guided Local-Global Fusion Framework for Single-Channel Sleep Staging

Sleep staging remains challenging due to long-range temporal dependencies, ambiguous stage transitions-particularly in N1-and substantial distribution shifts across subjects, sampling rates, and EEG montages. These difficulties are further amplified in single-channel, low-latency scenarios required by wearable and real-world applications. To address these issues, we propose LGFNet, a CTC-guided sequence-to-sequence framework for robust sleep staging. LGFNet introduces a Local-Global Fusion encoder that jointly models fine-grained temporal dynamics and long-range sleep structure, overcoming the limitations of conventional serial hybrid architectures. A CTC-Attention joint training paradigm is adopted to unify temporal alignment with context-dependent modeling, enabling more accurate recognition of stage boundaries and transitions. Furthermore, a three-stage decoding strategy is devised, leveraging CTC-guided decoding and Viterbi-based smoothing to reduce error accumulation and enforce physiological consistency. Extensive cross-dataset evaluations on five public benchmarks demonstrate that LGFNet consistently outperforms state-of-the-art single-channel methods. In particular, on Sleep-EDF-78, LGFNet surpasses DMIN by +1.27% accuracy, +1.74% macro-F1, and +1.93% kappa, with pronounced gains on N1 and transition segments, highlighting its robustness and strong generalization across diverse sampling rates, montages, and recording environments.

cs.CV↗
arXiv2026

AsyncWebRL: Efficient Asynchronous Reinforcement Learning for Multi-Step Visual Web Agents

Training vision-language web agents with multi-step RL is compute-intensive, with two dominant forms of inefficiency: idle GPUs in synchronous RL, and trajectories that use more steps and tokens than necessary. We present AsyncWebRL, which addresses both. On the system side, an asynchronous design overlaps rollout, gradient update, and policy refresh across iterations, paired with two web-agent-specific adaptations, namely an everlasting rollout pool and lightweight screenshot handling, that together deliver up to a $2.9\times$ end-to-end training-throughput speedup over the previously fastest open synchronous pipeline (WebGym). On the algorithmic side, we identify the per-trajectory normalizer $1/|τ_i|$ in multi-step GRPO as the root cause of trajectory-level and token-level inefficiency: because failures are systematically longer than successes, it down-weights the negative gradient on failed tokens, so the policy keeps producing verbose memory schemas. Replacing $1/|τ_i|$ with a constant $1/k$ breaks this coupling, contracting trajectories while preserving aggregate success. Together, these contributions set a new open-source state of the art on the WebGym out-of-distribution test split (+5.8% relative over the 42.9% prior best), with the largest gains on the harder slices (+42% relative on Medium, +48% relative on Hard).

cs.LG↗
arXiv2026

Dual-Attention and Adversarial Transfer Networks for Sim-to-Real Cross-Orientation Wireless Sensing

Millimeter-wave human activity recognition suffers significant performance degradation when the user's orientation changes relative to the sensing system, yet collecting labeled multi-orientation data is labor-intensive and costly. To eliminate the need for exhaustive multi-orientation measured data, we develop a physics-guided simulator that synthesizes orientation-diverse wireless training data from single-orientation motion. Specifically, to suppress orientation-induced feature variations, we propose a dual-attention network that extracts activity-discriminative and orientation-robust representations from dual-link Doppler spectrograms. To bridge the simulation-to-reality gap, we introduce an adversarial unsupervised transfer learning mechanism that aligns feature distributions using only a small number of unlabeled target-domain samples. The S2M-Sense platform shows high fidelity in reproducing real-world signatures, validated against 60.48 GHz mmWave measured data with an average structural similarity index measure (SSIM) of 0.84 between simulated and measured Doppler spectrograms across all 4 activities and 4 orientations. Experimental results show that S2M-Sense achieves 88.33% recognition accuracy using only the dual-link multi-orientation simulated dataset, which improves to 95% after simulation-to-reality transfer learning with as few as 16 unlabeled measured samples. Both cases with and without transfer learning outperform state-of-the-art cross-domain sensing methods.

cs.CV↗
arXiv2026

First-Principles Quantum-Spectral framework for Elementary Vortex Pinning in superconductors

The critical current of a type-II superconductor is controlled by vortex pinning, whose microscopic input is the elementary pinning force. Scanning tunneling spectroscopy has shown that a defect pins a vortex by reorganizing the Caroli--de Gennes--Matricon (CdGM) states in its core, but why this spectral reorganization amounts to a pinning force has lacked a quantum-mechanical, first-principles account. Here we establish a transferable first-principles computational framework for elementary vortex pinning, in which defect-resolved DFT/Wannier electronic structures are embedded into a finite-box projected Bogoliubov--de Gennes free-energy formalism to convert quasiparticle spectral reorganization into vortex-pinning energies and forces. Using this framework, we confirm that the defect-induced reorganization of the vortex-core spectrum is the microscopic origin of the elementary pinning force. The force is evaluated as a finite-box vortex-insertion free energy whose four-configuration subtraction isolates the meV-scale interaction from much larger backgrounds. With the superconducting gap scale and vortex-core profile fixed from experiments, the FeSe Fe-site vacancy reproduces the microscopic STM force scale together with the measured spectral reorganization. All five point defects in FeSe and FeTe pin attractively, with FeTe Fe-site vacancy strongest. Elementary vortex pinning thereby becomes a computable electronic-structure quantity, opening the first-principles screening of point defects toward higher critical currents.

cond-mat.supr-con↗