arXiv subjects
Publications and source records attributed to Yi Zhu.
Let $X$ be a smooth toric variety over a number field $k$ such that the global sections of $X$ are constant. We show that $U(k)$ is dense in $U(\mathbf A_k)^{\text{Br}_1 (U)}$ for any open subset $U$ of $X$ such that the codimension of $X\setminus U$ in $X$ is at least 2.
Large language model (LLM)-based agent applications often incur high response time. Speculative decoding is a promising solution to improve the inference efficiency of LLM agents without impacting generation quality. However, state-of-the-art speculative decoding algorithms exhibit substantial speed degradation under large batch sizes, limiting their effectiveness to deploy in real-world agent applications. In this work, we first present a systematic analysis of speculative decoding for LLM agents and identify two dominant factors of speedup degradation: high rejection rate of speculative tokens, and under-utilization of dynamic token budgets.B ased on these observations, we propose AgentSpec, a speculative decoding algorithm that addresses the limitations of existing methods for LLM agents. AgentSpec incorporates structure-isolated drafting that constrains speculation to semantically coherent segments of the agent workflow, reducing the drafts of irrelevant semantic paths and achieving an extremely low rejection rate. Moreover, AgentSpec adopts redundancy-aware budget allocation that exploits agent-level information to better utilize the dynamically-free token budget during the agent inference. We implement and evaluate AgentSpec on five different workloads and four different models from four different LLM families in vLLM. Our results demonstrate the superiority of AgentSpec over state-of-the-arts.
The social interactions among crowds via \textit{Danmaku} (a.k.a., bullet comments) on modern multimedia platforms can facilitate both viewpoint conflicts and consensus, providing fine-grained discriminative social signals that can benefit fake news detection. However, the inherent accumulation latency of \textit{Danmaku} in real-world scenarios violates the real-time necessity of fake news detection, making the studies of \textit{Danmaku}-related fake news detection underexplored. To break this violation, we simulate this temporal-aware user interactive process by proposing a novel temporal \textbf{Gen}erative \textbf{da}nmaku framework, called \textbf{Genda}, which consists of: (1) a \textit{Danmaku} Trigger for predicting the timing and intensity of user reactions; and (2) a \textit{Danmaku} Generator for synthesizing corresponding semantic and emotional expressions, thereby mutually constructing a temporally aligned and human-like pseudo \textit{Danmaku} streams. To make the generated \textit{Danmaku} useful for identifying fake news videos, we further design a \textit{Danmaku}-guided Temporal Multimodal fake news detection model - \textbf{DM-FEND}, which enables fine-grained multimodal interactions among video, audio, text, and \textit{Danmaku}, enhancing dynamic modalities alignment and semantic noise inhibition. The experimental results demonstrate that \emph{DM-FEND} consistently outperforms state-of-the-art baselines across both Chinese (FakeSV) and English (FakeTT) benchmarks. Further ablations validate the crucial role of temporal \textit{Danmaku} modeling in enhancing robustness and discriminative capability. Finally, this study offers a bright and robust solution for multimodal fake news detection in modern social interactive fashions by bridging the temporal inconsistency between news and user behaviors.
The Jin--Xin relaxation of a nonlinear hyperbolic conservation law introduces a relaxation parameter that controls the width of the internal layer resolving a shock; the discontinuity of the limiting conservation law emerges only in the singular limit as this width vanishes. Physics-informed neural networks (PINNs) use smooth network approximations and are therefore not well suited to this limit, while relaxation PINNs with a fixed parameter resolve only a single scale and cannot follow the multiscale transition toward the limiting solution. We propose the Jin--Xin relaxation gradual convergence method (JXRGCM), which treats the relaxation parameter as a continuation variable, annealing it to zero along a schedule and warm-starting each stage from the previous one, so that the approximation follows the relaxation profile through progressively sharper scales. Under the sub-characteristic condition we establish a stability estimate whose constant is independent of the relaxation parameter; combined with the relaxation limit, it yields for scalar conservation laws an $L^2$ convergence rate of $\mathcal{O}(\varepsilon^{1/4})$ toward the entropy solution. Numerical experiments on the Burgers equation, the shallow-water dam-break problem, and the Sod shock tube show that JXRGCM improves shock and rarefaction resolution compared with fixed-parameter relaxation PINNs and other physics-informed approaches.
Recent advances in Large Language Models(LLMs) and agents have substantially improved the ability of AI systems to execute complex tasks. Yet existing benchmarks largely rely on researcher-selected tasks, leaving uncertain whether such progress extends to the work that real-world users actually demand from AI systems. We introduce \textbf{StartupBench}, an E2E agent benchmark grounded in market-validated AI startup products. Rather than defining tasks from pre-defined assumptions about useful agent capabilities, we systematically study AI products with demonstrated adoption, together with their product workflows and users, to identify real-world tasks for which AI has established practical demand across diverse professional domains. We translate these workflows into complete deliverable-oriented tasks and evaluate them with fine-grained rubrics capturing their complex requirements. Across representative models evaluated under a unified agent harness, even the strongest model successfully completes only approximately 30\% of StartupBench, despite making substantial partial progress on many tasks. Further analysis identifies aspects like complex instruction following and domain-specific expertise as major sources of failure. Our results reveal that many market-validated workflows remain beyond the reliable capabilities of current general-purpose agents, establishing StartupBench as an empirical measure of progress toward E2E completions of real-world user tasks.
Generalized honeycomb-structured materials have received increasing attention due to their novel topological properties. In this article, we investigate zero-energy edge states in tight-binding models for such materials with two different interface configurations: type-I and type-II, which are analog to zigzag and armchair interfaces for the honeycomb structure. We obtain the necessary and sufficient conditions for the existence of such edge states and rigorously prove the existence of spin-like zero-energy edge states. More specifically, type-II interfaces support two zero-energy states exclusively between topologically distinct materials. For type-I interfaces, zero-energy edge states exist between both topologically distinct and identical materials when hopping coefficients satisfy specific constraints. We further prove that the two energy curves for edge states exhibit strict crossing. We numerically simulate the dynamics of edge state wave packets along bending interfaces, which agree with the topologically protected motion of spin-like edge states in physics.
This paper investigates the non-resistive compressible magnetohydrodynamic (MHD) equations in $\mathbb{R}^2$. We establish the global existence and stability of classical solutions for initial data sufficiently close to a constant equilibrium state. A distinguishing feature of our result is that global stability is derived solely from pure $H^s$ energy estimates and an intrinsic $L^2$ time-decay mechanism, thereby bypassing the traditional requirement for the initial data of $L^1$ integrability or negative-order Sobolev norm regularity. To achieve this goal, we first introduce a specific quantity motivated by the effective viscous flux, which intrinsically couples the density and magnetic field perturbations. Secondly, to overcome the critical time-decay obstacle arising from the absence of negative-order regularity, we develop a novel pseudo-negative-derivative technique. Moreover, we regard the wildest nonlinear term as a whole and bypass the need to obtain time-decay estimate for individual components. These approaches enable us to close the higher-order energy estimate entirely within standard Sobolev spaces.
Existing methods mainly adapt pretrained autoregressive (AR) language models to masked diffusion, whereas we directly adapt them to uniform-noise diffusion, where every token remains editable during sampling. However, adapting AR checkpoints across corruption kernels remains challenging because existing DLMs use different objectives and prediction parameterizations. We establish connections among SEDD, MDLM/GIDD, M2S, and Neural CTMC by expressing their conditional losses as a single generalized Kullback--Leibler objective over model reverse rates. We further derive conversions from clean-token predictions to concrete-score, posterior-mean, and exit-rate/jump parameterizations, yielding a shared \(x_0\) interface that supports switching between mask and uniform kernels. Building on these connections, we propose \ours{}, a simple continual pre-training approach for directly adapting pretrained GPT2 checkpoints to uniform-noise diffusion. Through systematic evaluation of 124M- and 355M-parameter models, we show that \ours{} steadily improves the trade-off between generative perplexity (GenPPL) and unigram entropy as the sampling budget increases from 16 to 256 steps. At 256 steps, \ours{}-S and \ours{}-M achieve GenPPL/entropy pairs of \(97.783/5.2626\) and \(71.516/5.6669\), respectively; no evaluated model at the same scale simultaneously outperforms \ours{} on both metrics. At both scales, \ours{} also achieves the highest WinoGrande, SIQA, and BBH accuracy among the compared diffusion models.
Score Entropy Discrete Diffusion (SEDD) parameterizes discrete reverse processes with unconstrained positive score ratios. While positivity guarantees nonnegative reverse jump rates, it does not ensure Bayes realizability: ratios at a noisy state need not be jointly induced by any clean-token posterior under the forward kernel. The score-entropy loss has the correct population optimum but does not enforce this constraint away from it. In a trained pure-uniform SEDD checkpoint, roughly one quarter of complete score vectors violate the coordinate box, while more than half lie inside it yet remain materially incompatible with any valid posterior. Such violations can produce negative pre-normalization weights in finite-step sampling. Projecting raw scores onto the bridge polytope removes all observed negative weights and improves external generative PPL from $203.6$ to $175.1$ without changing the sampler. We introduce \emph{mean-to-score} (M2S), which predicts a clean-token posterior mean and converts it to the score through an exact kernel-dependent linear map. The construction applies to any known coordinate-wise continuous-time Markov chain (CTMC) satisfying a mild support condition. For uniform corruption, it maps the probability simplex onto the bridge polytope; for absorbing-mask corruption, the resulting objective recovers MD4 exactly. In a controlled 28.4M-parameter CIFAR-10 comparison, M2S lowers test BPD from $3.173$ to $3.129$ and FID-50k from $\CifarSEDDFID$ to $\CifarMtwoSFID$. A 170M-parameter M2S model trained on about 262B OpenWebText token slots outperforms the evaluated pure-uniform SEDD, GIDD, and Neural CTMC checkpoints at every tested sampling budget, reaching generative PPL $143.3$ at 128 steps versus $183.6$ for the strongest pure-uniform baseline.
Lossless acceleration schemes, such as speculative decoding, promise significant inference speedups by relying on dynamic token-level alignment between a draft and a target model. However, this guarantee of semantic equivalence masks a severe operational vulnerability: draft-target alignment can be systematically attacked. In this paper, we introduce ADSD, which, to the best of our knowledge, is the first prompt-suffix attack that collapses verifier acceptance by pushing draft probability mass toward tokens the target is unlikely to accept. ADSD uses Soft-Collapse, a verifier-aligned surrogate derived from the asymmetric speculative acceptance rule, together with a target-preservation objective that discourages obvious task corruption. ADSD successfully generates highly effective adversarial suffixes. On the GSM8K dataset, our attack increases the mean sample time by 62.3% while preserving the task quality. We further show that this vulnerability exists across different domains, speculative decoding strategies, and model architectures.
The development of organic light-emitting diode (OLED) materials faces the compounded challenges of an astronomically large chemical space, stringent quantum-chemical constraints, and a scarcity of labeled data. Although the question of OLED generation is important, few models have been trained effectively for this specific domain. We propose an inverse molecular design framework based on causal language models: given target optoelectronic properties (e.g., excitation energy, oscillator strength), our model directly generates OLED SMILES sequences satisfying the specified constraints. We employ a multi-stage strategy: first, we establish a foundational chemical language model using a LLaMA-style transformer architecture. To the best of our knowledge, this represents the first successful adaptation of LLMs specifically for the OLED domain, bridging the gap between generic molecular generation and the stringent structural requirements of optoelectronic materials. Second, we fine-tune property predictors based on a BERT model pre-trained on our large-scale OLED dataset. Then, we perform Reinforcement Learning on our fine-tuned model, leveraging our property predictor, for better SMILES generation. Finally, through DFT verification, we demonstrate that our framework can efficiently navigate the OLED chemical space, generating novel candidates with high structural validity and optimized optoelectronic properties.
Honey bees (Apis mellifera) play a crucial role in agriculture and ecosystem stability as key pollinators of crops and wild plants. As such, monitoring hive strength remotely with Internet of Things (IoT) sensors has become a crucial task. Previously, handcrafted features extracted from the modulation spectrum of audio IoT devices were shown to improve acoustic monitoring of colony strength. In this paper, we hypothesize that important discriminative information is present in the temporal dynamics of the modulation spectrum, but this information is discarded with prior methods. As such, we explore the use of a new modulation tensorgram where the time dimension is kept. This new representation is used as input to a convolutional neural network (CNN) and a convolutional recurrent deep neural networks (CRDNN). Using the public UrBAN dataset, which contains more than 3,000 hours of beehive audio recordings, we show that the proposed method improves both accuracy and cross-hive generalizability over prior benchmark methods, and the results further suggest improved robustness to noisy in-the-wild recording conditions. We use saliency maps and gradient-weighted class activation maps for explainability and show the importance of the modulation spectral temporal dynamics for the task at hand. Overall, our results suggest that accurate, generalizable, and robust acoustic monitoring of honey bee colony strength is possible.
Strain offers an effective method for generating pseudo-magnetic fields in optical and acoustic materials, thereby enabling precise manipulation of wave propagation. In this article, we investigate wave packets spectrally localized near Dirac points in strained honeycomb-structured media and rigorously justify their long-time effective dynamics. We show that the envelope dynamics is governed by a two-dimensional Dirac equation with nontrivial gauge fields and prove that the associated two-scale ansatz approximates the exact wave evolution with error $O(\varepsilon)$ in $H^s$ for $0\le t\le ρ\varepsilon^{-1}$. Two difficulties distinguish this problem from standard wave-packet justifications. First, strain deforms the principal part of the wave operator, so the residual contains second-order differential terms that are not controlled by the unperturbed wave energy. Second, for a vanishing potential, the spectrum of the strained operator is not bounded away from zero, and a direct Duhamel estimate on the low-energy spectral subspace produces an apparent secular growth. We overcome the first difficulty by evolving with the strained operator and comparing regularized spectral projections for the strained and unperturbed operators through norm-resolvent estimates and functional calculus. For the second, we isolate the leading low-energy forced response through an explicit resolvent construction. Together, these results establish a rigorous continuum-wave theory of strain-induced pseudo-magnetic Dirac dynamics for slowly deformed honeycomb media, including perturbations of the principal part and the physically relevant zero-potential regime. More broadly, the spectral strategy may be useful for other linear systems with perturbations acting at the highest differential order.
Recent years have witnessed the rapid evolution of AI agents toward handling increasingly complex, real-world tasks. However, existing benchmarks rarely evaluate whether agents can operate graphical user interfaces to complete long-horizon, high-value professional workflows across diverse domains. Current GUI benchmarks still predominantly focus on general-purpose software, relatively simple applications, and short-horizon tasks, leaving it largely unknown whether modern agents can follow user instructions to autonomously operate domain-specific professional software and accomplish economically valuable work in an end-to-end manner. To bridge this gap, we introduce Workflow-GYM, a benchmark for long-horizon GUI tasks centered on professional domains and specialized software environments. Through extensive experiments on state-of-the-art models, we find that even the strongest models achieve only slightly above 30% success rates, highlighting that professional long-horizon GUI workflows remain highly challenging for current GUI agents. Further analysis reveals that current agents struggle to maintain long-horizon workflow consistency, frequently exhibiting workflow stage omission, error propagation, objective drift, and insufficient understanding of professional software environments. Our findings provide important insights into the limitations of current agent systems and suggest key directions for the next generation of GUI-agent research.
Governing equations provide compact descriptions of physical systems, yet the variables in which they are simple are often hidden in high-dimensional measurements. This challenge is sharper for forced systems, whose responses depend on both intrinsic dynamics and time-dependent inputs. Here we introduce FLARE, a forced latent autoencoder for response equations that learns compact response coordinates, identifies sparse input-dependent latent dynamics and decodes equation rollouts to full responses. By estimating latent dimension from data and separating state estimation from external forcing, FLARE enables forecasts to be initialized from past responses and driven by prescribed future inputs. Across known dynamical systems, application-scale forced responses and visual observations, FLARE recovers compact forced dynamics and predicts long-horizon high-dimensional responses under inputs not used for training. By turning learned coordinates into a dynamical interface, FLARE extends equation discovery to systems whose effective states are hidden within complex observations, providing a route for interpretable modelling and prediction of high-dimensional responses in forced dynamical systems.
Controllable image-to-video (I2V) generation transforms a reference image into a coherent video guided by user-specified control signals. While precise control over camera motion, object motion, and lighting is essential for high-fidelity creation, existing methods often treat these factors independently. This overlooks the physical coupling among viewpoint, geometry, and illumination in dynamic scenes, leading to visual inconsistencies such as mismatched shadows and perspective drift under simultaneous changes. We present VidCRAFT3, a unified and flexible I2V framework that explicitly models cross-factor interactions among geometry, motion, and illumination, enabling both independent and joint control over camera motion, object motion, and lighting direction. Image2Cloud provides explicit 3D geometric priors for accurate camera motion control. ObjMotionNet encodes sparse object trajectories into multi-scale motion features to guide realistic object motion. A Spatial Triple-Attention Transformer integrates lighting direction through lighting cross-attention for consistent relighting. To address the scarcity of jointly annotated data, we construct the VideoLightingDirection (VLD) dataset with accurate per-frame lighting direction annotations, and introduce a three-stage progressive training strategy that enables robust learning without fully joint annotations. Extensive experiments demonstrate that VidCRAFT3 achieves state-of-the-art performance in control precision and visual coherence across diverse scenarios.
With the prevalence of artificial intelligence (AI)-generated content, such as audio deepfakes, a large body of recent work has focused on developing deepfake detection techniques. However, existing benchmarks employ a narrow set of datasets, leaving detector generalization to real-world conditions uncertain. In this paper, we systematically review 31 existing audio deepfake datasets and present an open-source benchmarking toolkit called AUDDT (https://github.com/MuSAELab/AUDDT). The goal of this toolkit is to automate the evaluation of pretrained detectors across a wide range of speech and non-speech audio datasets, giving users direct feedback on the advantages and shortcomings of their deepfake detectors under diverse manipulation types and recording conditions. We start by showcasing the usage of the developed toolkit, the composition of our benchmark, and the breakdown of different deepfake subgroups. Next, we highlight how AUDDT differs from existing benchmarking efforts by enabling large-scale, diverse evaluation across modern spoofing methods and richer attribute-level analysis through comprehensive metadata annotation. Using a widely adopted pretrained deepfake detector, we present in- and out-of-domain detection results, revealing notable performance variability across different conditions and audio manipulation types. Lastly, we also analyze the limitations of these existing datasets and their gaps relative to practical deployment scenarios.
Current full-duplex (FD) spoken dialogue systems can produce fluid interactions, yet it remains unclear whether they can adapt their turn-taking behavior when explicitly instructed. This is critical for real-world deployment, where conversational policies vary across applications (e.g., proactive tutoring vs. passive counseling). We introduce Instruct-FD, an instruction-conditioned benchmark for evaluating controllable turn management in FD systems. To enable this, we develop a human-validated, scalable synthetic pipeline that generates instruction-conditioned conversations, along with a deployment-agnostic multi-turn evaluation protocol and an LLM-based judge. Benchmarking six state-of-the-art full-duplex systems reveals a substantial gap in instruction-following turn management: the best model achieves only 64.4% adherence. Performance is highly uneven across behaviors and scenarios, with proactive behaviors such as model backchanneling and interruption remaining particularly challenging. These findings establish instruction-following turn management as a crucial direction for building adaptable and deployable full-duplex dialogue systems.