Searcharxiv⌕ Search

arXiv subjects

Wentao Zhang

Publications and source records attributed to Wentao Zhang.

At least 19 recordsLinked to original sources
arXiv2026

SkillLens: Visual Skill Cards for Retrieval-Augmented GUI Action Prediction and On-Policy Distillation

Graphical user interface (GUI) agents can reuse past interaction experience. We study how illustrated procedures support action prediction and students deployed without retrieval. We introduce SkillLens, which represents procedures, applicability cues, and interface views in a shared Visual Skill Card (VSC) schema. Retrieved cards provide a frozen executor with reference evidence alongside the current screen. CardDistill uses this schema to construct privileged teacher context from labeled training steps for on-policy distillation; the student learns from native inputs and requires no retrieval at deployment. With matched training data and update counts, students trained with full-card teachers improve step success rate by an average of 5.04 percentage points over those trained with compact-text teachers across three training seeds on 119 tasks from ten websites absent from training. An independent runtime control uses training-source memory and pre-action retrieval queries: full cards achieve 12/64 correct actions (18.75%) versus 4/64 (6.25%) for text-only cards. These results establish the value of the complete VSC teacher-context package for retrieval-free students and provide evidence for pre-action retrieval of illustrated procedures.

cs.AI↗
arXiv2026

PDRVLA: A Vision-Language-Action Model with Physical Dynamics Representations

Vision--Language--Action (VLA) models generate robot actions from visual observations and language instructions, but their understanding of physical dynamics remains limited. Video world models can capture scene dynamics, yet generating future videos introduces high latency into the control loop. We introduce PDRVLA, whose lightweight Physical Dynamics Module extracts physical tokens from vision--language features to condition action generation without online video generation. The module learns through direct prediction of future RGB, depth, and optical-flow features, with auxiliary supervision through a frozen pretrained video world model. Training relies entirely on publicly available data, expanded through human-to-robot synthesis, subtask annotation, and embodied chain-of-thought (ECoT) annotation. Across four simulation benchmarks, PDRVLA achieves success rates of 99.3% on LIBERO, 89.9% on LIBERO-Plus, 93.6% on RoboTwin~2.0, and 81.5% on SimplerEnv. Ablations show improved future prediction with world-model supervision and evaluate how subtask conditioning, future object-pose supervision, and ECoT training affect action success.

cs.RO↗
arXiv2026

Ask Before You Answer: A Question-Driven Evidence Reasoning Architecture (QERA) for In-Context Learning

AI assistants are increasingly asked to reason over a user's own web traces: group chats, community threads, activity logs and revision histories. On CL-bench Life, a benchmark of such real-life context learning, frontier LLMs solve fewer than one task in five. We find that the request is not the specification: over 80% of the benchmark's rubrics share less than 60% of their content words with the final request, as for a rule set two turns earlier or every record that meets the user's criterion, and LLMs answering directly pass only 67% of these implicit rubrics against 81% of the others. We propose QERA, a Question-driven Evidence Reasoning Architecture that asks before it answers. QERA first turns the request into a requirement checklist and typed information needs, filtered by code for faithfulness to the user's words; it then answers from the whole conversation, asks the same questions again with quotes verified by code, and applies a guarded local repair. It needs no training or rubrics and makes about four LLM calls per task. On all 405 tasks with four frontier LLMs, QERA attains the highest solving rate on every LLM (21.0% on average, against 18.2% for the strongest of three inference-time baselines and 14.6% for direct answering) and leads every baseline significantly when pooled across LLMs, while using fewer tokens than PSCI and 4-5% of the tokens of Ctx2Skill. Ablations on GPT-6-Luna show that the gain is settled before the answer, by a fixed answer protocol (+3.8 rubric points) and the request-specific questions (+1.6), and that it is concentrated on implicit rubrics, with checking afterwards refining mainly numeric requirements.

cs.CL↗
arXiv2026

VibeVideoBench: Evaluating Multimodal Agents on Code-Driven Video Generation

Diffusion video models produce striking clips, but they cannot reliably render an exact formula, a correct chart, a rule-consistent game or narration synchronized to on-screen events. A different route is emerging: multimodal agents that write code to render every frame and synthesize every sound. We introduce VibeVideoBench, a benchmark for evaluating agents on code-driven video generation. Its 239 self-contained tasks in English and Chinese span 12 categories, from data stories and music-driven pieces to classroom lessons and paper explainers; each asks an agent, running in its native harness inside a sandbox with no image or video generation model, to deliver a finished MP4 with sound within one hour. Evaluation is hierarchical and evidence-based: program checks of the delivery spec, about 13k task-specific rubric checks traced to the requirements they enforce, and five quality dimensions anchored per task, scored by three VLM judge panels. Across nine frontier agent systems and 2,151 scored rollouts, GPT-6 Astra and Claude Opus 5.5 rank first and second with both the judges and human annotators, and the strongest system satisfies every core requirement on 76.3% of tasks. Delivery is a second axis: counting missing videos as zero changes the ranking, and three quarters of the lost videos are final renders that started too late, ran too slowly or never started, a failure of how agents plan their hour. Agents build their own rendering pipelines and inspect their own frames, yet how much they look does not predict quality. The judges' ranking of the systems closely matches blind human preferences.

cs.CV↗
arXiv2026

Machine Learning Hamiltonians with Polarizations for Heterogeneous Systems

Machine learning Hamiltonians use local atomic structures to determine local Hamiltonian matrix elements without carrying out self-consistent calculations. They have been used to study large systems and to carry out high-throughput screening for their electronic and optical properties. The prediction of the Hamiltonian is essentially the prediction of the electron electrostatic potential energy; however, for heterogeneous systems, this potential energy is often determined by faraway features, such as interface or surface dipoles, or even an applied voltage in a nanodevice. Thus, it cannot be determined by local atomic structures alone. In this sense, one needs to solve the electron electrostatic potential energy self-consistently using intrinsic atomic dipoles, the applied external potential energy, and polarizabilities. Fortunately, the intrinsic dipoles and polarizabilities are local properties, whereas the applied potential energy is a prescribed input. In this work, we propose DeepH-Polar, a physically informed machine learning Hamiltonian model that explicitly predicts intrinsic atomic dipoles and polarizability tensors and uses them to solve the electron electrostatic potential energy self-consistently. Benchmarks on polar GaN systems and crystalline-Si/amorphous-SiO$_2$ interfaces demonstrate improved Hamiltonian accuracy and reproduction of thickness-dependent electrostatic variations, as well as spatially nonuniform field-induced screening that models without polarization fail to capture. Our model also allows us to study nanoscale electronic devices. By iteratively solving the Hamiltonian eigenproblem and self-consistently updating free-carrier occupations and the electrostatic potential, we simulate an 8,194-atom gated semiconductor device.

physics.chem-ph↗
arXiv2026

EnvSched: Environment-Aware Scheduling for Rollout-Efficient Agent Reinforcement Learning

As executable environments for language agents scale, generating and verifying their rollouts becomes increasingly expensive. Under finite rollout budgets, this raises an important question: can data management at the environment level make agentic reinforcement learning (RL) more efficient by organizing, evaluating, and scheduling task instances together with the environments that determine their tools, states, transitions, and verification? To address this question, we introduce EnvSched (Environment-aware Scheduling), an online scheduler that learns from rollout history. EnvSched assigns scheduling value through two complementary modeling choices: exploitation prioritizes tasks likely to provide informative reward differences, while coverage spreads each batch across underobserved environments and interaction patterns. EnvSched organizes environment--task pairs as nodes in a dual-relation graph, with within-environment edges capturing task semantics and shared tool, state, and verifier structure, and cross-environment edges capturing abstract execution and reward topology. By propagating historical rollout statistics along both edge types, it predicts within-task reward variance. These predictions guide a scheduler that allocates each batch to tasks with high predicted reward variance and underobserved environments and tasks. Environment quotas and a similarity penalty further limit concentration and redundancy, yielding informative and diverse batches without additional pilot rollouts. In RL on Qwen3-8B, EnvSched achieves an average score of 48.81 across BFCL-v3, tau-Bench, and ACEBench-Agent after 40 training steps, exceeding the highest score attained by random task selection after 120 steps. It also improves downstream performance across model sizes at the same training budget.

cs.LG↗
arXiv2026

FilingLoop: Evaluating and Evolving LLM Agents for Corporate Disclosure Review and Revision

Public disclosure repositories such as SEC EDGAR support informed investment through web access to corporate filings. Supporting their review with language model agents requires discovering unstated deficiencies across related documents, acquiring missing company facts, and verifying that requested revisions satisfy evidence-backed requirements. Existing financial question-answering evaluations provide limited evidence of these interactive capabilities. We introduce \textbf{FilingLoop}, an interactive benchmark and a self-evolving agent framework for corporate disclosure review and revision. Grounded in historical U.S.\ Securities and Exchange Commission correspondence, FilingLoop reconstructs 380 predominantly IPO review tasks from 96 companies under distinct information constraints. An evaluator separate from the acting agents scores final filings and responses against 6,099 eligible requirements and audits unsupported claims and harmful revisions. Our framework couples iterative review and revision within each case with procedural learning across cases, converting requirement-level evaluation feedback into role-specific guidance without updating model weights. Updates affect only subsequent cases, allowing procedural adaptation to be assessed without retrying scored tasks. On the evaluated stream, evolution increases requirement fulfillment from 24.28\% to 27.96\% and full comment resolution from 16.90\% to 20.75\%, with a 5.8\% increase in total cost over the fixed-policy variant. FilingLoop connects the concerns an agent discovers with the company facts it obtains and the revisions it commits, providing requirement-level feedback for procedures used in later review tasks.

cs.AI↗
arXiv2026

SkillLifeBench: Benchmarking Security Across the Lifecycle of Agent Skills

Web agents reuse skills whose stored code, selected version, and loaded implementation can diverge across tasks and updates. SkillLifeBench links these states to task outcomes in 388 paired scenarios spanning implementation, interface, runtime, composition, and version changes. We evaluate 14 models on two paired decision tracks: 89 runtime tasks and 60 autonomous lifecycle tasks spanning 12 templates derived from introduced cases. Lifecycle tasks allow agents to inspect, select, and reconfigure skills while completing verifiable user requests. Joint success requires task completion without the registered attack goal. GPT-5.6-Sol and Sonnet 4.6 both achieve 82.0% on runtime tasks, but reach 96.7% and 56.7% on lifecycle tasks. Across lifecycle evaluations, challenged task success is 77.5%, while joint success is 65.2%; 103 runs complete the task and the attack goal. Adding execution provenance in a matched recovery study raises utility joint success from 49.35% with logs and metadata to 65.58%. A deterministic execution-consistency rule reaches 64.94% with higher benign utility, while full restoration reaches 69.48%. Lifecycle tasks distinguish models with similar runtime scores, and execution evidence improves recovery.

cs.LG↗
arXiv2026

EmbodiedLab: Benchmarking Embodied Agents That Act to Know

Embodied agents are expected to accomplish real-world goals by deciding for themselves how and when to interact with their surroundings. For many such goals, changing the physical state of the world is the means rather than the end: the agent changes the state of a system through interaction, observes the feedback, and explores until it either achieves the goal itself (e.g., discovering physical laws) or learns what the goal requires; it acts to know. Manipulation benchmarks stop at acting to change, with the target state given; scientific discovery benchmarks evaluate knowing but replace physical experimentation with symbolic actions or numerical queries. We introduce EmbodiedLab, a benchmark in which a robot must act to know. Quantitative science experiments make this setting controllable: the parameters of the system are resampled for every instance, so the result cannot be known in advance, yet it can be scored exactly. Across 30 experiments in five disciplines on 14 apparatus platforms, the agent receives an objective and the apparatus, chooses its own procedure, and operates a physics-based simulated laboratory, reading the instruments through its cameras. Evaluation checks what the agent came to know, grounded in readings recorded during the episode, as well as the physical state the task requires; every experiment is verified to be solvable. We evaluate six frontier models in their developers' coding agents with native observation and control, with added perception and motion planning, and with manipulation skills distilled from the strongest agents. With native control, even the strongest agent succeeds less than half the time; for every agent, most failures arise in interacting with the apparatus, and manipulation skills help weaker agents more than perception and planning. The benchmark and its evaluation code are available at https://github.com/HexaFuture/EmbodiedLab.

cs.RO↗
arXiv2026

SteerVTE: Seamless Video Text Editing with Style and Glyph Control

Visual text editing aims to precisely modify text in images and videos while preserving stylistic consistency and visual realism. Despite significant advances in the image domain, video text editing remains largely unexplored: it is a localized task demanding stroke-level precision within small text regions, which compounds the challenges of cross-frame accuracy, temporal coherence, and stylistic fidelity. We introduce SteerVTE, a unified framework that steers a frozen video diffusion model to perform precise Video Text Editing through style and glyph control. Built on a frozen diffusion transformer, SteerVTE attaches a lightweight text context adapter with two complementary modules: a style encoder capturing the original text's visual attributes, and dual-granularity glyph encoders encoding the target text at both the line and character levels. To overcome the inherently weak text rendering priors of video foundation models, we further propose a glyph-aware spatial-focal loss and a three-stage progressive training curriculum that scales from image to video data. To support large-scale training, we also develop an automatic synthesis pipeline and construct SteerVTE-1M, a dataset of one million triplets spanning diverse scenes, fonts, and stylistic effects. Extensive experiments demonstrate that SteerVTE substantially outperforms existing video editing baselines across text accuracy, style consistency, and temporal coherence. Our project page is available at https://zengkaiya.github.io/SteerVTE/.

cs.CV↗
arXiv2026

SpatialScout: Can VLM Agents Actively Explore and Map 3D Scenes?

Spatial intelligence requires agents to actively acquire visual evidence and integrate observations into a coherent representation of the three-dimensional world. Yet spatial question answering offers limited evidence of whether models can construct accurate metric maps through autonomous exploration. We introduce SpatialScout, a benchmark for evaluating vision-language model (VLM) agents on active object-level 3D mapping. SpatialScout comprises 80 diverse indoor scenes containing 3,140 target objects. Given an object inventory, RGB observations, camera calibration, and camera poses, agents control a camera to explore each scene and estimate object centers in a shared coordinate frame. The task requires coordinating viewpoint selection, cross-view association, geometric estimation, and map revision under bounded interaction budgets. Evaluation jointly measures localization accuracy and target coverage through object-scale overlap scores that penalize missing predictions. Experiments with frontier VLMs reveal substantial limitations: the strongest evaluated model, Claude Opus 5.5, achieves a mean overlap score of 41.71%, with 47.23% of targets exceeding an intersection-over-union (IoU) threshold of 0.5 and only 3.25% exceeding 0.9. These results reveal a substantial gap between recovering coarse object layouts and constructing complete, precise metric maps. SpatialScout provides a quantitative testbed for developing agents that build reliable 3D spatial representations through active perception. Code is available at https://github.com/JesseZZZZZ/SpatialScout.

cs.CV↗
arXiv2026

Long-MDR: Long-Context Reinforcement Learning for Multimodal Deep-Research Agents

The next generation of multimodal research agents must reason over long-lived research histories rather than short model completions. During a single task, an agent may repeatedly search the web, inspect visual evidence, revisit earlier hypotheses, and accumulate tens of thousands of tokens of multimodal context. Despite this trend, online RL for multimodal research agents remains largely confined to shorter contexts and interaction horizons. We push online RL training to 128k context and 75+ tool-interaction turns. To our knowledge, this is the first online multimodal deep-research RL study trained at 128k context, and the first trained with a 75 tool-turn horizon. Scaling to this regime exposes several practical limitations of conventional RL training. Early in training, weak policies make poor use of large interaction budgets, causing expensive rollouts with little reward improvement. Later, policy entropy can collapse before performance has saturated, prematurely ending useful learning. We introduce Long-MDR, a three-component training recipe designed specifically for this setting: On-Policy Distillation Warmup, Progressive Horizon Expansion, and Entropy-Triggered Rescue. Together, these techniques improve both the learning efficiency and stability of long-horizon RL, enabling continued gains in a regime where direct training is slow and costly. Code will be available at https://github.com/navileon/Long-MDR

cs.CV↗
arXiv2026

MemT-1.0: Towards Memory-Native Evolution

Memory-native evolution requires making memory an intrinsic capability of foundation models, enabling them to continuously accumulate and utilize information. To this end, we introduce MemT-1.0, a new architecture of memory foundation models that internalize memory as part of model computation. Building on Metis, MemT-1.0 employs Gated DeltaNet-2 with a stable token-level memory update mechanism within each turn. We further introduce Mixture-of-Storage to scale native memory capacity. Following a deliberate backbone preparation, we develop a two-stage training pipeline consisting of continued pre-training and post-training, supported by 34B tokens of training data and efficient training and inference infrastructure. We release a model family comprising MemT-1.0-Nano and MemT-1.0-Flash, and conduct a comprehensive evaluation of memory and general capabilities. The results show that they outperform the parametric memory baselines on average and remain competitive with retrieval-augmented methods, while largely preserving general capabilities. At a history length of 128K tokens, MemT-1.0-Nano achieves approximately $1.86\times$ end-to-end inference speedup and a $66\times$ reduction in persistent storage relative to full context KV cache. By internalizing retrieved procedural experience, MemT-1.0 improves task success on agentic tasks, providing a practical step toward memory-native evolution.

cs.CL↗
arXiv2026

ProtGlycanDock: Towards Accurate Protein-Glycan Docking with Tailored Dataset, Benchmark and Model

Performing accurate protein-glycan docking is critical to understanding protein-glycan interactions which govern many biological processes. However, this important task is hampered by the lack of tailored datasets, benchmarks and models for protein-glycan docking. To fill this blank, we first curate a standard benchmark ProtGlycanDock for training and evaluating protein-glycan docking models under both protein and glycan generalization settings. On this benchmark, we evaluate ten models in three categories, physics-based docking methods, machine-learning-based docking models and molecular structure prediction models. To obtain more precise binding poses, we further propose two finetuning techniques to inject the knowledge of glycan conformations into a general-purpose molecular structure prediction model. These two techniques respectively augment model inputs with glycan topological features and refine glycan binding pose with a glycan energy loss. Benchmark results show the superiority of AlphaFold 3 among existing methods, and the two proposed finetuning techniques can be well combined to achieve a state-of-the-art model for protein-glycan docking. Also, we perform stratified benchmark analysis and case studies to better understand the strengths and weaknesses of current models, providing insights for future model improvements. We maintain datasets at https://huggingface.co/datasets/ProtGlycanDock/ProtGlycanDock and source codes at https://github.com/kasawa1234/ProtGlycanDock

cs.LG↗
arXiv2026

ReuF2F: Benchmarking Coding Agents on Formalized Undergraduate Research Problems

As AI systems begin to resolve open research problems, reliable verification becomes increasingly important. Formal mathematics provides a trusted framework for mathematical reasoning, enabling agents to produce machine-checkable proofs while conducting mathematical exploration. Existing formal mathematics benchmarks often focus on known results or a small set of prominent open problems, which can become difficult to evaluate reliably once solutions are discovered and widely disseminated. Existing evaluation protocols are also not always aligned with how modern coding agents operate. We therefore introduce ReuF2F, a Lean 4 benchmark of 244 formalized open problems drawn from Research Experiences for Undergraduates (REU), spanning entry-level open research problems across a broad range of mathematical subfields. We expect these less prominent open problems, which tend to attract less attention and may remain unsolved for longer, to be better suited for held-out evaluation of end-to-end open problem resolution in Lean. We adopt a standard coding-agent evaluation setting, placing each problem and its logical negation into a Lean project in a Docker container. We also build Leanimum-agent on mini-SWE-agent to provide a unified harness for fair comparisons across models. Our evaluation finds that even the best-performing model resolves only ~40% of the problems, leaving substantial room for improvement on formalized open research problems.

cs.AI↗
arXiv2026

MMSearch-OPD: Easy Made Hard, Hard Made Learnable for Long-Horizon Multimodal Search Agents

Multimodal deep search is moving from short queries to long-horizon investigations that continually acquire evidence. However, training such agents remains constrained by limited challenging training questions and insufficient process feedback. We introduce MMSearch-OPD, combining controllable data refinement with joint reinforcement learning (RL) and on-policy distillation (OPD). Our Refine pipeline performs targeted expansion of reasoning chains derived from existing simple multimodal questions. Parameterized transformations control search depth and cross-modal evidence requirements while reusing existing image--question pairs. For training, answer rewards and lightweight behavior penalties guide RL, while the reward gap modulates token-level teacher guidance on imperfect student trajectories. An exponentially decaying distillation coefficient reduces teacher dependence as training proceeds. Experiments show gains from data refinement and further improvements from joint training after SFT, indicating stronger long-horizon multimodal deep search capabilities.

cs.CV↗
arXiv2026

The World Won't Wait: Evaluating and Teaching Agents Under a Live Clock

Large language model (LLM) agents can solve complex interactive tasks, yet reasoning and tool calls take time. In continuously changing environments, an otherwise correct action may arrive too late. We introduce RealtimeBench, with 180 tasks across six game types, to evaluate agents while games continue running during inference. Paired evaluations show that GPT-6.1-sol outperforms the faster Jev when games are paused, but falls behind when they keep running. We propose Two-Timescale Offline Recursive Self-Improvement (TORSI): a slow teacher studies failures and tests corrections in paused games between attempts, then writes reusable instructions for a fast student. The student acts without teacher calls during gameplay, and validation scores determine which updates are adopted. Two rounds raise Jev's average test score from 13.51 to 24.19 on a 0--100 progress scale, improving all six game types. The instructions also improve average scores for three other models without retuning. These findings point toward real-time intelligence: enabling powerful LLM agents to remain effective in a continuously changing world by organizing deliberation, learning, and action across timescales.

cs.AI↗
arXiv2026

Flux-OPD: On-Policy Distillation with Evolving Contexts

Large language model training in open-ended domains lacks verifiable rewards, making task preferences difficult to formalize as effective supervision. Contexts derived from task feedback can provide richer supervision, yet fixed contexts cannot adapt as the student evolves, motivating contexts that evolve with student performance. However, directly using evolving contexts as in-training supervision results in unstable distillation targets and conflicting distributions, requiring mechanisms to stabilize the target and downweight conflicts. In this paper, we characterize context distillation through a reverse-KL decomposition, showing that the student is distilled toward the geometric mean of context-conditioned teachers, while the objective contains a conflict term measuring disagreement among these teachers. Motivated by this characterization, we propose Flux-OPD, an OPD paradigm that uses evolving contexts as in-training supervision to incorporate contextual signals derived from task feedback. Flux-OPD treats differences between context-conditioned and context-free teachers as contextual difference signals, injects them as contextual corrections into the context-free teacher, and weights the correction strength using the conflict term. Experiments show that Flux-OPD achieves the best results among OPD paradigms, highlighting the potential of combining teacher supervision with evolving contexts as in-training supervision.

cs.LG↗