arXiv subjects

Zihao Wang

Publications and source records attributed to Zihao Wang.

At least 163 records · Page 9Linked to original sources

arXiv2025

OpenHA: A Series of Open-Source Hierarchical Agentic Models in Minecraft

The choice of action spaces is a critical yet unresolved challenge in developing capable, end-to-end trainable agents. This paper first presents a large-scale, systematic comparison of prominent abstracted action spaces and tokenizers for Vision-Language-Action (VLA) or hierarchical agent models in the open-ended Minecraft. Our analysis reveals that no single action space is universally optimal; instead, the most effective abstraction is highly task-dependent, creating a dilemma for building generalist agents. To resolve this, we introduce Chain of Action (CoA), a novel framework that unifies high-level planning and low-level control within a single, monolithic VLA model. CoA treats an abstracted action not as a command for a separate policy, but as an intermediate reasoning step--akin to a chain of thought--that guides the generation of the final, executable action. Furthermore, we demonstrate that an All-in-One agent trained on a diverse mixture of action spaces using the CoA paradigm learns a more robust and generalizable policy. This unified agent achieves a new state-of-the-art, improving the overall task success rate over strong, specialized baselines. To foster reproducible research, we release the OpenHA (Open Hierarchical Agents) suite, which includes our comprehensive benchmark of over 800 distinct tasks, curated datasets, source code, and all pretrained model checkpoints at https://github.com/CraftJarvis/OpenHA

cs.AI↗

arXiv2025

JARVIS-VLA: Post-Training Large-Scale Vision Language Models to Play Visual Games with Keyboards and Mouse

Recently, action-based decision-making in open-world environments has gained significant attention. Visual Language Action (VLA) models, pretrained on large-scale web datasets, have shown promise in decision-making tasks. However, previous work has primarily focused on action post-training, often neglecting enhancements to the foundational model itself. In response, we introduce a novel approach, Act from Visual Language Post-Training, which refines Visual Language Models (VLMs) through visual and linguistic guidance in a self-supervised manner. This enhancement improves the models' capabilities in world knowledge, visual recognition, and spatial grounding in open-world environments. Following the above post-training paradigms, we obtain the first VLA models in Minecraft that can follow human instructions on over 1k different atomic tasks, including crafting, smelting, cooking, mining, and killing. Our experiments demonstrate that post-training on non-trajectory tasks leads to a significant 40% improvement over the best agent baseline on a diverse set of atomic tasks. Furthermore, we demonstrate that our approach surpasses traditional imitation learning-based policies in Minecraft, achieving state-of-the-art performance. We have open-sourced the code, models, and datasets to foster further research. The project page can be found in https://craftjarvis.github.io/JarvisVLA.

cs.CV↗

arXiv2025

FPI-Det: a face--phone Interaction Dataset for phone-use detection and understanding

The widespread use of mobile devices has created new challenges for vision systems in safety monitoring, workplace productivity assessment, and attention management. Detecting whether a person is using a phone requires not only object recognition but also an understanding of behavioral context, which involves reasoning about the relationship between faces, hands, and devices under diverse conditions. Existing generic benchmarks do not fully capture such fine-grained human--device interactions. To address this gap, we introduce the FPI-Det, containing 22{,}879 images with synchronized annotations for faces and phones across workplace, education, transportation, and public scenarios. The dataset features extreme scale variation, frequent occlusions, and varied capture conditions. We evaluate representative YOLO and DETR detectors, providing baseline results and an analysis of performance across object sizes, occlusion levels, and environments. Source code and dataset is available at https://github.com/KvCgRv/FPI-Det.

cs.CV↗

arXiv2025

Meissner Effect and Nonreciprocal Charge Transport in Non-Topological 1T-CrTe2/FeTe Heterostructures

Interface-induced superconductivity has recently been achieved by stacking a magnetic topological insulator layer on an antiferromagnetic FeTe layer. However, the mechanism driving this emergent superconductivity remains unclear. Here, we employ molecular beam epitaxy to grow a 1T-CrTe2 layer, a two-dimensional ferromagnet with a Curie temperature up to room temperature, on a FeTe layer. These 1T-CrTe2/FeTe heterostructures show superconductivity with a critical temperature of ~12 K. Through magnetic force microscopy measurements, we observe the Meissner effect on the surface of the 1T-CrTe2 layer. Our electrical transport measurements reveal that the 1T-CrTe2/FeTe heterostructures exhibit nonreciprocal charge transport behavior, characterized by a large magneto-chiral anisotropy coefficient. The enhanced nonreciprocal charge transport in 1T-CrTe2/FeTe heterostructures provides a promising platform for exploring the magnetically controllable superconducting diode effect.

cond-mat.supr-con↗

arXiv2025

Orbital Hybridization-Induced Ising-Type Superconductivity in a Confined Gallium Layer

In low-dimensional superconductors, the interplay between quantum confinement and interfacial hybridization effects can reshape Cooper pair wavefunctions and induce novel forms of unconventional superconductivity. In this work, we employ a plasma-free, carbon buffer layer-assisted confinement epitaxy method to synthesize trilayer gallium (Ga) sandwiched between a graphene layer and a 6H-SiC(0001) substrate, forming an air-stable graphene/trilayer Ga/SiC heterostructure. In this confined light-element Ga layer, we demonstrate interfacial Ising-type superconductivity driven by atomic orbital hybridization between the Ga layer and the SiC substrate. Electrical transport measurements reveal that the in-plane upper critical magnetic field u0Hc2,|| reaches ~21.98T at T=400 mK, approximately 3.38 times the Pauli paramagnetic limit (~6.51T). Angle-resolved photoemission spectroscopy (ARPES) measurements combined with theoretical calculations confirm the presence of split Fermi surfaces with Ising-type spin textures at the K and K' valleys of the confined Ga layer strongly hybridized with SiC. Moreover, by incorporating finite relaxation time induced by impurity scattering into an Ising-type superconductivity model, we reproduce the entire temperature-dependent u0Hc2,|| phase diagram. This work establishes a new strategy to realize unconventional pairing wavefunctions by combining quantum confinement and interfacial hybridization effects in superconducting thin films. It also opens new avenues for designing scalable superconducting quantum electronic and spintronic devices through interfacial engineering.

cond-mat.mes-hall↗

arXiv2025

UI-TARS-2 Technical Report: Advancing GUI Agent with Multi-Turn Reinforcement Learning

The development of autonomous agents for graphical user interfaces (GUIs) presents major challenges in artificial intelligence. While recent advances in native agent models have shown promise by unifying perception, reasoning, action, and memory through end-to-end learning, open problems remain in data scalability, multi-turn reinforcement learning (RL), the limitations of GUI-only operation, and environment stability. In this technical report, we present UI-TARS-2, a native GUI-centered agent model that addresses these challenges through a systematic training methodology: a data flywheel for scalable data generation, a stabilized multi-turn RL framework, a hybrid GUI environment that integrates file systems and terminals, and a unified sandbox platform for large-scale rollouts. Empirical evaluation demonstrates that UI-TARS-2 achieves significant improvements over its predecessor UI-TARS-1.5. On GUI benchmarks, it reaches 88.2 on Online-Mind2Web, 47.5 on OSWorld, 50.6 on WindowsAgentArena, and 73.3 on AndroidWorld, outperforming strong baselines such as Claude and OpenAI agents. In game environments, it attains a mean normalized score of 59.8 across a 15-game suite-roughly 60% of human-level performance-and remains competitive with frontier proprietary models (e.g., OpenAI o3) on LMGame-Bench. Additionally, the model can generalize to long-horizon information-seeking tasks and software engineering benchmarks, highlighting its robustness across diverse agent tasks. Detailed analyses of training dynamics further provide insights into achieving stability and efficiency in large-scale agent RL. These results underscore UI-TARS-2's potential to advance the state of GUI agents and exhibit strong generalization to real-world interactive scenarios.

cs.AI↗

arXiv2025

Model Unmerging: Making Your Models Unmergeable for Secure Model Sharing

Model merging leverages multiple finetuned expert models to construct a multi-task model with low cost, and is gaining increasing attention. However, as a growing number of finetuned models become publicly available, concerns about the safety of model merging have emerged. Unauthorized merging may infringe on developers' rights and risk leaking sensitive personal information. Most existing methods focus on detecting whether a merged model originates from a specific source model, but fail to effectively prevent illegal merging. In this paper, we propose MergeLock, an active protection mechanism that disrupts model parameters to render them unmergeable, thereby directly preventing unauthorized model merging. Specifically, leveraging the inherent symmetry of the attention mechanism in Transformer-based models, we randomly sample two pairs of invertible matrices and apply them to the Query-Key (QK) and Value-Output (VO) branches. This transformation keeps the model's output unchanged while pushing it away from the shared parameter space of other finetuned models. Extensive experiments across both vision and language tasks demonstrate that MergeLock can degrade the performance of merged models by over 95% when a protected model is involved in most cases, demonstrating its effectiveness. Moreover, we further demonstrate that merged models protected by MergeLock cannot be effectively recovered using low-cost restoration methods, further enhancing robustness against unauthorized merging. The code is available at https://github.com/hetailang/Merge-Lock.

cs.LG↗

arXiv2025

Open-World Skill Discovery from Unsegmented Demonstrations

Learning skills in open-world environments is essential for developing agents capable of handling a variety of tasks by combining basic skills. Online demonstration videos are typically long but unsegmented, making them difficult to segment and label with skill identifiers. Unlike existing methods that rely on sequence sampling or human labeling, we have developed a self-supervised learning-based approach to segment these long videos into a series of semantic-aware and skill-consistent segments. Drawing inspiration from human cognitive event segmentation theory, we introduce Skill Boundary Detection (SBD), an annotation-free temporal video segmentation algorithm. SBD detects skill boundaries in a video by leveraging prediction errors from a pretrained unconditional action-prediction model. This approach is based on the assumption that a significant increase in prediction error indicates a shift in the skill being executed. We evaluated our method in Minecraft, a rich open-world simulator with extensive gameplay videos available online. Our SBD-generated segments improved the average performance of conditioned policies by 63.7% and 52.1% on short-term atomic skill tasks, and their corresponding hierarchical agents by 11.3% and 20.8% on long-horizon tasks. Our method can leverage the diverse YouTube videos to train instruction-following agents. The project page can be found in https://craftjarvis.github.io/SkillDiscovery.

cs.CV↗

arXiv2025

UQ: Assessing Language Models on Unsolved Questions

Benchmarks shape progress in AI research. A useful benchmark should be both difficult and realistic: questions should challenge frontier models while also reflecting real-world usage. Yet, current paradigms face a difficulty-realism tension: exam-style benchmarks are often made artificially difficult with limited real-world value, while benchmarks based on real user interaction often skew toward easy, high-frequency problems. In this work, we explore a radically different paradigm: assessing models on unsolved questions. Rather than a static benchmark scored once, we curate unsolved questions and evaluate models asynchronously over time with validator-assisted screening and community verification. We introduce UQ, a testbed of 500 challenging, diverse questions sourced from Stack Exchange, spanning topics from CS theory and math to sci-fi and history, probing capabilities including reasoning, factuality, and browsing. UQ is difficult and realistic by construction: unsolved questions are often hard and naturally arise when humans seek answers, thus solving them yields direct real-world value. Our contributions are threefold: (1) UQ-Dataset and its collection pipeline combining rule-based filters, LLM judges, and human review to ensure question quality (e.g., well-defined and difficult); (2) UQ-Validators, compound validation strategies that leverage the generator-validator gap to provide evaluation signals and pre-screen candidate solutions for human review; and (3) UQ-Platform, an open platform where experts collectively verify questions and solutions. The top model passes UQ-validation on only 15% of questions, and preliminary human verification has already identified correct answers among those that passed. UQ charts a path for evaluating frontier models on real-world, open-ended challenges, where success pushes the frontier of human knowledge. We release UQ at https://uq.stanford.edu.

cs.CL↗

arXiv2025

Zeolitic imidazolate framework glasses emit white light

Zeolitic imidazolate framework (ZIF) glasses represent a newly emerged class of melt-quenched glasses, characterized by their intrinsic nanoporous structure, good processability, and multifunctionalities such as gas separation and energy storage. However, creating photonic functionalities in Zn-based ZIF glasses remains elusive. Here we show a remarkable broadband white light-emitting behavior in a Zn-based ZIF glass, which can be enhanced by annealing. Furthermore, we discovered a sharp red shift upon increasing annealing temperature above the critical temperature of 1.07Tg, where Tg is the glass transition temperature, for a short duration of 30 min. Finally, we achieved a high absolute internal photoluminescence quantum yield of 12.2% upon annealing of ZIF glass at 1.13Tg. Based on the optimally annealed ZIF glass, we fabricated a white light-emitting diode (LED) with the luminous efficacy of 4.2 lm/W and high operational stability, retaining 74.1% of its initial luminous efficacy after 180 min of continuous operation. These results not only demonstrate the feasibility of utilizing ZIF glasses in LED applications but also mark a significant advancement in the development of durable, efficient, and multifunctional photonic materials.

physics.optics↗

arXiv2025

Echo-4o: Harnessing the Power of GPT-4o Synthetic Images for Improved Image Generation

Recently, GPT-4o has garnered significant attention for its strong performance in image generation, yet open-source models still lag behind. Several studies have explored distilling image data from GPT-4o to enhance open-source models, achieving notable progress. However, a key question remains: given that real-world image datasets already constitute a natural source of high-quality data, why should we use GPT-4o-generated synthetic data? In this work, we identify two key advantages of synthetic images. First, they can complement rare scenarios in real-world datasets, such as surreal fantasy or multi-reference image generation, which frequently occur in user queries. Second, they provide clean and controllable supervision. Real-world data often contains complex background noise and inherent misalignment between text descriptions and image content, whereas synthetic images offer pure backgrounds and long-tailed supervision signals, facilitating more accurate text-to-image alignment. Building on these insights, we introduce Echo-4o-Image, a 180K-scale synthetic dataset generated by GPT-4o, harnessing the power of synthetic image data to address blind spots in real-world coverage. Using this dataset, we fine-tune the unified multimodal generation baseline Bagel to obtain Echo-4o. In addition, we propose two new evaluation benchmarks for a more accurate and challenging assessment of image generation capabilities: GenEval++, which increases instruction complexity to mitigate score saturation, and Imagine-Bench, which focuses on evaluating both the understanding and generation of imaginative content. Echo-4o demonstrates strong performance across standard benchmarks. Moreover, applying Echo-4o-Image to other foundation models (e.g., OmniGen2, BLIP3-o) yields consistent performance gains across multiple metrics, highlighting the datasets strong transferability.

cs.CV↗

arXiv2025

Meta-Learning for Speeding Up Large Model Inference in Decentralized Environments

The deployment of large-scale models, such as large language models (LLMs), incurs substantial costs due to their computational demands. To mitigate these costs and address challenges related to scalability and data security, there is a growing shift towards decentralized systems for model deployment, where choosing efficient inference acceleration schemes become crucial to manage computational resources effectively and enhance system responsiveness. In this work, we address the challenge of selecting optimal acceleration methods in decentralized systems by introducing a meta-learning-based framework. This framework automates the selection process by learning from historical performance data of various acceleration techniques across different tasks. Unlike traditional methods that rely on random selection or expert intuition, our approach systematically identifies the best acceleration strategies based on the specific characteristics of each task. We demonstrate that our meta-learning framework not only streamlines the decision-making process but also consistently outperforms conventional methods in terms of efficiency and performance. Our results highlight the potential of inference acceleration in decentralized AI systems, offering a path towards more democratic and economically feasible artificial intelligence solutions.

cs.LG↗

arXiv2025

Model Reveals What to Cache: Profiling-Based Feature Reuse for Video Diffusion Models

Recent advances in diffusion models have demonstrated remarkable capabilities in video generation. However, the computational intensity remains a significant challenge for practical applications. While feature caching has been proposed to reduce the computational burden of diffusion models, existing methods typically overlook the heterogeneous significance of individual blocks, resulting in suboptimal reuse and degraded output quality. To this end, we address this gap by introducing ProfilingDiT, a novel adaptive caching strategy that explicitly disentangles foreground and background-focused blocks. Through a systematic analysis of attention distributions in diffusion models, we reveal a key observation: 1) Most layers exhibit a consistent preference for either foreground or background regions. 2) Predicted noise shows low inter-step similarity initially, which stabilizes as denoising progresses. This finding inspires us to formulate a selective caching strategy that preserves full computation for dynamic foreground elements while efficiently caching static background features. Our approach substantially reduces computational overhead while preserving visual fidelity. Extensive experiments demonstrate that our framework achieves significant acceleration (e.g., 2.01 times speedup for Wan2.1) while maintaining visual fidelity across comprehensive quality metrics, establishing a viable method for efficient video generation.

cs.CV↗

arXiv2025

Quantum thermodynamics in a rotating BTZ black hole spacetime

We address the problem of the thermalization process for an Unruh-DeWitt (UDW) detector outside a BTZ black hole, from a perspective of quantum thermodynamics. In the context of an open quantum system, we derive the complete dynamics of the detector, which encodes a complicated response to scalar background fields. Using various information theory tools, such as quantum relative entropy, quantum heat, coherence, quantum Fisher information, and quantum speed of evolution, we examined three quantum thermodynamic laws for the UDW detector, where the influences from BTZ angular momentum and Hawking radiation are investigated. In particular, based on information geometry theory, we find an intrinsic asymmetry in the detector's thermolization process as it undergoes Hawking radiation from the BTZ black hole. In particular, we find that the detector consistently heats faster than it cools, analogous to the quantum Mpemba effect for nonequilibrium systems. Moreover, we demonstrate that the spin of a black hole significantly influences the magnitude of the asymmetry, while preserving the dominance of heating over cooling.

hep-th↗

arXiv2025

DailyLLM: Context-Aware Activity Log Generation Using Multi-Modal Sensors and LLMs

Rich and context-aware activity logs facilitate user behavior analysis and health monitoring, making them a key research focus in ubiquitous computing. The remarkable semantic understanding and generation capabilities of Large Language Models (LLMs) have recently created new opportunities for activity log generation. However, existing methods continue to exhibit notable limitations in terms of accuracy, efficiency, and semantic richness. To address these challenges, we propose DailyLLM. To the best of our knowledge, this is the first log generation and summarization system that comprehensively integrates contextual activity information across four dimensions: location, motion, environment, and physiology, using only sensors commonly available on smartphones and smartwatches. To achieve this, DailyLLM introduces a lightweight LLM-based framework that integrates structured prompting with efficient feature extraction to enable high-level activity understanding. Extensive experiments demonstrate that DailyLLM outperforms state-of-the-art (SOTA) log generation methods and can be efficiently deployed on personal computers and Raspberry Pi. Utilizing only a 1.5B-parameter LLM model, DailyLLM achieves a 17% improvement in log generation BERTScore precision compared to the 70B-parameter SOTA baseline, while delivering nearly 10x faster inference speed.

cs.AI↗

arXiv2025

Enhancing Transformers for Generalizable First-Order Logical Entailment

Transformers, as the fundamental deep learning architecture, have demonstrated great capability in reasoning. This paper studies the generalizable first-order logical reasoning ability of transformers with their parameterized knowledge and how to improve it. Transformers' capability of first-order reasoning is further captured by whether they can conduct first-order logical entailment, which is quantitatively measured by their performance in answering knowledge graph queries. We establish the connections between (1) two types of distribution shifts studied in out-of-distribution generalization and (2) unseen knowledge and query settings discussed in the task of knowledge graph query answering, which makes it possible to characterize the fine-grained generalizability. Results on our comprehensive dataset showed that transformers \textit{outperform} previous methods designed particularly for this task and provided detailed empirical evidence about the impact of the input query syntax, token embedding, and transformer architectures on their reasoning capability. Interestingly, our results revealed the mismatch of positional encoding and other design choices of transformer architectures in previous practices. Motivated by this, we propose TEGA, a logic-aware architecture that significantly improves the performance in generalizable first-order logical entailment.

cs.CL↗

arXiv2025

Learning Federated Neural Graph Databases for Answering Complex Queries from Distributed Knowledge Graphs

The increasing demand for deep learning-based foundation models has highlighted the importance of efficient data retrieval mechanisms. Neural graph databases (NGDBs) offer a compelling solution, leveraging neural spaces to store and query graph-structured data, thereby enabling LLMs to access precise and contextually relevant information. However, current NGDBs are constrained to single-graph operation, limiting their capacity to reason across multiple, distributed graphs. Furthermore, the lack of support for multi-source graph data in existing NGDBs hinders their ability to capture the complexity and diversity of real-world data. In many applications, data is distributed across multiple sources, and the ability to reason across these sources is crucial for making informed decisions. This limitation is particularly problematic when dealing with sensitive graph data, as directly sharing and aggregating such data poses significant privacy risks. As a result, many applications that rely on NGDBs are forced to choose between compromising data privacy or sacrificing the ability to reason across multiple graphs. To address these limitations, we propose to learn Federated Neural Graph DataBase (FedNGDB), a pioneering systematic framework that empowers privacy-preserving reasoning over multi-source graph data. FedNGDB leverages federated learning to collaboratively learn graph representations across multiple sources, enriching relationships between entities, and improving the overall quality of graph data.

cs.LG↗

arXiv2025

A Recipe for Causal Graph Regression: Confounding Effects Revisited

Through recognizing causal subgraphs, causal graph learning (CGL) has risen to be a promising approach for improving the generalizability of graph neural networks under out-of-distribution (OOD) scenarios. However, the empirical successes of CGL techniques are mostly exemplified in classification settings, while regression tasks, a more challenging setting in graph learning, are overlooked. We thus devote this work to tackling causal graph regression (CGR); to this end we reshape the processing of confounding effects in existing CGL studies, which mainly deal with classification. Specifically, we reflect on the predictive power of confounders in graph-level regression, and generalize classification-specific causal intervention techniques to regression through a lens of contrastive learning. Extensive experiments on graph OOD benchmarks validate the efficacy of our proposals for CGR. The model implementation and the code are provided on https://github.com/causal-graph/CGR.

cs.LG↗