arXiv subjects

Boxuan Li

Publications and source records attributed to Boxuan Li.

At least 19 recordsLinked to original sources

arXiv2026

Vision as Unified Multimodal Generation

We formulate computer vision as unified multimodal generation, where heterogeneous visual tasks are expressed in the native text and image generation spaces of a unified multimodal model, without task-specific architectures. Under this formulation, SenseNova-Vision uses natural-language instructions and optional visual prompts to specify tasks, target regions or views, and decoding conventions, and generates responses as text for symbolic outputs, images for dense spatial predictions, or mixed text-and-image outputs for compositional tasks. To support large-scale training, we convert diverse computer vision annotations into instruction-response examples compatible with these generation spaces, resulting in the SenseNova-Vision Corpus, a computer-vision instruction-response corpus spanning text, image, and mixed targets. Starting from an off-the-shelf pretrained unified multimodal model, SenseNova-Vision is trained primarily on this corpus, with auxiliary multimodal data used as a capability-preserving mixture, and requires no task-specific prediction heads or architectural modifications. The resulting model covers a broad range of vision tasks, including detection, OCR, keypoint estimation, segmentation, depth estimation, surface normal prediction, point maps, and camera pose estimation, while supporting language-defined variants that combine category, color, region, and other visual cues. Experiments show that a single unified model can match leading task-specialized systems across structured visual understanding, dense geometric prediction, segmentation, and multi-view visual geometry. These results suggest unified multimodal generation as a scalable route for integrating computer vision capabilities into general-purpose foundation models. The model and corpus are publicly available.

cs.CV↗

arXiv2026

OpenThoughts-Agent: Data Recipes for Agentic Models

Agentic language models dramatically expand the applications of AI yet little is publicly known about how to curate training data for broadly capable agents. Existing open efforts such as SWE-Smith, SERA, and Nemotron-Terminal typically target a single benchmark, leaving open the question of how to train models that generalize across diverse agentic tasks. The OpenThoughts-Agent (OT-Agent) project addresses this gap with a fully open data curation pipeline for training agentic models. We conduct more than 100 controlled ablation experiments to systematically investigate each stage of the pipeline, yielding insights on the importance of task sources and diversity. We then assemble a training set of 100K examples from our pipeline and fine-tune Qwen3-32B on this dataset, which yields an average accuracy of 44.8% across seven agentic benchmarks and a 3.9 percentage point improvement over the strongest existing open data agentic model (Nemotron-Terminal-32B, 40.9%). Moreover, our training data exhibits strong scaling properties, outperforming alternative open datasets at every training set size in compute-controlled comparisons. We publicly release our training sets, data pipeline, experimental data, and models at openthoughts.ai to support future open research on agentic model training.

cs.AI↗

arXiv2026

Hystar: Hypernetwork-driven Style-adaptive Retrieval via Dynamic SVD Modulation

Query-based image retrieval (QBIR) requires retrieving relevant images given diverse and often stylistically heterogeneous queries, such as sketches, artworks, or low-resolution previews. While large-scale vision--language representation models (VLRMs) like CLIP offer strong zero-shot retrieval performance, they struggle with distribution shifts caused by unseen query styles. In this paper, we propose the Hypernetwork-driven Style-adaptive Retrieval (Hystar), a lightweight framework that dynamically adapts model weights to each query's style. Hystar employs a hypernetwork to generate singular-value perturbations ($\Delta S$) for attention layers, enabling flexible per-input adaptation, while static singular-value offsets on MLP layers ensure cross-style stability. To better handle semantic confusions across styles, we design StyleNCE as part of Hystar, an optimal-transport-weighted contrastive loss that emphasizes hard cross-style negatives. Extensive experiments on multi-style retrieval and cross-style classification benchmarks demonstrate that Hystar consistently outperforms strong baselines, achieving state-of-the-art performance while being parameter-efficient and stable across styles.

cs.CV↗

arXiv2026

Terminal-Bench: Benchmarking Agents on Hard, Realistic Tasks in Command Line Interfaces

AI agents may soon become capable of autonomously completing valuable, long-horizon tasks in diverse domains. Current benchmarks either do not measure real-world tasks, or are not sufficiently difficult to meaningfully measure frontier models. To this end, we present Terminal-Bench 2.0: a carefully curated hard benchmark composed of 89 tasks in computer terminal environments inspired by problems from real workflows. Each task features a unique environment, human-written solution, and comprehensive tests for verification. We show that frontier models and agents score less than 65\% on the benchmark and conduct an error analysis to identify areas for model and agent improvement. We publish the dataset and evaluation harness to assist developers and researchers in future work at https://www.tbench.ai/ .

cs.SE↗

arXiv2025

The Impact of Bitcoin ETF Approval on Bitcoin's Hedging Properties Against Traditional Assets

The approval of the Bitcoin Spot ETF in January 2024 marked a transformative event in cryptocurrency markets, signaling increased institutional adoption and integration into traditional finance. This study examines Bitcoin's changing relationships with traditional assets, including equities, gold, and fiat currencies, following this milestone. Using rolling correlation analysis, Chow tests, and DCC-GARCH models, we found that Bitcoin's correlation with the S\&P 500 increased significantly post-ETF approval, indicating stronger alignment with equities. Its relationship with gold stabilized near zero, while its correlation with the U.S. Dollar Index remained consistently negative, reflecting its continued independence from fiat currencies. These findings offer insights into Bitcoin's evolving role in portfolios, implications for market stability, and future research opportunities on cryptocurrency integration into traditional financial systems.

q-fin.GN↗

arXiv2025

ConsistCompose: Unified Multimodal Layout Control for Image Composition

Unified multimodal models that couple visual understanding with image generation have advanced rapidly, yet most systems still focus on visual grounding-aligning language with image regions-while their generative counterpart, linguistic-embedded layout-grounded generation (LELG) for layout-controllable multi-instance generation, remains underexplored and limits precise compositional control. We present ConsistCompose, a unified multimodal framework that embeds layout coordinates directly into language prompts, enabling layout-controlled multi-instance image generation from Interleaved Image-Text within a single generative interface. We further construct ConsistCompose3M, a 3.4M multi-instance generation dataset with layout and identity annotations (2.6M text-guided and 0.8M image-guided data pairs) that provides large-scale supervision for layout-conditioned generation. Within this framework, LELG is instantiated through instance-coordinate binding prompts and coordinate-aware classifier-free guidance, which translate linguistic layout cues into precise spatial control without task-specific branches. Experiments on COCO-Position and MS-Bench show that ConsistCompose substantially improves spatial accuracy over layout-controlled baselines while preserving identity fidelity and competitive general multimodal understanding, establishing a unified paradigm for layout-controllable multimodal image generation.

cs.CV↗

arXiv2025

Unveiling two-dimensional electron systems on ultra-wide bandgap semiconductor $\mathrm{\beta}$-Ga$_2$O$_3$

Ultra-wide bandgap (UWBG) semiconductors promise to revolutionize power electronics, yet a fundamental understanding of their interfacial electronic structure has been hindered by the absence of direct experimental observation. Here, we report the first momentum-resolved observation of two-dimensional electron systems on a UWBG material, enabled by angle resolved photoemission spectroscopy (ARPES) on high-purity $\beta$-Ga$_2$O$_3$ single crystals. Alkaline-metal-induced electron doping forms an isotropic circular Fermi surface, achieving a sheet carrier density of up to $1.0\times10^{14}$ $\mathrm{cm}^{-2}$. Self-consistent Poisson-Schr\"odinger calculations show that the electrons are confined within 1.2 nm of the surface and reveal an internal electric field of $18$ MV cm$^{-1}$. Crucially, our measurements reveal a pronounced renormalization of the electronic band structure: a series of carrier-density-dependent ARPES measurements shows that as the carrier density increases from $2\times10^{13}$ to $1.0\times10^{14}$ $\mathrm{cm}^{-2}$, the effective mass anomalously increases, nearly doubling to a final value of 0.48 $\textit{m}_{\mathrm{e}}$. This trend is notably opposite to that reported for other oxide semiconductors, pointing towards a unique renormalization mechanism in $\beta$-Ga$_2$O$_3$. Our findings establish the interfacial electronic structure of $\beta$-Ga$_2$O$_3$ and demonstrate that UWBG materials provide fertile ground for exploring carrier-density-driven electronic phenomena, opening new avenues for future quantum and power devices.

cond-mat.mtrl-sci↗

arXiv2025

Coding Agents with Multimodal Browsing are Generalist Problem Solvers

Modern human labor is characterized by specialization; we train for years and develop particular tools that allow us to perform well across a variety of tasks. In addition, AI agents have been specialized for domains such as software engineering, web navigation, and workflow automation. However, this results in agents that are good for one thing but fail to generalize beyond their intended scope. One reason for this is that agent developers provide a highly specialized set of tools or make architectural decisions optimized for a specific use case or benchmark. In this work, we ask the question: what is the minimal set of general tools that can be used to achieve high performance across a diverse set of tasks? Our answer is OpenHands-Versa, a generalist agent built with a modest number of general tools: code editing and execution, web search, as well as multimodal web browsing and file access. Importantly, OpenHands-Versa demonstrates superior or competitive performance over leading specialized agents across three diverse and challenging benchmarks: SWE-Bench Multimodal, GAIA, and The Agent Company, outperforming the best-performing previously published results with absolute improvements in success rate of 9.1, 1.3, and 9.1 points respectively. Further, we show how existing state-of-the-art multi-agent systems fail to generalize beyond their target domains. These results demonstrate the feasibility of developing a generalist agent to solve diverse tasks and establish OpenHands-Versa as a strong baseline for future research.

cs.CL↗

arXiv2025

Bernstein Bounds for Caustics

Systematically simulating specular light transport requires an exhaustive search for primitive tuples containing admissible paths. Given the extreme inefficiency of enumerating all combinations, we propose to significantly reduce the search domain by sampling such tuples. The challenge is to design proper sampling probabilities that keep the noise level controllable. Our key insight is that by bounding the range of irradiance contributed by each primitive tuple at a given position, we can sample a subset of primitive tuples with potentially high contributions. Although low-contribution tuples are assigned a negligible probability, the overall variance remains low. Therefore, we derive vertex position and irradiance bounds for each primitive tuple, introducing a bounding property of rational functions on the Bernstein basis. When formulating position and irradiance expressions into rational functions, we handle non-rational components through remainder variables to maintain validity. Finally, we carefully design the sampling probabilities by optimizing the upper bound of the variance, expressed only using the position and irradiance bound. The proposed primitive sampling is intrinsically unbiased. It can be seamlessly combined with various unbiased and biased root-finding techniques within a local primitive domain. Extensive evaluations show that our method enables fast and reliable rendering of complex caustic effects.

cs.GR↗

arXiv2025

Enhancing Contrastive Learning for Retinal Imaging via Adjusted Augmentation Scales

Contrastive learning, a prominent approach within self-supervised learning, has demonstrated significant effectiveness in developing generalizable models for various applications involving natural images. However, recent research indicates that these successes do not necessarily extend to the medical imaging domain. In this paper, we investigate the reasons for this suboptimal performance and hypothesize that the dense distribution of medical images poses challenges to the pretext tasks in contrastive learning, particularly in constructing positive and negative pairs. We explore model performance under different augmentation strategies and compare the results to those achieved with strong augmentations. Our study includes six publicly available datasets covering multiple clinically relevant tasks. We further assess the model's generalizability through external evaluations. The model pre-trained with weak augmentation outperforms those with strong augmentation, improving AUROC from 0.838 to 0.848 and AUPR from 0.523 to 0.597 on MESSIDOR2, and showing similar enhancements across other datasets. Our findings suggest that optimizing the scale of augmentation is critical for enhancing the efficacy of contrastive learning in medical imaging.

cs.CV↗

arXiv2024

TheAgentCompany: Benchmarking LLM Agents on Consequential Real World Tasks

We interact with computers on an everyday basis, be it in everyday life or work, and many aspects of work can be done entirely with access to a computer and the Internet. At the same time, thanks to improvements in large language models (LLMs), there has also been a rapid development in AI agents that interact with and affect change in their surrounding environments. But how performant are AI agents at accelerating or even autonomously performing work-related tasks? The answer to this question has important implications both for industry looking to adopt AI into their workflows and for economic policy to understand the effects that adoption of AI may have on the labor market. To measure the progress of these LLM agents' performance on performing real-world professional tasks, in this paper we introduce TheAgentCompany, an extensible benchmark for evaluating AI agents that interact with the world in similar ways to those of a digital worker: by browsing the Web, writing code, running programs, and communicating with other coworkers. We build a self-contained environment with internal web sites and data that mimics a small software company environment, and create a variety of tasks that may be performed by workers in such a company. We test baseline agents powered by both closed API-based and open-weights language models (LMs), and find that the most competitive agent can complete 30% of tasks autonomously. This paints a nuanced picture on task automation with LM agents--in a setting simulating a real workplace, a good portion of simpler tasks could be solved autonomously, but more difficult long-horizon tasks are still beyond the reach of current systems. We release code, data, environment, and experiments on https://the-agent-company.com.

cs.CL↗

arXiv2024

Spectroscopic signatures of magnetization-induced band renormalization and strong spin-charge-lattice coupling in EuZn$_2$As$_2$

We report an infrared spectroscopy study of the antiferromagnetic (AFM) insulator EuZn$_2$As$_2$ over a broad frequency range, spanning temperatures both above and below the AFM transition $T_{\rm N} \simeq$ 20 K. The optical response reveals an insulating behavior, featuring two prominent infrared-active phonon modes at around 95 and 190 cm$^{-1}$, and two subtle absorption peaks at around 130 ($\alpha$ peak) and 2700 cm$^{-1}$ ($\beta$ peak), along with a strong absorption edge rising around 9000 cm$^{-1}$ ($\gamma$ peak). Significantly, the temperature-dependent changes in these peaks show noticeable anomalies across the AFM transition, particularly the emergence of the $\alpha$ peak and an unusual redshift of the $\gamma$ peak, suggesting a strong interaction between the charge excitations and the AFM order. Band structure calculations reveal that these anomalies arise from magnetization-induced band renormalizations, including shifts and foldings. Additionally, both phonon modes feature asymmetric Fano line shapes at low temperatures, with the 95 cm$^{-1}$ phonon mode exhibiting strong coupling to the fluctuations of Eu spins. These findings highlight a complex interplay of spin, charge, and lattice degrees of freedom in EuZn$_2$As$_2$.

cond-mat.mtrl-sci↗

arXiv2024

Deformed Intersections of Half-spaces

This paper is devoted to the classification problems concerning extended deformations of convex polyhedra and real hyperplane arrangements in the following senses: combinatorial equivalence of face posets, normal equivalence on normal fans of convex polyhedra, and sign equivalence on half-spaces. The extended deformations of convex polyhedra arise from parallel translations of given half-spaces and hyperplanes, whose normal vectors give rise to the so-called ``derived arrangement'' proposed by Rota as well as Crapo in different forms. We show that two extended deformations of convex polyhedra are normally (combinatorially, as a consequence) equivalent if they are parameterized by the same open face of the derived arrangement. Note that these extended deformations are based on parallel translations of the given hyperplanes. It allows us to study three deformations of real hyperplane arrangements: parallel translations, conings, and elementary lifts, whose configuration spaces are parameterized by open faces of the derived arrangement. Consequently, it gives a characterization of the normal, combinatorial, and sign equivalences of those three deformations via the derived arrangement. Additionally, the relationships among these three equivalence relations are discussed, and several new descriptions of real derived arrangements associated with faces and sign vectors of real hyperplane arrangements are provided.

math.CO↗

arXiv2024

OpenHands: An Open Platform for AI Software Developers as Generalist Agents

Software is one of the most powerful tools that we humans have at our disposal; it allows a skilled programmer to interact with the world in complex and profound ways. At the same time, thanks to improvements in large language models (LLMs), there has also been a rapid development in AI agents that interact with and affect change in their surrounding environments. In this paper, we introduce OpenHands (f.k.a. OpenDevin), a platform for the development of powerful and flexible AI agents that interact with the world in similar ways to those of a human developer: by writing code, interacting with a command line, and browsing the web. We describe how the platform allows for the implementation of new agents, safe interaction with sandboxed environments for code execution, coordination between multiple agents, and incorporation of evaluation benchmarks. Based on our currently incorporated benchmarks, we perform an evaluation of agents over 15 challenging tasks, including software engineering (e.g., SWE-BENCH) and web browsing (e.g., WEBARENA), among others. Released under the permissive MIT license, OpenHands is a community project spanning academia and industry with more than 2.1K contributions from over 188 contributors.

cs.SE↗

arXiv2024

PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

In this study, we introduce the safety human preference dataset, PKU-SafeRLHF, designed to promote research on safety alignment in large language models (LLMs). As a sibling project to SafeRLHF and BeaverTails, we separate annotations of helpfulness and harmlessness for question-answering pairs, providing distinct perspectives on these coupled attributes. Overall, we provide 44.6k refined prompts and 265k question-answer pairs with safety meta-labels for 19 harm categories and three severity levels ranging from minor to severe, with answers generated by Llama-family models. Based on this, we collected 166.8k preference data, including dual-preference (helpfulness and harmlessness decoupled) and single-preference data (trade-off the helpfulness and harmlessness from scratch), respectively. Using the large-scale annotation data, we further train severity-sensitive moderation for the risk control of LLMs and safety-centric RLHF algorithms for the safety alignment of LLMs. We believe this dataset will be a valuable resource for the community, aiding in the safe deployment of LLMs. Data is available at https://huggingface.co/datasets/PKU-Alignment/PKU-SafeRLHF.

cs.AI↗

arXiv2024

GUIOdyssey: A Comprehensive Dataset for Cross-App GUI Navigation on Mobile Devices

Autonomous Graphical User Interface (GUI) navigation agents can enhance user experience in communication, entertainment, and productivity by streamlining workflows and reducing manual intervention. However, prior GUI agents often trained with datasets comprising tasks that can be completed within a single app, leading to poor performance in cross-app navigation. To address this problem, we present GUIOdyssey, a comprehensive dataset for cross-app mobile GUI navigation. GUIOdyssey comprises 8,334 episodes with an average of 15.3 steps per episode, covering 6 mobile devices, 212 distinct apps, and 1,357 app combinations. Each step is enriched with detailed semantic reasoning annotations, which aid the model in building cognitive processes and enhancing its reasoning abilities for complex cross-app tasks. Building on GUIOdyssey, we develop OdysseyAgent, an exploratory multimodal agent for long-step cross-app navigation equipped with a history resampler module that efficiently attends to historical screenshot tokens, balancing performance and inference speed. Extensive experiments conducted in both in-domain and out-of-domain scenarios validate the effectiveness of our approach. Moreover, we demonstrate that historial information involving actions, screenshots and context in our dataset can significantly enhances OdysseyAgent's performance on complex cross-app tasks.

cs.CV↗

arXiv2023

Structure, physical properties, and magnetically tunable topological phases in topological semimetal EuCuBi

A single material achieving multiple topological phases can provide potential application for topological spintronics, whereas the candidate materials are very limited. Here, we report the structure, physical properties, and possible emergence of multiple topological phases in the newly discovered, air-stable EuCuBi single crystal. EuCuBi crystallizes in a hexagonal space group P63/mmc (No. 194) in ZrBeSi-type structure with an antiferromagnetic (AFM) ground state below TN = 11.2 K. There is a competition between AFM and ferromagnetic (FM) interactions below TN revealed by electrical resistivity and magnetic susceptibility measurements. With the increasing magnetic field, EuCuBi evolves from the AFM ground state with a small amount of FM component, going through two possible metamagnetic phases, finally reaches the field-induced FM phase. Based on the first-principles calculations, we demonstrate that the Dirac, Weyl, and possible mirror Chern insulator can be achieved in EuCuBi by tuning the temperature and applying magnetic field, making EuCuBi a promising candidate for exploring multiple topological phases.

cond-mat.mtrl-sci↗

arXiv2022

FasterRisk: Fast and Accurate Interpretable Risk Scores

Over the last century, risk scores have been the most popular form of predictive model used in healthcare and criminal justice. Risk scores are sparse linear models with integer coefficients; often these models can be memorized or placed on an index card. Typically, risk scores have been created either without data or by rounding logistic regression coefficients, but these methods do not reliably produce high-quality risk scores. Recent work used mathematical programming, which is computationally slow. We introduce an approach for efficiently producing a collection of high-quality risk scores learned from data. Specifically, our approach produces a pool of almost-optimal sparse continuous solutions, each with a different support set, using a beam-search algorithm. Each of these continuous solutions is transformed into a separate risk score through a "star ray" search, where a range of multipliers are considered before rounding the coefficients sequentially to maintain low logistic loss. Our algorithm returns all of these high-quality risk scores for the user to consider. This method completes within minutes and can be valuable in a broad variety of applications.

cs.LG↗