arXiv subjects

Longxiang Liu

Publications and source records attributed to Longxiang Liu.

At least 19 recordsLinked to original sources

arXiv2026

StartupBench: Benchmarking General-Purpose Agents on Market-Validated End-to-End Workflows

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.

cs.AI↗

arXiv2026

Agent-World: Scaling Real-World Environment Synthesis for Evolving General Agent Intelligence

Large language models are increasingly expected to serve as general-purpose agents that interact with external, stateful tool environments. The Model Context Protocol (MCP) and broader agent skills offer a unified interface for connecting agents with scalable real-world services, but training robust agents remains limited by the lack of realistic environments and principled mechanisms for life-long learning. In this paper, we present \textbf{Agent-World}, a self-evolving training arena for advancing general agent intelligence through scalable environments. Agent-World has two main components: (1) Agentic Environment-Task Discovery, which autonomously explores topic-aligned databases and executable tool ecosystems from thousands of real-world environment themes and synthesizes verifiable tasks with controllable difficulty; and (2) Continuous Self-Evolving Agent Training, which combines multi-environment reinforcement learning with a self-evolving agent arena that automatically identifies capability gaps through dynamic task synthesis and drives targeted learning, enabling the co-evolution of agent policies and environments. Across 23 challenging agent benchmarks, Agent-World-8B and 14B consistently outperforms strong proprietary models and environment scaling baselines. Further analyses reveal scaling trends in relation to environment diversity and self-evolution rounds, offering insights for building general agent intelligence.

cs.AI↗

arXiv2026

High-Precision Measurement of D($γ$, $n$)$p$ Photodisintegration Reaction and Implications for Big-Bang Nucleosynthesis

We report on a high-precision measurement of the D($γ$,\,$n$)$p$ photodisintegration reaction at the newly commissioned Shanghai Laser Electron Gamma Source (SLEGS), employing a quasi-monochromatic $γ$-ray beam from Laser Compton Scattering. The cross sections were determined over $E_γ$=2.327--7.089 MeV, achieving up to a factor of 2.2 improvement in precision near the neutron separation threshold. Combined with previous data in a global Markov chain Monte Carlo (MCMC) analysis using dibaryon effective field theory, we obtained the unprecedentedly precise $p$($n$,\,$γ$)D cross sections and thermonuclear rate, with a precision up to $\approx$4 times higher than previous evaluations. Implemented in a standard Big-Bang Nucleosynthesis (BBN) framework, this new rate decreases uncertainty of the key cosmological parameter of baryon density $Ω_b h^2$ by up to $\approx$16\% relative to the LUNA result. A residual $\approx$1.2$σ$ tension between $Ω_b h^2$ constrained from primordial D/H observations and CMB measurements persists, highlighting the need for improved $dd$ reaction rates and offering potential hints of new physics beyond the standard model of cosmology.

nucl-ex↗

arXiv2025

MME-CC: A Challenging Multi-Modal Evaluation Benchmark of Cognitive Capacity

As reasoning models scale rapidly, the essential role of multimodality in human cognition has come into sharp relief, driving a growing need to probe vision-centric cognitive behaviors. Yet, existing multimodal benchmarks either overemphasize textual reasoning or fall short of systematically capturing vision-centric cognitive behaviors, leaving the cognitive capacity of MLLMs insufficiently assessed. To address this limitation, we introduce MME-CC (Multi-Modal Evaluation benchmark of Cognitive Capacity), a vision-grounded benchmark that organizes 11 representative reasoning tasks into three fundamental categories of visual information: spatial, geometric, and knowledge-based reasoning, and provides fine-grained analyses of MLLMs' cognitive capacity across these dimensions. Based on MME-CC, we conduct extensive experiments over 16 representative MLLMs. Our study reveals that closed-source models currently lead overall (e.g., 42.66 for Gemini-2.5-Pro vs. 30.45 for GLM-4.5V), while spatial and geometric reasoning remain broadly weak (less than or equal to 30%). We further identify common error patterns, including orientation mistakes, fragile cross-view identity persistence, and poor adherence to counterfactual instructions, and observe that Chain-of-Thought typically follows a three-stage process (extract -> reason -> verify) with heavy reliance on visual extraction. We hope this work catalyzes a shift toward treating the cognitive capacity of MLLMs as central to both evaluation and model design.

cs.CL↗

arXiv2025

Game-TARS: Pretrained Foundation Models for Scalable Generalist Multimodal Game Agents

We present Game-TARS, a generalist game agent trained with a unified, scalable action space anchored to human-aligned native keyboard-mouse inputs. Unlike API- or GUI-based approaches, this paradigm enables large-scale continual pre-training across heterogeneous domains, including OS, web, and simulation games. Game-TARS is pre-trained on over 500B tokens with diverse trajectories and multimodal data. Key techniques include a decaying continual loss to reduce causal confusion and an efficient Sparse-Thinking strategy that balances reasoning depth and inference cost. Experiments show that Game-TARS achieves about 2 times the success rate over the previous sota model on open-world Minecraft tasks, is close to the generality of fresh humans in unseen web 3d games, and outperforms GPT-5, Gemini-2.5-Pro, and Claude-4-Sonnet in FPS benchmarks. Scaling results on training-time and test-time confirm that the unified action space sustains improvements when scaled to cross-game and multimodal data. Our results demonstrate that simple, scalable action representations combined with large-scale pre-training provide a promising path toward generalist agents with broad computer-use abilities.

cs.AI↗

arXiv2025

MARS-Bench: A Multi-turn Athletic Real-world Scenario Benchmark for Dialogue Evaluation

Large Language Models (\textbf{LLMs}), e.g. ChatGPT, have been widely adopted in real-world dialogue applications. However, LLMs' robustness, especially in handling long complex dialogue sessions, including frequent motivation transfer, sophisticated cross-turn dependency, is criticized all along. Nevertheless, no existing benchmarks can fully reflect these weaknesses. We present \textbf{MARS-Bench}, a \textbf{M}ulti-turn \textbf{A}thletic \textbf{R}eal-world \textbf{S}cenario Dialogue \textbf{Bench}mark, designed to remedy the gap. MARS-Bench is constructed from play-by-play text commentary so to feature realistic dialogues specifically designed to evaluate three critical aspects of multi-turn conversations: Ultra Multi-turn, Interactive Multi-turn, and Cross-turn Tasks. Extensive experiments on MARS-Bench also reveal that closed-source LLMs significantly outperform open-source alternatives, explicit reasoning significantly boosts LLMs' robustness on handling long complex dialogue sessions, and LLMs indeed face significant challenges when handling motivation transfer and sophisticated cross-turn dependency. Moreover, we provide mechanistic interpretability on how attention sinks due to special tokens lead to LLMs' performance degradation when handling long complex dialogue sessions based on attention visualization experiment in Qwen2.5-7B-Instruction.

cs.CL↗

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

First systematic experimental 2D mapping of linearly polarized $γ$-ray polarimetric distribution in relativistic Compton scattering

The interaction of photons with relativistic electrons constitutes a fundamental electromagnetic process whose polarization transfer mechanics remain incompletely characterized. We report the first systematic measurement of spatial polarization distribution for $γ$-rays generated via \SI{45}{\degree} slant inverse Compton scattering (ICS) between linearly polarized \SI{0.117}{\eV} photons and \SI{3.5}{\GeV} electrons, performing full 2D mapping of intensity, polarization angle (AOP), and degree of polarization (DOP). Measurements reveal an asymmetric beam profile along the laser's polarization direction that resembles \SI{180}{\degree} backward ICS observations. The central beam region exhibits DOP $\approx$ 1.0 with AOP rigidly aligned at \SI{45}{\degree}, while peripheral regions display complex non-uniform polarization distributions. These findings confirm quantum electrodynamics predictions of near-complete polarization transfer along the beam axis in slant geometries, thus establishing slant scattering as a viable alternative to head-on configurations for generating high DOP $γ$-rays.

nucl-ex↗

arXiv2025

UI-TARS: Pioneering Automated GUI Interaction with Native Agents

This paper introduces UI-TARS, a native GUI agent model that solely perceives the screenshots as input and performs human-like interactions (e.g., keyboard and mouse operations). Unlike prevailing agent frameworks that depend on heavily wrapped commercial models (e.g., GPT-4o) with expert-crafted prompts and workflows, UI-TARS is an end-to-end model that outperforms these sophisticated frameworks. Experiments demonstrate its superior performance: UI-TARS achieves SOTA performance in 10+ GUI agent benchmarks evaluating perception, grounding, and GUI task execution. Notably, in the OSWorld benchmark, UI-TARS achieves scores of 24.6 with 50 steps and 22.7 with 15 steps, outperforming Claude (22.0 and 14.9 respectively). In AndroidWorld, UI-TARS achieves 46.6, surpassing GPT-4o (34.5). UI-TARS incorporates several key innovations: (1) Enhanced Perception: leveraging a large-scale dataset of GUI screenshots for context-aware understanding of UI elements and precise captioning; (2) Unified Action Modeling, which standardizes actions into a unified space across platforms and achieves precise grounding and interaction through large-scale action traces; (3) System-2 Reasoning, which incorporates deliberate reasoning into multi-step decision making, involving multiple reasoning patterns such as task decomposition, reflection thinking, milestone recognition, etc. (4) Iterative Training with Reflective Online Traces, which addresses the data bottleneck by automatically collecting, filtering, and reflectively refining new interaction traces on hundreds of virtual machines. Through iterative training and reflection tuning, UI-TARS continuously learns from its mistakes and adapts to unforeseen situations with minimal human intervention. We also analyze the evolution path of GUI agents to guide the further development of this domain.

cs.AI↗

arXiv2024

Differential cross-section measurements for neutron-induced $α$ production reactions on carbon across neutron energy range of 6.2 to 76 MeV

Angle-differential cross sections for neutron-induced $α$ production in carbon were determined at thirty discrete neutron energy levels ranging from 6.2 to 76 MeV at the Back-n white neutron source of the China Spallation Neutron Source. Utilizing the $ΔE-E$ telescopes within the Light-charged Particle Detector Array spanning angular measurements from 24.5°to 155.5° in the laboratory frame, the $^{12}C(n,α)x$ reaction cross sections were obtained. These experimental findings exhibit a strong concordance with prior results and have been benchmarked against theoretical estimates from codes such as TALYS, Geant4, and assessments from the ENDF/B-VIII.0 database. Remarkably, distinct resonance-like features were observed at neutron energies of 13.7, 22.4, 29.5, and 61.8 MeV, marking their first-time identification in the literature.Furthermore, a comparative analysis involving the theoretical Distorted Wave Born Approximation was conducted.

nucl-ex↗

arXiv2024

TA&AT: Enhancing Task-Oriented Dialog with Turn-Level Auxiliary Tasks and Action-Tree Based Scheduled Sampling

Task-oriented dialog systems have witnessed substantial progress due to conversational pre-training techniques. Yet, two significant challenges persist. First, most systems primarily utilize the latest turn's state label for the generator. This practice overlooks the comprehensive value of state labels in boosting the model's understanding for future generations. Second, an overreliance on generated policy often leads to error accumulation, resulting in suboptimal responses when adhering to incorrect actions. To combat these challenges, we propose turn-level multi-task objectives for the encoder. With the guidance of essential information from labeled intermediate states, we establish a more robust representation for both understanding and generation. For the decoder, we introduce an action tree-based scheduled sampling technique. Specifically, we model the hierarchical policy as trees and utilize the similarity between trees to sample negative policy based on scheduled sampling, hoping the model to generate invariant responses under perturbations. This method simulates potential pitfalls by sampling similar negative policy, bridging the gap between task-oriented dialog training and inference. Among methods without continual pre-training, our approach achieved state-of-the-art (SOTA) performance on the MultiWOZ dataset series and was also competitive with pre-trained SOTA methods.

cs.CL↗

arXiv2023

Graphical Representations and Worm Algorithms for the O($N$) Spin Model

We present a family of graphical representations for the O($N$) spin model, where $N \ge 1$ represents the spin dimension, and $N=1,2,3$ corresponds to the Ising, XY and Heisenberg models, respectively. With an integer parameter $0 \le \ell \le N/2$, each configuration is the coupling of $\ell$ copies of subgraphs consisting of directed flows and $N -2\ell$ copies of subgraphs constructed by undirected loops, which we call the XY and Ising subgraphs, respectively. On each lattice site, the XY subgraphs satisfy the Kirchhoff flow-conservation law and the Ising subgraphs obey the Eulerian bond condition. Then, we formulate worm-type algorithms and simulate the O($N$) model on the simple-cubic lattice for $N$ from 2 to 6 at all possible $\ell$. It is observed that the worm algorithm has much higher efficiency than the Metropolis method, and, for a given $N$, the efficiency is an increasing function of $\ell$. Beside Monte Carlo simulations, we expect that these graphical representations would provide a convenient basis for the study of the O($N$) spin model by other state-of-the-art methods like the tensor network renormalization.

cond-mat.stat-mech↗

arXiv2023

Channel-aware Decoupling Network for Multi-turn Dialogue Comprehension

Training machines to understand natural language and interact with humans is one of the major goals of artificial intelligence. Recent years have witnessed an evolution from matching networks to pre-trained language models (PrLMs). In contrast to the plain-text modeling as the focus of the PrLMs, dialogue texts involve multiple speakers and reflect special characteristics such as topic transitions and structure dependencies between distant utterances. However, the related PrLM models commonly represent dialogues sequentially by processing the pairwise dialogue history as a whole. Thus the hierarchical information on either utterance interrelation or speaker roles coupled in such representations is not well addressed. In this work, we propose compositional learning for holistic interaction across the utterances beyond the sequential contextualization from PrLMs, in order to capture the utterance-aware and speaker-aware representations entailed in a dialogue history. We decouple the contextualized word representations by masking mechanisms in Transformer-based PrLM, making each word only focus on the words in current utterance, other utterances, and two speaker roles (i.e., utterances of sender and utterances of the receiver), respectively. In addition, we employ domain-adaptive training strategies to help the model adapt to the dialogue domains. Experimental results show that our method substantially boosts the strong PrLM baselines in four public benchmark datasets, achieving new state-of-the-art performance over previous methods.

cs.CL↗

arXiv2020

Filling the Gap of Utterance-aware and Speaker-aware Representation for Multi-turn Dialogue

A multi-turn dialogue is composed of multiple utterances from two or more different speaker roles. Thus utterance- and speaker-aware clues are supposed to be well captured in models. However, in the existing retrieval-based multi-turn dialogue modeling, the pre-trained language models (PrLMs) as encoder represent the dialogues coarsely by taking the pairwise dialogue history and candidate response as a whole, the hierarchical information on either utterance interrelation or speaker roles coupled in such representations is not well addressed. In this work, we propose a novel model to fill such a gap by modeling the effective utterance-aware and speaker-aware representations entailed in a dialogue history. In detail, we decouple the contextualized word representations by masking mechanisms in Transformer-based PrLM, making each word only focus on the words in current utterance, other utterances, two speaker roles (i.e., utterances of sender and utterances of receiver), respectively. Experimental results show that our method boosts the strong ELECTRA baseline substantially in four public benchmark datasets, and achieves various new state-of-the-art performance over previous methods. A series of ablation studies are conducted to demonstrate the effectiveness of our method.

cs.CL↗

arXiv2020

Worm-algorithm-type Simulation of Quantum Transverse-Field Ising Model

We apply a worm algorithm to simulate the quantum transverse-field Ising model in a path-integral representation of which the expansion basis is taken as the spin component along the external-field direction. In such a representation, a configuration can be regarded as a set of non-intersecting loops constructed by "kinks" for pairwise interactions and spin-down (or -up) imaginary-time segments. The wrapping probability for spin-down loops, a dimensionless quantity characterizing the loop topology on a torus, is observed to exhibit small finite-size corrections and yields a high-precision critical point in two dimensions (2D) as $h_c \! =\! 3.044\, 330(6)$, significantly improving over the existing results and nearly excluding the best one $h_c \! =\! 3.044\, 38 (2)$. At criticality, the fractal dimensions of the loops are estimated as $d_{\ell \downarrow} (1{\rm D}) \! = \! 1.37(1) \! \approx \! 11/8 $ and $d_{\ell \downarrow} (2{\rm D}) \! = \! 1.75 (3)$, consistent with those for the classical 2D and 3D O(1) loop model, respectively. An interesting feature is that in one dimension (1D), both the spin-down and -up loops display the critical behavior in the whole disordered phase ($ 0 \! \leq \! h \! < \! h_c$), having a fractal dimension $d_{\ell} \! = \! 1.750 (7)$ that is consistent with the hull dimension $d_{\rm H} \! = \! 7/4$ for critical 2D percolation clusters. The current worm algorithm can be applied to simulate other quantum systems like hard-core boson models with pairing interactions.

cond-mat.stat-mech↗

arXiv2019

Evidences for interaction-induced Haldane fractional exclusion statistics in one and higher dimensions

Haldane fractional exclusion statistics (FES) has a long history of intense studies, but its realization in physical systems is rare. Here we study repulsively interacting Bose gases at and near a quantum critical point, and find evidences that such strongly correlated gases obey simple non-mutual FES over a wide range of interaction strengths in both one and two dimensions. Based on exact solutions in one dimension, quantum Monte Carlo simulations and experiments in both dimensions, we show that the thermodynamic properties of these interacting gases, including entropy per particle, density and pressure, are essentially equivalent to those of non-interacting particles with FES. Accordingly, we establish a simple interaction-to-FES mapping that reveals the statistical nature of particle-hole symmetry breaking induced by interaction in such quantum many-body systems. Whereas strongly interacting Bose gases reach full fermionization in one dimension, they exhibit incomplete fermionization in two dimensions. Our results open a route to understanding correlated interacting systems via non-interacting particles with FES in arbitrary dimensions.

cond-mat.quant-gas↗

arXiv2018

Phases of superclimbing dislocation with long-range interaction between jogs

The main candidate for the superfluid pathways in solid Helium-4 are dislocations with Burgers vector along the hcp symmetry axis. Here we focus on quantum behavior of a generic edge dislocation which can perform superclimb -- climb supported by the superflow along its core. The role of the long range elastic interactions between jogs is addressed by Monte Carlo simulations. It is found that such interactions do not change qualitatively the phase diagram found without accounting for such forces. Their main effect consists of renormalizing the effective scale determining compressibility of the dislocation in the Tomonaga-Luttinger Liquid phase. It is also found that the quantum rough phase of the dislocation can be well described within the gaussian approximation which features off-diagonal long range order in 1D for the superfluid order parameter along the core.

cond-mat.other↗

arXiv2017

Simulation of Physical Parameters for a Photoneutron Source

A compact photoneutron source (PNS), based on an electron linac was designed and constructed to provide required nuclear data for the design of Thorium Molten Salt Reactor (TMSR). Many local shielding are built to reduce the background of neutron and γ rays, making the location of the time of flight (TOF) detector be fixed at 6.2 m place. Under the existing layout, some physical parameters are very difficult to get by the experiments, while can be obtained by the Monte Carlo simulation method. However, for the deep penetration problem of the neutron and γ rays transport in the channel of PNS with complex geometry, the normal Monte Carlo method is inefficient since electron transport calculation need a large amount of computing time and neutrons have little contribution to the detector in far-source region. In this work, the subsection method is applied in the simulation for PNS, which divide the simulation process in two steps, recording the neutron and γ rays information passing through the source window in the first step and adopting the covariance reduction techniques in the second step. The simulated neutron flux and energy spectrum at the TOF detector place with the relative error 1.6% are well agreement with the experimental results, achieving an efficiency 23 times better than the normal method. This method is fast and efficient in predicting the physical parameters, providing a required verification and initiating the foreseen physics experiment.

physics.comp-ph↗