arXiv subjects

Xueyan Li

Publications and source records attributed to Xueyan Li.

15 recordsLinked to original sources

arXiv2026

Multi-Stream LLMs: Unblocking Language Models with Parallel Streams of Thoughts, Inputs and Outputs

The continued improvements in language model capability have unlocked their widespread use as drivers of autonomous agents, for example in coding or computer use applications. However, the core of these systems has not changed much since early instruction-tuned models like ChatGPT. Even advanced AI agents function on message exchange formats, successively exchanging messages with users, systems, with itself (i.e. chain-of-thought) and tools in a single stream of computation. This bottleneck to a single stream in chat models leads to a number of limitations: the agent cannot act (generate output) while reading, and in reverse, cannot react to new information while writing. Similarly, the agent cannot act while thinking and cannot think while reading or acting on information. In this work, we show that models can be unblocked by switching from instruction-tuning for sequential message formats to instruction-tuning for multiple, parallel streams of computation, splitting each role into a separate stream. Every forward pass of the language model then simultaneously reads from multiple input streams and generates tokens in multiple output streams, all of which causally depend on earlier timesteps. We argue that this data-driven change remedies a number of usability limitations as outlined above, improves model efficiency through parallelization, improves model security through better separation of concerns and can further improve model monitorability.

cs.LG↗

arXiv2026

Efficient Test-Time Inference via Deterministic Exploration of Truncated Decoding Trees

Self-consistency boosts inference-time performance by sampling multiple reasoning traces in parallel and voting. However, in constrained domains like math and code, this strategy is compute-inefficient because it samples with replacement, repeatedly revisiting the same high-probability prefixes and duplicate completions. We propose Distinct Leaf Enumeration (DLE), a deterministic decoding method that treats truncated sampling as traversal of a pruned decoding tree and systematically enumerates distinct leaves instead of sampling with replacement. This strategy improves inference efficiency in two ways. Algorithmically, it increases coverage of the truncated search space under a fixed budget by exploring previously unvisited high-probability branches. Systemically, it reuses shared prefixes and reduces redundant token generation. Empirically, DLE explores higher-quality reasoning traces than stochastic self-consistency, yielding better performance on math, coding, and general reasoning tasks.

cs.LG↗

arXiv2026

Reimagining Legal Fact Verification with GenAI: Toward Effective Human-AI Collaboration

Fact verification is a critical yet underexplored component of non-litigation legal practice. While existing research has examined automation in legal workflow and human-AI collaboration in high-stakes domains, little is known about how GenAI can support fact verification, a task that demands prudent judgment and strict accountability. To address this, we conducted semi-structured interviews with 18 lawyers to understand their current verification practices, attitudes toward GenAI adoption, and expectations for future systems. We found that while lawyers use GenAI for low-risk tasks like drafting and language optimization, concerns over accuracy, confidentiality, and liability are currently limiting its adoption for fact verification. These concerns translate into core design requirements for AI systems that are trustworthy and accountable. Based on these, we contribute design insights for human-AI collaboration in legal fact verification, emphasizing the development of auditable systems that balance efficiency with professional judgment and uphold ethical and legal accountability in high-stakes practice.

cs.HC↗

arXiv2025

From Perception to Punchline: Empowering VLM with the Art of In-the-wild Meme

Generating humorous memes is a challenging multimodal task that moves beyond direct image-to-caption supervision. It requires a nuanced reasoning over visual content, contextual cues, and subjective humor. To bridge this gap between visual perception and humorous punchline creation, we propose HUMOR}, a novel framework that guides VLMs through hierarchical reasoning and aligns them with group-wise human preferences. First, HUMOR employs a hierarchical, multi-path Chain-of-Thought (CoT): the model begins by identifying a template-level intent, then explores diverse reasoning paths under different contexts, and finally anchors onto a high-quality, context-specific path. This CoT supervision, which traces back from ground-truth captions, enhances reasoning diversity. We further analyze that this multi-path exploration with anchoring maintains a high expected humor quality, under the practical condition that high-quality paths retain significant probability mass. Second, to capture subjective humor, we train a pairwise reward model that operates within groups of memes sharing the same template. Following established theory, this approach ensures a consistent and robust proxy for human preference, even with subjective and noisy labels. The reward model then enables a group-wise reinforcement learning optimization, guaranteeing providing a theoretical guarantee for monotonic improvement within the trust region. Extensive experiments show that HUMOR empowers various VLMs with superior reasoning diversity, more reliable preference alignment, and higher overall meme quality. Beyond memes, our work presents a general training paradigm for open-ended, human-aligned multimodal generation, where success is guided by comparative judgment within coherent output group.

cs.LG↗

arXiv2025

Generative Multi-modal Feedback for Singing Voice Synthesis Evaluation

Singing voice synthesis (SVS) has advanced significantly, enabling models to generate vocals with accurate pitch and consistent style. As these capabilities improve, the need for reliable evaluation and optimization becomes increasingly critical. However, current methods like reward systems often rely on single numerical scores, struggle to capture various dimensions such as phrasing or expressiveness, and require costly annotations, limiting interpretability and generalization. To address these issues, we propose a generative feedback (i.e., reward model) framework that provides multi-dimensional language and audio feedback for SVS assessment. Our approach leverages an audio-language model to generate text and audio critiques-covering aspects such as melody, content, and auditory quality. The model is fine-tuned on a hybrid dataset combining human music reactions and synthetic critiques from a MLLMs, enhancing diversity and linguistic richness. Quantitative experiments validate the effectiveness of the proposed dataset and training strategy, demonstrating that the framework produces musically accurate and interpretable evaluations suitable for guiding generative model improvement. The code is at [https://github.com/opendilab/VocalCritic](https://github.com/opendilab/VocalCritic)

cs.SD↗

arXiv2025

Sample Smart, Not Hard: Correctness-First Decoding for Better Reasoning in LLMs

Large Language Models (LLMs) are increasingly applied to complex tasks that require extended reasoning. In such settings, models often benefit from diverse chains-of-thought to arrive at multiple candidate solutions. This requires two competing objectives: to inject enough stochasticity to explore multiple reasoning chains, and to ensure sufficient accuracy and quality in each path. Existing works pursue the first objective by increasing exploration at highly uncertain steps with higher temperature or larger candidate token sets, while others improve reliability by rejecting samples with low confidence post-generation, implying that low confidence correlates with low answer quality. These two lines of thought are in conflict, as they conflate different sources of uncertainty. To resolve this, we argue that the decoding rule should be calibrated by correctness, not confidence alone. We should sample from tokens with higher estimated correctness, and reduce sampling where expected correctness is low. We propose simple strategies that achieve this goal: Greedy-Threshold makes sampling greedy at very low confidence steps. Calibrated-TopK and Calibrated-epsilon set truncation threshold based on estimated rank-wise correctness. Together, our findings challenge prevailing heuristics about decoding under uncertainty and show gains across math and general reasoning benchmarks.

cs.LG↗

arXiv2025

SafeWork-R1: Coevolving Safety and Intelligence under the AI-45$^{\circ}$ Law

We introduce SafeWork-R1, a cutting-edge multimodal reasoning model that demonstrates the coevolution of capabilities and safety. It is developed by our proposed SafeLadder framework, which incorporates large-scale, progressive, safety-oriented reinforcement learning post-training, supported by a suite of multi-principled verifiers. Unlike previous alignment methods such as RLHF that simply learn human preferences, SafeLadder enables SafeWork-R1 to develop intrinsic safety reasoning and self-reflection abilities, giving rise to safety `aha' moments. Notably, SafeWork-R1 achieves an average improvement of $46.54\%$ over its base model Qwen2.5-VL-72B on safety-related benchmarks without compromising general capabilities, and delivers state-of-the-art safety performance compared to leading proprietary models such as GPT-4.1 and Claude Opus 4. To further bolster its reliability, we implement two distinct inference-time intervention methods and a deliberative search mechanism, enforcing step-level verification. Finally, we further develop SafeWork-R1-InternVL3-78B, SafeWork-R1-DeepSeek-70B, and SafeWork-R1-Qwen2.5VL-7B. All resulting models demonstrate that safety and capability can co-evolve synergistically, highlighting the generalizability of our framework in building robust, reliable, and trustworthy general-purpose AI.

cs.AI↗

arXiv2025

Empowering LLMs in Decision Games through Algorithmic Data Synthesis

Large Language Models (LLMs) have exhibited impressive capabilities across numerous domains, yet they often struggle with complex reasoning and decision-making tasks. Decision-making games, which inherently require multifaceted reasoning logic, serve as ideal sandboxes for evaluating and enhancing the reasoning abilities of LLMs. In this work, we first explore whether LLMs can master complex decision-making games through targeted post-training. To this end, we design data synthesis strategies and curate extensive offline datasets from two classic games, Doudizhu and Go. We further develop a suite of techniques to effectively incorporate this data into LLM training, resulting in two novel agents: Mastermind-Dou and Mastermind-Go. Our experimental results demonstrate that these Mastermind LLMs achieve competitive performance in their respective games. Additionally, we explore whether integrating decision-making data can enhance the general reasoning abilities of LLMs. Our findings suggest that such post-training improves certain aspects of reasoning, providing valuable insights for optimizing LLM data collection and synthesis strategies.

cs.LG↗

arXiv2024

PsyDI: Towards a Personalized and Progressively In-depth Chatbot for Psychological Measurements

In the field of psychology, traditional assessment methods, such as standardized scales, are frequently critiqued for their static nature, lack of personalization, and reduced participant engagement, while comprehensive counseling evaluations are often inaccessible. The complexity of quantifying psychological traits further limits these methods. Despite advances with large language models (LLMs), many still depend on single-round Question-and-Answer interactions. To bridge this gap, we introduce PsyDI, a personalized and progressively in-depth chatbot designed for psychological measurements, exemplified by its application in the Myers-Briggs Type Indicator (MBTI) framework. PsyDI leverages user-related multi-modal information and engages in customized, multi-turn interactions to provide personalized, easily accessible measurements, while ensuring precise MBTI type determination. To address the challenge of unquantifiable psychological traits, we introduce a novel training paradigm that involves learning the ranking of proxy variables associated with these traits, culminating in a robust score model for MBTI measurements. The score model enables PsyDI to conduct comprehensive and precise measurements through multi-turn interactions within a unified estimation context. Through various experiments, we validate the efficacy of both the score model and the PsyDI pipeline, demonstrating its potential to serve as a general framework for psychological measurements. Furthermore, the online deployment of PsyDI has garnered substantial user engagement, with over 3,000 visits, resulting in the collection of numerous multi-turn dialogues annotated with MBTI types, which facilitates further research. The source code for the training and web service components is publicly available as a part of OpenDILab at: https://github.com/opendilab/PsyDI

cs.HC↗

arXiv2023

LightZero: A Unified Benchmark for Monte Carlo Tree Search in General Sequential Decision Scenarios

Building agents based on tree-search planning capabilities with learned models has achieved remarkable success in classic decision-making problems, such as Go and Atari. However, it has been deemed challenging or even infeasible to extend Monte Carlo Tree Search (MCTS) based algorithms to diverse real-world applications, especially when these environments involve complex action spaces and significant simulation costs, or inherent stochasticity. In this work, we introduce LightZero, the first unified benchmark for deploying MCTS/MuZero in general sequential decision scenarios. Specificially, we summarize the most critical challenges in designing a general MCTS-style decision-making solver, then decompose the tightly-coupled algorithm and system design of tree-search RL methods into distinct sub-modules. By incorporating more appropriate exploration and optimization strategies, we can significantly enhance these sub-modules and construct powerful LightZero agents to tackle tasks across a wide range of domains, such as board games, Atari, MuJoCo, MiniGrid and GoBigger. Detailed benchmark results reveal the significant potential of such methods in building scalable and efficient decision intelligence. The code is available as part of OpenDILab at https://github.com/opendilab/LightZero.

cs.LG↗

arXiv2022

Core-shell enhanced single particle model for lithium iron phosphate batteries: model formulation and analysis of numerical solutions

In this paper, a core-shell enhanced single particle model for iron-phosphate battery cells is formulated, implemented, and verified. Starting from the description of the positive and negative electrodes charge and mass transport dynamics, the positive electrode intercalation and deintercalation phenomena and associated phase transitions are described with the core-shell modeling paradigm. Assuming two phases are formed in the positive electrode, one rich and one poor in lithium, a core-shrinking problem is formulated and the phase transition is modeled through a shell phase that covers the core one. A careful discretization of the coupled partial differential equations is proposed and used to convert the model into a system of ordinary differential equations. To ensure robust and accurate numerical solutions of the governing equations, a sensitivity analysis of numerical solutions is performed and the best setting, in terms of solver tolerances, solid phase concentration discretization points, and input current sampling time, is determined in a newly developed probabilistic framework. Finally, unknown model parameters are identified at different C-rate scenarios and the model is verified against experimental data.

eess.SY↗

arXiv2022

Core-shell enhanced single particle model for LiFePO$_4$ batteries

In this paper, a novel electrochemical model for LiFePO$_4$ battery cells that accounts for the positive particle lithium intercalation and deintercalation dynamics is proposed. Starting from the enhanced single particle model, mass transport and balance equations along with suitable boundary conditions are introduced to model the phase transformation phenomena during lithiation and delithiation in the positive electrode material. The lithium-poor and lithium-rich phases are modeled using the core-shell principle, where a core composition is encapsulated with a shell composition. The coupled partial differential equations describing the phase transformation are discretized using the finite difference method, from which a system of ordinary differential equations written in state-space representation is obtained. Finally, model parameter identification is performed using experimental data from a 49Ah LFP pouch cell.

eess.SY↗

arXiv2022

FastCover: An Unsupervised Learning Framework for Multi-Hop Influence Maximization in Social Networks

Finding influential users in social networks is a fundamental problem with many possible useful applications. Viewing the social network as a graph, the influence of a set of users can be measured by the number of neighbors located within a given number of hops in the network, where each hop marks a step of influence diffusion. In this paper, we reduce the problem of IM to a budget-constrained d-hop dominating set problem (kdDSP). We propose a unified machine learning (ML) framework, FastCover, to solve kdDSP by learning an efficient greedy strategy in an unsupervised way. As one critical component of the framework, we devise a novel graph neural network (GNN) architecture, graph reversed attention network (GRAT), that captures the diffusion process among neighbors. Unlike most heuristic algorithms and concurrent ML frameworks for combinatorial optimization problems, FastCover determines the entire seed set from the nodes' scores computed with only one forward propagation of the GNN and has a time complexity quasi-linear in the graph size. Experiments on synthetic graphs and real-world social networks demonstrate that FastCover finds solutions with better or comparable quality rendered by the concurrent algorithms while achieving a speedup of over 1000x.

cs.SI↗

arXiv2020

Synthesis and temperature-dependent photoluminescence of high density GeSe triangular nanoplate arrays on Si substrates

We have grown germanium selenide (GeSe) triangular nanoplate arrays (TNAs) with a high density (3.82E+6 / mm2) on the Si (111) substrate using a simple thermal evaporation method. The thickness and trilateral lengths of a single triangular nanoplate were statistically estimated by atomic force microscopy (AFM) as 44 nm, 365 nm, 458 nm and 605 nm, respectively. Transmission electron microscopy (TEM) images and X-ray diffraction (XRD) patterns show that the TNAs were composed of single crystalline GeSe phase. The Se-related defects in the lattice were also revealed by TEM images and Raman vibration modes. Unlike previously reported GeSe compounds, the GeSe TNAs exhibited temperature-dependent photoluminescence (PL). In addition, not previously reported PL peak (1.25 eV) of the 44 nm thick TNAs at 5 K was in the gaps between those of GeSe monolayers (1.5 nm) and thin films (400 nm), revealing a close relationship between the PL peak and the thickness of GeSe. The high-density structure and temperature-dependent PL of the TNAs on the Si substrate may be useful for temperature controllable semiconductor nanodevices.

cond-mat.mes-hall↗

arXiv2018

Maximizing Specific Loss Power for Magnetic Hyperthermia by Hard-Soft Mixed Ferrites

We report maximized specific loss power and intrinsic loss power approaching theoretical limits for AC magnetic field heating of nanoparticles. This is achieved by engineering the effective magnetic anisotropy barrier of nanoparticles via alloying of hard and soft ferrites. 22 nm Co0.03Mn0.28Fe2.7O4/SiO2 NPs reached a specific loss power value of 3417 W/gmetal at a field of 33 kA/m and 380 kHz. Biocompatible Zn0.3Fe2.7O4/SiO2 nanoparticles achieved specific loss power of 500 W/gmetal and intrinsic loss power of 26.8 nHm2/kg at field parameters of 7 kA/m and 380 kHz, below the clinical safety limit. Magnetic bone cement achieved heating adequate for bone tumor hyperthermia, incorporating ultralow dosage of just 1 wt% of nanoparticles. In cellular hyperthermia experiments, these nanoparticles demonstrated high cell death rate at low field parameters. Zn0.3Fe2.7O4/SiO2 nanoparticles show cell viabilities above 97% at concentrations up to 0.5 mg/ml within 48 hrs, suggesting toxicity lower than that of magnetite.

cond-mat.mes-hall↗