arXiv subjects

Peng Li

Publications and source records attributed to Peng Li.

At least 109 records · Page 6Linked to original sources

arXiv2026

CoreEditor: Correspondence-constrained Diffusion for Consistent 3D Editing

Text-driven 3D editing seeks to modify 3D scenes according to textual descriptions, and most existing approaches tackle this by adapting pre-trained 2D image editors to multi-view inputs. However, without explicit control over multi-view information exchange, they often fail to maintain cross-view consistency, leading to insufficient edits and blurry details. We introduce CoreEditor, a novel framework for consistent text-to-3D editing. The key innovation is a correspondence-constrained attention mechanism that enforces precise interactions between pixels expected to remain consistent throughout the diffusion denoising process. Beyond relying solely on geometric alignment, we further incorporate semantic similarity estimated during denoising, enabling more reliable correspondence modeling and robust multi-view editing. In addition, we design a selective editing pipeline that allows users to choose preferred results from multiple candidates, offering greater flexibility and user control. Extensive experiments show that CoreEditor produces high-quality, 3D-consistent edits with sharper details, significantly outperforming prior methods.

cs.CV↗

arXiv2026

Beyond Words: Evaluating and Bridging Epistemic Divergence in User-Agent Interaction via Theory of Mind

Large Language Models (LLMs) have developed rapidly and are widely applied to both general-purpose and professional tasks to assist human users. However, they still struggle to comprehend and respond to the true user needs when intentions and instructions are imprecisely conveyed, leading to a divergence between subjective user believes and true environment states. Resolving this epistemic divergence requires Theory of Mind (ToM), yet existing ToM evaluations for LLMs primarily focus on isolated belief inference, overlooking its functional utility in real-world interaction. To this end, we formalize ToM for LLMs as a mechanism for epistemic divergence detection and resolution, and propose a benchmark, \benchname, to assess how models reconcile user beliefs and profiles in practice. Results across 11 leading models reveal a significant limitation to identify underlying cognitive gaps that impede task success. To bridge this gap, we further curate a trajectory-based ToM dataset linking belief tracking with task-related state inference. The model trained on this data via reinforcement learning shows consistent improvement in reasoning about user mental states, leading to enhanced downstream performance. Our work highlights the practical value of ToM as an essential interaction-level mechanism rather than as a standalone reasoning skill.

cs.CL↗

arXiv2026

DeepPrep: An LLM-Powered Agentic System for Autonomous Data Preparation

Data preparation, which aims to transform heterogeneous and noisy raw tables into analysis-ready data, remains a major bottleneck in data science. Recent approaches leverage large language models (LLMs) to automate data preparation from natural language specifications. However, existing LLM-powered methods either make decisions without grounding in intermediate execution results, or rely on linear interaction processes that offer limited support for revising earlier decisions. To address these limitations, we propose DeepPrep, an LLM-powered agentic system for autonomous data preparation. DeepPrep constructs data preparation pipelines through iterative, execution-grounded interaction with an environment that materializes intermediate table states and returns runtime feedback. To overcome the limitations of linear interaction, DeepPrep organizes pipeline construction with tree-based agentic reasoning, enabling structured exploration and non-local revision based on execution feedback. To enable effective learning of such behaviors, we propose a progressive agentic training framework, together with data synthesis that supplies diverse and complex ADP tasks. Extensive experiments show that DeepPrep achieves data preparation accuracy comparable to strong closed-source models (e.g., GPT-5) while incurring 15x lower inference cost, while establishing state-of-the-art performance among open-source baselines and generalizing effectively across diverse datasets.

cs.DB↗

arXiv2026

Optimized adiabatic-impulse protocol preserving Kibble-Zurek scaling with attenuated anti-Kibble-Zurek behavior

We propose an optimized adiabatic-impulse (OAI) protocol that substantially reduces the evolution time for crossing a quantum phase transition while preserving Kibble-Zurek (KZ) scaling. Near criticality, the control parameter is ramped linearly across the critical point at a rate characterized by a quench time $τ_Q$. Away from criticality, the evolution remains adiabatic and is tuned close to the threshold of adiabatic breakdown, as quantified by an adiabatic coefficient $ζ$ that scales as $τ_Q^α$. As a consequence, the total evolution time exhibits a sublinear power-law dependence on $τ_Q$, and the conventional linear quench is recovered in the limit $α\rightarrow\infty$. We apply the OAI protocol to the transverse Ising chain and numerically determine the minimal $ζ$ required for KZ scaling. We further investigate the nonequilibrium dynamics in the presence of a noisy field that can induce anti-Kibble-Zurek (AKZ) behavior. Within the OAI protocol, noise-induced defects is significantly attenuated due to the shorter evolution time. The optimal quench time at which the defect density is minimized obeys an altered universal power-law scaling with the noise strength. Finally, we generalize the OAI protocol to the nonlinear quenches and numerically demonstrate a marked reduction in noise-induced defects.

quant-ph↗

arXiv2026

$D^*_{(s)}\to P$ form factors and their applications to semi-leptonic and non-leptonic weak decays

Similar to other heavy flavor mesons, the weak decays of $D^*_{(s)}$ mesons can provide a platform to verify the standard model, explore new physics, and understand the mechanisms of weak interactions. At present, the theoretical and experimental studies on $D^*_{(s)}$ mesons are relatively limited. In addition to the dominant electromagnetic decays, the $D^*_{(s)}$ weak decays should be feasible to explore the $D^*_{(s)}$ mesons. In this study, we used the covariant light-front quark model to study the form factors of the transitions $D^*_{(s)}\to π, K, η_{q,s}$, then calculated the branching ratios of the semi-leptonic decays $D^*_{(s)}\to P\ell^{+}ν_{\ell}$ and the non-leptonic decays $D^*_{(s)}\to PP, PV$ with $P=π, K, η^{(\prime)}, V=ρ, K^*, ϕ$ and $\ell=e, μ$. The channels $D_{s}^{*+}\toη\ell^{+}ν_{\ell}$ and $D^{*+}_{s}\to ηρ^{+}$ possess the largest branching ratios, which can reach an order of $10^{-6}$ among these decays, and are most likely to be accessible in experiments at future high-luminosity colliders. Furthermore, we predict and discuss the longitudinal polarization fraction $f_{L}$ and the forward-backward asymmetry $A_{FB}$ for the considered semi-leptonic $D^*_{(s)}$ decays.

hep-ph↗

arXiv2026

InstructDiff: Domain-Adaptive Data Selection via Differential Entropy for Efficient LLM Fine-Tuning

Supervised fine-tuning (SFT) is fundamental to adapting large language models, yet training on complete datasets incurs prohibitive costs with diminishing returns. Existing data selection methods suffer from severe domain specificity: techniques optimized for general instruction-following fail on reasoning tasks, and vice versa. We observe that measuring entropy differences between base models and minimally instruction-tuned calibrated models reveals a pattern -- samples with the lowest differential entropy consistently yield optimal performance across domains, yet this principle manifests domain-adaptively: reasoning tasks favor entropy increase (cognitive expansion), while general tasks favor entropy decrease (cognitive compression). We introduce InstructDiff, a unified framework that operationalizes differential entropy as a domain-adaptive selection criterion through warmup calibration, bi-directional NLL filtering, and entropy-based ranking. Extensive experiments show that InstructDiff achieves 17\% relative improvement over full data training on mathematical reasoning and 52\% for general instruction-following, outperforming prior baselines while using only 10\% of the data.

cs.CL↗

arXiv2026

FRoM-W1: Towards General Humanoid Whole-Body Control with Language Instructions

Humanoid robots are capable of performing various actions such as greeting, dancing and even backflipping. However, these motions are often hard-coded or specifically trained, which limits their versatility. In this work, we present FRoM-W1, an open-source framework designed to achieve general humanoid whole-body motion control using natural language. To universally understand natural language and generate corresponding motions, as well as enable various humanoid robots to stably execute these motions in the physical world under gravity, FRoM-W1 operates in two stages: (a) H-GPT: utilizing massive human data, a large-scale language-driven human whole-body motion generation model is trained to generate diverse natural behaviors. We further leverage the Chain-of-Thought technique to improve the model's generalization in instruction understanding. (b) H-ACT: After retargeting generated human whole-body motions into robot-specific actions, a motion controller that is pretrained and further fine-tuned through reinforcement learning in physical simulation enables humanoid robots to accurately and stably perform corresponding actions. It is then deployed on real robots via a modular simulation-to-reality module. We extensively evaluate FRoM-W1 on Unitree H1 and G1 robots. Results demonstrate superior performance on the HumanML3D-X benchmark for human whole-body motion generation, and our introduced reinforcement learning fine-tuning consistently improves both motion tracking accuracy and task success rates of these humanoid robots. We open-source the entire FRoM-W1 framework and hope it will advance the development of humanoid intelligence.

cs.RO↗

arXiv2026

SPEC-RL: Accelerating On-Policy Reinforcement Learning with Speculative Rollouts

Large Language Models (LLMs) increasingly rely on reinforcement learning with verifiable rewards (RLVR) to elicit reliable chain-of-thought reasoning. However, the training process remains bottlenecked by the computationally expensive rollout stage. Existing acceleration methods-such as parallelization, objective- and data-driven modifications, and replay buffers-either incur diminishing returns, introduce bias, or overlook redundancy across iterations. We identify that rollouts from consecutive training epochs frequently share a large portion of overlapping segments, wasting computation. To address this, we propose SPEC-RL, a novel framework that integrates SPECulative decoding with the RL rollout process. SPEC-RL reuses prior trajectory segments as speculative prefixes and extends them via a draft-and-verify mechanism, avoiding redundant generation while ensuring policy consistency. Experiments on diverse math reasoning and generalization benchmarks, including AIME24, MATH-500, OlympiadBench, MMLU-STEM, and others, demonstrate that SPEC-RL reduces rollout time by 2-3x without compromising policy quality. As a purely rollout-stage enhancement, SPEC-RL integrates seamlessly with mainstream algorithms (e.g., PPO, GRPO, DAPO), offering a general and practical path to scale RLVR for large reasoning models. Our code is available at https://github.com/ShopeeLLM/Spec-RL

cs.LG↗

arXiv2026

Browse and Concentrate: Comprehending Multimodal Content via prior-LLM Context Fusion

With the bloom of Large Language Models (LLMs), Multimodal Large Language Models (MLLMs) that incorporate LLMs with pre-trained vision models have recently demonstrated impressive performance across diverse vision-language tasks. However, they fall short to comprehend context involving multiple images. A primary reason for this shortcoming is that the visual features for each images are encoded individually by frozen encoders before feeding into the LLM backbone, lacking awareness of other images and the multimodal instructions. We term this issue as prior-LLM modality isolation and propose a two phase paradigm, browse-and-concentrate, to enable in-depth multimodal context fusion prior to feeding the features into LLMs. This paradigm initially "browses" through the inputs for essential insights, and then revisits the inputs to "concentrate" on crucial details, guided by these insights, to achieve a more comprehensive understanding of the multimodal inputs. Additionally, we develop training strategies specifically to enhance the understanding of multi-image inputs. Our method markedly boosts the performance on 7 multi-image scenarios, contributing to increments on average accuracy by 2.13% and 7.60% against strong MLLMs baselines with 3B and 11B LLMs, respectively.

cs.CL↗

arXiv2026

SDCD: Structure-Disrupted Contrastive Decoding for Mitigating Hallucinations in Large Vision-Language Models

Large Vision-Language Models (LVLMs) demonstrate significant progress in multimodal understanding and reasoning, yet object hallucination remains a critical challenge. While existing research focuses on mitigating language priors or high-level statistical biases, they often overlook the internal complexities of the visual encoding process. We identify that visual statistical bias, arising from the inherent Bag-of-Patches behavior of Vision Encoders under weak structural supervision, acts as a contributing factor of object hallucinations. Under this bias, models prioritize local texture features within individual patches over holistic geometric structures. This tendency may induce spurious visual confidence and result in hallucinations. To address this, we introduce a training-free algorithm called Structure-Disrupted Contrastive Decoding (SDCD), which performs contrastive calibration of the output distribution by introducing a shuffled structure-disrupted view. By penalizing tokens that maintain high confidence under this structure-less view, SDCD effectively suppresses the texture-driven bias. Experimental results demonstrate that SDCD significantly mitigates hallucinations across multiple benchmarks and enhances the overall multimodal capabilities of LVLMs.

cs.CV↗

arXiv2026

UniSH: Unifying Scene and Human Reconstruction in a Feed-Forward Pass

We present UniSH, a unified, feed-forward framework for joint metric-scale 3D scene and human reconstruction. A key challenge in this domain is the scarcity of large-scale, annotated real-world data, forcing a reliance on synthetic datasets. This reliance introduces a significant sim-to-real domain gap, leading to poor generalization, low-fidelity human geometry, and poor alignment on in-the-wild videos. To address this, we propose an innovative training paradigm that effectively leverages unlabeled in-the-wild data. Our framework bridges strong, disparate priors from scene reconstruction and HMR, and is trained with two core components: (1) a robust distillation strategy to refine human surface details by distilling high-frequency details from an expert depth model, and (2) a two-stage supervision scheme, which first learns coarse localization on synthetic data, then fine-tunes on real data by directly optimizing the geometric correspondence between the SMPL mesh and the human point cloud. This approach enables our feed-forward model to jointly recover high-fidelity scene geometry, human point clouds, camera parameters, and coherent, metric-scale SMPL bodies, all in a single forward pass. Extensive experiments demonstrate that our model achieves state-of-the-art performance on human-centric scene reconstruction and delivers highly competitive results on global human motion estimation, comparing favorably against both optimization-based frameworks and HMR-only methods. Project page: https://murphylmf.github.io/UniSH/

cs.CV↗

arXiv2025

A Survey of Efficient Reasoning for Large Reasoning Models: Language, Multimodality, and Beyond

Recent Large Reasoning Models (LRMs), such as DeepSeek-R1 and OpenAI o1, have demonstrated strong performance gains by scaling up the length of Chain-of-Thought (CoT) reasoning during inference. However, a growing concern lies in their tendency to produce excessively long reasoning traces, which are often filled with redundant content (e.g., repeated definitions), over-analysis of simple problems, and superficial exploration of multiple reasoning paths for harder tasks. This inefficiency introduces significant challenges for training, inference, and real-world deployment (e.g., in agent-based systems), where token economy is critical. In this survey, we provide a comprehensive overview of recent efforts aimed at improving reasoning efficiency in LRMs, with a particular focus on the unique challenges that arise in this new paradigm. We identify common patterns of inefficiency, examine methods proposed across the LRM lifecycle, i.e., from pretraining to inference, and discuss promising future directions for research. To support ongoing development, we also maintain a real-time GitHub repository tracking recent progress in the field. We hope this survey serves as a foundation for further exploration and inspires innovation in this rapidly evolving area.

cs.CL↗

arXiv2025

U-Net-Like Spiking Neural Networks for Single Image Dehazing

Image dehazing is a critical challenge in computer vision, essential for enhancing image clarity in hazy conditions. Traditional methods often rely on atmospheric scattering models, while recent deep learning techniques, specifically Convolutional Neural Networks (CNNs) and Transformers, have improved performance by effectively analyzing image features. However, CNNs struggle with long-range dependencies, and Transformers demand significant computational resources. To address these limitations, we propose DehazeSNN, an innovative architecture that integrates a U-Net-like design with Spiking Neural Networks (SNNs). DehazeSNN captures multi-scale image features while efficiently managing local and long-range dependencies. The introduction of the Orthogonal Leaky-Integrate-and-Fire Block (OLIFBlock) enhances cross-channel communication, resulting in superior dehazing performance with reduced computational burden. Our extensive experiments show that DehazeSNN is highly competitive to state-of-the-art methods on benchmark datasets, delivering high-quality haze-free images with a smaller model size and less multiply-accumulate operations. The proposed dehazing method is publicly available at https://github.com/HaoranLiu507/DehazeSNN.

cs.CV↗

arXiv2025

LVAgent: Long Video Understanding by Multi-Round Dynamical Collaboration of MLLM Agents

Existing MLLMs encounter significant challenges in modeling the temporal context within long videos. Currently, mainstream Agent-based methods use external tools to assist a single MLLM in answering long video questions. Despite such tool-based support, a solitary MLLM still offers only a partial understanding of long videos, resulting in limited performance. In order to better address long video tasks, we introduce LVAgent, the first framework enabling multi-round dynamic collaboration of MLLM agents in long video understanding. Our method consists of four key steps: 1) Selection: We pre-select appropriate agents from the model library to form optimal agent teams based on different tasks. 2) Perception: We design an effective retrieval scheme for long videos to improve the coverage of critical temporal segments while maintaining computational efficiency. 3) Action: Agents answer long video questions and exchange reasons. 4) Reflection: We evaluate each agent's performance in each round of discussion and optimize the agent team for dynamic collaboration. The agents iteratively refine their answers by multi-round dynamical collaboration of MLLM agents. LVAgent is the first agent system method that outperforms all closed-source models (like GPT-4o) and open-source models (like InternVL-2.5 and Qwen2-VL) in the long video understanding tasks. Our LVAgent achieves an accuracy of 80\% on four mainstream long video understanding tasks. Notably, LVAgent improves accuracy by 13.3\% on LongVideoBench. Code is available at https://github.com/64327069/LVAgent.

cs.CV↗

arXiv2025

Thinking with Visual Abstract: Enhancing Multimodal Reasoning via Visual Abstraction

Images usually convey richer detail than text, but often include redundant information, which potentially downgrades multimodal reasoning performance. When faced with lengthy or complex messages, humans tend to employ abstract thinking to convert them into simple and concise abstracts. Inspired by this cognitive strategy, we introduce a novel paradigm to elicit the ability to Think with Visual Abstract (VAT), by prompting Multimodal Large Language Models (MLLMs) with visual abstract instead of explicit verbal thoughts or elaborate guidance, permitting a more efficient visual reasoning mechanism via concentrated perception. VAT encourages models to focus on more essential visual elements, concepts and structural features by undermining redundant information compared with explicit thinking methods, such as Chain-of-thought (CoT) and tool-using approaches, that increase the complexity of reasoning process via inserting verbose intermediate steps and external knowledge. Experimental results show that VAT consistently empowers different MLLMs in visual perception and reasoning tasks. VAT achieves an average gain of $2.21\%$ over GPT-5 baseline, surpassing the gain of CoT, demonstrating that VAT better enhances multimodal task performance of MLLMs. Additionally, VAT spends fewer tokens while achieving higher performance. These findings highlight the effectiveness of visual abstract thinking and encourage further exploration of more diverse reasoning paradigms from the perspective of human cognition.

cs.CL↗

arXiv2025

Visual-Friendly Concept Protection via Selective Adversarial Perturbations

Personalized concept generation by tuning diffusion models with a few images raises potential legal and ethical concerns regarding privacy and intellectual property rights. Researchers attempt to prevent malicious personalization using adversarial perturbations. However, previous efforts have mainly focused on the effectiveness of protection while neglecting the visibility of perturbations. They utilize global adversarial perturbations, which introduce noticeable alterations to original images and significantly degrade visual quality. In this work, we propose the Visual-Friendly Concept Protection (VCPro) framework, which prioritizes the protection of key concepts chosen by the image owner through adversarial perturbations with lower perceptibility. To ensure these perturbations are as inconspicuous as possible, we introduce a relaxed optimization objective to identify the least perceptible yet effective adversarial perturbations, solved using the Lagrangian multiplier method. Qualitative and quantitative experiments validate that VCPro achieves a better trade-off between the visibility of perturbations and protection effectiveness, effectively prioritizing the protection of target concepts in images with less perceptible perturbations.

cs.CV↗

arXiv2025

Exceptional Alkaline Methanol Electrooxidation on Bi-modified Pt3M Intermetallics: Kinetic Origins and an OH Binding Energy Descriptor

The exploration of advanced CO-free catalysts and clarifying the ambiguous kinetic origins and governing factors would undoubtedly open up opportunities to overcome the sluggish kinetics of methanol electrooxidation and promote the development of direct methanol fuel cells. Herein, we constructed a family of Bi-modified Pt3M intermetallic catalysts (Bi-Pt3M/C, M=Cr, Mn, Co, Zn, In, Ga, and Sn) that follow CO-free dominated pathway and exhibit exceptional catalytic activity. More significantly, leveraging this platform, we have identified the pivotal factor governing the reaction kinetics in CO-free pathway, namely OH binding energy (OHBE). This arises because the rate-determining step (RDS) encompasses both C-H bond activation and water dissociation, whose respective barriers can be reflected by the OHBE. Accordingly, OHBE can act as an activity descriptor. Specifically, Bi-Pt3In/C stands out from other Bi-Pt3M/C and delivers the unprecedented mass activity of 36.7 A mgPt-1 at peak potential, far exceeding state-of-the-art Pt-based catalysts reported to date. Taking Bi-Pt3In/C as a proof of concept, we clearly elucidate the origin of enhanced MOR activity by combining theoretical calculations, kinetic isotope effects, and formaldehyde electrooxidation. Moreover, there exhibits a volcano-type trend between OHBE and the activity of Bi-Pt3M/C. Beyond the discovery of ultrahigh-performance catalysts, these findings provide a detailed mechanistic picture of RDS and offer an innovative design principle for advanced catalysts.

physics.chem-ph↗

arXiv2025

VEGAS: Mitigating Hallucinations in Large Vision-Language Models via Vision-Encoder Attention Guided Adaptive Steering

Large vision-language models (LVLMs) exhibit impressive ability to jointly reason over visual and textual inputs. However, they often produce outputs that are linguistically fluent but factually inconsistent with the visual evidence, i.e., they hallucinate. Despite growing efforts to mitigate such hallucinations, a key question remains: what form of visual attention can effectively suppress hallucinations during decoding? In this work, we provide a simple answer: the vision encoder's own attention map. We show that LVLMs tend to hallucinate when their final visual-attention maps fail to concentrate on key image objects, whereas the vision encoder's more concentrated attention maps substantially reduce hallucinations. To further investigate the cause, we analyze vision-text conflicts during decoding and find that these conflicts peak in the language model's middle layers. Injecting the vision encoder's attention maps into these layers effectively suppresses hallucinations. Building on these insights, we introduce VEGAS, a simple yet effective inference-time method that integrates the vision encoder's attention maps into the language model's mid-layers and adaptively steers tokens which fail to concentrate on key image objects. Extensive experiments across multiple benchmarks demonstrate that VEGAS consistently achieves state-of-the-art performance in reducing hallucinations.

cs.CV↗