arXiv subjects

Yang Li

Publications and source records attributed to Yang Li.

At least 217 records · Page 12Linked to original sources

arXiv2026

Beyond Mode Collapse: Distribution Matching for Diverse Reasoning

On-policy reinforcement learning methods like GRPO suffer from mode collapse: they exhibit reduced solution diversity, concentrating probability mass on a single solution once discovered and ceasing exploration of alternative strategies. We show this stems from reverse KL minimization's mode-seeking behavior, which reinforces the first high-reward trajectory found rather than maintaining a distribution over multiple diverse solutions. We propose DMPO (Distribution-Matching Policy Optimization), which prevents mode collapse through principled approximation of forward KL minimization. DMPO constructs a group level target distribution over sampled trajectories proportional to their rewards, then aligns the policy distribution to this target. This provides mode-covering behavior without requiring sampling from the intractable global target distribution, enabling sustained exploration throughout training. We validate DMPO on NP-hard combinatorial optimization, where exponentially many feasible solutions exist but only a few approach optimality, an ideal testbed for evaluating exploration. DMPO achieves 43.9% Quality Ratio on text-based NP-Bench (vs. GRPO's 40.1%) and 43.1% on vision-based NP-Bench (vs. 38.4%), demonstrating 9% and 12% relative improvements respectively. These gains generalize to mathematical reasoning (+2.0%) and out-of-domain tasks (+2.3%), showing that diversity-preserving training enhances general reasoning capabilities across modalities. Our work establishes distribution matching as a practical, principled approach to preventing mode collapse in on-policy RL, with consistent quality improvements demonstrating sustained exploration across diverse reasoning tasks.

cs.AI↗

arXiv2026

ac strain based thermodynamic criterion for vortex lattice in type-II superconductors

In type-I superconductors, zero electrical resistivity and perfect diamagnetism define two fundamental criteria for superconducting behavior. In contrast, type-II superconductors exhibit more complex mixed state physics, where magnetic flux penetrates the material above the lower critical field Hc1 in the form of quantized vortices, each carrying a single flux quantum. These vortices form a two dimensional lattice which persists up to another irreversible field (Hirr) and then melts into a dissipative liquid phase. The vortex lattice is fundamental to the magnetic and electrical properties of type II superconductors, ac strain susceptibility-a thermodynamic criterion-for identifying this phase has remained elusive. Here, we report the discovery of a dynamic magnetostrictive effect, wherein the geometry of the superconductor oscillates only under an applied alternating magnetic field due to the disturbance of the vortex lattice. This effect is detected by a thin piezoelectric transducer, which converts the excited geometric deformation into an in-phase ac voltage. Notably, we find a direct and nearly linear relationship between the signal amplitude and the vortex density in lattice across several representative type-II superconductors. In the vortex liquid phase above Hirr, the signal amplitude rapidly decays to zero near the upper critical field (Hc2), accompanied by a pronounced out-of-phase component due to enhanced dissipation. This dynamic magnetostrictive effect not only reveals an unexplored magnetoelastic property of the vortex lattice but also establishes a fundamental criterion for identifying the type-II superconductors.

cond-mat.supr-con↗

arXiv2026

Beyond the Final Actor: Modeling the Dual Roles of Creator and Editor for Fine-Grained LLM-Generated Text Detection

The misuse of large language models (LLMs) requires precise detection of synthetic text. Existing works mainly follow binary or ternary classification settings, which can only distinguish pure human/LLM text or collaborative text at best. This remains insufficient for the nuanced regulation, as the LLM-polished human text and humanized LLM text often trigger different policy consequences. In this paper, we explore fine-grained LLM-generated text detection under a rigorous four-class setting. To handle such complexities, we propose RACE (Rhetorical Analysis for Creator-Editor Modeling), a fine-grained detection method that characterizes the distinct signatures of creator and editor. Specifically, RACE utilizes Rhetorical Structure Theory (RST) to construct a logic graph for the creator's foundation while extracting Elementary Discourse Unit (EDU)-level features for the editor's style. Experiments show that RACE outperforms 12 baselines in identifying fine-grained types with low false alarms, offering a policy-aligned solution for LLM regulation.

cs.CL↗

arXiv2026

Learning from Language Feedback via Variational Policy Distillation

Reinforcement learning from verifiable rewards (RLVR) suffers from sparse outcome signals, creating severe exploration bottlenecks on complex reasoning tasks. Recent on-policy self-distillation methods attempt to address this by utilizing language feedback to generate dense, token-level supervision. However, these approaches rely on a fixed, passive teacher to interpret the feedback. As the student policy improves, the teacher's zero-shot assessment capabilities plateau, ultimately halting further learning. To overcome this, we propose Variational Policy Distillation (VPD), a framework that formalizes learning from language feedback as a Variational Expectation-Maximization (EM) problem. VPD co-evolves both policies: in the E-step, the teacher is actively refined on trajectory outcomes via an adaptive trust-region update, translating textual feedback into a dynamically improved target token distribution. In the M-step, the student internalizes this dense distributional guidance on its own on-policy rollouts. By continuously improving the teacher's ability to extract actionable signals from textual critique, VPD overcomes the limitations of passive distillation. Evaluated across diverse sources of diagnostic feedback on scientific reasoning and code generation tasks, VPD consistently outperforms both standard RLVR and existing self-distillation baselines. Finally, by stress-testing our framework on rigid mathematical reasoning and cold-start regimes, we illuminate the fundamental bounds of feedback-driven self-distillation compared to pure environment-driven RL.

cs.LG↗

arXiv2026

Stabilizing, Scaling & Enhancing MeanFlow for Large-scale Diffusion Distillation

Diffusion models exhibit remarkable generative capability, but their high latency limits practical deployment. Many studies have attempted to reduce sampling steps to accelerate inference. Among them, MeanFlow has attracted considerable attention due to its concise formulation and remarkable performance. Nevertheless, the instability of its optimization objective and the ''mean-seeking bias'' have limited its applicability to distill large-scale industrial models. To stabilize MeanFlow for distilling large-scale models, we first introduce a warm-up technique, in which the original differential solution of MeanFlow is replaced by a discrete solution. This design avoids training collapse caused by the MeanFlow target containing a stop-gradient term from an undertrained model. Once the model acquires a preliminary ability to fit the average velocity field, we switch the optimization objective back to the differential solution, enabling further refinement. Meanwhile, to alleviate the ''mean-seeking bias'' of MeanFlow under extremely few-step inference with complex target distributions, we incorporate trajectory distribution alignment as an auxiliary objective, encouraging the student model's trajectory distribution to align more closely with that of the teacher model. Our proposed distillation framework achieves superior performance compared to existing distillation approaches when applied to the text-to-image (T2I) model FLUX.1-dev (up to 12B parameters). Furthermore, when extended to the 80B-parameter state-of-the-art (SOTA) T2I model HunyuanImage 3.0, our method continues to demonstrate robust generalization and strong performance.

cs.CV↗

arXiv2026

One Model to Translate Them All: Universal Any-to-Any Translation for Heterogeneous Collaborative Perception

By sharing intermediate features, collaborative perception extends each agent's sensing beyond standalone limits, but real-world feature modality heterogeneity remains a key barrier to effective fusion. Most existing methods, including direct adaption and protocol-based transformation, typically rely on training adapters for newly emerging feature modalities and often require additional retraining or fine-tuning. Such repeated training is costly and is often infeasible across manufacturers due to model and data privacy constraints, limiting real-world scalability. To address this issue, we propose UniTrans, a universal any-to-any feature modality translation model that instantiates translators on the fly for arbitrary modalities. UniTrans pretrains a bank of translator expert parameters and learns their combination coefficients as a function of source-to-target modality mapping. The mapping is measured in a modality-intrinsic latent space, where an intrinsic encoder extracts modality-specific yet scene-invariant codes from single-frame intermediate features, enabling UniTrans to instantiate translators in a zero-shot manner. Experiments on OPV2V-H and DAIR-V2X demonstrate that UniTrans consistently outperforms state-of-the-art methods in both simulated and real-world settings, enabling efficient any-to-any translation through a universal model. The code is available at https://github.com/CheeryLeeyy/UniTrans.

cs.CV↗

arXiv2026

TabQL: In-Context Q-Learning with Tabular Foundation Models

We propose Tabular Q-Learning (TabQL), a reinforcement learning framework that replaces the conventional parametric Q-network in Deep Q-Learning (DQN) with a tabular foundation model endowed with in-context learning capabilities. The key idea is to represent Q-values through a sequence-to-sequence foundation model operating over a tabularized representation of state-action-Q-value tuples, enabling rapid adaptation from limited online interaction by conditioning on recent experience. TabQL departs from classical DQN by leveraging (i) zero- or few-shot Q-value inference via in-context updates, and (ii) a warm-up phase using standard DQN to bootstrap high-quality context. Particularly, to enhance the context quality, new transitions are generated by executing actions output by TabQL with predicted Q values from DQN. We formalize TabQL, analyze its convergence and sample complexity under mild assumptions, and show that TabQL interpolates between vanilla Q-learning and DQN with in-context learning. Our analysis demonstrates that TabQL achieves improved efficiency compared to DQN by amortizing Bellman updates through in-context learning. Extensive numerical experiments with several benchmarks showcase the effectiveness and efficacy of the proposed TabQL.

cs.LG↗

arXiv2026

Heterogeneity-Aware Dataset Scheduling for Efficient Audio Large Language Model Training

Training general-purpose Audio Large Language Models (ALLMs) across diverse datasets is essential for holistic audio understanding, yet it faces significant challenges due to dataset heterogeneity, which often leads to conflicting gradients and slow convergence. Despite its impact, how to explicitly manage this heterogeneity during training remains underexplored, with current practices relying primarily on uniform mixture. In this work, we analyze multi-dataset AudioQA training from a convergence perspective and propose Grouped Sequential Training (GST). GST strategically organizes datasets into affinity-aware groups and introduces them via a progressive scheduling protocol, effectively balancing the stability of parallel training with the efficiency of sequential optimization. To ensure scalability, we develop gradient-based affinity metrics that capture inter-dataset relationships without the prohibitive cost of empirical transferability estimation. Extensive evaluations on 14 AudioQA datasets spanning speech, music, and environmental sounds demonstrate that GST achieves 30--40\% faster convergence than standard parallel training while maintaining or even surpassing the performance of mix-all training. Our results provide both theoretical insights and a practical, model-agnostic framework for efficient large-scale ALLM optimization.

cs.SD↗

arXiv2026

Unlocking Biological Workflows for Robust Protein-Text Question Answering: A Dual-Dimensional RAG Framework

Protein-Text Question Answering (QA) is crucial for interpreting biological sequences through natural language. The integration of Large Language Models (LLMs) with Retrieval-Augmented Generation (RAG) that efficiently leverages biological databases and facilitates reasoning offers a potent approach for it. However, constrained by the standard RAG pipeline, these models often rely on curated, static datasets instead of expert-proven biological workflows, lacking the fine-grained information processing and struggling to generalize to novel (OOD) proteins. To bridge this gap, we propose 2D-ProteinRAG, a novel framework that empowers LLMs to operate within the gold-standard biological research workflow (BLAST). To further extract high-quality information from noisy retrieval contexts, we introduce a dual-dimensional (2D) filtering strategy following the expert analytical paradigms. Horizontal Fine-grained Attribute Alignment utilizes a lightweight, intent-aware discriminative filter to prune irrelevant metadata and align database entries with specific user queries. Vertical Homology-based Semantic Denoising resolves functional contradictions and redundancy across multiple homologs via hierarchical clustering. Extensive evaluations on both In-Distribution and diverse biological OOD benchmarks demonstrate that 2D-ProteinRAG consistently achieves state-of-the-art performance, outperforming fine-tuned baselines and other RAG methods. Our results validate the framework's robustness and scalability, providing a practical solution for interpreting protein functions in real-world scientific scenarios.

cs.IR↗

arXiv2026

Quantum entanglement between partons in a strongly coupled quantum field theory

We perform a first-principles, non-perturbative investigation of quantum entanglement between partonic constituents in a strongly coupled 3+1-dimensional scalar Yukawa theory, using light-front Hamiltonian methods with controlled Fock-space truncations. By explicitly constructing reduced density matrices for (mock) nucleon, pion, and anti-nucleon subsystems from light-front wave functions, we compute key entanglement witnesses, including von Neumann entropy, mutual information, and linear entropy, in both quenched (no sea pairs) and unquenched frameworks. We find that the entanglement entropy is closely related to the Shannon entropy of the transverse momentum dependent distribution, establishing a link between quantum information and parton structure. In contrast, the unquenched theory reveals genuinely non-classical correlations: the entanglement entropy cannot be reduced to any Shannon entropy of normalized parton distributions, demonstrating that the full hadronic wave function encodes quantum information beyond classical probabilities. Our findings highlight the role of entanglement as a fundamental probe of non-perturbative dynamics in relativistic quantum field theory and lay the groundwork for extending these concepts to QCD and future collider phenomenology.

hep-ph↗

arXiv2026

Rethinking Predictive Modeling for LLM Routing: When Simple kNN Beats Complex Learned Routers

As large language models (LLMs) grow in scale and specialization, routing--selecting the best model for a given input--has become essential for efficient and effective deployment. While recent methods rely on complex learned routing strategies, their dependence on disparate training data and evaluation setups makes comparison and generalization difficult. In this work, we revisit LLM routing through the lens of simplicity. We show that a well-tuned k-Nearest Neighbors (kNN) approach not only matches but often outperforms state-of-the-art learned routers across diverse tasks. To support systematic evaluation, we introduce a suite of standardized routing benchmarks spanning instruction-following, question-answering, and reasoning tasks, as well as the first multi-modal routing dataset involving visual inputs. Our findings reveal that the locality properties of model performance in embedding space enable simple non-parametric methods to achieve strong routing decisions with lower sample complexity than parametric approaches. This challenges the prevailing trend toward sophisticated architectures and highlights the importance of thoroughly evaluating simple baselines before investing in complex solutions. To support reproducibility and further exploration, we will release all benchmarks and code upon publication.

cs.LG↗

arXiv2026

KAST-BAR: Knowledge-Anchored Semantically-Dynamic Topology Brain Autoregressive Modeling for Universal Neural Interpretation

While EEG foundation models have shown significant potential in universal neural decoding across tasks, their advancement remains constrained by the inadequacy modeling of complex spatiotemporal topology, as well as the inherent modality gap between low-level physiological signals and high-level textual semantics. To address these challenges, we propose a Knowledge-Anchored Semantically-Dynamic Topology Brain Autoregressive Model (KAST-BAR), which dynamically aligns physiological representations derived from multi-level brain topology with an expert-level semantic space. Specifically, we design a Dual-Stream Hierarchical Attention (DSHA) encoder that accurately captures the brain's intrinsic non-Euclidean topology by modeling local temporal dynamics with global spatial contexts. On this basis, a Knowledge-Anchored Semantic Profiler (KASP) is proposed to synthesize physically-grounded and instance-level textual profiles, which subsequently drive a Semantic Text-Aware Refiner (STAR) to dynamically reconstruct EEG representations using Latent Expert Queries. By conducting large-scale pre-training on 21 diverse datasets to build a foundation model, KAST-BAR effectively integrates expert-level medical knowledge into EEG signal representations, consistently achieving superior performance across six downstream tasks. Our code is available at https://github.com/KAST-BAR/KAST-BAR

cs.LG↗

arXiv2026

Introducing Environmental Constraints to Grasping Strategies for Paper-Like Flexible Materials Using a Soft Gripper

Robotic manipulation of flexible objects is widely required in both industrial and service applications. Among such objects, paper-like materials exhibit distinct mechanical characteristics compared to cloth, being more sensitive to compressive stress, where minor variations in physical properties can significantly affect grasping. This study systematically investigates grasping strategies for paper-like materials using a universal soft gripper by exploiting environmental constraints. Based on manipulation primitives employed in existing grasping strategies, we proposed systematic grasping strategies for flexible materials by exploiting environmental constraints and analyzed their mechanical and kinematic models. To investigate the influence of materials and working conditions on grasping, an evaluation system for measuring grasping force and success rate was defined and experimentally evaluated. Finally, we summarized the specific workspaces and characteristics of different strategies that can satisfy various task requirements and lead to potential applications in household service robots for grasping planar flexible objects.

cs.RO↗

arXiv2026

A framework for constructing non-GRS MDS-NMDS codes from deep holes and its application

Maximum distance separable (MDS) codes and near MDS (NMDS) codes are of particular interest in coding theory due to their optimal error-correcting capabilities and wide applications in communication, cryptography, and storage systems. A family of linear codes is called a family of non-GRS MDS-NMDS codes if for each $[n,k]_q$ code in the family, it is either an $[n,k,n-k+1]_q$ MDS code that is not monomially equivalent to any GRS code or extended GRS code, or an $[n,k,n-k]_q$ NMDS code. This paper develops a unified framework for constructing new families of non-GRS MDS-NMDS codes via deep holes. We show that, starting from a family of $[n,k]_q$ non-GRS MDS-NMDS codes with covering radius $n-k$, one can systematically obtain more $[n+1,k+1]_q$ non-GRS MDS-NMDS codes. The proposed framework is further reformulated in terms of the second kind of extended codes. This reformulation recovers a main result of Wu, Ding, and Chen (IEEE Trans. Inf. Theory, 71(1): 263-272, 2025), provides a provable reduction in the computational complexity compared with the approach of Ma, Kai, and Zhu (Finite Fields Appl., 114, 102844, 2026), and reveals additional structural properties of the resulting codes. As an application, we determine the covering radius and characterize two classes of deep holes of extended subcodes of GRS codes. By applying our framework, we obtain three new families of non-GRS MDS-NMDS codes and investigate the monomial equivalence between the resulting codes and Roth-Lempel codes.

cs.IT↗

arXiv2026

LACE: Lattice Attention for Cross-thread Exploration

Current large language models reason in isolation. Although it is common to sample multiple reasoning paths in parallel, these trajectories do not interact, and often fail in the same redundant ways. We introduce LACE, a framework that transforms reasoning from a collection of independent trials into a coordinated, parallel process. By repurposing the model architecture to enable cross-thread attention, LACE allows concurrent reasoning paths to share intermediate insights and correct one another during inference. A central challenge is the absence of natural training data that exhibits such collaborative behavior. We address this gap with a synthetic data pipeline that explicitly teaches models to communicate and error-correct across threads. Experiments show that this unified exploration substantially outperforms standard parallel search, improving reasoning accuracy by over 7 points. Our results suggest that large language models can be more effective when parallel reasoning paths are allowed to interact.

cs.AI↗

arXiv2026

Generative Giants, Retrieval Weaklings: Why do Multimodal Large Language Models Fail at Multimodal Retrieval?

Despite the remarkable success of multimodal large language models (MLLMs) in generative tasks, we observe that they exhibit a counterintuitive deficiency in the zero-shot multimodal retrieval task. In this work, we investigate the underlying mechanisms that hinder MLLMs from being effective retrievers. With the help of sparse autoencoders (SAEs), we decompose MLLM output representations into interpretable semantic concepts to probe their intrinsic behavior. Our analysis reveals that the representation space of MLLMs is overwhelmingly dominated by textual semantics; and the visual semantics essential for multimodal retrieval only constitute a small portion. We find that this imbalance is compounded by the heavy focus of MLLMs on bridging image-text modalities, which facilitates generation but homogenizes embeddings and finally diminishes the discriminative power required for multimodal retrieval. We further discover that the specific feature components that contribute most to the similarity computations of MLLMs are actually distractors that greatly reduce retrieval performance. Building on these insights, we propose ReAlign, a test-time adaptation approach that applies a whitening transformation to adjust the geometry of MLLM representation spaces. Empirical results show that this simple intervention consistently improves zero-shot multimodal retrieval performance across diverse MLLMs without fine-tuning efforts. The code is available at https://github.com/Heinz217/mllm-retrieval-analysis.

cs.CV↗

arXiv2026

Forge: Quality-Aware Reinforcement Learning for NP-Hard Optimization in LLMs

Large Language Models (LLMs) have achieved remarkable success on reasoning benchmarks through Reinforcement Learning with Verifiable Rewards (RLVR), excelling at tasks such as math, coding, logic, and puzzles. However, existing benchmarks evaluate only correctness, while overlooking optimality, namely the ability to find the best solutions under constraints. We propose OPT-BENCH, the first comprehensive framework for training and evaluating LLMs on NP-hard optimization problems through quality-aware RLVR. OPT-BENCH provides three key components: a scalable training infrastructure with instance generators, quality verifiers, and optimal baselines across 10 tasks; a rigorous benchmark with 1,000 instances evaluating both feasibility, measured by Success Rate, and quality, measured by Quality Ratio; and quality-aware rewards that enable continuous improvement beyond binary correctness. Training on Qwen2.5-7B-Instruct-1M with 15K examples achieves 93.1% SR and 46.6% QR, significantly outperforming GPT-4o, which achieves 29.6% SR and 14.6% QR. Beyond optimization, training on OPT-BENCH transfers to diverse tasks, including mathematics (+2.2%), logic (+1.2%), knowledge (+4.1%), and instruction following (+6.1%). Our analysis reveals that quality-aware rewards improve solutions by 28.8% over binary rewards, and that task diversity drives generalization more than data quantity, offering insights into RLVR scaling for complex reasoning.

cs.AI↗

arXiv2026

A Hybrid Graph Neural Network for Enhanced EEG-Based Depression Detection

Graph neural networks (GNNs) are becoming increasingly popular for EEG-based depression detection. However, previous GNN-based methods fail to sufficiently consider the characteristics of depression, thus limiting their performance. Firstly, studies in neuroscience indicate that depression patients exhibit both common and individualized brain abnormal patterns. Previous GNN-based approaches typically focus either on fixed graph connections to capture common abnormal brain patterns or on adaptive connections to capture individualized patterns, which is inadequate for depression detection. Secondly, brain network exhibits a hierarchical structure, which includes the arrangement from channel-level graph to region-level graph. This hierarchical structure varies among individuals and contains significant information relevant to detecting depression. Nonetheless, previous GNN-based methods overlook these individualized hierarchical information. To address these issues, we propose a Hybrid GNN (HGNN) that merges a Common Graph Neural Network (CGNN) branch utilizing fixed connection and an Individualized Graph Neural Network (IGNN) branch employing adaptive connections. The two branches capture common and individualized depression patterns respectively, complementing each other. Furthermore, we enhance the IGNN branch with a Graph Pooling and Unpooling Module (GPUM) to extract individualized hierarchical information. Extensive experiments on two public datasets show that our model achieves state-of-the-art performance.

eess.SP↗