Searcharxiv⌕ Search

arXiv subjects

Jian Li

Publications and source records attributed to Jian Li.

At least 109 records · Page 6Linked to original sources
arXiv2026

The Diminishing Returns of Early-Exit Decoding in Modern LLMs

In Large Language Model (LLM) inference, early-exit refers to stopping computation at an intermediate layer once the prediction is sufficiently confident, thereby reducing latency and cost. However, recent LLMs adopt improved pretraining recipes and architectures that reduce layer redundancy, potentially limiting early-exit opportunities. We re-evaluate layer-wise early-exit in modern LLMs and analyze how intermediate representations evolve during training. We introduce a metric to quantify a model's intrinsic suitability for early-exit and propose a benchmark for researchers to explore the potential early-exit benefits on different models and workloads. Our results show a diminishing trend in early-exit effectiveness across newer model generations. We further find that dense transformers generally offer greater early-exit potential than Mixture-of-Experts and State Space Models. In addition, larger models, particularly those with more than 20 billion parameters, and base pretrained models without specialized tuning tend to exhibit higher early-exit potential.

cs.CL↗
arXiv2026

Finite-nuclear-size effect for hydrogenlike ions under high external pressure

The influence of pressure on finite-nuclear-size corrections to atomic energy levels and electron-capture decay rate is investigated in confined hydrogenlike ions. The ions are modeled inside an impenetrable spherical cavity, with a Gaussian distribution used to represent the nuclear charge distribution. For each confinement radius used to simulate external pressure, the energies and wave functions of the lowest-lying bound states are determined by numerically solving the Dirac equation via the kinetically balanced generalized pseudospectral method. In contrast to unconfined ions, both the FNS corrections and electron-capture decay rates increase markedly under pressure and exhibit parallel trends with increasing confinement. Pressure also removes level degeneracies and alters the relative magnitudes of FNS corrections across different bound states. Moreover, the nuclear charge radius is found to significantly affect the pressure-enhanced electron-capture decay rate.

physics.atom-ph↗
arXiv2026

Collective Dynamics in Circuit Quantum Acoustodynamics with a Macroscopic Resonator

Collective dynamics in engineered quantum systems offer a unique and versatile platform for exploring how many-body correlations bridge microscopic entanglement and macroscopic behavior. In this work, we report collective Dicke dynamics of acoustic modes in a macroscopic high-overtone bulk acoustic resonator (HBAR). To achieve this, we engineer a hybrid quantum acoustodynamic system comprising an HBAR strongly coupled to a superconducting transmon qubit. The HBAR device is distinctive in the sense that its narrow mode spacing, together with enhanced qubit-mode coupling strength, gives rise to efficient coupling between the transmon and clusters of near-resonant modes. By harnessing the system properties, we observe collective dynamics involving clusters composed by two or three mechanical modes, where their non-resonant spectrum allows for the observation of the transition between the Dicke static regime to dynamically induced timed-Dicke one. The coherent collective behavior of the system is supported by time-domain measurements of the qubit's purity, indicating the quantum nature of the collective dynamics. Overall, our work establishes HBAR-based hybrid quantum system as a promising platform for exploring many-body collective dynamics in macroscopic mechanical systems.

quant-ph↗
arXiv2026

L2V-CoT: Cross-Modal Transfer of Chain-of-Thought Reasoning via Latent Intervention

Recently, Chain-of-Thought (CoT) reasoning has significantly enhanced the capabilities of large language models (LLMs), but Vision-Language Models (VLMs) still struggle with multi-step reasoning tasks due to limited multimodal reasoning data. To bridge this gap, researchers have explored methods to transfer CoT reasoning from LLMs to VLMs. However, existing approaches either need high training costs or require architectural alignment. In this paper, we use Linear Artificial Tomography (LAT) to empirically show that LLMs and VLMs share similar low-frequency latent representations of CoT reasoning despite architectural differences. Based on this insight, we propose L2V-CoT, a novel training-free latent intervention approach that transfers CoT reasoning from LLMs to VLMs. L2V-CoT extracts and resamples low-frequency CoT representations from LLMs in the frequency domain, enabling dimension matching and latent injection into VLMs during inference to enhance reasoning capabilities. Extensive experiments demonstrate that our approach consistently outperforms training-free baselines and even surpasses supervised methods.

cs.CL↗
arXiv2026

Experimental Demonstration of Twin-Field Quantum Digital Signatures over 504 km

Digital signatures are one of the security cornerstones of the current information age. Compared with classical digital signatures based on computational complexity, quantum digital signatures (QDS) theoretically guarantee data integrity, authenticity, and non-repudiation by quantum mechanics, showing great potential for development in cryptography and thus attracting widespread attention. However, the performance of existing QDS systems are still limited in rate and distance. Here we report the first experimental demonstration of twin-field QDS (TF-QDS) using a GHz system. We achieve a maximum transmission distance of 504 km fiber spools for both single-bit and multi-bit schemes, surpassing all existing state-of-the-art QDS experiments more than 200 km. Furthermore, by combining the one-time universal hash method, we achieve a maximum signature rate of 21.1 times per second for a 1 Mbit file over fiber distances up to 302 km. In this work, the signature rates of both single-bit scheme and multi-bit scheme are more than two orders of magnitude higher than that of previous works at similar distance. Our work provides a new record for long-distance and high-rate QDS, representing a significant step in the development of QDS.

quant-ph↗
arXiv2026

Bridging Theory and Data: Correcting Nuclear Mass Models with Interpretable Machine Learning

Nuclear mass prediction is one of the core issues in nuclear physics research, yet it faces the challenge of small-sample datasets with high complexity. This study introduces the Kolmogorov-Arnold Network (KAN) into the refinement of nuclear mass models, proposing an efficient and interpretable solution. By constructing the KAN-WS4 hybrid model, the prediction accuracy is significantly improved (the root mean square error is reduced from 0.3 MeV to 0.16 MeV). Furthermore, leveraging the intrinsic interpretability of KAN, feature importance analysis reveals that the proton number is the most critical factor influencing residuals, indicating potential systematic biases in proton-related terms within existing theoretical models. The method's generality is demonstrated across five mass models. This study shows that KAN provides a novel approach to small-sample, high-complexity scientific problems. Its interpretability facilitates the data-driven discovery of physical laws, promising broad applicability to key nuclear physics issues.

nucl-th↗
arXiv2026

Beyond Frame-wise Tracking: A Trajectory-based Paradigm for Efficient Point Cloud Tracking

LiDAR-based 3D single object tracking (3D SOT) is a critical task in robotics and autonomous systems. Existing methods typically follow frame-wise motion estimation or a sequence-based paradigm. However, the two-frame methods are efficient but lack long-term temporal context, making them vulnerable in sparse or occluded scenes, while sequence-based methods that process multiple point clouds gain robustness at a significant computational cost. To resolve this dilemma, we propose a novel trajectory-based paradigm and its instantiation, TrajTrack. TrajTrack is a lightweight framework that enhances a base two-frame tracker by implicitly learning motion continuity from historical bounding box trajectories alone-without requiring additional, costly point cloud inputs. It first generates a fast, explicit motion proposal and then uses an implicit motion modeling module to predict the future trajectory, which in turn refines and corrects the initial proposal. Extensive experiments on the large-scale NuScenes benchmark show that TrajTrack achieves new state-of-the-art performance, dramatically improving tracking precision by 3.02% over a strong baseline while running at 55 FPS. Besides, we also demonstrate the strong generalizability of TrajTrack across different base trackers. Code is available at https://github.com/FiBonaCci225/TrajTrack.

cs.CV↗
arXiv2026

Improving Search Agent with One Line of Code

Tool-based Agentic Reinforcement Learning (TARL) has emerged as a promising paradigm for training search agents to interact with external tools for a multi-turn information-seeking process autonomously. However, we identify a critical training instability that leads to catastrophic model collapse: Importance Sampling Distribution Drift(ISDD). In Group Relative Policy Optimization(GRPO), a widely adopted TARL algorithm, ISDD manifests as a precipitous decline in the importance sampling ratios, which nullifies gradient updates and triggers irreversible training failure. To address this, we propose \textbf{S}earch \textbf{A}gent \textbf{P}olicy \textbf{O}ptimization (\textbf{SAPO}), which stabilizes training via a conditional token-level KL constraint. Unlike hard clipping, which ignores distributional divergence, SAPO selectively penalizes the KL divergence between the current and old policies. Crucially, this penalty is applied only to positive tokens with low probabilities where the policy has shifted excessively, thereby preventing distribution drift while preserving gradient flow. Remarkably, SAPO requires only one-line code modification to standard GRPO, ensuring immediate deployability. Extensive experiments across seven QA benchmarks demonstrate that SAPO achieves \textbf{+10.6\% absolute improvement} (+31.5\% relative) over Search-R1, yielding consistent gains across varying model scales (1.5B, 14B) and families (Qwen, LLaMA).

cs.LG↗
arXiv2026

CloDS: Visual-Only Unsupervised Cloth Dynamics Learning in Unknown Conditions

Deep learning has demonstrated remarkable capabilities in simulating complex dynamic systems. However, existing methods require known physical properties as supervision or inputs, limiting their applicability under unknown conditions. To explore this challenge, we introduce Cloth Dynamics Grounding (CDG), a novel scenario for unsupervised learning of cloth dynamics from multi-view visual observations. We further propose Cloth Dynamics Splatting (CloDS), an unsupervised dynamic learning framework designed for CDG. CloDS adopts a three-stage pipeline that first performs video-to-geometry grounding and then trains a dynamics model on the grounded meshes. To cope with large non-linear deformations and severe self-occlusions during grounding, we introduce a dual-position opacity modulation that supports bidirectional mapping between 2D observations and 3D geometry via mesh-based Gaussian splatting in video-to-geometry grounding stage. It jointly considers the absolute and relative position of Gaussian components. Comprehensive experimental evaluations demonstrate that CloDS effectively learns cloth dynamics from visual data while maintaining strong generalization capabilities for unseen configurations. Our code is available at https://github.com/whynot-zyl/CloDS. Visualization results are available at https://github.com/whynot-zyl/CloDS_video}.%\footnote{As in this example.

cs.CV↗
arXiv2026

Vibration induced transparency and absorption with two ion ensembles in a linear trap

We study the spectra of collective low excitations of two atomic ion ensembles which are confined in a liner trap by addressing lases. When the left ensemble is driven by an external optical field, its corresponding response spectrum to the incident optical light shows a vibration-induced transparency phenomenon when the detuning of the laser addressing the ion is tuned to the first red sideband. In the case of the detuning tuned to the first blue sideband, the response spectrum shows a conversion from the absorption peak to the transparency window. Furthermore, we investigate the fluctuation spectra of the collective excitation modes of ion ensemble and show the similar phenomena.

quant-ph↗
arXiv2026

RLHFless: Serverless Computing for Efficient RLHF

Reinforcement Learning from Human Feedback (RLHF) has been widely applied to Large Language Model (LLM) post-training to align model outputs with human preferences. Recent models, such as DeepSeek-R1, have also shown RLHF's potential to improve LLM reasoning on complex tasks. In RL, inference and training co-exist, creating dynamic resource demands throughout the workflow. Compared to traditional RL, RLHF further challenges training efficiency due to expanding model sizes and resource consumption. Several RLHF frameworks aim to balance flexible abstraction and efficient execution. However, they rely on serverful infrastructures, which struggle with fine-grained resource variability. As a result, during synchronous RLHF training, idle time between or within RL components often causes overhead and resource wastage. To address these issues, we present RLHFless, the first scalable training framework for synchronous RLHF, built on serverless computing environments. RLHFless adapts to dynamic resource demands throughout the RLHF pipeline, pre-computes shared prefixes to avoid repeated computation, and uses a cost-aware actor scaling strategy that accounts for response length variation to find sweet spots with lower cost and higher speed. In addition, RLHFless assigns workloads efficiently to reduce intra-function imbalance and idle time. Experiments on both physical testbeds and a large-scale simulated cluster show that RLHFless achieves up to 1.35x speedup and 44.8% cost reduction compared to the state-of-the-art baseline.

cs.AI↗
arXiv2026

SAM4Dcap: Training-free Biomechanical Twin System from Monocular Video

Quantitative biomechanical analysis is essential for clinical diagnosis and injury prevention but is often restricted to laboratories due to the high cost of optical motion capture systems. While multi-view video approaches have lowered barriers, they remain impractical for home-based scenarios requiring monocular capture. This paper presents SAM4Dcap, an open-source, end-to-end framework for estimating biomechanical metrics from monocular video without additional training. SAM4Dcap integrates the temporally consistent 4D human mesh recovery of SAM-Body4D with the OpenSim biomechanical solver. The pipeline converts reconstructed meshes into trajectory files compatible with diverse musculoskeletal models. We introduce automated prompting strategies and a Linux-native build for processing. Preliminary evaluations on walking and drop-jump tasks indicate that SAM4Dcap has the potential to achieve knee kinematic predictions comparable to multi-view systems, although some discrepancies in hip flexion and residual jitter remain. By bridging advanced computer vision with established biomechanical simulation, SAM4Dcap provides a flexible, accessible foundation for non-laboratory motion analysis.

cs.CV↗
arXiv2026

Geo-Code: A Code Framework for Reverse Code Generation from Geometric Images Based on Two-Stage Multi-Agent Evolution

Program code serves as a bridge linking vision and logic, providing a feasible supervisory approach for enhancing the multimodal reasoning capability of large models through geometric operations such as auxiliary line construction and perspective transformation. Nevertheless, current inverse graphics methods face tremendous challenges in accurately reconstructing complex geometric details, which often results in the loss of key geometric constraints or structural distortion. To address this bottleneck, we propose Geo-coder -- the first inverse programming framework for geometric images based on a multi-agent system. Our method innovatively decouples the process into geometric modeling via pixel-wise anchoring and metric-driven code evolution: Stage 1 leverages the complementary advantages of visual operators and large models to achieve precise capture of pixel coordinates and visual attributes; Stage 2 introduces a synthesis-rendering-validation closed loop, where bidirectional visual feedback drives the self-correction of code. Extensive experiments demonstrate that Geo-coder achieves a substantial lead in both geometric reconstruction accuracy and visual consistency. Notably, by effectively preserving the core geometric semantics, the images reconstructed with our method exhibit equivalent performance to the original ones in multimodal reasoning tasks, which fully validates the robustness of the framework. Finally, to further reduce research costs, we have open-sourced the Geo-coder dataset constructed on the GeoCode framework, which contains more than 1,500 samples. On this basis, we have also open-sourced the GeocodeLM model, laying a solid data and model foundation for subsequent research in this field.

cs.AI↗
arXiv2026

SE-Search: Self-Evolving Search Agent via Memory and Dense Reward

Retrieval augmented generation (RAG) reduces hallucinations and factual errors in large language models (LLMs) by conditioning generation on retrieved external knowledge. Recent search agents further cast RAG as an autonomous, multi-turn information-seeking process. However, existing methods often accumulate irrelevant or noisy documents and rely on sparse reinforcement learning signals. We propose \textbf{S}elf-\textbf{E}volving \textbf{Search}, a Self-Evolving Search agent that improves online search behavior through three components, memory purification, atomic query training, and dense rewards. SE-Search follows a \textit{Think-Search-Memorize} strategy that retains salient evidence while filtering irrelevant content. Atomic query training promotes shorter and more diverse queries, improving evidence acquisition. Dense rewards provide fine-grained feedback that speeds training. Experiments on single-hop and multi-hop question answering benchmarks show that \texttt{SE-Search-3B} outperforms strong baselines, yielding a $10.8$ point absolute improvement and a $33.8\%$ relative gain over Search-R1.\footnote{We will make the code and model weights publicly available upon acceptance.}

cs.CL↗
arXiv2026

Return time sets and product recurrence

Let $G$ be a countable infinite discrete group. We show that a subset $F$ of $G$ contains a return time set of some piecewise syndetic recurrent point $x$ in a compact Hausdorff space $X$ with a $G$-action if and only if $F$ is a quasi-central set. As an application, we show that if a nonempty closed subsemigroup $S$ of the Stone-Čech compactification $βG$ contains the smallest ideal $K(βG)$ of $βG$ then $S$-product recurrent is equivalent to distality, which partially answers a question of Auslander and Furstenberg (Trans. Amer. Math. Soc. 343, 1994, 221--232).

math.DS↗
arXiv2026

Near-Optimal Online Deployment and Routing for Streaming LLMs

The rapid pace at which new large language models (LLMs) appear, and older ones become obsolete, forces providers to manage a streaming inventory under a strict concurrency cap and per-query cost budgets. We cast this as an online decision problem that couples stage-wise deployment (at fixed maintenance windows) with per-query routing among live models. We introduce StageRoute, a hierarchical algorithm that (i) optimistically selects up to $M_{\max}$ models for the next stage using reward upper-confidence and cost lower-confidence bounds, and (ii) routes each incoming query by solving a budget- and throughput-constrained bandit subproblem over the deployed set. We prove a regret of $\tilde{\mathcal{O}}(T^{2/3})$ with a matching lower bound, establishing near-optimality, and validate the theory empirically: StageRoute tracks a strong oracle under tight budgets across diverse workloads.

cs.LG↗
arXiv2026

RareAlert: Aligning heterogeneous large language model reasoning for early rare disease risk screening

Missed and delayed diagnosis remains a major challenge in rare disease care. At the initial clinical encounters, physicians assess rare disease risk using only limited information under high uncertainty. When high-risk patients are not recognised at this stage, targeted diagnostic testing is often not initiated, resulting in missed diagnosis. Existing primary care triage processes are structurally insufficient to reliably identify patients with rare diseases at initial clinical presentation and universal screening is needed to reduce diagnostic delay. Here we present RareAlert, an early screening system which predict patient-level rare disease risk from routinely available primary-visit information. RareAlert integrates reasoning generated by ten LLMs, calibrates and weights these signals using machine learning, and distils the aligned reasoning into a single locally deployable model. To develop and evaluate RareAlert, we curated RareBench, a real-world dataset of 158,666 cases covering 33 Orphanet disease categories and more than 7,000 rare conditions, including both rare and non-rare presentations. The results showed that rare disease identification can be reconceptualised as a universal uncertainty resolution process applied to the general patient population. On an independent test set, RareAlert, a Qwen3-4B based model trained with calibrated reasoning signals, achieved an AUC of 0.917, outperforming the best machine learning ensemble and all evaluated LLMs, including GPT-5, DeepSeek-R1, Claude-3.7-Sonnet, o3-mini, Gemini-2.5-Pro, and Qwen3-235B. These findings demonstrate the diversity in LLM medical reasoning and the effectiveness of aligning such reasoning in highly uncertain clinical tasks. By incorporating calibrated reasoning into a single model, RareAlert enables accurate, privacy-preserving, and scalable rare disease risk screening suitable for large-scale local deployment.

cs.AI↗
arXiv2026

Exotic vortex states at high magnetic fields in a quasi-two-dimensional FeSe-based superconductor

Owing to strong electronic correlations, high-temperature superconductivity always exhibits intricate intertwinement with various competing electronic orders in phase diagrams, such as spin/charge density waves (S/CDWs). In cuprate superconductors,the intertwinement of superconductivity and CDW order could strongly affect the fundamental properties of superconductivity, such as the critical temperature(Tc) and critical magnetic field(Hc). Recent high-field transport measurements indicate that when quantum fluctuations become important at low temperatures and high magnetic fields, the CDW order also reshapes the vortex states, which leads to fragile superconductivity with extremely low critical current(Jc). Here, by performing comprehensive high-field transport measurements, the H-T phase diagram of vortex states is mapped to H = 33 T in a quasi-two-dimensional FeSe-based superconductor (TBA+)xFeSe with a zero-resistivity transition temperature above 40 K. Our results indicate that (TBA+)xFeSe is an extremely type II superconductor with significant thermal fluctuations.At low temperatures, high magnetic fields cause the vortex solid state to exhibit similar current-dependent zero-resistance behavior as the fragile superconductivity in cuprate superconductors with CDW order. When the vortex solid state is melted with increasing temperature, a superconducting regime with vortex-like phase fluctuations emerges as an intermediate state, which features finite longitudinal resistance and vanishing Hall resistance. At higher temperatures, a vortex liquid state with finite Hall resistance eventually appears due to thermal fluctuations. All these observations suggest exotic vortex states beyond the classical paradigm of vortex matter.

cond-mat.supr-con↗