arXiv subjects

Jing Gao

Publications and source records attributed to Jing Gao.

At least 37 records · Page 2Linked to original sources

arXiv2026

When the Same Coefficients Reach Different Places: Asymmetric Realizability in Transplanting Tokenizers across Large Language Models

Tokenizer transplant in cross-vocabulary model composition reconstructs donor-only embedding rows as weighted combinations over shared lexical anchors and reuses those coefficients on the base. We identify a structural geometric property of this reconstruction: the same coefficient vector reaches different sets in the donor and base anchor spans, an \emph{asymmetric realizability} gap. Across 65 donor-base pairs under OMP, with cross-operator validation on CLP, WECHSEL, and FOCUS, we construct \textit{breaker tokens}: single coefficient vectors that remain statistically inert in the donor anchor span while producing a high-salience reconstruction in the base. The same Gemma-2-2B donor checkpoint admits this construction against 13 different downstream bases drawn from five model families. The planted direction passes weight-merging with a clean reference unchanged. In a deployer case study, standard LoRA fine-tuning suppresses the breaker primarily on prompts whose distribution matches the training corpus and is not a sufficient mitigation against this attack family in our setting. The tested spectral filters miss the asymmetry. We discuss potential misuse in the open-weight composition supply chain.

cs.LG↗

arXiv2026

Determination of $B$-meson distribution amplitudes from $B\to π,K,D$ transition form factors

Recent work on $B \to π$, $K$ and $B\to D$ form factors from lattice QCD and light-cone sum rules has made it possible to constrain the inverse moment $λ_B$ of the $B$-meson light-cone distribution amplitudes by performing a global fit of $B\to π,K,D$ form factors. We have compiled the $B\to π,K,D$ form factors calculated by the HPQCD, MILC, and RBC/UKQCD collaborations in the large $q^2$ region. By employing an three-parameter ansatz of the $B$-meson light-cone distribution amplitudes, we express the $B\to π,K,D$ form factors at $q^2=0$ that are calculated from light-cone sum rules, in terms of the inverse moment $λ_B$ of the leading-twist $B$-meson light-cone distribution amplitude. In the $B \to π\ell ν$ channel, we also include the available $q^2$-binned experimental data from the BaBar, Belle, and Belle~II collaborations. Using the Bourrely-Caprini-Lellouch parametrization, we perform a global fit and obtain $λ_B=217(19)_{-17}^{+82}$~MeV and $|V_{\text{ub}}|=3.68(13)_{-1}^{+0}\times10^{-3}$. The second uncertainty is obtained by constraining $λ_B>200$ MeV and varying the inverse logarithmic moments $\hatσ_1\in[-0.7,0.7]$ and $\hatσ_2\in[-6,6]$, which represents the model-dependent uncertainty from the $B$-meson light-cone distribution amplitudes. When taking into account $λ_B$ and $\hatσ_1$ as fitting parameters simultaneously, the intervals of our preditions are $λ_B=[208, 324]$~MeV and $\hatσ_1=[-0.7, 0.27]$.

hep-ph↗

arXiv2026

Strategic Over-Parameterization for Generalizable Low-Rank Adaptation

Adapting large language models (LLMs) to downstream tasks via full fine-tuning is increasingly impractical due to its computational and memory demands. Parameter-efficient fine-tuning (PEFT) approaches such as Low-Rank Adaptation (LoRA) mitigate this by confining updates to a compact set of trainable parameters, but this aggressive reduction often sacrifices generalization, especially under transfer across heterogeneous tasks and domains. We revisit the tension between parameter efficiency and adaptation capacity, and ask whether the two are truly at odds. We answer in the negative by introducing LoRA-Over, a framework grounded in a simple principle: enrich the optimization landscape during training, then collapse the enrichment at inference. LoRA-Over injects auxiliary parameters into the low-rank adapters during training to broaden the effective hypothesis space, and through a decomposition-based reformulation folds them back into a standard low-rank structure with negligible reconstruction error, keeping inference cost identical to vanilla LoRA. Since not all weight matrices benefit equally from added capacity, we further propose two scheduling strategies, one statically predefined and one dynamically determined at runtime, that direct extra capacity where most needed. We evaluate LoRA-Over on language understanding (GLUE, T5-Base), dialogue (MT-Bench), arithmetic reasoning (GSM8K), and code generation (HumanEval), using LLaMA 2-7B and LLaMA 3.1-8B. Across all benchmarks and scales, LoRA-Over consistently outperforms vanilla LoRA, showing that principled over-parameterization designed to vanish at inference is an effective lever for improving PEFT generalization. Code will be released upon acceptance.

cs.LG↗

arXiv2026

Uncovering Latent Pathological Signatures in Pulmonary CT via Cross-Window Knowledge Distillation

Multi-window CT imaging captures complementary pathological information across anatomical structures of differing densities, yet existing deep learning methods fuse representations only at later stages, missing cross-density interactions. We propose a cross-window knowledge distillation framework in which student encoders learn latent clinical priors from a teacher trained on the most informative window. Evaluated retrospectively on three cohorts - COPD-CT-DF (n=719), RSNA PE (n=1,433), and an in-house CTEPD dataset (n=161) - distillation improved per-window AUC by 10.1-16.5 percentage points on COPD-CT-DF (0.75-0.81 to 0.90-0.94; all P<0.001), with ensemble AUC reaching 0.9960. Similar gains were observed on RSNA PE (0.80-0.83 to 0.90-0.92) and CTEPD (AUC 0.7481 vs. 0.6264). Cross-window distillation internalises pathological signatures invisible to supervised approaches, offering a generalisable solution for multi-window pulmonary CT analysis.

eess.IV↗

arXiv2026

PORTool: Importance-Aware Policy Optimization with Rewarded Tree for Multi-Tool-Integrated Reasoning

Multi-tool-integrated reasoning enables LLM-empowered tool-use agents to solve complex tasks by interleaving natural-language reasoning with calls to external tools. However, training such agents from outcome-only rewards suffers from credit-assignment ambiguity, obscuring which intermediate tool-use decisions drive success or failure. In this paper, we propose PORTool, an importance-aware policy-optimization algorithm that reinforces agents' tool-use competence from outcome-level supervision while assigning reward at the step level. Specifically, PORTool generates a rewarded rollout tree in which trajectories share prefixes before branching, enabling direct comparisons among alternative tool-use decisions within the same context. It then estimates each step's importance by a correctness-dominant signal, i.e., whether descendants of that step can ultimately produce a correct final answer, plus an auxiliary term indicating whether the step's tool calls satisfy formatting constraints and execute successfully. Using these step-wise importance estimates, PORTool updates the policy to generate efficient tool-call steps, guided by both local comparisons within each branching decision and the overall quality of entire trajectories. Experiments show that PORTool improves final-answer accuracy while reducing tool-call steps compared with state-of-the-art policy-optimization baselines, and ablation studies confirm the robustness of the proposed step-wise importance estimates.

cs.CL↗

arXiv2026

LegalDrill: Diagnosis-Driven Synthesis for Legal Reasoning in Small Language Models

Small language models (SLMs) are promising for real-world deployment due to their efficiency and low operational cost. However, their limited capacity struggles with high-stakes legal reasoning tasks that require coherent statute interpretation and logically consistent deduction. Furthermore, training SLMs for such tasks demands high-quality, concise reasoning trajectories, which are prohibitively expensive to manually collect and difficult to curate via standard rejection sampling, lacking granularity beyond final verdicts. To address these challenges, we propose {LegalDrill}, a diagnosis-driven synthesis framework that extracts and iteratively refines reasoning trajectories from a capable teacher via fine-grained prompting, then a self-reflective verification is employed to adaptively select the most effective data for the SLM student. The resulting data empower SLM training through supervised fine-tuning and direct preference optimization. Extensive experiments on several legal benchmarks demonstrate that {LegalDrill} significantly bolsters the legal reasoning capabilities of representative SLMs while bypassing the need for scarce expert annotations, paving a scalable path toward practical legal reasoning systems.

cs.CL↗

arXiv2026

Cattle-CLIP: A Multimodal Framework for Cattle Behaviour Recognition from Video

Robust behaviour recognition in real-world farm environments remains challenging due to several data-related limitations, including the scarcity of well-annotated livestock video datasets and the substantial domain gap between large-scale pre-training corpora and agricultural surveillance footage. To address these challenges, we propose Cattle-CLIP, a domain-adaptive vision-language framework that reformulates cattle behaviour recognition as cross-modal semantic alignment rather than purely visual classification. Instead of directly fine-tuning visual backbones, Cattle-CLIP incorporates a temporal integration module to extend image-level contrastive pre-training to video-based behaviour understanding, enabling consistent semantic alignment across time. To mitigate the distribution shift between web-scale image-text data used for the pre-trained model and real-world cattle surveillance footage, we further introduce tailored augmentation strategies and specialised behaviour prompts. Furthermore, we construct CattleBehaviours6, a curated and behaviour-consistent video dataset comprising 1905 annotated clips across six indoor behaviours to support model training and evaluation. Beyond serving as a benchmark for our proposed method, the dataset provides a standardised ethogram definition, offering a practical resource for future research in livestock behaviour analysis. Cattle-CLIP is evaluated under both fully-supervised and few-shot learning scenarios, with a particular focus on data-scarce behaviour recognition, an important yet under-explored goal in livestock monitoring. Experiments show that Cattle-CLIP achieves 96.1% overall accuracy across six behaviours in supervised settings, with near-perfect recall for feeding, drinking and standing-ruminating behaviours, and demonstrates robust generalisation with limited data in few-shot scenarios.

cs.CV↗

arXiv2026

Talk to Right Specialists: Iterative Routing in Multi-agent Systems for Question Answering

Retrieval-augmented generation (RAG) agents are increasingly deployed to answer questions over local knowledge bases that cannot be centralized due to knowledge-sovereignty constraints. This results in two recurring failures in production: users do not know which agent to consult, and complex questions require evidence distributed across multiple agents. To overcome these challenges, we propose RIRS, a training-free orchestration framework to enable a multi-agent system for question answering. In detail, RIRS summarizes each agent's local corpus in an embedding space, enabling a user-facing server to route queries only to the most relevant agents, reducing latency and avoiding noisy "broadcast-to-all" contexts. For complicated questions, the server can iteratively aggregate responses to derive intermediate results and refine the question to bridge the gap toward a comprehensive answer. Extensive experiments demonstrate the effectiveness of RIRS, including its ability to precisely select agents and provide accurate responses to single-hop queries, and its use of an iterative strategy to achieve accurate, multi-step resolutions for complex queries.

cs.MA↗

arXiv2026

Your VAR Model is Secretly an Efficient and Explainable Generative Classifier

Generative classifiers, which leverage conditional generative models for classification, have recently demonstrated desirable properties such as robustness to distribution shifts. However, recent progress in this area has been largely driven by diffusion-based models, whose substantial computational cost severely limits scalability. This exclusive focus on diffusion-based methods has also constrained our understanding of generative classifiers. In this work, we propose a novel generative classifier built on recent advances in visual autoregressive (VAR) modeling, which offers a new perspective for studying generative classifiers. To further enhance its performance, we introduce the Adaptive VAR Classifier$^+$ (A-VARC$^+$), which achieves a superior trade-off between accuracy and inference speed, thereby significantly improving practical applicability. Moreover, we show that the VAR-based method exhibits fundamentally different properties from diffusion-based methods. In particular, due to its tractable likelihood, the VAR-based classifier enables visual explainability via token-wise mutual information and demonstrates inherent resistance to catastrophic forgetting in class-incremental learning tasks.

cs.LG↗

arXiv2026

Experimental Powder X-ray Diffraction Crystal Structure Determination with RealPXRD-Solver

Determining crystal structures from experimental powder X-ray diffraction data remains challenging because peak overlap, preferred orientation, and impurity phases obscure atomic arrangements. We present RealPXRD-Solver, a generative model trained on 6,250,238 theoretical structures with experiment-mimicking augmentations and a universal encoder of d-spacing--intensity fingerprints, enabling both lattice-conditioned and lattice-free inference. RealPXRD-Solver reaches a 98.3% Top-20 match rate on a 10,000-structure theoretical benchmark and achieves Top-1/Top-20 accuracies of 77.9%/91.9% on CNRS and 78.8%/92.9% on RRUFF experimental datasets, and it solved 39 previously unreported Powder Diffraction File entries.

cond-mat.mtrl-sci↗

arXiv2026

Study of $CP$ violation in $Λ_b^0/Ξ^-_b\rightarrow Λ(1520)M$ decays with the final-state rescattering mechanism

Recently, the LHCb collaboration reported the first observation of $CP$ violation in baryon decays, with a significance of more than $5σ$. This strongly motivates us to investigate the $CP$ violation in more baryon decay processes. In this work, we employ the final-state rescattering mechanism with introducing two model parameters, $Λ_{charm}$ and $Λ_{charmless}$, and calculate two-body non-leptonic baryon decays $Λ^0_b \rightarrow Λ(1520)\,π^0/κ(700)/f_0(500, 980)/ρ^0/K^{*0}/ϕ$ and $Ξ^-_b \rightarrow Λ(1520)\,K^-$. Consequently, we evaluate the corresponding branching ratios, $CP$ asymmetries, and interference effects between different decay amplitudes. Our theoretical predictions for certain decay channels are in good agreement with current experimental measurements, while the remaining processes--particularly the remarkably large $CP$ violation observable revealed by the kinematic analysis are expected to be tested in future experiments.

hep-ph↗

arXiv2026

A Vehicle-portable Ultra-stable Laser for Operating on Highways

Portable ultra-stable lasers are essential for high-precision measurements. This study presents a 1550 nm vehicle-portable ultra-stable laser designed for continuous real-time operation on highways. We implement several measures to mitigate environmental impacts, including active temperature control with a standard deviation of mK/day to reduce frequency drift of the optical reference cavity, all-polarization-maintaining fiber devices to enhance the robustness of the optical path, and highly integrated electronic units to diminish thermal effects. The performance of the ultra-stable laser is evaluated through real-time beat frequency measurements with another similar ultra-stable laser over a transport distance of approximately 100 km, encompassing rural roads, national roads, urban roads, and expressways. The results indicate frequency stability of approximately 10-12/(0.01s-100 s) during transport, about 5E-14/s while the vehicle is stationary with the engine running, and around 3E-15/s with the engine off, all without active vibration isolation. This work marks the first recorded instance of a portable ultra-stable laser achieving continuous real-time operation on highways and lays a crucial foundation for non-laboratory applications, such as mobile laser communication and dynamic free-space time-frequency comparison.

physics.optics↗

arXiv2026

Study of $CP$ violation in $Λ_b^0\rightarrow N^*M$ decays with the final-state rescattering mechanism

In this work, we investigate the charmless non-leptonic two-body $Λ_b$ decays within the framework of final-state rescattering mechanism. In contrast to the Cutkosky cutting method, we compute both the absorptive and dispersive parts of the hadronic rescattering triangle diagrams. Based on the established formalism, we analyze the $Λ_b \to N^*(1535,1520)M$ decay processes with $M =K_S, K^*_0(700)$, $f_0(500,980), ρ(770), \bar{K}^{*0}$, $ϕ$, and predict various physical observables, such as their branching ratios, direct and partial-wave $CP$ asymmetries, as well as decay asymmetry parameters. These two-body decay processes are expected to contribute primarily to the subsequent four-body decay channels, such as $Λ_b^0 \to p\,π^-\,π^+\,π^-$, whose $CP$ asymmetry measurements will be accessible at the LHCb experiment.

hep-ph↗

arXiv2025

Towards Privacy-Preserving and Heterogeneity-aware Split Federated Learning via Probabilistic Masking

Split Federated Learning (SFL) has emerged as an efficient alternative to traditional Federated Learning (FL) by reducing client-side computation through model partitioning. However, exchanging of intermediate activations and model updates introduces significant privacy risks, especially from data reconstruction attacks that recover original inputs from intermediate representations. Existing defenses using noise injection often degrade model performance. To overcome these challenges, we present PM-SFL, a scalable and privacy-preserving SFL framework that incorporates Probabilistic Mask training to add structured randomness without relying on explicit noise. This mitigates data reconstruction risks while maintaining model utility. To address data heterogeneity, PM-SFL employs personalized mask learning that tailors submodel structures to each client's local data. For system heterogeneity, we introduce a layer-wise knowledge compensation mechanism, enabling clients with varying resources to participate effectively under adaptive model splitting. Theoretical analysis confirms its privacy protection, and experiments on image and wireless sensing tasks demonstrate that PM-SFL consistently improves accuracy, communication efficiency, and robustness to privacy attacks, with particularly strong performance under data and system heterogeneity.

cs.LG↗

arXiv2025

UniLabOS: An AI-Native Operating System for Autonomous Laboratories

Autonomous laboratories promise to accelerate discovery by coupling learning algorithms with robotic experimentation, yet adoption remains limited by fragmented software that separates high-level planning from low-level execution. Here we present UniLabOS, an AI-native operating system for autonomous laboratories that bridges digital decision-making and embodied experimentation through typed, stateful abstractions and transactional safeguards. UniLabOS unifies laboratory elements via an Action/Resource/Action&Resource (A/R/A&R) model, represents laboratory structure with a dual-topology of logical ownership and physical connectivity, and reconciles digital state with material motion using a transactional CRUTD protocol. Built on a distributed edge-cloud architecture with decentralized discovery, UniLabOS enables protocol mobility across reconfigurable topologies while supporting human-in-the-loop governance. We demonstrate the system in four real-world settings -- a liquid-handling workstation, a modular organic synthesis platform, a distributed electrolyte foundry, and a decentralized computation-intensive closed-loop system -- showing robust orchestration across heterogeneous instruments and multi-node coordination. UniLabOS establishes a scalable foundation for agent-ready, reproducible, and provenance-aware autonomous experimentation.

cs.CE↗

arXiv2025

Behavioral Machine Learning? Regularization and Forecast Bias

Standard forecast efficiency tests interpret violations as evidence of behavioral bias. We show theoretically and empirically that rational forecasters using optimal regularization systematically violate these tests. Machine learning forecasts show near zero bias at one year horizon, but strong overreaction at two years, consistent with predictions from a model of regularization and measurement noise. We provide three complementary tests: experimental variation in regularization parameters, cross-sectional heterogeneity in firm signal quality, and quasi-experimental evidence from ML adoption around 2013. Technically trained analysts shift sharply toward overreaction post-2013. Our findings suggest reported violations may reflect statistical sophistication rather than cognitive failure.

q-fin.ST↗

arXiv2025

Bohrium + SciMaster: Building the Infrastructure and Ecosystem for Agentic Science at Scale

AI agents are emerging as a practical way to run multi-step scientific workflows that interleave reasoning with tool use and verification, pointing to a shift from isolated AI-assisted steps toward \emph{agentic science at scale}. This shift is increasingly feasible, as scientific tools and models can be invoked through stable interfaces and verified with recorded execution traces, and increasingly necessary, as AI accelerates scientific output and stresses the peer-review and publication pipeline, raising the bar for traceability and credible evaluation. However, scaling agentic science remains difficult: workflows are hard to observe and reproduce; many tools and laboratory systems are not agent-ready; execution is hard to trace and govern; and prototype AI Scientist systems are often bespoke, limiting reuse and systematic improvement from real workflow signals. We argue that scaling agentic science requires an infrastructure-and-ecosystem approach, instantiated in Bohrium+SciMaster. Bohrium acts as a managed, traceable hub for AI4S assets -- akin to a HuggingFace of AI for Science -- that turns diverse scientific data, software, compute, and laboratory systems into agent-ready capabilities. SciMaster orchestrates these capabilities into long-horizon scientific workflows, on which scientific agents can be composed and executed. Between infrastructure and orchestration, a \emph{scientific intelligence substrate} organizes reusable models, knowledge, and components into executable building blocks for workflow reasoning and action, enabling composition, auditability, and improvement through use. We demonstrate this stack with eleven representative master agents in real workflows, achieving orders-of-magnitude reductions in end-to-end scientific cycle time and generating execution-grounded signals from real workloads at multi-million scale.

cs.AI↗

arXiv2025

ChiMDQA: Towards Comprehensive Chinese Document QA with Fine-grained Evaluation

With the rapid advancement of natural language processing (NLP) technologies, the demand for high-quality Chinese document question-answering datasets is steadily growing. To address this issue, we present the Chinese Multi-Document Question Answering Dataset(ChiMDQA), specifically designed for downstream business scenarios across prevalent domains including academic, education, finance, law, medical treatment, and news. ChiMDQA encompasses long-form documents from six distinct fields, consisting of 6,068 rigorously curated, high-quality question-answer (QA) pairs further classified into ten fine-grained categories. Through meticulous document screening and a systematic question-design methodology, the dataset guarantees both diversity and high quality, rendering it applicable to various NLP tasks such as document comprehension, knowledge extraction, and intelligent QA systems. Additionally, this paper offers a comprehensive overview of the dataset's design objectives, construction methodologies, and fine-grained evaluation system, supplying a substantial foundation for future research and practical applications in Chinese QA. The code and data are available at: https://anonymous.4open.science/r/Foxit-CHiMDQA/.

cs.CL↗