Search arXiv⌕ Search

arXiv subjects

Bo Zhang

Publications and source records attributed to Bo Zhang.

At least 145 records · Page 8Linked to original sources
arXiv2026

MTRouter: Cost-Aware Multi-Turn LLM Routing with History-Model Joint Embeddings

Multi-turn, long-horizon tasks are increasingly common for large language models (LLMs), but solving them typically requires many sequential model invocations, accumulating substantial inference costs. Here, we study cost-aware multi-turn LLM routing: selecting which model to invoke at each turn from a model pool, given a fixed cost budget. We propose MTRouter, which encodes the interaction history and candidate models into joint history-model embeddings, and learns an outcome estimator from logged trajectories to predict turn-level model utility. Experiments show that MTRouter improves the performance-cost trade-off: on ScienceWorld, it surpasses GPT-5 while reducing total cost by 58.7%; on Humanity's Last Exam (HLE), it achieves competitive accuracy while reducing total cost by 43.4% relative to GPT-5, and these gains even carry over to held-out tasks. Further analyses reveal several mechanisms underlying its effectiveness: relative to prior multi-turn routers, MTRouter makes fewer model switches, is more tolerant to transient errors, and exhibits emergent specialization across models. Code: https://github.com/ZhangYiqun018/MTRouter

cs.CL↗
arXiv2026

Design-based nested instrumental variable analysis

Two binary instrumental variables (IVs) are nested if individuals who comply under one binary IV also comply under the other. This situation often arises when the two IVs represent different intensities of encouragement or discouragement to take the treatment, with one stronger than the other. In a nested IV structure, treatment effects can be identified for two latent subgroups: always-compliers and switchers. Always-compliers are individuals who comply even under the weaker IV, while switchers are those who do not comply under the weaker IV but do under the stronger IV. We introduce a novel pair-of-pairs nested IV design, where each matched stratum consists of four units organized in two pairs. We develop design-based inference for the always-complier sample average treatment effect and switcher sample average treatment effect. In a nested IV analysis, IV assignment is randomized within each IV pair; however, whether a study unit receives the weaker or stronger IV may not be randomized. To address this complication, we then propose a novel partly biased randomization scheme and study design-based inference under this new scheme. Using extensive simulation studies, we demonstrate the validity of the proposed method even in challenging scenarios with small sample sizes and a low proportion of switchers. Applying the nested IV framework, we estimated that 52.2% (95% CI: 50.4%-53.9%) of participants enrolled at the Henry Ford Health System in the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial were always-compliers, while 26.7% (95% CI: 24.5%-28.9%) were switchers. Among always-compliers, flexible sigmoidoscopy was associated with a trend toward a decreased colorectal cancer rate. No effect was detected among switchers. This offers a richer interpretation of why no increase in the intention-to-treat effect was observed after 1997, even though the compliance rate rose.

stat.ME↗
arXiv2026

Magnon Damping as a Probe of Kondo Coupling in Magnetically Ordered Systems

In $d$-electron systems, there can also be intricate interplay between Kondo coupling and magnetic interactions as that in $f$-electron systems, but the underlying mechanism remains elusive. Here, using inelastic neutron scattering, we investigate the temperature evolution of the low-energy spin waves (magnons) in a metallic van der Waals ferromagnet Fe$_{3-x}$GeTe$_{2}$, and observe that the magnon damping diverges at both low and high temperatures and exhibits a minimum at an intermediate temperature. These behaviours are described by a formula that combines logarithmic and power-law terms, representing the dominant contributions from Kondo coupling and thermal fluctuations, respectively. These findings can be explained by considering electron-magnon scattering of spin-flip type within the ferromagnetic Kondo-Heisenberg lattice model, distinct from the original Kondo effect which only considers the coupling between itinerant electrons and isolated impurity spins. These results unveil the intriguing interplay between itinerant electrons and spin waves in metallic 3$d$-electron systems with magnetic order, and provide magnon damping as a new effective probe of Kondo coupling in metallic quantum magnets, thereby opening new avenues for exploring Kondo physics from the magnon perspective.

cond-mat.str-el↗
arXiv2026

Musical Score Understanding Benchmark: Evaluating Large Language Models' Comprehension of Complete Musical Scores

Understanding complete musical scores entails integrated reasoning over pitch, rhythm, harmony, and large-scale structure, yet the ability of Large Language Models and Vision--Language Models to interpret full musical notation remains insufficiently examined. We introduce Musical Score Understanding Benchmark (MSU-Bench), a human-curated benchmark for score-level musical understanding across textual (ABC notation) and visual (PDF) modalities. MSU-Bench contains 1,800 generative question-answer pairs from works by Bach, Beethoven, Chopin, Debussy, and others, organised into four levels of increasing difficulty, ranging from onset information to texture and form. Evaluations of more than fifteen state-of-the-art models, in both zero-shot and fine-tuned settings, reveal pronounced modality gaps, unstable level-wise performance, and challenges in maintaining multilevel correctness. Fine-tuning substantially improves results across modalities while preserving general knowledge, positioning MSU-Bench as a robust foundation for future research in multimodal reasoning. The benchmark and code are available at https://github.com/Congren-Dai/MSU-Bench.

cs.SD↗
arXiv2026

AgentDoG: A Diagnostic Guardrail Framework for AI Agent Safety and Security

The rise of AI agents introduces complex safety and security challenges arising from autonomous tool use and environmental interactions. Current guardrail models lack agentic risk awareness and transparency in risk diagnosis. To introduce an agentic guardrail that covers complex and numerous risky behaviors, we first propose a unified three-dimensional taxonomy that orthogonally categorizes agentic risks by their source (where), failure mode (how), and consequence (what). Guided by this structured and hierarchical taxonomy, we introduce a new fine-grained agentic safety benchmark (ATBench) and a Diagnostic Guardrail framework for agent safety and security (AgentDoG). AgentDoG provides fine-grained and contextual monitoring across agent trajectories. More Crucially, AgentDoG can diagnose the root causes of unsafe actions and seemingly safe but unreasonable actions, offering provenance and transparency beyond binary labels to facilitate effective agent alignment. AgentDoG variants are available in three sizes (4B, 7B, and 8B parameters) across Qwen and Llama model families. Extensive experimental results demonstrate that AgentDoG achieves state-of-the-art performance in agentic safety moderation in diverse and complex interactive scenarios. All models and datasets are openly released.

cs.AI↗
arXiv2026

Amodal SAM: A Unified Amodal Segmentation Framework with Generalization

Amodal segmentation is a challenging task that aims to predict the complete geometric shape of objects, including their occluded regions. Although existing methods primarily focus on amodal segmentation within the training domain, these approaches often lack the generalization capacity to extend effectively to novel object categories and unseen contexts. This paper introduces Amodal SAM, a unified framework that leverages SAM (Segment Anything Model) for both amodal image and amodal video segmentation. Amodal SAM preserves the powerful generalization ability of SAM while extending its inherent capabilities to the amodal segmentation task. The improvements lie in three aspects: (1) a lightweight Spatial Completion Adapter that enables occluded region reconstruction, (2) a Target-Aware Occlusion Synthesis (TAOS) pipeline that addresses the scarcity of amodal annotations by generating diverse synthetic training data, and (3) novel learning objectives that enforce regional consistency and topological regularization. Extensive experiments demonstrate that Amodal SAM achieves state-of-the-art performance on standard benchmarks, while simultaneously exhibiting robust generalization to novel scenarios. We anticipate that this research will advance the field toward practical amodal segmentation systems capable of operating effectively in unconstrained real-world environments.

cs.CV↗
arXiv2026

OmniGen2: Towards Instruction-Aligned Multimodal Generation

In this work, we introduce OmniGen2, a versatile and open-source generative model designed to provide a unified solution for diverse generation tasks, including text-to-image, image editing, and in-context generation. Unlike OmniGen v1, OmniGen2 features two distinct decoding pathways for text and image modalities, utilizing unshared parameters and a decoupled image tokenizer. This design enables OmniGen2 to build upon existing multimodal understanding models without the need to re-adapt VAE inputs, thereby preserving the original text generation capabilities. To facilitate the training of OmniGen2, we developed comprehensive data construction pipelines, encompassing image editing and in-context generation data. Additionally, we introduce a reflection mechanism tailored for image generation tasks and curate a dedicated reflection dataset based on OmniGen2. Despite its relatively modest parameter size, OmniGen2 achieves competitive results on multiple task benchmarks, including text-to-image and image editing. To further evaluate in-context generation, also referred to as subject-driven tasks, we introduce a new benchmark named OmniContext. OmniGen2 achieves state-of-the-art performance among open-source models in terms of consistency. We will release our models, training code, datasets, and data construction pipeline to support future research in this field. Project Page: https://vectorspacelab.github.io/OmniGen2; GitHub Link: https://github.com/VectorSpaceLab/OmniGen2

cs.CV↗
arXiv2026

A Survey of Reinforcement Learning for Large Language Models under Data Scarcity: Challenges and Solutions

Reinforcement learning (RL) has emerged as a powerful post-training paradigm for enhancing the reasoning capabilities of large language models (LLMs). However, reinforcement learning for LLMs faces substantial data scarcity challenges, including the limited availability of high-quality external supervision and the constrained volume of model-generated experience. These limitations make data-efficient reinforcement learning a critical research direction. In this survey, we present the first systematic review of reinforcement learning for LLMs under data scarcity. We propose a bottom-up hierarchical framework built around three complementary perspectives: the data-centric perspective, the training-centric perspective, and the framework-centric perspective. We develop a taxonomy of existing methods, summarize representative approaches in each category, and analyze their strengths and limitations. Our taxonomy aims to provide a clear conceptual foundation for understanding the design space of data-efficient RL for LLMs and to guide researchers working in this emerging area. We hope this survey offers a comprehensive roadmap for future research and inspires new directions toward more efficient and scalable reinforcement learning post-training for LLMs.

cs.LG↗
arXiv2026

Easy Samples Are All You Need: Self-Evolving LLMs via Data-Efficient Reinforcement Learning

Previous LLMs-based RL studies typically follow either supervised learning with high annotation costs, or unsupervised paradigms using voting or entropy-based rewards. However, their performance remains far from satisfactory due to the substantial annotation cost and issues such as model collapse or reward hacking. To address these issues, we introduce a new perspective inspired by cognitive learning theory and propose a novel approach called EasyRL. The core of EasyRL is to simulate the human cognitive acquisition curve by integrating reliable knowledge transfer from easy labeled data with a progressive divide-and-conquer strategy that tackles increasingly difficult unlabeled data. Specifically, we initialize a warm-up model using supervised RL with few-shot labeled data. This is followed by a divide-and-conquer pseudo-labeling strategy on difficult unlabeled data, combining consistency-based selection for low-uncertainty cases and reflection-based resolution for medium-uncertainty cases. Finally, difficulty-progressive self-training with iterative pseudo-labeling and RL further strengthens the model's reasoning capability. EasyRL provides a unified self-evolving framework that facilitates data-efficient post-training of LLMs. Experimental results on mathematical and scientific benchmarks demonstrate that EasyRL, using only 10% of easy labeled data, consistently outperforms state-of-the-art baselines.

cs.LG↗
arXiv2026

Progress-Think: Semantic Progress Reasoning for Vision-Language Navigation

Vision-Language Navigation requires agents to act coherently over long horizons by understanding not only local visual context but also how far they have advanced within a multi-step instruction. However, recent Vision-Language-Action models focus on direct action prediction and earlier progress methods predict numeric achievements; both overlook the monotonic co-progression property of the observation and instruction sequences. Building on this insight, Progress-Think introduces semantic progress reasoning, predicting instruction-style progress from visual observations to enable more accurate navigation. To achieve this without expensive annotations, we propose a three-stage framework. In the initial stage, Self-Aligned Progress Pretraining bootstraps a reasoning module via a novel differentiable alignment between visual history and instruction prefixes. Then, Progress-Guided Policy Pretraining injects learned progress states into the navigation context, guiding the policy toward consistent actions. Finally, Progress-Policy Co-Finetuning jointly optimizes both modules with tailored progress-aware reinforcement objectives. Experiments on R2R-CE and RxR-CE show state-of-the-art success and efficiency, demonstrating that semantic progress yields a more consistent representation of navigation advancement.

cs.RO↗
arXiv2026

Unveiling the Surprising Efficacy of Navigation Understanding in End-to-End Autonomous Driving

Global navigation information and local scene understanding are two crucial components of autonomous driving systems. However, our experimental results indicate that many end-to-end autonomous driving systems tend to over-rely on local scene understanding while failing to utilize global navigation information. These systems exhibit weak correlation between their planning capabilities and navigation input, and struggle to perform navigation-following in complex scenarios. To overcome this limitation, we propose the Sequential Navigation Guidance (SNG) framework, an efficient representation of global navigation information based on real-world navigation patterns. The SNG encompasses both navigation paths for constraining long-term trajectories and turn-by-turn (TBT) information for real-time decision-making logic. We constructed the SNG-QA dataset, a visual question answering (VQA) dataset based on SNG that aligns global and local planning. Additionally, we introduce an efficient model SNG-VLA that fuses local planning with global planning. The SNG-VLA achieves state-of-the-art performance through precise navigation information modeling without requiring auxiliary loss functions from perception tasks. Project page: SNG-VLA

cs.RO↗
arXiv2026

Efficient estimation of cumulative incidence curves via data fusion with surrogates: application to integrated analysis of vaccine trial and immunobridging data

Refined vaccine regimens containing variant-matched inserts are often authorized based on historical phase 3 efficacy trials together with immunobridging studies. Phase 3 trials are essential for establishing immune biomarkers that reliably predict disease risk or vaccine efficacy against clinical endpoints. Once such immune correlates are identified, updated vaccine regimens can be approved through immunobridging designs that compare the immunogenicity of the updated regimen to that of an already-approved vaccine. We develop methods of inference for the counterfactual cumulative incidence curve using participant-level data from both a historical vaccine efficacy trial and an immunobridging study. We further extend these methods to pathogens with multiple serotypes -- such as dengue virus and influenza -- by estimating cause-specific cumulative incidence curves. We describe the identification assumptions, propose efficient and multiply robust estimators, and assess their finite-sample performance through simulation studies. We then apply the proposed methods to (1) estimating the hypothetical cumulative incidence curve for a bivalent mRNA booster and (2) testing a key assumption of no controlled direct effects, using data from the COVID-19 Variant Immunologic Landscape (COVAIL) Trial, a multistage randomized clinical study evaluating the safety and immunogenicity of a second COVID-19 booster dose.

stat.ME↗
arXiv2026

NTIRE 2026 The 3rd Restore Any Image Model (RAIM) Challenge: AI Flash Portrait (Track 3)

In this paper, we present a comprehensive overview of the NTIRE 2026 3rd Restore Any Image Model (RAIM) challenge, with a specific focus on Track 3: AI Flash Portrait. Despite significant advancements in deep learning for image restoration, existing models still encounter substantial challenges in real-world low-light portrait scenarios. Specifically, they struggle to achieve an optimal balance among noise suppression, detail preservation, and faithful illumination and color reproduction. To bridge this gap, this challenge aims to establish a novel benchmark for real-world low-light portrait restoration. We comprehensively evaluate the proposed algorithms utilizing a hybrid evaluation system that integrates objective quantitative metrics with rigorous subjective assessment protocols. For this competition, we provide a dataset containing 800 groups of real-captured low-light portrait data. Each group consists of a 1K-resolution low-light input image, a 1K ground truth (GT), and a 1K person mask. This challenge has garnered widespread attention from both academia and industry, attracting over 100 participating teams and receiving more than 3,000 valid submissions. This report details the motivation behind the challenge, the dataset construction process, the evaluation metrics, and the various phases of the competition. The released dataset and baseline code for this track are publicly available from the same \href{https://github.com/zsn1434/AI_Flash-BaseLine/tree/main}{GitHub repository}, and the official challenge webpage is hosted on \href{https://www.codabench.org/competitions/12885/}{CodaBench}.

cs.CV↗
arXiv2026

NTIRE 2026 The 3rd Restore Any Image Model (RAIM) Challenge: Multi-Exposure Image Fusion in Dynamic Scenes (Track 2)

This paper presents NTIRE 2026, the 3rd Restore Any Image Model (RAIM) challenge on multi-exposure image fusion in dynamic scenes. We introduce a benchmark that targets a practical yet difficult HDR imaging setting, where exposure bracketing must be fused under scene motion, illumination variation, and handheld camera jitter. The challenge data contains 100 training sequences with 7 exposure levels and 100 test sequences with 5 exposure levels, reflecting real-world scenarios that frequently cause misalignment and ghosting artefacts. We evaluate submissions with a leaderboard score derived from PSNR, SSIM, and LPIPS, while also considering perceptual quality, efficiency, and reproducibility during the final review. This track attracted 114 participating teams and received 987 submissions. The winning methods significantly improved the ability to remove artifacts from multi-exposure fusion and recover fine details. The dataset and the code of each team can be found at the repository: https://github.com/qulishen/RAIM-HDR.

cs.CV↗
arXiv2026

Scheduling Cause-Effect Chains without Timing Anomalies in End-to-End Latency

In real-time systems, both individual task execution and data propagation must meet strict timing constraints. Cause-effect (CE) chains are widely used to analyze such behaviors by end-to-end latency. However, timing anomalies (TAs) can distort it, where a local reduction in execution times leads to an increase in the overall end-to-end latency. As a result, precisely analyzing the upper bounds of the latency becomes challenging, and such systems typically exhibit larger upper bounds than TA-eliminated systems. Existing studies either eliminate TAs by completely sacrificing average latency to simplify analysis or, despite adopting complex safe analysis methods, do not eliminate TAs effectively, still having high latencies. To address this issue, we identify two basic causes of TAs in end-to-end latency. Based on these causes, we propose the first treatment that eliminates TAs in the latency with negligible average latency loss using Deterministic Data Flow (DDF). We further formally prove its TA-free property. Therefore, we can get a precise upper bound for latency when all jobs execute with their worst-case execution times. Experimental results show that it effectively reduces the maximum end-to-end latency, the average latency, and latency jitter compared with the state-of-the-art (SOTA) method.

eess.SY↗
arXiv2026

JoyAI-LLM Flash: Advancing Mid-Scale LLMs with Token Efficiency

We introduce JoyAI-LLM Flash, an efficient Mixture-of-Experts (MoE) language model designed to redefine the trade-off between strong performance and token efficiency in the sub-50B parameter regime. JoyAI-LLM Flash is pretrained on a massive corpus of 20 trillion tokens and further optimized through a rigorous post-training pipeline, including supervised fine-tuning (SFT), Direct Preference Optimization (DPO), and large-scale reinforcement learning (RL) across diverse environments. To improve token efficiency, JoyAI-LLM Flash strategically balances \emph{thinking} and \emph{non-thinking} cognitive modes and introduces FiberPO, a novel RL algorithm inspired by fibration theory that decomposes trust-region maintenance into global and local components, providing unified multi-scale stability control for LLM policy optimization. To enhance architectural sparsity, the model comprises 48B total parameters while activating only 2.7B parameters per forward pass, achieving a substantially higher sparsity ratio than contemporary industry leading models of comparable scale. To further improve inference throughput, we adopt a joint training-inference co-design that incorporates dense Multi-Token Prediction (MTP) and Quantization-Aware Training (QAT). We release the checkpoints for both JoyAI-LLM-48B-A3B Base and its post-trained variants on Hugging Face to support the open-source community.

cs.CL↗
arXiv2026

Dealing with positivity violations in mediation analysis via weighted controlled effects, with application to assessing immune correlates of protection in antigen-experienced participants

Causal mediation analysis has become an important and increasingly used framework for evaluating candidate immune response biomarkers in vaccine research. A controlled effects approach has been proposed to estimate controlled risk curves under a counterfactual scenario in which the entire study population is vaccinated and their post-vaccination immune responses are set to a range of fixed levels. This framework performs well when the study population is antigenically naïve, that is, individuals have not been previously exposed to the antigen, as is common in HIV-1 vaccine research and during the early phases of the COVID-19 pandemic. However, the controlled effects framework becomes more challenging to apply in antigen-experienced populations, where prior vaccination or infection has occurred, as in the case of influenza, dengue, and more recent phases of the COVID-19 pandemic. In such settings, a key identification assumption for valid causal mediation analysis, the positivity assumption, is violated: it is no longer plausible to conceive of a hypothetical intervention that sets a post-vaccination immune marker to a fixed level below an individual's baseline immune level. In this article, we introduce a weighted controlled risk approach that targets a subpopulation for whom there is a prespecified probability of attaining a post-vaccination immune marker level. We further generalize this framework to study contrasts of controlled risks for relevant subpopulations. We demonstrate the validity of the proposed estimators through simulation studies and apply the method to reanalyze post-vaccination neutralizing antibody titers against Omicron BA.4/BA.5 as an immune correlate of COVID-19 in the Coronavirus Variant Immunologic Landscape (COVAIL) trial. R code to implement the proposed method can be found on Github: https://github.com/Qijia-He/weighted_CVE.

stat.ME↗
arXiv2026

R2-Write: Reflection and Revision for Open-Ended Writing with Deep Reasoning

While deep reasoning with long chain-of-thought has dramatically improved large language models in verifiable domains like mathematics, its effectiveness for open-ended tasks such as writing remains unexplored. In this paper, we conduct a systematic investigation revealing that existing mainstream reasoning models achieve limited gains on open-ended writing tasks. Our further analysis shows that these models lack deep reflection and revision patterns in open-ended writing, resulting in substantially smaller improvements compared to mathematical reasoning tasks. To address this limitation, we introduce R2-Write: an automated framework that synthesizes high-quality thinking trajectories enriched with explicit reflection and revision patterns through iterative writer-judge interaction. To prevent redundant reflections, we design a process reward mechanism that supervises reflection quality during reinforcement learning, improving both performance and token efficiency. Extensive experiments across multiple creative writing and deep-research benchmarks demonstrate significant improvements, validating that explicitly incorporating reflection and revision patterns unlocks deep reasoning capabilities for open-ended writing tasks.

cs.CL↗