Large language models (LLMs) excel at semantic understanding, yet their ability to reconstruct internal structure from scrambled inputs remains underexplored. Sentence-level restoration is difficult to evaluate automatically because scrambled sentences often admit multiple valid reorderings. We introduce OrderProbe, a deterministic benchmark for structural reconstruction using fixed four-character expressions in Chinese, Japanese, and Korean, which have a unique canonical order and thus support exact-match scoring. We further propose a diagnostic framework that evaluates models beyond recovery accuracy, including Semantic Accuracy, Logical Validity, Structural Consistency, Robustness, and Information Density. Experiments on twelve widely used LLMs show that structural reconstruction remains difficult even for frontier systems: zero-shot recovery frequently falls below 35%. We also observe a consistent gap between meaning-oriented generation and exact structural reconstruction, suggesting that structural robustness is not an automatic byproduct of semantic competence.

The deployment of Large Language Models (LLMs) in interactive systems necessitates a deep alignment with the nuanced and dynamic preferences of individual users. Current alignment techniques predominantly address universal human values or static, single-turn preferences, thereby failing to address the critical needs of long-term personalization and the initial user cold-start problem. To bridge this gap, we propose PersonalAgent, a novel user-centric lifelong agent designed to continuously infer and adapt to user preferences. PersonalAgent constructs and dynamically refines a unified user profile by decomposing dialogues into single-turn interactions, framing preference inference as a sequential decision-making task. Experiments show that PersonalAgent achieves superior performance over strong prompt-based and policy optimization baselines, not only in idealized but also in noisy conversational contexts, while preserving cross-session preference consistency. Furthermore, human evaluation confirms that PersonalAgent excels at capturing user preferences naturally and coherently. Our findings underscore the importance of lifelong personalization for developing more inclusive and adaptive conversational agents. Our code and ALOE-Unseen dataset are released here.

Practical deployment of medical vision-language models requires a single model that interprets heterogeneous clinical visual inputs: 2D images, 3D volumes, and medical videos. Yet how to train such unified models remains underexplored, as these modalities differ in spatial-temporal structure, token length, and data availability. We present OmniV-Med, a scalable training recipe built on four empirical findings. First, broad modality coverage requires balancing modality diversity with task-level supervision, motivating OmniV-Med-Instruct, a 252K corpus spanning 14 medical modalities and 11 clinical tasks. Second, we identify a modality-ordering effect: introducing 3D and video too early destabilizes optimization and weakens 2D grounding, motivating a progressive curriculum that first builds 2D alignment before extending to spatial-temporal inputs. Third, we observe an emergent spatiotemporal transfer effect: once strong 2D patch representations are in place, lightweight rotary position adaptation extrapolates them to 3D slices and video frames, letting a single 2D ViT replace modality-specific encoder stacks. Fourth, the imbalance in sequence length across modalities and the redundancy of long volumes and videos hinder scaling; our Redundancy-Guided Token Pruning (RGTP) exploits inter-slice and inter-frame redundancy to cut visual tokens by 60% while preserving performance. Across nine 2D, 3D, and video medical benchmarks, OmniV-Med achieves robust performance. Extensive component-wise ablations systematically refine the design space, yielding a simple yet efficient and scalable recipe for training high-performance unified Med-VLMs. Code is released at https://github.com/ZJUI-AI4H/Hulu-Med.

Retrieval-Augmented Generation (RAG) has advanced open-domain question answering by incorporating external information into model reasoning. However, raw retrieved content often suffers from low signal-to-noise ratio, where answer-supportive information is diluted by irrelevant material, and error accumulation, where incomplete or misleading content degrades multi-hop reasoning. To address these challenges, we propose EviNoteRAG, a framework that enhances RAG through quality-augmented evidence notes. The model is trained to generate Supportive-Evidence Notes (SENs), concise abstractions that preserve only answer-critical information and explicitly highlight key and uncertain content to improve reasoning accuracy. To further ensure note quality, we introduce an entailment-based Evidence Quality Reward (EQR), which incentivizes SENs to be logically sufficient for deriving the correct answer. Experiments on both in-domain and out-of-domain benchmarks demonstrate that EviNoteRAG achieves state-of-the-art performance, with relative F1 gains of 20% on HotpotQA (+0.093), 40% on Bamboogle (+0.151), and 91% on 2Wiki (+0.256), while also improving training stability and efficiency.

Criminal judgment prediction requires models to infer statutory articles, charges, and sentencing outcomes from case facts. Unlike standard classification tasks, it involves a structured reasoning process in which statutes should be matched with facts, charges should be justified by statutes, and sentencing outcomes should remain consistent with charges. Existing approaches optimize final labels, and while some have attempted to evaluate reasoning quality, their evaluations are indirect, often relying on LLM-generated rubrics that reflect model-internal preferences rather than the inherent logical structure of legal adjudication. We propose Juris Policy Optimization (JPO), a post-training framework for structured legal reasoning in Chinese criminal judgment prediction. JPO first uses teacher-generated rationales to supervise a standardized four-step reasoning process, and then applies reinforcement learning with a composite reward over legal prediction quality, reasoning structure completeness, and cross-step consistency. JPO further introduces token-level advantage reweighting and adaptive clipping for legally salient reasoning segments. Experiments on multiple open-source language models and three Chinese legal benchmarks show that JPO consistently improves both judgment prediction and reasoning quality over supervised fine-tuning and reinforcement learning baselines.

While reinforcement learning from verifiable rewards (RLVR) has been proven highly effective for enhancing reasoning, its application to medical visual question answering (Med-VQA) is hampered by models producing reasoning inconsistent with either the visual evidence or the final answer. Our analysis reveals a critical flaw in RLVR training: it paradoxically encourages models to disregard visual evidence and generate answers that contradict their own reasoning. This degradation is most pronounced in specialized medical modalities (e.g., Fundus, Ultrasound) where base VLMs lack robust understanding, a failure we attribute to a flawed reward mechanism exacerbated by the scarcity of diverse training data. To tackle this, we introduce Med-Zero-17K, a large-scale dataset spanning over 30 modalities and 24 clinically relevant tasks, and the Multi-Consistency Reward (MCR) framework, which explicitly rewards both perceptual grounding and logical coherence. Extensive experiments validate our approach: integrating MCR into the RLVR framework delivers robust performance gains. This success stems from our crucial finding that rewarding internal consistency is significantly more effective than attempting to judge reasoning correctness. Furthermore, MCR proves highly versatile, exhibiting strong generalization across diverse VLM backbones, compatibility with RL algorithms like GRPO and DPO, and extending its effectiveness to 3D VQA tasks and R1-style training paradigms. Code and dataset will be released.

Reinforcement Learning with Verifiable Rewards (RLVR) has significantly advanced reasoning capabilities in Large Language Models. However, adapting RLVR to multimodal domains suffers from a critical perception-reasoning decoupling. Existing paradigms, driven by text-centric outcome rewards, reasoning in language medium, inadvertently encourage models to bypass visual perception. We empirically validate this through blind experiments: paradoxically, state-of-the-art policies maintain or even improve performance even when visual inputs are entirely removed. This reveals that these models degenerate into blind reasoners, exploiting linguistic priors to generate plausible answers instead of attending to visual evidence. In response, we propose Thinking with Deltas, a framework driven by a Differential Visual Reasoning Policy (DVRP). DVRP introduces intrinsic supervision via visual triplets, comprising original, masked, and perturbed inputs. It optimizes the model to maximize reasoning divergence from masked inputs (enforcing visual sensitivity) while minimizing divergence from perturbed inputs (ensuring visual robustness). By aligning reasoning variations strictly with the Delta of visual information, DVRP inherently bolsters visual understanding capabilities and significantly outperforms state-of-the-art methods on both general and medical benchmarks, without requiring external annotations or auxiliary tools.

Artificial intelligence has advanced in Medical Visual Question Answering (Med-VQA), but prevalent research tends to focus on the accuracy of the answers, often overlooking the reasoning paths and interpretability, which are crucial in clinical settings. Besides, current Med-VQA algorithms, typically reliant on singular models, lack the robustness needed for real-world medical diagnostics which usually require collaborative expert evaluation. To address these shortcomings, this paper presents MedCoT, a novel hierarchical expert verification reasoning chain method designed to enhance interpretability and accuracy in biomedical imaging inquiries. MedCoT is predicated on two principles: The necessity for explicit reasoning paths in Med-VQA and the requirement for multi-expert review to formulate accurate conclusions. The methodology involves an Initial Specialist proposing diagnostic rationales, followed by a Follow-up Specialist who validates these rationales, and finally, a consensus is reached through a vote among a sparse Mixture of Experts within the locally deployed Diagnostic Specialist, which then provides the definitive diagnosis. Experimental evaluations on four standard Med-VQA datasets demonstrate that MedCoT surpasses existing state-of-the-art approaches, providing significant improvements in performance and interpretability.