In this work, we introduce SPLICE, a human-curated benchmark derived from the COIN instructional video dataset, designed to probe event-based reasoning across multiple dimensions: temporal, causal, spatial, contextual, and general knowledge. SPLICE includes 3,381 human-filtered videos spanning 12 categories and 180 sub-categories, such as sports, engineering, and housework. These videos are segmented into a total of 11,423 event clips. We evaluate both human participants and state-of-the-art vision-language models (VLMs) on the task of rearranging these clips into coherent event sequences to assess visual reasoning capabilities. Results reveal a significant gap: VLMs struggle to match human performance. While human-annotated textual descriptions improve model accuracy, they do not affect human performance, suggesting that models rely more on language priors than on visual understanding. Even with annotations, VLMs fall short of human-level reasoning, underscoring persistent challenges in visual reasoning. A deeper analysis across sub-categories shows that VLMs perform relatively better on videos where temporal and causal reasoning are dominant, compared to those where contextual and spatial reasoning are dominant. They also perform better on everyday tasks than on specialized ones.

In science, promotional language ('hype') is increasing and can undermine objective evaluation of evidence, impede research development, and erode trust in science. In this paper, we introduce the task of automatic detection of hype, which we define as hyperbolic or subjective language that authors use to glamorize, promote, embellish, or exaggerate aspects of their research. We propose formalized guidelines for identifying hype language and apply them to annotate a portion of the National Institutes of Health (NIH) grant application corpus. We then evaluate traditional text classifiers and language models on this task, comparing their performance with a human baseline. Our experiments show that formalizing annotation guidelines can help humans reliably annotate candidate hype adjectives and that using our annotated dataset to train machine learning models yields promising results. Our findings highlight the linguistic complexity of the task, and the potential need for domain knowledge and temporal awareness of the facts. While some linguistic works address hype detection, to the best of our knowledge, we are the first to approach it as a natural language processing task.

Abstract-- Diffusion-based methods have been acknowledged as a powerful paradigm for end-to-end visuomotor control in robotics. Most existing approaches adopt a Diffusion Policy in U-Net architecture (DP-U), which, while effective, suffers from limited global context modeling and over-smoothing artifacts. T o address these issues, we propose U-DiT Policy, a novel U-shaped Diffusion Transformer framework. U-DiT preserves the multi-scale feature fusion advantages of U-Net while integrating the global context modeling capability of Transformers, thereby enhancing representational power and policy expressiveness. We evaluate U-DiT extensively across both simulation and real-world robotic manipulation tasks. In simulation, U-DiT achieves an average performance gain of 10% over baseline methods and surpasses Transformer-based diffusion policies (DP-T) that use AdaLN blocks by 6% under comparable parameter budgets. On real-world robotic tasks, U-DiT demonstrates superior generalization and robustness, achieving an average improvement of 22.5% over DP-U. Imitation learning [1] has emerged as a prominent data-driven and sample-efficient approach for programming robots from expert demonstrations. Within this paradigm, behavior cloning is typically formulated as a supervised regression task that maps observations to corresponding actions.

Recent advances in inference time scaling with extended long chain-of thought have significantly improved the reasoning capabilities of both general and medical large language models (LLMs). However, these models tend to engage in lengthy reasoning processes regardless of the difficulty of the input question, leading to increased inference costs in real-world applications. Therefore, enabling adaptive thinking where models think less for simpler questions and think more for complex ones is critical for the effective use of medical LLMs in practice. Despite its importance, there is a lack of end-to-end approaches designed to enhance the adaptive thinking capabilities of medical LLMs while providing a comprehensive examination of the trade-off between performance and computational cost. To bridge this gap, we propose AdaThink-Med, the first end-to-end framework designed to enhance adaptive thinking ability in medical reasoning models with uncertainty-guided length calibration. AdaThink-Med first generates multiple candidate outputs for each question, evaluates the correctness and uncertainty of each candidate, and then estimates problem difficulty via an uncertainty-guided length calibration module. For outputs with low difficulty and correct answers, the framework penalizes longer reasoning paths; whereas for those with high difficulty and incorrect answers, it encourages extending the chain of thought to explore alternative solutions. On six public medical QA benchmarks, AdaThink-Med achieves up to 6.4x length reduction on average while retaining performance with only minimal degradation. Intriguingly, we observe that AdaThink-Med spontaneously develops two distinct reasoning modes, which we characterize as "non-thinking" and "thinking", demonstrating the model's ability to suppress redundant reasoning processes dynamically.

Few-shot Continual Event Detection (FCED) poses the dual challenges of learning from limited data and mitigating catastrophic forgetting across sequential tasks. Existing approaches often suffer from severe forgetting due to the full fine-tuning of a shared base model, which leads to knowledge interference between tasks. Moreover, they frequently rely on data augmentation strategies that can introduce unnatural or semantically distorted inputs. To address these limitations, we propose LEAF, a novel and robust expert-based framework for FCED. LEAF integrates a specialized mixture of experts architecture into the base model, where each expert is parameterized with low-rank adaptation (LoRA) matrices. A semantic-aware expert selection mechanism dynamically routes instances to the most relevant experts, enabling expert specialization and reducing knowledge interference. To improve generalization in limited-data settings, LEAF incorporates a contrastive learning objective guided by label descriptions, which capture high-level semantic information about event types. Furthermore, to prevent overfitting on the memory buffer, our framework employs a knowledge distillation strategy that transfers knowledge from previous models to the current one. Extensive experiments on multiple FCED benchmarks demonstrate that LEAF consistently achieves state-of-the-art performance.

While LLM-based specification generation is gaining traction, existing tools primarily focus on mainstream programming languages like C, Java, and even Solidity, leaving emerging and yet verification-oriented languages like Move underexplored. In this paper, we introduce MSG, an automated specification generation tool designed for Move smart contracts. MSG aims to highlight key insights that uniquely present when applying LLM-based specification generation to a new ecosystem. Specifically, MSG demonstrates that LLMs exhibit robust code comprehension and generation capabilities even for non-mainstream languages. MSG successfully generates verifiable specifications for 84% of tested Move functions and even identifies clauses previously overlooked by experts. Additionally, MSG shows that explicitly leveraging specification language features through an agentic, modular design improves specification quality substantially (generating 57% more verifiable clauses than conventional designs). Incorporating feedback from the verification toolchain further enhances the effectiveness of MSG, leading to a 30% increase in generated verifiable specifications.

Multimodal Large Language Models (MLLMs) have significantly improved the performance of various tasks, but continue to suffer from visual hallucinations, a critical issue where generated responses contradict visual evidence. While Direct Preference Optimization (DPO) is widely used for alignment, its application to MLLMs often fails to capture fine-grained semantic differences and encourages shortcut learning. To address these challenges, we propose Semantic Curriculum Preference Optimization (SCPO), a novel framework for MLLM alignment. SCPO employs a progressive, easy-to-hard curriculum built upon our Semantic Curriculum Preference Pairs dataset, which provides fine-grained semantic contrasts sorted by difficulty. This curriculum is trained with a dynamic reference model and a novel symmetric, bidirectional objective to facilitate simultaneous learning from both textual and visual preferences. To our knowledge, SCPO is the first framework to unify semantics, symmetry, and curriculum for MLLMs alignment, effectively mitigating visual hallucinations. Extensive experiments on LLaV A models across various scales and versions validate that SCPO demonstrates superior performance compared to baseline models on multiple hallucination benchmarks, reducing the hallucination rate by up to 62.9%. Moreover, evaluations on generalized benchmarks show that SCPO improves factuality while preserving general capabilities, with its performance remaining stable across general vision-language benchmarks. Multimodal large language models (MLLMs) (Liu et al., 2024a; OpenAI, 2023; Liu et al., 2024b) have demonstrated remarkable capabilities in tasks requiring fine-grained perception, such as visual question answering and image captioning (Y u et al., 2025). Despite these advances, their reliability is compromised by the persistent challenge of visual hallucination. During inference, MLLMs often rely too heavily on the semantic priors embedded in the language model, while failing to faithfully ground these priors in fine-grained visual evidence from the input image (Li et al., 2022; Kalai et al., 2025; Xu et al., 2024b). This misalignment between language and vision produces outputs that contradict the actual visual content, thereby seriously limiting their applicability in high-stakes scenarios such as autonomous driving and medical image diagnosis (Y u et al., 2024a).

Large language models (LLMs) exhibit social biases, prompting the development of various debiasing methods. However, debiasing methods may degrade the capabilities of LLMs. Previous research has evaluated the impact of bias mitigation primarily through tasks measuring general language understanding, which are often unrelated to social biases. In contrast, cultural commonsense is closely related to social biases, as both are rooted in social norms and values. The impact of bias mitigation on cultural commonsense in LLMs has not been well investigated. Considering this gap, we propose SOBACO (SOcial BiAs and Cultural cOmmonsense benchmark), a Japanese benchmark designed to evaluate social biases and cultural commonsense in LLMs in a unified format. We evaluate several LLMs on SOBACO to examine how debiasing methods affect cultural commonsense in LLMs. Our results reveal that the debiasing methods degrade the performance of the LLMs on the cultural commonsense task (up to 75% accuracy deterioration). These results highlight the importance of developing debiasing methods that consider the trade-off with cultural commonsense to improve fairness and utility of LLMs.

Process reward models (PRMs) have demonstrated significant efficacy in enhancing the mathematical reasoning capabilities of large language models (LLMs) by leveraging test-time scaling (TTS). However, while most PRMs exhibit substantial gains in mathematical domains, the scarcity of domain-specific training data and knowledge-based learning patterns limits their generalization ability when faced with other domains. T o address this limitation, we shift the learning objective from verifying domain-specific knowledge to modeling domain-agnostic logical flow. Centering on contextual coherence between chain-of-thought (CoT) steps, our approach is realized through a novel data annotation and training framework, which enhances the model's generalization capabilities across diverse domains. F or instance, our resulting model, ContextPRM, achieves a notable 6.5% average accuracy improvement over the majority voting baseline via weighted majority voting across nine non-mathematical domains in MMLU-Pro, including law, history, and philosophy, significantly surpassing the 2.2% improvement from V ersaPRM and 0.5% gains from other mathematics-focused PRMs, demonstrating consistent performance across both mathematical and non-mathematical domains.

The performance of Video Question Answering (VideoQA) models is fundamentally constrained by the nature of their supervision, which typically consists of isolated, factual question-answer pairs. This "bag-of-facts" approach fails to capture the underlying narrative and causal structure of events, limiting models to a shallow understanding of video content. To move beyond this paradigm, we introduce a framework to synthesize richer supervisory signals. We propose two complementary strategies: Question-Based Paraphrasing (QBP), which synthesizes the diverse inquiries (what, how, why) from a video's existing set of question-answer pairs into a holistic narrative paragraph that reconstructs the video's event structure; and Question-Based Captioning (QBC), which generates fine-grained visual rationales, grounding the answer to each question in specific, relevant evidence. Leveraging powerful generative models, we use this synthetic data to train VideoQA models under a unified next-token prediction objective. Extensive experiments on STAR and NExT-QA validate our approach, demonstrating significant accuracy gains and establishing new state-of-the-art results, such as improving a 3B model to 72.5\% on STAR (+4.9\%) and a 7B model to 80.8\% on NExT-QA. Beyond accuracy, our analysis reveals that both QBP and QBC substantially enhance cross-dataset generalization, with QBP additionally accelerating model convergence by over 2.5x. These results demonstrate that shifting data synthesis from isolated facts to narrative coherence and grounded rationales yields a more accurate, efficient, and generalizable training paradigm.