Vision-Language Models (VLMs) continue to struggle to make morally salient judgments in multimodal and socially ambiguous contexts. Prior works typically rely on binary or pairwise supervision, which often fail to capture the continuous and pluralistic nature of human moral reasoning. We present MM-SCALE (Multimodal Moral Scale), a large-scale dataset for aligning VLMs with human moral preferences through 5-point scalar ratings and explicit modality grounding. Each image-scenario pair is annotated with moral acceptability scores and grounded reasoning labels by humans using an interface we tailored for data collection, enabling listwise preference optimization over ranked scenario sets. By moving from discrete to scalar supervision, our framework provides richer alignment signals and finer calibration of multimodal moral reasoning. Experiments show that VLMs fine-tuned on MM-SCALE achieve higher ranking fidelity and more stable safety calibration than those trained with binary signals.

Turn-taking is a central component of full-duplex interaction. Which turn-taking behaviors are appropriate varies with the scenario, yet current models apply a single norm regardless of context. This limitation originates in their training data: human-human speech corpora capture natural timing phenomena but provide little role grounding or scenario-specific norms, while heuristic or prompted synthesis methods inject turn-taking behaviors without basing them on human preferences. We introduce DuplexGen, a framework for generating dialogues with scenario-adaptive turn-taking by calibrating LLM predictions against a small set of slot-level human preference annotations. In six cooperative and competitive tasks, human turn-taking preferences differ systematically, and DuplexGen aligns substantially more closely with those preferences than uncalibrated prompting or training solely on generic human-human data; a full-duplex model trained on DuplexGen-generated data exhibits distinctive, human-preferred turn-taking behaviors. These results show that human calibration, not corpus scale or prompt design alone, is what allows turn-taking synthesis to be scenario-specific.

Large audio language models (LALMs) extend language understanding into the auditory domain, yet their ability to perform low-level listening, such as pitch and duration detection, remains underexplored. However, low-level listening is critical for real-world, out-of-distribution tasks where models must reason about unfamiliar sounds based on fine-grained acoustic cues. To address this gap, we introduce the World-of-Whale benchmark (WoW-Bench) to evaluate low-level auditory perception and cognition using marine mammal vocalizations. WoW-bench is composed of a Perception benchmark for categorizing novel sounds and a Cognition benchmark, inspired by Bloom's taxonomy, to assess the abilities to remember, understand, apply, and analyze sound events. For the Cognition benchmark, we additionally introduce distractor questions to evaluate whether models are truly solving problems through listening rather than relying on other heuristics. Experiments with state-of-the-art LALMs show performance far below human levels, indicating a need for stronger auditory grounding in LALMs.

Learning robust dexterous grasping requires real-world data that records the physical outcomes of grasp attempts. Such data is hard to obtain at scale: teleoperation yields valid physical outcomes but is slow and operator-biased, while simulation-based generation is cheap and scalable but cannot certify contact validity. A natural solution is to generate candidate grasps and verify them on real hardware, but this scales only if the entire collection loop (perception, execution, labeling, and reset) runs without human intervention. We present AutoDex, an automated real-world data-collection system that closes this loop: for each candidate from a replaceable generator, it localizes the object under severe hand-object occlusion with dense 20-camera perception, executes collision-monitored robot motions, labels lift-and-hold success or failure, and actively resets the object between trials to expose additional candidates across stable poses. The result is a reusable database of physically labeled grasp trials that downstream systems can query by retrieval and feasibility filtering. Using AutoDex, we collect 3,593 grasp trials across Allegro and Inspire hands on 100 diverse objects, with synchronized multi-view observations and robot-state logs. For a matched 500-trajectory collection, AutoDex requires 10.3 h versus 49.4 h for teleoperation, yielding a 4.8x throughput improvement, and grasps retrieved from the AutoDex-validated database succeed 76

Vision-Language Models (VLMs) exhibit strong visual reasoning capabilities, yet they still struggle with 3D understanding. In particular, VLMs often fail to infer a text-consistent goal 6D pose of a target object in a 3D scene. However, we find that with some inference-time techniques and iterative reasoning, VLMs can achieve dramatic performance gains. Concretely, given a 3D scene represented by an RGB-D image (or a compositional scene of 3D meshes) and a text instruction specifying a desired state change, we repeat the following loop: observe the current scene; evaluate whether it is faithful to the instruction; propose a pose update for the target object; apply the update; and render the updated scene. Through this closed-loop interaction, the VLM effectively acts as an agent. We further introduce three inference-time techniques that are essential to this closed-loop process: (i) multi-view reasoning with supporting view selection, (ii) object-centered coordinate system visualization, and (iii) single-axis rotation prediction. Without any additional fine-tuning or new modules, our approach surpasses prior methods at predicting the text-guided goal 6D pose of the target object. It works consistently across both closed-source and open-source VLMs. Moreover, when combining our 6D pose prediction with simple robot motion planning, it enables more successful robot manipulation than existing methods. Finally, we conduct an ablation study to demonstrate the necessity of each proposed technique.

Spatial audio is essential for enhancing the immersiveness of audio-visual experiences, yet its production typically demands complex recording systems and specialized expertise. In this work, we address a novel problem of generating first-order ambisonics, a widely used spatial audio format, directly from silent videos. To support this task, we develop comprehensive evaluation metrics that capture both standard video-to-audio generation quality and spatial coherence among multiple channels. We introduce YT-Ambigen, a dataset comprising 102K YouTube video clips paired with first-order ambisonics tailored for audio generation, and its expanded version YT-Ambigen+ containing 3x more clips with a rigorously validated high-quality test subset of 19.3K clips. Furthermore, we present Video-to-Spatial Audio Generation (ViSAGe), an end-to-end framework that generates first-order ambisonics from silent videos by leveraging CLIP features, patchwise energy maps, and neural audio codecs with rotation augmentation. To address efficiency challenges, we propose a variant coined ViSAGe-SC (Single Codebook), which replaces complex residual codebooks with an optimized single codebook approach, achieving 4x faster training and 5x faster inference while maintaining superior performance. ViSAGe-SC incorporates heterogeneous codec chaining for postprocessing and candidate reranking for inference-time refinement. Experimental results demonstrate that our approach outperforms several V2A models across spatial metrics and displays competitive performance in semantic quality, generating high-quality spatial audio from video input.

Multimodal Continual Instruction Tuning (MCIT) is crucial for adapting Multimodal Large Language Models (MLLMs) to evolving a sequence of downstream tasks. Prior methods mostly utilize Mixture of Experts or expansion merge approach, primarily focusing on catastrophic forgetting, yet they still suffer from negative interference during inference, where newly learned updates overwrite useful prior knowledge and degrade overall performance. To address this, we propose SiGMA (Sign Guided Merging and Adaptation), a simple yet effective framework that mitigates negative interference with two components: sign guided adaptive tuning during training and sign guided merging at inference. Sign guided adaptive tuning reduces collisions with past knowledge and learns the current task with minimal drift, mitigating severe forgetting. Sign guided merging further improves consolidation by selectively scaling salient parameters to preserve and amplify useful task specific knowledge. Experiments on UCIT and DCL benchmarks show that SiGMA significantly reduces negative interference and outperforms state of the art MCIT methods. Our code is available at SiGMA.

The recent introduction of 3D Gaussian Splatting (3DGS) has significantly advanced novel view synthesis. Several studies have further improved the rendering quality of 3DGS, yet they still exhibit noticeable visual discrepancies when synthesizing views at sampling rates unseen during training. Specifically, they suffer from (i) erosion-induced blurring artifacts when zooming in and (ii) dilation-induced staircase artifacts when zooming out. We speculate that these artifacts arise from the fundamental limitation of the alpha blending adopted in 3DGS methods. Instead of the conventional alpha blending that computes alpha and transmittance as scalar quantities over a pixel, we propose to replace it with our novel Gaussian Blending that treats alpha and transmittance as spatially varying distributions. Thus, transmittances can be updated considering the spatial distribution of alpha values across the pixel area, allowing nearby background splats to contribute to the final rendering. Our Gaussian Blending maintains real-time rendering speed and requires no additional memory cost, while being easily integrated as a drop-in replacement into existing 3DGS-based or other NVS frameworks. Extensive experiments demonstrate that Gaussian Blending effectively captures fine details at various sampling rates unseen during training, consistently outperforming existing novel view synthesis models across both unseen and seen sampling rates.

Recent search agents leverage multi-turn reasoning and search tools to achieve strong performance on multi-hop and long-horizon benchmarks. Yet it remains unclear whether they reliably reason across all requirements by tracking, verifying, and maintaining multiple conditions in these questions. We study this capability under multi-constraint problems, where valid answers must satisfy several constraints simultaneously. We find that illusory completion frequently occurs, wherein agents believe tasks are complete despite unresolved or violated constraints, leading to underverified answers. To diagnose this behavior, we introduce the Epistemic Ledger, an evaluation framework that tracks evidential support and agents' beliefs for each constraint throughout multi-turn reasoning. Our analysis reveals four recurring failure patterns: bare assertions, overlooked refutations, stagnation, and premature exit. Motivated by these findings, we examine whether explicit constraint-state tracking during execution mitigates these failures via LiveLedger, an inference-time tracker. This simple intervention consistently improves performance, substantially reducing underverified answers (by up to 26.5

Large reasoning models (LRMs) combined with retrieval-augmented generation (RAG) have enabled deep research agents capable of multi-step reasoning with external knowledge retrieval. However, previous methods that extend reasoning with single-query search steps struggle to scale to complex tasks demanding broad document exploration. Meanwhile, approaches that generate multiple independent queries simultaneously may limit deeper, sequential reasoning. To address these limitations, we propose HybridDeepSearcher that dynamically integrates parallel and sequential search strategies to enable effective search scaling. To support training, we introduce HDS-QA, a novel dataset that seamlessly integrates broad parallel search with sequential search reasoning, providing answer trajectories in the form of reasoning-query-retrieval loops with parallel sub-queries. Across all five benchmarks, our approach significantly outperforms the state-of-the-art, improving F1 scores by +15.9 on FanOutQA and +11.5 on a subset of BrowseComp. Further analysis reveals that HybridDeepSearcher effectively scales performance with additional test-time search resources and demonstrates robustness on questions requiring more evidence, achieving higher evidence coverage. We include the code in the supplementary materials and will release the dataset and code publicly.

We introduce PhysGaia, a novel physics-aware dataset specifically designed for Dynamic Novel View Synthesis (DyNVS), encompassing both structured objects and unstructured physical phenomena.Unlike existing datasets that primarily focus on photorealistic reconstruction, PhysGaia is created to actively support physics-aware dynamic scene modeling.Our dataset provides complex dynamic scenarios with rich interactions among multiple objects, where they realistically collide with each other and exchange forces.Furthermore, it contains a diverse range of physical materials, such as liquid, gas, textile, and rheological substances, which moves beyond the rigid bodies prevalent in existing datasets.All scenes in PhysGaia are faithfully generated to strictly adhere to physical laws, leveraging carefully selected material-specific physics solvers. To enable quantitative evaluation of physical modeling, our dataset provides essential ground-truth information, including 3D particle trajectories and physics parameters, e.g., viscosity.To facilitate research adoption, we also provide essential integration pipelines for using recent 4D Gaussian Splatting models with our dataset and report their results.By addressing the critical lack of datasets for physics-aware modeling, PhysGaia will significantly advance research in dynamic view synthesis, physics-based scene understanding, and deep learning models integrated with physical simulation-ultimately enabling more faithful reconstruction and interpretation of complex dynamic scenes.

Large language models (LLMs) have enabled web agents that follow natural language goals through multi-step browser interactions. However, agents fine-tuned on specific trajectories and domain often struggle to generalize out of domain, and offline training can be compute-inefficient due to noisy, redundant trajectories and long accessibility-tree (AXTree) states. To address both issues, we propose Weasel, a trajectory selection method for offline training of web agents. Weasel selects a fixed-budget subset of trajectory steps by optimizing an objective that balances unary importance with pairwise diversity over states, websites, and interaction patterns, solving efficiently with a greedy algorithm. We further improve efficiency with target-centered AXTree pruning that keeps only content around the ground-truth action target, and we mitigate style mismatch for reasoning-native models by replacing expert traces with model-generated, style-consistent rationales. Across AgentTrek and NNetNav training datasets, evaluations in WebArena, WorkArena, and MiniWob, and experiments with Qwen2.5-7B, Gemma3-4B, and Qwen3-8B, Weasel improves out-of-domain performance while reducing training cost, producing roughly 9.7-12.5× training speedups over standard fine-tuning. We make the code available at https://github.com/fatemehpesaran310/weasel.

We present Task 5 of the DCASE 2025 Challenge: an Audio Question Answering (AQA) benchmark spanning multiple domains of sound understanding. This task defines three QA subsets (Bioacoustics, Temporal Soundscapes, and Complex QA) to test audio-language models on interactive question-answering over diverse acoustic scenes. We describe the dataset composition (from marine mammal calls to soundscapes and complex real-world clips), the evaluation protocol (top-1 accuracy with answer-shuffling robustness), and baseline systems (Qwen2-Audio-7B, AudioFlamingo 2, Gemini-2-Flash). Preliminary results on the development set are compared, showing strong variation across models and subsets. This challenge aims to advance the audio understanding and reasoning capabilities of audio-language models toward human-level acuity, which are crucial for enabling AI agents to perceive and interact about the world effectively.

Information retrieval (IR) increasingly targets open-ended queries that admit diverse perspectives. Existing IR benchmarks, however, focus primarily on closed-ended queries, while even open-ended benchmarks largely consist of queries whose supporting documents span a single subject domain and modality. We introduce Multi^3IR, a benchmark that evaluates how well retrievers cover the multifaceted perspectives of open-ended queries across diverse domains and modalities. It comprises 104.9K Stack Exchange queries, each annotated with perspective descriptions that capture the query's implicit viewpoints. We further propose SPIN, a parameter- and label-efficient method that learns noise vectors to steer embeddings toward diverse yet meaningful semantic directions. Experiments show that existing multimodal retrievers suffer from single-perspective bias, while SPIN substantially improves perspective coverage on Multi^3IR and generalizes well to unseen open-ended IR benchmarks. The dataset and experimental code are available at https://github.com/seokwon99/Multi3IR.

Source attribution aims to enhance the reliability of AI-generated answers by including references for each statement, helping users validate the provided answers. However, existing work has primarily focused on text-only scenario and largely overlooked the role of multimodality. We introduce MAVIS, the first benchmark designed to evaluate multimodal source attribution systems that understand user intent behind visual questions, retrieve multimodal evidence, and generate long-form answers with citations. Our dataset comprises 157K visual QA instances, where each answer is annotated with fact-level citations referring to multimodal documents. We develop fine-grained automatic metrics along three dimensions of informativeness, groundedness, and fluency, and demonstrate their strong correlation with human judgments. Our key findings are threefold: (1) LVLMs with multimodal RAG generate more informative and fluent answers than unimodal RAG, but they exhibit weaker groundedness for image documents than for text documents, a gap amplified in multimodal settings. (2) Given the same multimodal documents, there is a trade-off between informativeness and groundedness across different prompting methods. (3) Our proposed method highlights mitigating contextual bias in interpreting image documents as a crucial direction for future research.

Strategic deception by LLM and VLM agents has emerged as a central AI alignment and safety concern. Social-deduction games (where each player holds a hidden role and communicates with others to deduce identities) serve as the canonical testbed, particularly in multi-agent settings. Existing testbeds, however, are text-only and run on a single fixed agent configuration, missing the non-verbal sensorimotor channels treated as core by deception taxonomies and leaving it ambiguous whether an observed behavior reflects the underlying model or the surrounding harness. We introduce MineAmongUs, a 3D multimodal Among Us sandbox where imposter agents must deceive crewmates through joint verbal and non-verbal action. We also propose ARIA, a configurable VLM-agent harness that exposes five cognitive-component ablation axes; and an atom- and arc-level annotation scheme grounded in deception taxonomies and operationalized at scale by an LLM-as-a-Judge reaching near-human atom-labeling agreement. Empirical results show that VLM agents pursue imposter wins through joint verbal and non-verbal deception, with non-verbal channels emerging as the more decisive winning contributors across both harness ablation and cross-VLM evaluation. Taken together, our work opens a new path for embodied VLM-agent alignment research.

Spoken dialogue systems increasingly employ large language models (LLMs) to leverage their advanced reasoning capabilities. However, direct application of LLMs in spoken communication often yield suboptimal results due to mismatches between optimal textual and verbal delivery. While existing approaches adapt LLMs to produce speech-friendly outputs, their impact on reasoning performance remains underexplored. In this work, we propose Think-Verbalize-Speak, a framework that decouples reasoning from spoken delivery to preserve the full reasoning capacity of LLMs. Central to our method is verbalizing, an intermediate step that translates thoughts into natural, speech-ready text. We also introduce ReVerT, a latency-efficient verbalizer based on incremental and asynchronous summarization. Experiments across multiple benchmarks show that our method enhances speech naturalness and conciseness with minimal impact on reasoning. The project page with the dataset and the source code is available at https://yhytoto12.github.io/TVS-ReVerT

Aligning large VLMs with human preferences is a challenging task, as methods like RLHF and DPO often overfit to textual information or exacerbate hallucinations. Although augmenting negative image samples partially addresses these pitfalls, no prior work has employed listwise preference optimization for VLMs, due to the complexity and cost of constructing listwise image samples. In this work, we propose LPOI, the first object-aware listwise preference optimization developed for reducing hallucinations in VLMs. LPOI identifies and masks a critical object in the image, and then interpolates the masked region between the positive and negative images to form a sequence of incrementally more complete images. The model is trained to rank these images in ascending order of object visibility, effectively reducing hallucinations while retaining visual fidelity. LPOI requires no extra annotations beyond standard pairwise preference data, as it automatically constructs the ranked lists through object masking and interpolation. Comprehensive experiments on MMHalBench, AMBER, and Object HalBench confirm that LPOI outperforms existing preference optimization methods in reducing hallucinations and enhancing VLM performance. We make the code available at https://github.com/fatemehpesaran310/lpoi.

Conceptual combination is a cognitive process that merges basic concepts, enabling the creation of complex expressions. During this process, the properties of combination (e.g., the whiteness of a peeled apple) can be inherited from basic concepts, newly emerge, or be canceled. However, previous studies have evaluated a limited set of properties and have not examined the generative process. To address this gap, we introduce the Conceptual Combination with Property Type dataset (CCPT), which consists of 12.3K annotated triplets of noun phrases, properties, and property types. Using CCPT, we establish three types of tasks to evaluate LLMs for conceptual combination thoroughly. Our key findings are threefold: (1) Our automatic metric grading property emergence and cancellation closely corresponds with human judgments. (2) LLMs, including OpenAI's o1, struggle to generate noun phrases which possess given emergent properties. (3) Our proposed method, inspired by cognitive psychology model that explains how relationships between concepts are formed, improves performances in all generative tasks. The dataset and experimental code are available at https://github.com/seokwon99/CCPT.git.