The Necessity of a Unified Framework for LLM-Based Agent Evaluation

With the advent of Large Language Models (LLMs), general-purpose agents have seen fundamental advancements. However, evaluating these agents presents unique challenges that distinguish them from static QA benchmarks. We observe that current agent benchmarks are heavily confounded by extraneous factors, including system prompts, toolset configurations, and environmental dynamics. Existing evaluations often rely on fragmented, researcher-specific frameworks where the prompt engineering for reasoning and tool usage varies significantly, making it difficult to attribute performance gains to the model itself. Additionally, the lack of standardized environmental data leads to untraceable errors and non-reproducible results. This lack of standardization introduces substantial unfairness and opacity into the field. We propose that a unified evaluation framework is essential for the rigorous advancement of agent evaluation. To this end, we introduce a proposal aimed at standardizing agent evaluation.

VVLoc: Prior-free 3-DoF Vehicle Visual Localization

Localization is a critical technology in autonomous driving, encompassing both topological localization, which identifies the most similar map keyframe to the current observation, and metric localization, which provides precise spatial coordinates. Conventional methods typically address these tasks independently, rely on single-camera setups, and often require additional 3D semantic or pose priors, while lacking mechanisms to quantify the confidence of localization results, making them less feasible for real industrial applications. In this paper, we propose VVLoc, a unified pipeline that employs a single neural network to concurrently achieve topological and metric vehicle localization using multi-camera system. VVLoc first evaluates the geo-proximity between visual observations, then estimates their relative metric poses using a matching strategy, while also providing a confidence measure. Additionally, the training process for VVLoc is highly efficient, requiring only pairs of visual data and corresponding ground-truth poses, eliminating the need for complex supplementary data. We evaluate VVLoc not only on the publicly available datasets, but also on a more challenging self-collected dataset, demonstrating its ability to deliver state-of-the-art localization accuracy across a wide range of localization tasks.

A Cautionary Tale of Self-Supervised Learning for Imaging Biomarkers: Alzheimer's Disease Case Study

Discovery of sensitive and biologically grounded biomarkers is essential for early detection and monitoring of Alzheimer's disease (AD). Structural MRI is widely available but typically relies on hand-crafted features such as cortical thickness or volume. We ask whether self-supervised learning (SSL) can uncover more powerful biomarkers from the same data. Existing SSL methods underperform FreeSurfer-derived features in disease classification, conversion prediction, and amyloid status prediction. We introduce Residual Noise Contrastive Estimation (R-NCE), a new SSL framework that integrates auxiliary FreeSurfer features while maximizing additional augmentation-invariant information. R-NCE outperforms traditional features and existing SSL methods across multiple benchmarks, including AD conversion prediction. To assess biological relevance, we derive Brain Age Gap (BAG) measures and perform genome-wide association studies. R-NCE-BAG shows high heritability and associations with MAPT and IRAG1, with enrichment in astrocytes and oligodendrocytes, indicating sensitivity to neurodegenerative and cerebrovascular processes.

SEE: Signal Embedding Energy for Quantifying Noise Interference in Large Audio Language Models

Large Audio Language Models (LALMs) have been widely applied in real-time scenarios, such as in-car assistants and online meeting comprehension. In practice, audio inputs are often corrupted by device and environmental noise, leading to performance degradation. However, existing LALM studies on noise lack quantitative analysis and rely mainly on intuition and empirical observation, thus failing to understand practical robustness. To address this issue, we introduce Signal Embedding Energy (SEE), a method for quantifying the impact of noise intensity on LALM inputs, enabling the differentiation of LALM robustness in real-world deployments. SEE introduces a perspective based on structured activation subspaces derived from the model's internal representations, which more accurately captures its perception of noise than raw audio features. Across experiments, SEE exhibits a strong correlation with LALM performance, achieving a correlation of 0.98. Surprisingly, traditional audio denoising methods are only marginally effective for LALMs, and, in some cases, even increase SEE and impair performance. This suggests a mismatch between speech-centric denoising objectives and the noise sensitivity of modern LALMs. Therefore, we propose a mitigation strategy derived from SEE to denoise LALM inputs, outperforming existing denoising methods. This paper introduces a novel metric for noise quantification in LALMs, providing guidance for robustness improvements in real-world deployments.

HearSay Benchmark: Do Audio LLMs Leak What They Hear?

While Audio Large Language Models (ALLMs) have achieved remarkable progress in understanding and generation, their potential privacy implications remain largely unexplored. This paper takes the first step to investigate whether ALLMs inadvertently leak user privacy solely through acoustic voiceprints and introduces $\textit{HearSay}$, a comprehensive benchmark constructed from over 22,000 real-world audio clips. To ensure data quality, the benchmark is meticulously curated through a rigorous pipeline involving automated profiling and human verification, guaranteeing that all privacy labels are grounded in factual records. Extensive experiments on $\textit{HearSay}$ yield three critical findings: $\textbf{Significant Privacy Leakage}$: ALLMs inherently extract private attributes from voiceprints, reaching 92.89% accuracy on gender and effectively profiling social attributes. $\textbf{Insufficient Safety Mechanisms}$: Alarmingly, existing safeguards are severely inadequate; most models fail to refuse privacy-intruding requests, exhibiting near-zero refusal rates for physiological traits. $\textbf{Reasoning Amplifies Risk}$: Chain-of-Thought (CoT) reasoning exacerbates privacy risks in capable models by uncovering deeper acoustic correlations. These findings expose critical vulnerabilities in ALLMs, underscoring the urgent need for targeted privacy alignment. The codes and dataset are available at https://github.com/JinWang79/HearSay_Benchmark

Seedance 1.5 pro: A Native Audio-Visual Joint Generation Foundation Model

Recent strides in video generation have paved the way for unified audio-visual generation. In this work, we present Seedance 1.5 pro, a foundational model engineered specifically for native, joint audio-video generation. Leveraging a dual-branch Diffusion Transformer architecture, the model integrates a cross-modal joint module with a specialized multi-stage data pipeline, achieving exceptional audio-visual synchronization and superior generation quality. To ensure practical utility, we implement meticulous post-training optimizations, including Supervised Fine-Tuning (SFT) on high-quality datasets and Reinforcement Learning from Human Feedback (RLHF) with multi-dimensional reward models. Furthermore, we introduce an acceleration framework that boosts inference speed by over 10X. Seedance 1.5 pro distinguishes itself through precise multilingual and dialect lip-syncing, dynamic cinematic camera control, and enhanced narrative coherence, positioning it as a robust engine for professional-grade content creation. Seedance 1.5 pro is now accessible on Volcano Engine at https://console.volcengine.com/ark/region:ark+cn-beijing/experience/vision?type=GenVideo.

Wireless Channel Foundation Model with Embedded Noise-Plus-Interference Suppression Structure

Wireless channel foundation model (WCFM) is a task-agnostic AI model that is pretrained on large-scale wireless channel datasets to learn a universal channel feature representation that can be used for a wide range of downstream tasks related to communications and sensing. While existing works on WCFM have demonstrated its great potentials in various tasks including beam prediction, channel prediction, localization, etc, the models are all trained using perfect (i.e., error-free and complete) channel information state (CSI) data which are generated with simulation tools. However, in practical systems where the WCFM is deployed, perfect CSI is not available. Instead, channel estimation needs to be first performed based on pilot signals over a subset of the resource elements (REs) to acquire a noisy version of the CSI (termed as degraded CSI), which significantly differs from the perfect CSI in some real-world environments with severe noise and interference. As a result, the feature representation generated by the WCFM is unable to reflect the characteristics of the true channel, yielding performance degradation in downstream tasks. To address this issue, in this paper we propose an enhanced wireless channel foundation model architecture with noise-plus-interference (NPI) suppression capability. In our approach, coarse estimates of the CSIs are first obtained. With these information, two projection matrices are computed to extract the NPI terms in the received signals, which are further processed by a NPI estimation and subtraction module. Finally, the resultant signal is passed through a CSI completion network to get a clean version of the CSI, which is used for feature extraction. Simulation results demonstrated that compared to the state-of-the-art solutions, WCFM with NPI suppression structure achieves improved performance on channel prediction task.

MoSE: Unveiling Structural Patterns in Graphs via Mixture of Subgraph Experts

While graph neural networks (GNNs) have achieved great success in learning from graph-structured data, their reliance on local, pairwise message passing restricts their ability to capture complex, high-order subgraph patterns. leading to insufficient structural expressiveness. Recent efforts have attempted to enhance structural expressiveness by integrating random walk kernels into GNNs. However, these methods are inherently designed for graph-level tasks, which limits their applicability to other downstream tasks such as node classification. Moreover, their fixed kernel configurations hinder the model's flexibility in capturing diverse subgraph structures. To address these limitations, this paper proposes a novel Mixture of Subgraph Experts (MoSE) framework for flexible and expressive subgraph-based representation learning across diverse graph tasks. Specifically, MoSE extracts informative subgraphs via anonymous walks and dynamically routes them to specialized experts based on structural semantics, enabling the model to capture diverse subgraph patterns with improved flexibility and interpretability. We further provide a theoretical analysis of MoSE's expressivity within the Subgraph Weisfeiler-Lehman (SWL) Test, proving that it is more powerful than SWL. Extensive experiments, together with visualizations of learned subgraph experts, demonstrate that MoSE not only outperforms competitive baselines but also provides interpretable insights into structural patterns learned by the model.

HyperTASR: Hypernetwork-Driven Task-Aware Scene Representations for Robust Manipulation

Effective policy learning for robotic manipulation requires scene representations that selectively capture task-relevant environmental features. Current approaches typically employ task-agnostic representation extraction, failing to emulate the dynamic perceptual adaptation observed in human cognition. We present HyperTASR, a hypernetwork-driven framework that modulates scene representations based on both task objectives and the execution phase. Our architecture dynamically generates representation transformation parameters conditioned on task specifications and progression state, enabling representations to evolve contextually throughout task execution. This approach maintains architectural compatibility with existing policy learning frameworks while fundamentally reconfiguring how visual features are processed. Unlike methods that simply concatenate or fuse task embeddings with task-agnostic representations, HyperTASR establishes computational separation between task-contextual and state-dependent processing paths, enhancing learning efficiency and representational quality. Comprehensive evaluations in both simulation and real-world environments demonstrate substantial performance improvements across different representation paradigms. Through ablation studies and attention visualization, we confirm that our approach selectively prioritizes task-relevant scene information, closely mirroring human adaptive perception during manipulation tasks. The project website is at \href{https://lisunphil.github.io/HyperTASR_projectpage/}{lisunphil.github.io/HyperTASR\_projectpage}.

Seedance 1.0: Exploring the Boundaries of Video Generation Models

Notable breakthroughs in diffusion modeling have propelled rapid improvements in video generation, yet current foundational model still face critical challenges in simultaneously balancing prompt following, motion plausibility, and visual quality. In this report, we introduce Seedance 1.0, a high-performance and inference-efficient video foundation generation model that integrates several core technical improvements: (i) multi-source data curation augmented with precision and meaningful video captioning, enabling comprehensive learning across diverse scenarios; (ii) an efficient architecture design with proposed training paradigm, which allows for natively supporting multi-shot generation and jointly learning of both text-to-video and image-to-video tasks. (iii) carefully-optimized post-training approaches leveraging fine-grained supervised fine-tuning, and video-specific RLHF with multi-dimensional reward mechanisms for comprehensive performance improvements; (iv) excellent model acceleration achieving ~10x inference speedup through multi-stage distillation strategies and system-level optimizations. Seedance 1.0 can generate a 5-second video at 1080p resolution only with 41.4 seconds (NVIDIA-L20). Compared to state-of-the-art video generation models, Seedance 1.0 stands out with high-quality and fast video generation having superior spatiotemporal fluidity with structural stability, precise instruction adherence in complex multi-subject contexts, native multi-shot narrative coherence with consistent subject representation.