arXiv:2603.03714v1 Announce Type: cross Abstract: We study a systematic bias in modern image generation models: the mention order of entities in text spuriously determines spatial layout and entity--role binding. We term this phenomenon Order-to-Space Bias (OTS) and show that it arises in both text-to-image and image-to-image generation, often overriding grounded cues and causing incorrect layouts or swapped assignments. To quantify OTS, we introduce OTS-Bench, which isolates order effects with paired prompts differing only in entity order and evaluates models along two dimensions: homogenization and correctness. Experiments show that Order-to-Space Bias (OTS) is widespread in modern image generation models, and provide evidence that it is primarily data-driven and manifests during the early stages of layout formation. Motivated by this insight, we show that both targeted fine-tuning and early-stage intervention strategies can substantially reduce OTS, while preserving generation quality.

arXiv:2509.21782v2 Announce Type: replace Abstract: Multimodal large language models (MLLMs) are increasingly deployed as the core reasoning engine for web-facing systems, powering GUI agents and front-end automation that must interpret page structure, select actionable widgets, and execute multi-step interactions reliably. However, existing benchmarks largely emphasize visual perception or UI code generation, showing insufficient evaluation on the reasoning, robustness and safety capability required for end-to-end web applications. To bridge the gap, we introduce a comprehensive web understanding benchmark, named WebRRSBench, that jointly evaluates Reasoning, Robustness, and Safety across eight tasks, such as position relationship reasoning, color robustness, and safety critical detection, etc. The benchmark is constructed from 729 websites and contains 3799 QA pairs that probe multi-step inference over page structure, text, widgets, and safety-critical interactions. To ensure reliable measurement, we adopt standardized prompts, a protocolized and deterministic evaluation pipeline, and multi-stage quality control combining automatic checks with targeted human verification. We evaluate 11 MLLMs on WebRRSBench. The results reveal significant gaps: models still struggle with compositional and cross-element reasoning over realistic layouts, show limited robustness when facing perturbations in user interfaces and content such as layout rearrangements or visual style shifts, and are rather conservative in recognizing and avoiding safety critical or irreversible actions. Our code and appendix are available at https: //github.com/annoy-worker/WebRSSBench.

arXiv:2603.02230v1 Announce Type: cross Abstract: Self-correction is an effective technique for maintaining parallel sampling in discrete diffusion models with minimal performance degradation. Prior work has explored self-correction at inference time or during post-training; however, such approaches often suffer from limited generalization and may impair reasoning performance. GIDD pioneers pretraining-based self-correction via a multi-step BERT-style uniform-absorbing objective. However, GIDD relies on a continuous interpolation-based pipeline with opaque interactions between uniform transitions and absorbing masks, which complicates hyperparameter tuning and hinders practical performance. In this work, we propose a Self-Correcting Discrete Diffusion (SCDD) model to reformulate pretrained self-correction with explicit state transitions and learn directly in discrete time. Our framework also simplifies the training noise schedule, eliminates a redundant remasking step, and relies exclusively on uniform transitions to learn self-correction. Experiments at the GPT-2 scale demonstrate that our method enables more efficient parallel decoding while preserving generation quality.

arXiv:2603.03047v1 Announce Type: cross Abstract: While Large Language Models (LLMs) demonstrate significant potential in providing accessible mental health support, their practical deployment raises critical trustworthiness concerns due to the domains high-stakes and safety-sensitive nature. Existing evaluation paradigms for general-purpose LLMs fail to capture mental health-specific requirements, highlighting an urgent need to prioritize and enhance their trustworthiness. To address this, we propose TrustMH-Bench, a holistic framework designed to systematically quantify the trustworthiness of mental health LLMs. By establishing a deep mapping from domain-specific norms to quantitative evaluation metrics, TrustMH-Bench evaluates models across eight core pillars: Reliability, Crisis Identification and Escalation, Safety, Fairness, Privacy, Robustness, Anti-sycophancy, and Ethics. We conduct extensive experiments across six general-purpose LLMs and six specialized mental health models. Experimental results indicate that the evaluated models underperform across various trustworthiness dimensions in mental health scenarios, revealing significant deficiencies. Notably, even generally powerful models (e.g., GPT-5.1) fail to maintain consistently high performance across all dimensions. Consequently, systematically improving the trustworthiness of LLMs has become a critical task. Our data and code are released.

arXiv:2412.17596v4 Announce Type: replace-cross Abstract: While Large Language Models (LLMs) demonstrate remarkable capabilities in scientific tasks such as literature analysis and experimental design (e.g., accurately extracting key findings from papers or generating coherent experimental procedures), existing evaluation benchmarks primarily assess performance using rich contextual inputs. We introduce LiveIdeaBench, a comprehensive benchmark evaluating LLMs' scientific idea generation by assessing divergent thinking capabilities using single-keyword prompts. Drawing from Guilford's creativity theory, our benchmark employs a dynamic panel of state-of-the-art LLMs to assess generated ideas across five key dimensions: originality, feasibility, fluency, flexibility, and clarity. Through extensive experimentation with over 40 leading models across 1,180 keywords spanning 22 scientific domains, we reveal that the scientific idea generation capabilities measured by our benchmark, are poorly predicted by standard metrics of general intelligence. Our results demonstrate that models like QwQ-32B-preview achieve creative performance comparable to top-tier models such as claude-3.7-sonnet:thinking, despite significant gaps in their general intelligence scores. These findings highlight the need for specialized evaluation benchmarks for scientific idea generation and suggest that enhancing these idea generation capabilities in LLMs may require different training strategies than those used for improving general problem-solving abilities, potentially enabling a wider range of AI tools tailored for different stages of the scientific process.

arXiv:2601.18650v2 Announce Type: replace-cross Abstract: Machine unlearning, which aims to efficiently remove the influence of specific data from trained models, is crucial for upholding data privacy regulations like the ``right to be forgotten". However, existing research predominantly evaluates unlearning methods on relatively balanced forget sets. This overlooks a common real-world scenario where data to be forgotten, such as a user's activity records, follows a long-tailed distribution. Our work is the first to investigate this critical research gap. We find that in such long-tailed settings, existing methods suffer from two key issues: \textit{Heterogeneous Unlearning Deviation} and \textit{Skewed Unlearning Deviation}. To address these challenges, we propose FaLW, a plug-and-play, instance-wise dynamic loss reweighting method. FaLW innovatively assesses the unlearning state of each sample by comparing its predictive probability to the distribution of unseen data from the same class. Based on this, it uses a forgetting-aware reweighting scheme, modulated by a balancing factor, to adaptively adjust the unlearning intensity for each sample. Extensive experiments demonstrate that FaLW achieves superior performance.

arXiv:2510.10689v2 Announce Type: replace Abstract: Recent advances in multimodal large language models (MLLMs) have demonstrated substantial potential in video understanding. However, existing benchmarks fail to comprehensively evaluate synergistic reasoning capabilities across audio and visual modalities, often neglecting either one of the modalities or integrating them in a logically inconsistent manner. To bridge this gap, we introduce OmniVideoBench, a large-scale and rigorously designed benchmark dedicated to assessing synergistic audio-visual understanding, with a strong emphasis on modality complementarity and logical consistency. Specifically, OmniVideoBench comprises 1000 high-quality question-answer(QA) pairs, each annotated with step-by-step reasoning traces, derived from 628 diverse videos ranging from several seconds to 30 minutes, and manually verified to guarantee complete correctness and uniqueness. Moreover, OmniVideoBench encompasses 13 carefully designed question types, covering temporal reasoning, spatial localization, counting, causal inference, summarization, and beyond, thereby capturing the essential challenges of video understanding. Evaluation of multiple MLLMs on OmniVideoBench reveals a pronounced gap between model performance and human reasoning, with open-source models lagging significantly behind their closed-source counterparts, underscoring the inherent difficulty of genuine audio-visual reasoning. We will release OmniVideoBench to foster the development of MLLMs with stronger and more generalizable reasoning capabilities.