Traditional augmented reality (AR) systems predominantly rely on fixed class detectors or fiducial markers, limiting their ability to interpret complex, open-vocabulary natural language queries. We present a modular AR agent system that integrates multimodal large language models (MLLMs) with grounded vision models to enable relational reasoning in space and language-conditioned spatial retrieval in physical environments. Our adaptive task agent coordinates MLLMs and coordinate-aware perception tools to address varying query complexities, ranging from simple object identification to multi-object relational reasoning, while returning meter-accurate 3D anchors. It constructs dynamic AR scene graphs encoding nine typed relations (spatial, structural-semantic, causal-functional), enabling MLLMs to understand not just what objects exist, but how they relate and interact in 3D space. Through task-adaptive region-of-interest highlighting and contextual spatial retrieval, the system guides human attention to information-dense areas while supporting human-in-the-loop refinement. The agent dynamically invokes coordinate-aware tools for complex queries-selection, measurement, comparison, and actuation-grounding language understanding in physical operations. The modular architecture supports plug-and-use vision-language models without retraining, establishing AR agents as intermediaries that augment MLLMs with real-world spatial intelligence for interactive scene understanding. We also introduce GroundedAR-Bench, an evaluation framework for language-driven real world localization and relation grounding across diverse environments.

Time series forecasting is central to data analysis and web technologies. The recent success of Large Language Models (LLMs) offers significant potential for this field, especially from the cross-modality aspect. Most methods adopt an LLM-as-Predictor paradigm, using LLM as the forecasting backbone and designing modality alignment mechanisms to enable LLM to understand time series data. However, the semantic information in the two modalities of time series and text differs significantly, making it challenging for LLM to fully understand time series data. To mitigate this challenge, our work follows an LLM-as-Enhancer paradigm to fully utilize the advantage of LLM in text understanding, where LLM is only used to encode text modality to complement time series modality. Based on this paradigm, we propose FiCoTS, an LLM-enhanced fine-to-coarse framework for multimodal time series forecasting. Specifically, the framework facilitates progressive cross-modality interaction by three levels in a fine-to-coarse scheme: First, in the token-level modality alignment module, a dynamic heterogeneous graph is constructed to filter noise and align time series patches with text tokens; Second, in the feature-level modality interaction module, a global cross-attention mechanism is introduced to enable each time series variable to connect with relevant textual contexts; Third, in the decision-level modality fusion module, we design a gated network to adaptively fuse the results of the two modalities for robust predictions. These three modules work synergistically to let the two modalities interact comprehensively across three semantic levels, enabling textual information to effectively support temporal prediction. Extensive experiments on seven real-world benchmarks demonstrate that our model achieves state-of-the-art performance. The codes will be released publicly.

Large language models (LLMs) demonstrate significant potential for educational applications. However, their unscrutinized deployment poses risks to educational standards, underscoring the need for rigorous evaluation. We introduce EduEval, a comprehensive hierarchical benchmark for evaluating LLMs in Chinese K-12 education. This benchmark makes three key contributions: (1) Cognitive Framework: We propose the EduAbility Taxonomy, which unifies Bloom's Taxonomy and Webb's Depth of Knowledge to organize tasks across six cognitive dimensions including Memorization, Understanding, Application, Reasoning, Creativity, and Ethics. (2) Authenticity: Our benchmark integrates real exam questions, classroom conversation, student essays, and expert-designed prompts to reflect genuine educational challenges; (3) Scale: EduEval comprises 24 distinct task types with over 11,000 questions spanning primary to high school levels. We evaluate 14 leading LLMs under both zero-shot and few-shot settings, revealing that while models perform well on factual tasks, they struggle with classroom dialogue classification and exhibit inconsistent results in creative content generation. Interestingly, several open source models outperform proprietary systems on complex educational reasoning. Few-shot prompting shows varying effectiveness across cognitive dimensions, suggesting that different educational objectives require tailored approaches. These findings provide targeted benchmarking metrics for developing LLMs specifically optimized for diverse Chinese educational tasks.

Existing appliance assets suffer from poor rendering, incomplete mechanisms, and misalignment with manuals, leading to simulation-reality gaps that hinder appliance manipulation development. In this work, we introduce the RealAppliance dataset, comprising 100 high-fidelity appliances with complete physical, electronic mechanisms, and program logic aligned with their manuals. Based on these assets, we propose the RealAppliance-Bench benchmark, which evaluates multimodal large language models and embodied manipulation planning models across key tasks in appliance manipulation planning: manual page retrieval, appliance part grounding, open-loop manipulation planning, and closed-loop planning adjustment. Our analysis of model performances on RealAppliance-Bench provides insights for advancing appliance manipulation research.

Background: Large Language Models show promise in psychiatry but are English-centric. Their ability to understand mood states in other languages is unclear, as different languages have their own idioms of distress. Aim: To quantify the ability of language models to faithfully represent phrases (idioms of distress) of four distinct mood states (depression, euthymia, euphoric mania, dysphoric mania) expressed in Indian languages. Methods: We collected 247 unique phrases for the four mood states across 11 Indic languages. We tested seven experimental conditions, comparing k-means clustering performance on: (a) direct embeddings of native and Romanised scripts (using multilingual and Indic-specific models) and (b) embeddings of phrases translated to English and Chinese. Performance was measured using a composite score based on Adjusted Rand Index, Normalised Mutual Information, Homogeneity and Completeness. Results: Direct embedding of Indic languages failed to cluster mood states (Composite Score = 0.002). All translation-based approaches showed significant improvement. High performance was achieved using Gemini-translated English (Composite=0.60) and human-translated English (Composite=0.61) embedded with gemini-001. Surprisingly, human-translated English, further translated into Chinese and embedded with a Chinese model, performed best (Composite = 0.67). Specialised Indic models (IndicBERT and Sarvam-M) performed poorly. Conclusion: Current models cannot meaningfully represent mood states directly from Indic languages, posing a fundamental barrier to their psychiatric application for diagnostic or therapeutic purposes in India. While high-quality translation bridges this gap, reliance on proprietary models or complex translation pipelines is unsustainable. Models must first be built to understand diverse local languages to be effective in global mental health.

Purpose: To benchmark open-source or commercial medical image-specific VLMs against real-world radiologist-written reports. Methods: This retrospective study included adult patients who presented to the emergency department between January 2022 and April 2025 and underwent same-day CXR and CT for febrile or respiratory symptoms. Reports from five VLMs (AIRead, Lingshu, MAIRA-2, MedGemma, and MedVersa) and radiologist-written reports were randomly presented and blindly evaluated by three thoracic radiologists using four criteria: RADPEER, clinical acceptability, hallucination, and language clarity. Comparative performance was assessed using generalized linear mixed models, with radiologist-written reports treated as the reference. Finding-level analyses were also performed with CT as the reference. Results: A total of 478 patients (median age, 67 years [interquartile range, 50-78]; 282 men [59.0%]) were included. AIRead demonstrated the lowest RADPEER 3b rate (5.3% [76/1434] vs. radiologists 13.9% [200/1434]; P<.001), whereas other VLMs showed higher disagreement rates (16.8-43.0%; P<.05). Clinical acceptability was the highest with AIRead (84.5% [1212/1434] vs. radiologists 74.3% [1065/1434]; P<.001), while other VLMs performed worse (41.1-71.4%; P<.05). Hallucinations were rare with AIRead, comparable to radiologists (0.3% [4/1425]) vs. 0.1% [1/1425]; P=.21), but frequent with other models (5.4-17.4%; P<.05). Language clarity was higher with AIRead (82.9% [1189/1434]), Lingshu (88.0% [1262/1434]), and MedVersa (88.4% [1268/1434]) compared with radiologists (78.1% [1120/1434]; P<.05). Sensitivity varied substantially across VLMs for the common findings: AIRead, 15.5-86.7%; Lingshu, 2.4-86.7%; MAIRA-2, 6.0-72.0%; MedGemma, 4.8-76.7%; and MedVersa, 20.2-69.3%. Conclusion: Medical VLMs for CXR report generation exhibited variable performance in report quality and diagnostic measures.

There has been significant progress in open-source text-only translation large language models (LLMs) with better language coverage and quality. However, these models can be only used in cascaded pipelines for speech translation (ST), performing automatic speech recognition first followed by translation. This introduces additional latency, which is particularly critical in simultaneous ST (SimulST), and prevents the model from exploiting multimodal context, such as images, which can aid disambiguation. Pretrained multimodal foundation models (MMFMs) already possess strong perception and reasoning capabilities across multiple modalities, but generally lack the multilingual coverage and specialized translation performance of dedicated translation LLMs. To build an effective multimodal translation system, we propose an end-to-end approach that fuses MMFMs with translation LLMs. We introduce a novel fusion strategy that connects hidden states from multiple layers of a pretrained MMFM to a translation LLM, enabling joint end-to-end training. The resulting model, OmniFusion, built on Omni 2.5-7B as the MMFM and SeedX PPO-7B as the translation LLM, can perform speech-to-text, speech-and-image-to-text, and text-and-image-to-text translation. Experiments demonstrate that OmniFusion effectively leverages both audio and visual inputs, achieves a 1-second latency reduction in SimulST compared to cascaded pipelines and also improves the overall translation quality\footnote{Code is available at https://github.com/saikoneru/OmniFusion}.

CodeT5+: CodeT5+ was initially chosen as one of the baselines because it was among the top-performing models in our experiments on defect prediction (Monteiro et al., 2025). Although CodeT5+ does not contain an explicit [CLS] token, as in BERT-based language models, we still use the first encoded token as the head of the classification layer. Therefore, we maintain the default practice of inspecting the weights of the first token attention heads. UniXCoder: In the same way as in CodeT5+, UniXCoder was also among the top performers in defect prediction experiments (Monteiro et al., 2025), so we keep the same default strategy of using the first encoded token attention weights. We also initially considered JIT-Block (Huang et al., 2024) and JIT-CF (Ju et al., 2025). Regarding JIT-Block, its authors reconstructed the dataset (JIT-Defects4J) into the changed block format, which preserves the relative positional information between added and deleted code lines -- information lost in traditional datasets -- thus facilitating the model's ability to learn the semantic meaning of code changes. So, as the dataset was changed, it would not be possible to conduct a fair comparison. Finally, according to its published results, JIT-CF does not achieve better results than JIT-Smart. A consolidated overview of the baseline classifiers is presented in Table 2. 3.4 Description of the Experiments RQ1 How do pre-trained language models perform in comparison to traditional machine learning approaches for continual within-project and cross-project Just-in-Time Software Defect Prediction (JIT-SDP)?

Empirical grammar research has become increasingly data-driven, but the systematic analysis of annotated corpora still requires substantial methodological and technical effort. We explore how agentic large language models (LLMs) can streamline this process by reasoning over annotated corpora and producing interpretable, data-grounded answers to linguistic questions. We introduce an agentic framework for corpus-grounded grammatical analysis that integrates concepts such as natural-language task interpretation, code generation, and data-driven reasoning. As a proof of concept, we apply it to Universal Dependencies (UD) corpora, testing it on multilingual grammatical tasks inspired by the World Atlas of Language Structures (WALS). The evaluation spans 13 word-order features and over 170 languages, assessing system performance across three complementary dimensions - dominant-order accuracy, order-coverage completeness, and distributional fidelity - which reflect how well the system generalizes, identifies, and quantifies word-order variations. The results demonstrate the feasibility of combining LLM reasoning with structured linguistic data, offering a first step toward interpretable, scalable automation of corpus-based grammatical inquiry.

The success of large language models (LLMs) can be attributed in part to their ability to efficiently store factual knowledge as key-value mappings within their MLP parameters. Recent work has proposed explicit weight constructions to build such fact-storing MLPs, providing an improved understanding of LLM fact storage mechanisms. In this paper, we introduce an MLP construction framework that improves over previous constructions in three areas: it 1) works for all but a measure-zero set of feasible input-output pairs, 2) achieves asymptotically optimal parameter efficiency matching information-theoretic bounds for some embeddings, and 3) maintains usability within Transformers for factual recall. Through our improvements, we 1) discover a metric on value embeddings that characterizes facts-per-parameter scaling for both constructed and gradient-descent-trained MLPs, 2) identify a simple encoder-decoder mechanism that empirically matches gradient-descent MLP facts-per-parameter asymptotics across all the inputs and outputs we test, and 3) uncover a fundamental tradeoff between an MLP's fact-storage capacity and its usability within Transformers. Finally, we demonstrate a proof-of-concept application of fact-storing MLPs: modular fact editing on one-layer Transformers by \textit{replacing entire MLPs at once}.