Developing accurate biomedical Question Answering (QA) systems in low-resource languages remains a major challenge, limiting equitable access to reliable medical knowledge. This paper introduces BanglaMedQA and BanglaMMedBench, the first large-scale Bangla biomedical Multiple Choice Question (MCQ) datasets designed to evaluate reasoning and retrieval in medical artificial intelligence (AI). The study applies and benchmarks several Retrieval-Augmented Generation (RAG) strategies, including Traditional, Zero-Shot Fallback, Agentic, Iterative Feedback, and Aggregate RAG, combining textbook-based and web retrieval with generative reasoning to improve factual accuracy. A key novelty lies in integrating a Bangla medical textbook corpus through Optical Character Recognition (OCR) and implementing an Agentic RAG pipeline that dynamically selects between retrieval and reasoning strategies. Experimental results show that the Agentic RAG achieved the highest accuracy 89.54% with openai/gpt-oss-120b, outperforming other configurations and demonstrating superior rationale quality. These findings highlight the potential of RAG-based methods to enhance the reliability and accessibility of Bangla medical QA, establishing a foundation for future research in multilingual medical artificial intelligence.

In domains such as healthcare, finance, and e-commerce, the temporal dynamics of relational data emerge from complex interactions-such as those between patients and providers, or users and products across diverse categories. To be broadly useful, models operating on these data must integrate long-range spatial and temporal dependencies across diverse types of entities, while also supporting multiple predictive tasks. However, existing graph models for relational data primarily focus on spatial structure, treating temporal information merely as a filtering constraint to exclude future events rather than a modeling signal, and are typically designed for single-task prediction. To address these gaps, we introduce a temporal subgraph sampler that enhances global context by retrieving nodes beyond the immediate neighborhood to capture temporally relevant relationships. In addition, we propose the Relational Graph Perceiver (RGP), a graph transformer architecture for relational deep learning that leverages a cross-attention-based latent bottleneck to efficiently integrate information from both structural and temporal contexts. This latent bottleneck integrates signals from different node and edge types into a common latent space, enabling the model to build global context across the entire relational system. RGP also incorporates a flexible cross-attention decoder that supports joint learning across tasks with disjoint label spaces within a single model. Experiments on RelBench, SALT, and CTU show that RGP delivers state-of-the-art performance, offering a general and scalable solution for relational deep learning with support for diverse predictive tasks.

As the role of Large Language Models (LLM)-based coding assistants in software development becomes more critical, so does the role of the bugs they generate in the overall cybersecurity landscape. While a number of LLM code security benchmarks have been proposed alongside approaches to improve the security of generated code, it remains unclear to what extent they have impacted widely used coding LLMs. Here, we show that even the latest open-weight models are vulnerable in the earliest reported vulnerability scenarios in a realistic use setting, suggesting that the safety-functionality trade-off has until now prevented effective patching of vulnerabilities. To help address this issue, we introduce a new severity metric that reflects the risk posed by an LLM-generated vulnerability, accounting for vulnerability severity, generation chance, and the formulation of the prompt that induces vulnerable code generation - Prompt Exposure (PE). To encourage the mitigation of the most serious and prevalent vulnerabilities, we use PE to define the Model Exposure (ME) score, which indicates the severity and prevalence of vulnerabilities a model generates.

When a language model generates text, the selection of individual tokens might lead it down very different reasoning paths, making uncertainty difficult to quantify. In this work, we consider whether reasoning language models represent the alternate paths that they could take during generation. To test this hypothesis, we use hidden activations to control and predict a language model's uncertainty during chain-of-thought reasoning. In our experiments, we find a clear correlation between how uncertain a model is at different tokens, and how easily the model can be steered by controlling its activations. This suggests that activation interventions are most effective when there are alternate paths available to the model -- in other words, when it has not yet committed to a particular final answer. We also find that hidden activations can predict a model's future outcome distribution, demonstrating that models implicitly represent the space of possible paths.

Abstract--Retrieval-Augmented Generation (RAG) is a critical technique for grounding Large Language Models (LLMs) in factual evidence, yet evaluating RAG systems in specialized, safety-critical domains remains a significant challenge. Existing evaluation frameworks often rely on heuristic-based metrics that fail to capture domain-specific nuances and other works utilize LLM-as-a-Judge approaches that lack validated alignment with human judgment. This paper introduces RAGalyst, an automated, human-aligned agentic framework designed for the rigorous evaluation of domain-specific RAG systems. RAGalyst features an agentic pipeline that generates high-quality, synthetic question-answering (QA) datasets from source documents, incorporating an agentic filtering step to ensure data fidelity. The framework refines two key LLM-as-a-Judge metrics--Answer Correctness and Answerability--using prompt optimization to achieve a strong correlation with human annotations. Applying this framework to evaluate various RAG components across three distinct domains (military operations, cybersecurity, and bridge engineering), we find that performance is highly context-dependent. No single embedding model, LLM, or hyperparam-eter configuration proves universally optimal. Additionally, we provide an analysis on the most common low Answer Correctness reasons in RAG. These findings highlight the necessity of a systematic evaluation framework like RAGalyst, which empowers practitioners to uncover domain-specific trade-offs and make informed design choices for building reliable and effective RAG systems. RAGalyst is available on our Github. Although modern Large Language Models (LLMs) are great synthesizers of information, they still suffer from hallucinations [1], [2], which refers to the generation of content that appears plausible but is factually incorrect or unsupported by evidence. Mitigating hallucinations is especially important in safety-critical applications (e.g., military operations, cybersecurity, and bridge engineering) where inaccurate information can lead to serious consequences and undermine trust in artificial intelligence (AI) systems [3], [4]. Retrieval-Augmented Generation (RAG) has been widely adopted to mitigate hallucinations by grounding responses in provided context [5], [6]. A key advantage of RAG is its ability to provide models with dynamic, inference-time access to private and domain-relevant documents [5], [7].

Existing tabular reasoning benchmarks mostly test models on small, uniform tables, underrepresenting the complexity of real-world data and giving an incomplete view of Large Language Models' (LLMs) reasoning abilities. Real tables are long, heterogeneous, and domain-specific, mixing structured fields with free text and requiring multi-hop reasoning across thousands of tokens. To address this gap, we introduce RUST-BENCH, a benchmark of 7966 questions from 2031 real-world tables spanning two domains: i) RB-Science (NSF grant records) and ii) RB-Sports (NBA statistics). Unlike prior work, RUST-BENCH evaluates LLMs jointly across scale, heterogeneity, domain specificity, and reasoning complexity. Experiments with open-source and proprietary models show that LLMs struggle with heterogeneous schemas and complex multi-hop inference, revealing persistent weaknesses in current architectures and prompting strategies. RUST-BENCH establishes a challenging new testbed for advancing tabular reasoning research.

Parameter-efficient training, based on low-rank optimization, has become a highly successful tool for fine-tuning large deep-learning models. However, these methods fail at low-rank pre-training tasks where maintaining the low-rank structure and the objective remains a challenging task. We propose the Quadratic Reweighted Rank Regularizer dubbed Q3R, which leads to a novel low-rank inducing training strategy inspired by the iteratively reweighted least squares (IRLS) framework. Q3R is based on a quadratic regularizer term which majorizes a smoothed log determinant serving as rank surrogate objective. Unlike other low-rank training techniques, Q3R is able to train weight matrices with prescribed, low target ranks of models that achieve comparable predictive performance as dense models, with small computational overhead, while remaining fully compatible with existing architectures. For example, we demonstrated one experiment where we are able to truncate $60\%$ and $80\%$ of the parameters of a ViT-Tiny model with $~1.3\%$ and $~4\%$ accuracy drop in CIFAR-10 performance respectively. The efficacy of Q3R is confirmed on Transformers across both image and language tasks, including for low-rank fine-tuning.

Web agents, like OpenAI's Operator and Google's Project Mariner, are powerful agentic systems pushing the boundaries of Large Language Models (LLM). They can autonomously interact with the internet at the user's behest, such as navigating websites, filling search masks, and comparing price lists. Though web agent research is thriving, induced sustainability issues remain largely unexplored. To highlight the urgency of this issue, we provide an initial exploration of the energy and $CO_2$ cost associated with web agents from both a theoretical -via estimation- and an empirical perspective -by benchmarking. Our results show how different philosophies in web agent creation can severely impact the associated expended energy, and that more energy consumed does not necessarily equate to better results. We highlight a lack of transparency regarding disclosing model parameters and processes used for some web agents as a limiting factor when estimating energy consumption. Our work contributes towards a change in thinking of how we evaluate web agents, advocating for dedicated metrics measuring energy consumption in benchmarks.

The LLM-as-a-judge paradigm enables flexible, user-defined evaluation, but its effectiveness is often limited by the scarcity of diverse, representative data for refining criteria. We present a tool that integrates synthetic data generation into the LLM-as-a-judge workflow, empowering users to create tailored and challenging test cases with configurable domains, personas, lengths, and desired outcomes, including borderline cases. The tool also supports AI-assisted inline editing of existing test cases. To enhance transparency and interpretability, it reveals the prompts and explanations behind each generation. In a user study (N=24), 83% of participants preferred the tool over manually creating or selecting test cases, as it allowed them to rapidly generate diverse synthetic data without additional workload. The generated synthetic data proved as effective as hand-crafted data for both refining evaluation criteria and aligning with human preferences. These findings highlight synthetic data as a promising alternative, particularly in contexts where efficiency and scalability are critical.

Accurate and interpretable predictions of depression severity are essential for clinical decision support, yet existing models often lack uncertainty estimates and temporal modeling. We propose PTTSD, a Probabilistic Textual Time Series Depression Detection framework that predicts PHQ-8 scores from utterance-level clinical interviews while modeling uncertainty over time. PTTSD includes sequence-to-sequence and sequence-to-one variants, both combining bidirectional LSTMs, self-attention, and residual connections with Gaussian or Student-t output heads trained via negative log-likelihood. Evaluated on E-DAIC and DAIC-WOZ, PTTSD achieves state-of-the-art performance among text-only systems (e.g., MAE = 3.85 on E-DAIC, 3.55 on DAIC) and produces well-calibrated prediction intervals. Ablations confirm the value of attention and probabilistic modeling, while comparisons with MentalBERT establish generality. A three-part calibration analysis and qualitative case studies further highlight the interpretability and clinical relevance of uncertainty-aware forecasting.