This paper introduces LLM-MARL, a unified framework that incorporates large language models (LLMs) into multi-agent reinforcement learning (MARL) to enhance coordination, communication, and generalization in simulated game environments. The framework features three modular components of Coordinator, Communicator, and Memory, which dynamically generate subgoals, facilitate symbolic inter-agent messaging, and support episodic recall. Training combines PPO with a language-conditioned loss and LLM query gating. LLM-MARL is evaluated in Google Research Football, MAgent Battle, and StarCraft II. Results show consistent improvements over MAPPO and QMIX in win rate, coordination score, and zero-shot generalization. Ablation studies demonstrate that subgoal generation and language-based messaging each contribute significantly to performance gains. Qualitative analysis reveals emergent behaviors such as role specialization and communication-driven tactics. By bridging language modeling and policy learning, this work contributes to the design of intelligent, cooperative agents in interactive simulations. It offers a path forward for leveraging LLMs in multi-agent systems used for training, games, and human-AI collaboration.

Trustworthiness in healthcare question-answering (QA) systems is important for ensuring patient safety, clinical effectiveness, and user confidence. As large language models (LLMs) become increasingly integrated into medical settings, the reliability of their responses directly influences clinical decision-making and patient outcomes. However, achieving comprehensive trustworthiness in medical QA poses significant challenges due to the inherent complexity of healthcare data, the critical nature of clinical scenarios, and the multifaceted dimensions of trustworthy AI. In this survey, we systematically examine six key dimensions of trustworthiness in medical QA, i.e., Factuality, Robustness, Fairness, Safety, Explainability, and Calibration. We review how each dimension is evaluated in existing LLM-based medical QA systems. We compile and compare major benchmarks designed to assess these dimensions and analyze evaluation-guided techniques that drive model improvements, such as retrieval-augmented grounding, adversarial fine-tuning, and safety alignment. Finally, we identify open challenges-such as scalable expert evaluation, integrated multi-dimensional metrics, and real-world deployment studies-and propose future research directions to advance the safe, reliable, and transparent deployment of LLM-powered medical QA.

Foundation Models (FMs) are large-scale, pre-trained artificial intelligence (AI) systems that have revolutionized natural language processing and computer vision, and are now advancing geospatial analysis and Earth Observation (EO). They promise improved generalization across tasks, scalability, and efficient adaptation with minimal labeled data. However, despite the rapid proliferation of geospatial FMs, their real-world utility and alignment with global sustainability goals remain underexplored. We introduce SustainFM, a comprehensive benchmarking framework grounded in the 17 Sustainable Development Goals with extremely diverse tasks ranging from asset wealth prediction to environmental hazard detection. This study provides a rigorous, interdisciplinary assessment of geospatial FMs and offers critical insights into their role in attaining sustainability goals. Our findings show: (1) While not universally superior, FMs often outperform traditional approaches across diverse tasks and datasets. (2) Evaluating FMs should go beyond accuracy to include transferability, generalization, and energy efficiency as key criteria for their responsible use. (3) FMs enable scalable, SDG-grounded solutions, offering broad utility for tackling complex sustainability challenges. Critically, we advocate for a paradigm shift from model-centric development to impact-driven deployment, and emphasize metrics such as energy efficiency, robustness to domain shifts, and ethical considerations.

Chain-of-thought (CoT) reasoning enhances performance of large language models, but questions remain about whether these reasoning traces faithfully reflect the internal processes of the model. We present the first comprehensive study of CoT faithfulness in large vision-language models (LVLMs), investigating how both text-based and previously unexplored image-based biases affect reasoning and bias articulation. Our work introduces a novel, fine-grained evaluation pipeline for categorizing bias articulation patterns, enabling significantly more precise analysis of CoT reasoning than previous methods. This framework reveals critical distinctions in how models process and respond to different types of biases, providing new insights into LVLM CoT faithfulness. Our findings reveal that subtle image-based biases are rarely articulated compared to explicit text-based ones, even in models specialized for reasoning. Additionally, many models exhibit a previously unidentified phenomenon we term ``inconsistent'' reasoning - correctly reasoning before abruptly changing answers, serving as a potential canary for detecting biased reasoning from unfaithful CoTs. We then apply the same evaluation pipeline to revisit CoT faithfulness in LLMs across various levels of implicit cues. Our findings reveal that current language-only reasoning models continue to struggle with articulating cues that are not overtly stated.

In-context learning, the ability of large language models to perform tasks using only examples provided in the prompt, has recently been adapted for time series forecasting. This paradigm enables zero-shot prediction, where past values serve as context for forecasting future values, making powerful forecasting tools accessible to non-experts and increasing the performance when training data are scarce. Most existing zero-shot forecasting approaches rely on transformer architectures, which, despite their success in language, often fall short of expectations in time series forecasting, where recurrent models like LSTMs frequently have the edge. Conversely, while LSTMs are well-suited for time series modeling due to their state-tracking capabilities, they lack strong in-context learning abilities. We introduce TiRex that closes this gap by leveraging xLSTM, an enhanced LSTM with competitive in-context learning skills. Unlike transformers, state-space models, or parallelizable RNNs such as RWKV, TiRex retains state-tracking, a critical property for long-horizon forecasting. To further facilitate its state-tracking ability, we propose a training-time masking strategy called CPM. TiRex sets a new state of the art in zero-shot time series forecasting on the HuggingFace benchmarks GiftEval and Chronos-ZS, outperforming significantly larger models including TabPFN-TS (Prior Labs), Chronos Bolt (Amazon), TimesFM (Google), and Moirai (Salesforce) across both short- and long-term forecasts.

The reasoning capabilities of large language models (LLMs) have advanced rapidly, particularly following the release of DeepSeek R1, which has inspired a surge of research into data quality and reinforcement learning (RL) algorithms. Despite the pivotal role diversity plays in RL, its influence on LLM reasoning remains largely underexplored. To bridge this gap, this work presents a systematic investigation into the impact of diversity in RL-based training for LLM reasoning, and proposes a novel diversity-aware policy optimization method. Across evaluations on 12 LLMs, we observe a strong positive correlation between the solution diversity and Potential at k (a novel metric quantifying an LLM's reasoning potential) in high-performing models. This finding motivates our method to explicitly promote diversity during RL training. Specifically, we design a token-level diversity and reformulate it into a practical objective, then we selectively apply it to positive samples. Integrated into the R1-zero training framework, our method achieves a 3.5 percent average improvement across four mathematical reasoning benchmarks, while generating more diverse and robust solutions.

Effective clinical decision-making depends on iterative, multimodal reasoning across diverse sources of evidence. The recent emergence of multimodal reasoning models has significantly transformed the landscape of solving complex tasks. Although such models have achieved notable success in mathematics and science, their application to medical domains remains underexplored. In this work, we propose \textit{MedE$^2$}, a two-stage post-training pipeline that elicits and then enhances multimodal reasoning for medical domains. In Stage-I, we fine-tune models using 2,000 text-only data samples containing precisely orchestrated reasoning demonstrations to elicit reasoning behaviors. In Stage-II, we further enhance the model's reasoning capabilities using 1,500 rigorously curated multimodal medical cases, aligning model reasoning outputs with our proposed multimodal medical reasoning preference. Extensive experiments demonstrate the efficacy and reliability of \textit{MedE$^2$} in improving the reasoning performance of medical multimodal models. Notably, models trained with \textit{MedE$^2$} consistently outperform baselines across multiple medical multimodal benchmarks. Additional validation on larger models and under inference-time scaling further confirms the robustness and practical utility of our approach.

Large Language Models (LLMs) can be understood as Collective Knowledge (CK): a condensation of human cultural and technical output, whose apparent intelligence emerges in dialogue. This perspective article, drawing on extended interaction with ChatGPT-4, postulates differential response modes that plausibly trace their origin to distinct model subnetworks. It argues that CK has no persistent internal state or ``spine'': it drifts, it complies, and its behaviour is shaped by the user and by fine-tuning. It develops the notion of co-augmentation, in which human judgement and CK's representational reach jointly produce forms of analysis that neither could generate alone. Finally, it suggests that CK offers a tractable object for neuroscience: unlike biological brains, these systems expose their architecture, training history, and activation dynamics, making the human--CK loop itself an experimental target.

Project-Based Learning (PBL) involves a variety of highly correlated multimodal data, making it a vital educational approach within STEM disciplines. With the rapid development of multimodal large language models (MLLMs), researchers have begun exploring their potential to enhance tasks such as information retrieval, knowledge comprehension, and data generation in educational settings. However, existing benchmarks fall short in providing both a free-form output structure and a rigorous human expert validation process, limiting their effectiveness in evaluating real-world educational tasks. Additionally, few methods have developed automated pipelines to assist with the complex responsibilities of teachers leveraging MLLMs, largely due to model hallucination and instability, which lead to unreliable implementation. To address this gap, we introduce PBLBench, a novel benchmark designed to evaluate complex reasoning grounded in domain-specific knowledge and long-context understanding, thereby challenging models with tasks that closely resemble those handled by human experts. To establish reliable ground truth, we adopt the Analytic Hierarchy Process (AHP), utilizing expert-driven pairwise comparisons to derive structured and weighted evaluation criteria. We assess the performance of 15 leading MLLMs/LLMs using PBLBench and demonstrate that even the most advanced models achieve only 59% rank accuracy, underscoring the significant challenges presented by this benchmark. We believe PBLBench will serve as a catalyst for the development of more capable AI agents, ultimately aiming to alleviate teacher workload and enhance educational productivity.

Privacy policies are statements that notify users of the services' data practices. However, few users are willing to read through policy texts due to the length and complexity. While automated tools based on machine learning exist for privacy policy analysis, to achieve high classification accuracy, classifiers need to be trained on a large labeled dataset. Most existing policy corpora are labeled by skilled human annotators, requiring significant amount of labor hours and effort. In this paper, we leverage active learning and crowdsourcing techniques to develop an automated classification tool named Calpric (Crowdsourcing Active Learning PRIvacy Policy Classifier), which is able to perform annotation equivalent to those done by skilled human annotators with high accuracy while minimizing the labeling cost. Specifically, active learning allows classifiers to proactively select the most informative segments to be labeled. On average, our model is able to achieve the same F1 score using only 62% of the original labeling effort. Calpric's use of active learning also addresses naturally occurring class imbalance in unlabeled privacy policy datasets as there are many more statements stating the collection of private information than stating the absence of collection. By selecting samples from the minority class for labeling, Calpric automatically creates a more balanced training set.