By contrast, epistemic uncertainty describes the inference model's lack of knowledge

By contrast, epistemic uncertainty describes the inference model's lack of knowledge

Large Language Models (LLMs) increasingly rely on external tools such as search engines to solve complex agentic tasks that require reasoning and external knowledge retrieval. Recently, reinforcement learning with verifiable rewards (RL VR) has demonstrated its effectiveness in advancing capabilities of LLMs by rewarding the final answers via outcome rewards. While straightforward to supervise, outcome rewards only provide sparse signals and delayed feedback, which limits their effectiveness on long trajectories. Process rewards address this by evaluating intermediate steps, providing fine-grained supervision and encouraging grounded problem solving. However, it is notoriously hard to annotate step-wise labels, especially in non-verifiable process without "golden" answers. Furthermore, stepwise judgment requires the balance between local quality with contribution to the final outcome, as optimizing towards higher process reward may not always align with better final outcomes. To address the above challenges, we introduce Principle Process Reward (PPR), an RL approach that unifies principled step-level assessment and outcome verification. We train a principle-based reward model to improve the transparency and reliability of process evaluation, and further introduce a Reward Normalization (ReNorm) strategy to calibrate outcome and process rewards. Experiment results show that PPR achieves state-of-the-art performance across a wide range of benchmarks, demonstrating its impressive robustness and generalization. Our code and model collection is available in this link.Figure 1: Performance of PPR on various benchmarks with other baselines Large Language Models (LLMs) have achieved remarkable progress across a wide range of tasks, from open-domain question answering to multi-step reasoning (Guo et al., 2025; OpenAI, 2025b; Comanici et al., 2025). A key factor for success is their abilities to leverage external tools such as search engines, calculators, code interpreters, and browsers (DeepMind, 2025; Guo et al., 2024; OpenAI, 2025a). In particular, the search engine is a linchpin tool that provides verifiable and up-to-date knowledge for LLMs, helping to ground their answers and reduce hallucinations. However, training LLM agents to leverage tools effectively still remains challenging, as the complex behavior involving task decomposition, query generation, information aggregation, and stopping decisions.

Process Reward Model (PRM) is widely used in the post-training of Large Language Model (LLM) because it can perform fine-grained evaluation of the reasoning steps of generated content. However, most PRMs lack long-term reasoning and deep thinking capabilities. On the other hand, although a few works have tried to introduce Chain-of-Thought capability into PRMs, the annotation cost of CoT-PRM data is too expensive to play a stable role in various tasks. To address the above challenges, we propose VRPRM, a process reward model via visual reasoning, and design an efficient two-stage training strategy. Experimental results show that using only 3.6K CoT-PRM SFT data and 50K non-CoT PRM RL training data, VRPRM can surpass the non-thinking PRM with a total data volume of 400K and achieved a relative performance improvement of up to 118\% over the base model in the BoN experiment. This result confirms that the proposed combined training strategy can achieve higher quality reasoning capabilities at a lower data annotation cost, thus providing a new paradigm for PRM training with more efficient data utilization.

Sudden Death Syndrome (SDS), caused by Fusarium virguliforme, poses a significant threat to soybean production. This study presents an AI-driven web application for early detection of SDS on soybean leaves using hyperspectral imaging, enabling diagnosis prior to visible symptom onset. Leaf samples from healthy and inoculated plants were scanned using a portable hyperspectral imaging system (398-1011 nm), and a Genetic Algorithm was employed to select five informative wavelengths (505.4, 563.7, 712.2, 812.9, and 908.4 nm) critical for discriminating infection status. These selected bands were fed into a lightweight Convolutional Neural Network (CNN) to extract spatial-spectral features, which were subsequently classified using ten classical machine learning models. Ensemble classifiers (Random Forest, AdaBoost), Linear SVM, and Neural Net achieved the highest accuracy (>98%) and minimal error across all folds, as confirmed by confusion matrices and cross-validation metrics. Poor performance by Gaussian Process and QDA highlighted their unsuitability for this dataset. The trained models were deployed within a web application that enables users to upload hyperspectral leaf images, visualize spectral profiles, and receive real-time classification results. This system supports rapid and accessible plant disease diagnostics, contributing to precision agriculture practices. Future work will expand the training dataset to encompass diverse genotypes, field conditions, and disease stages, and will extend the system for multiclass disease classification and broader crop applicability.

Climate system models (CSMs), through integrating cross-sphere interactions among the atmosphere, ocean, land, and cryosphere, have emerged as pivotal tools for deciphering climate dynamics and improving forecasting capabilities. Recent breakthroughs in artificial intelligence (AI)-driven meteorological modeling have demonstrated remarkable success in single-sphere systems and partially spheres coupled systems. However, the development of a fully coupled AI-based climate system model encompassing atmosphere-ocean-land-sea ice interactions has remained an unresolved challenge. This paper introduces FengShun-CSM, an AI-based CSM model that provides 60-day global daily forecasts for 29 critical variables across atmospheric, oceanic, terrestrial, and cryospheric domains. The model significantly outperforms the European Centre for Medium-Range Weather Forecasts (ECMWF) subseasonal-to-seasonal (S2S) model in predicting most variables, particularly precipitation, land surface, and oceanic components. This enhanced capability is primarily attributed to its improved representation of intra-seasonal variability modes, most notably the Madden-Julian Oscillation (MJO). Remarkably, FengShun-CSM exhibits substantial potential in predicting subseasonal extreme events. Such breakthroughs will advance its applications in meteorological disaster mitigation, marine ecosystem conservation, and agricultural productivity enhancement. Furthermore, it validates the feasibility of developing AI-powered CSMs through machine learning technologies, establishing a transformative paradigm for next-generation Earth system modeling.

An attorney with the American Civil Liberties Union (ACLU) who's overseeing a high-profile deportation case in Louisiana says she was stripped of her electronics moments before a pivotal hearing, preventing her from accessing evidence and court records that remained available to the three US government attorneys in the room, each of whom were allowed use of a laptop by the court. Louisiana immigration judge Jamee Comans ruled late last month that Columbia graduate student Mahmoud Khalil was eligible for deportation. During that hearing, however, Khalil's attorney Nora Ahmed says she was barred from bringing her laptop into the courtroom, despite having filed the proper paperwork in advance and being a frequent visitor to the immigration facility. "There should not be an advantage, no matter how small or how large, provided to a particular party over the other," says Ahmed. "Because that starts to infect the proceedings themselves and the notion of fundamental fairness that we all uphold in courtroom proceedings." The Justice Department did not respond to a request for comment.

Large language models afford opportunities for using computers for intensive tasks, realizing research opportunities that have not been considered before. One such opportunity could be a systematic interrogation of the scientific literature. Here, we show how a large language model can be used to construct a literature review of 2699 publications associated with microphysics parametrizations in the Weather and Research Forecasting (WRF) model, with the goal of learning how they were used and their systematic biases, when simulating precipitation. The database was constructed of publications identified from Web of Science and Scopus searches. The large language model GPT-4 Turbo was used to extract information about model configurations and performance from the text of 2699 publications. Our results reveal the landscape of how nine of the most popular microphysics parameterizations have been used around the world: Lin, Ferrier, WRF Single-Moment, Goddard Cumulus Ensemble, Morrison, Thompson, and WRF Double-Moment. More studies used one-moment parameterizations before 2020 and two-moment parameterizations after 2020. Seven out of nine parameterizations tended to overestimate precipitation. However, systematic biases of parameterizations differed in various regions. Except simulations using the Lin, Ferrier, and Goddard parameterizations that tended to underestimate precipitation over almost all locations, the remaining six parameterizations tended to overestimate, particularly over China, southeast Asia, western United States, and central Africa. This method could be used by other researchers to help understand how the increasingly massive body of scientific literature can be harnessed through the power of artificial intelligence to solve their research problems.

This paper introduces Indiana Jones, an innovative approach to jailbreaking Large Language Models (LLMs) by leveraging inter-model dialogues and keyword-driven prompts. Through orchestrating interactions among three specialised LLMs, the method achieves near-perfect success rates in bypassing content safeguards in both white-box and black-box LLMs. The research exposes systemic vulnerabilities within contemporary models, particularly their susceptibility to producing harmful or unethical outputs when guided by ostensibly innocuous prompts framed in historical or contextual contexts. Experimental evaluations highlight the efficacy and adaptability of Indiana Jones, demonstrating its superiority over existing jailbreak methods. These findings emphasise the urgent need for enhanced ethical safeguards and robust security measures in the development of LLMs. Moreover, this work provides a critical foundation for future studies aimed at fortifying LLMs against adversarial exploitation while preserving their utility and flexibility.

How well can Multimodal Large Language Models (MLLMs) understand composite images? Composite images (CIs) are synthetic visuals created by merging multiple visual elements, such as charts, posters, or screenshots, rather than being captured directly by a camera. While CIs are prevalent in real-world applications, recent MLLM developments have primarily focused on interpreting natural images (NIs). Our research reveals that current MLLMs face significant challenges in accurately understanding CIs, often struggling to extract information or perform complex reasoning based on these images. We find that existing training data for CIs are mostly formatted for question-answer tasks (e.g., in datasets like ChartQA and ScienceQA), while high-quality image-caption datasets, critical for robust vision-language alignment, are only available for NIs. To bridge this gap, we introduce Composite Captions (CompCap), a flexible framework that leverages Large Language Models (LLMs) and automation tools to synthesize CIs with accurate and detailed captions. Using CompCap, we curate CompCap-118K, a dataset containing 118K image-caption pairs across six CI types. We validate the effectiveness of CompCap-118K by supervised fine-tuning MLLMs of three sizes: xGen-MM-inst.-4B and LLaVA-NeXT-Vicuna-7B/13B. Empirical results show that CompCap-118K significantly enhances MLLMs' understanding of CIs, yielding average gains of 1.7%, 2.0%, and 2.9% across eleven benchmarks, respectively.