Synthetic seismograms play a crucial role in understanding Earth's internal structure and focal mechanisms. Although established methods remain fundamental, advancements in computational tools and optimizations are crucial for improving stability, efficiency, and accessibility in modern seismic modeling. Here, we introduce PyGRT, an open-source Python package designed to calculate synthetic seismograms for a point source in a layered half-space model. Built on the generalized reflection-transmission matrix method and the discrete wavenumber method, PyGRT integrates the peak-trough averaging method and Filon's integration method to handle diverse source-receiver distributions. To optimize performance, PyGRT uses C for its core computational tasks, accelerated with OpenMP for parallel processing, whereas Python provides a user-friendly interface. This dual-language approach combines efficiency with ease of use, offering a flexible and extensible framework for seismic modeling. The modular design includes both low-level C and high-level Python interfaces, enabling high-performance computing, easy customization, and streamlined data processing. PyGRT outperforms traditional tools in stability for high-frequency calculations, computational speed, and accessibility. It serves as a practical and versatile resource, providing both researchers and beginners with a robust tool for seismic-wave simulation and theoretical exploration.

Reinforcement learning with verifiable rewards has been especially effective in mathematics and coding, where answers can be checked automatically. Many open-ended medical questions lack comparably cheap outcome verifiers: responses may be partly correct, incomplete, or contain clinically consequential errors. Rubrics written or validated by physicians offer strong clinical grounding, but involving experts in every instance is costly. Model-generated rubrics make this supervision scalable. We introduce ConRub-Med to preserve useful distinctions as rubric feedback moves from construction to policy optimization. For each prompt, three heterogeneous language models propose atomic criteria independently; a separate model reviews them, retaining only criteria with semantic support from all three generators. Three-State scoring distinguishes correct coverage, missing information, and incorrect claims. Errors receive negative rather than zero credit. When every response in a complete Group Relative Policy Optimization (GRPO) group receives the same final reward, a pairwise judge provides sequence advantages only if both candidate orders agree, without changing the scalar rewards. Groups without ties use vanilla GRPO. In a blinded study matched by question, two medical experts rate panels from the full pipeline as more clinically relevant than panels produced by one generator. Across the evaluated open models, ConRub-Med ranks first on six of nine benchmarks and achieves the highest medical and generalization averages. Using the resulting rubric dataset of 5,166 prompts, it scores 38.98 ± 1.04 (mean ± SD) on HealthBench-Hard, compared with InfiMed-ORBIT's 33.60 with 8,000 samples and 37.30 with 28,000.

The Short-Offset Transient Electromagnetic method (SOTEM) offers distinct theoretical advantages for high-resolution exploration of subsurface targets shallower than 3000 meters. Its effectiveness largely depends on the quality of data processing, imaging, inversion, and anomaly interpretation method. Considering the broadband and information-rich nature of SOTEM signals, this paper reviews recent advancements in intelligent processing techniques of SOTEM method, covering advances in noise reduction, estimation of induced polarization parameters, and three-dimensional inversion. The discussion includes integrated multi-source noise suppression, hybrid inversion approaches that combine data-driven and model-driven strategies. This study points out that constructing a cross-modal network integrating "SOTEM data + " multi-source datasets is an important research direction to advance from physical property interpretation to geological interpretation.

Continual Pre-training (CPT) serves as a fundamental approach for adapting foundation models to domain-specific applications. Scaling laws for pre-training define a power-law relationship between dataset size and the test loss of an LLM. However, the marginal gains from simply increasing data for CPT diminish rapidly, yielding suboptimal data utilization and inefficient training. To address this challenge, we propose a novel perplexity-aware data scaling law to establish a predictive relationship between the perplexity landscape of domain-specific data and the test loss. Our approach leverages the pre-trained model’s own perplexity on domain data as a proxy for estimating the knowledge gap, effectively quantifying the informational perplexity landscape of candidate training samples. By fitting this scaling law across diverse perplexity regimes, we enable adaptive selection of high-utility data subsets, prioritizing content that maximizes knowledge absorption while minimizing redundancy and noise. Extensive experiments on both medical and general-domain benchmarks demonstrate that our method consistently identifies near-optimal training subsets, achieving superior performance with significantly reduced data consumption.

Large Language Models (LLMs) have emerged as powerful tools for passage reranking in information retrieval, leveraging their superior reasoning capabilities to address the limitations of conventional models on complex queries. However, current LLM-based reranking paradigms are fundamentally constrained by an efficiency-accuracy trade-off: (1) pointwise methods are efficient but ignore inter-document comparison, yielding suboptimal accuracy; (2) listwise methods capture global context but suffer from context-window constraints and prohibitive inference latency. To address these issues, we propose GroupRank, a novel paradigm that balances flexibility and context awareness. To unlock the full potential of groupwise reranking, we propose an answer-free data synthesis pipeline that fuses local pointwise signals with global listwise rankings. These samples facilitate supervised fine-tuning and reinforcement learning, with the latter guided by a specialized group-ranking reward comprising ranking-utility and group-alignment. These complementary components synergistically optimize document ordering and score calibration to reflect intrinsic query-document relevance. Experimental results show GroupRank achieves a state-of-the-art 65.2 NDCG@10 on BRIGHT and surpasses baselines by 2.1 points on R2MED, while delivering a 6.4× inference speedup.

The proliferation of Large Language Models (LLMs) presents transformative potential for healthcare, yet practical deployment is hindered by the absence of frameworks that assess real-world clinical utility. Existing benchmarks test static knowledge, failing to capture the dynamic, application-oriented capabilities required in clinical practice. To bridge this gap, we introduce a Medical LLM Benchmark MLB, a comprehensive benchmark evaluating LLMs on both foundational knowledge and scenario-based reasoning. MLB is structured around five core dimensions: Medical Knowledge (MedKQA), Safety and Ethics (MedSE), Medical Record Understanding (MedRU), Smart Services (SmartServ), and Smart Healthcare (SmartCare). The benchmark integrates 22 datasets (17 newly curated) from diverse Chinese clinical sources, covering 64 clinical specialties. Its design features a rigorous curation pipeline involving 300 licensed physicians. Besides, we provide a scalable evaluation methodology, centered on a specialized judge model trained via Supervised Fine-Tuning (SFT) on expert annotations. Our comprehensive evaluation of 10 leading models reveals a critical translational gap: while the top-ranked model, Kimi-K2-Instruct (77.3

Unsupervised image anomaly detection has been widely used in the industrial field, yet data imbalance and overfitting limit its performance. Therefore, this study proposes an industrial image anomaly localization network based on coordinate dual memory banks and difference subsection evolution (DmseNet). Firstly, to enable diversified generation of anomaly masks, the foreground coordinate dual memory banks strategy is proposed, and the corrosion factor is introduced for generating more realistic anomalies. Secondly, to effectively guide the learning process, a group channel migrated attention module was designed as a projector. It directly propagates and refines multi-scale teacher features through joint channel-spatial optimization, providing precise reconstruction guidance for the student network. In addition, a regularization repair module is proposed, which enhances the generalization ability and feature repair ability of the model. Finally, the entire system is refined under the optimization of the proposed difference subsection evolution loss. It dynamically assigns weights and loss levels according to the degree of feature deviation and enables the model to focus on the key information areas more accurately and improves the model’s discrimination ability. Comprehensive experiments on four challenging datasets, a self-built printed circuit board (PCB) anomaly detection platform and multi-view inspection application verify the portability and practicability of DmseNet in different scenarios, although its capability for image-level logical anomaly detection remains relatively limited. The research results not only promote the application of unsupervised learning in industrial image analysis, but also provide a new solution for the landing of artificial intelligence.

Short-offset Transient Electromagnetics (SOTEM) is a recently developed time-domain, near-source controlled-source electromagnetic (CSEM) method. Although its response mechanisms have attracted substantial interest, they have also prompted skepticism regarding the feasibility of near-source depth sounding. To address these concerns, this paper reviews the historical evolution of EM exploration methods and establishes the theoretical basis of SOTEM by systematically comparing time- and frequency-domain approaches in terms of transfer-function representation, acquisition strategy, spatial-region definitions, the physical meaning of offset, and treatment of primary (source) currents. Our analysis shows that conventional frequency-domain EM sounding is inherently tied to the far-zone plane-wave approximation; in the near zone, the strong primary field can dominate the measured response and obscure the secondary field, thereby hindering effective depth sounding. In contrast, SOTEM leverages the off-time window following the shut-off of a bipolar pulse to achieve a natural temporal separation of primary and secondary fields, enabling observation of diffusion-dominated responses controlled by induced eddy currents. These characteristics make time-domain transient EM methods not only theoretically sound under near-source conditions but also practically advantageous, facilitating deep exploration in complex terrains or where transmitter–receiver offsets are constrained. Overall, SOTEM provides a new technical route and a robust theoretical framework for deep mineral exploration.

Clinical agents promise to democratize access to electronic health records (EHRs), yet existing benchmarks fail to reflect the complexity of practical EHR analysis, e.g., often operating on idealized, clean EHRs via static SQL generation rather than interactive execution. In this work, we introduce EHR-Complex, a large-scale benchmark designed for interactive clinical database reasoning. Built on the large MIMIC-IV substrate (365K patients, 31 tables, 500M+ records), EHR-Complex comprises about 52K tasks spanning six clinical intents, supporting both patient-level and population-level queries, where each task requires an agent to interact with a sandboxed environment by executing SQL queries or Python code. Notably, EHR-Complex considers the real-world SQL task complexity for longitudinal multi-table aggregation and compositional reasoning, resulting in 31.93 SQL structural components per query on average. Evaluation results on EHR-Complex reveal the clinical difficulty of these EHR reasoning scenarios, with the top-performing model achieving only 62.3

Personalized generation with frozen large language models requires a conditioning signal that is both compact and current. Existing personalization methods typically retrieve or summarize user histories in text, or compress them into static latent profiles and soft prompts. These approaches are efficient, but they treat a user's past behavior as an aggregate profile and therefore mix stable identity, recent drift, and item content in the same representation. We propose LAtent Trajectory Tracking and Extrapolation (LATTE), a framework that represents personalization as forecasting a peer anchored relative preference state. For each historical session, LATTE subtracts a time masked baseline formed from comparable users who responded to the same item, producing a state that measures how the target user differs from peers under a shared item context. A lightweight sequence predictor then forecasts the next state in this trajectory, and a State to Token Bridge injects the forecast into a frozen instruction tuned LLM through a single anchored soft token. We provide a latent factor analysis showing when peer anchoring cancels shared item variation and why temporal forecasting trades off stale averages against noisy recent states. Experiments on Amazon Reviews 2023 and MemoryCD show that LATTE consistently outperforms retrieval, summary memory, static latent profiles, difference aware latent profiles, and soft prompt compression baselines. On Amazon Reviews 2023, LATTE improves average ROUGE-L from 0.219 for a static latent profile and 0.245 for the strongest added latent compression baseline to 0.259. Additional pairwise comparisons and diagnostic analyses suggest that the improvement is mainly due to forecasting user-specific trajectory information, rather than merely adding a soft prompt interface.

Deep research agents have demonstrated remarkable capabilities in complex information-seeking tasks, yet this power comes at a steep computational cost. Driven by accuracy-focused training paradigms, current models adopt brute-force strategies characterized by blind tool dependency and performative reasoning-generating long, redundant trajectories that are far from necessary for resolving these tasks, leading to wasteful tool calls and excessive token consumption. To overcome this efficiency trap, we propose SlimSearcher, a principled framework that pushes the Pareto frontier between accuracy and computational cost across both Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL). In the SFT stage, SlimSearcher employs Pareto-efficient filtration to distill trajectories that are both successful and economical, guiding the model toward inherently efficiency-aware search behaviors. During RL, we introduce Adaptive Reward Gating, a dynamic reward-shaping mechanism that evaluates relative tool and token efficiency within a sampled cohort. By cascading these adaptive efficiency metrics with a strict correctness gate, our approach effectively avoids the brevity bias associated with absolute penalties and mitigates reward hacking. Extensive experiments on long-horizon benchmarks, including GAIA, BrowseComp, and XBenchDeepSearch, demonstrate that SlimSearcher reduces average tool-call rounds by 17

Retrieval-Augmented Generation (RAG) enhances large language models (LLMs) by integrating external knowledge, where the LLM's ability to generate responses based on the combination of a given query and retrieved documents is crucial. However, most benchmarks focus on overall RAG system performance, rarely assessing LLM-specific capabilities. Current benchmarks emphasize broad aspects such as noise robustness, but lack a systematic and granular evaluation framework on document utilization. To this end, we introduce Placeholder-RAG-Benchmark, a multi-level fine-grained benchmark, emphasizing the following progressive dimensions: (1) multi-level filtering abilities, (2) combination abilities, and (3) reference reasoning. To provide a more nuanced understanding of LLMs' roles in RAG systems, we formulate an innovative placeholder-based approach to decouple the contributions of the LLM's parametric knowledge and the external knowledge. Experiments demonstrate the limitations of representative LLMs in the RAG system's generation capabilities, particularly in error resilience and context faithfulness. Our benchmark provides a reproducible framework for developing more reliable and efficient RAG systems.

Although large language models (LLMs) demonstrate expert-level medical knowledge, aligning their open-ended outputs with fine-grained clinician preferences remains challenging. Existing methods often rely on coarse objectives or unreliable automated judges that are weakly grounded in professional guidelines. We propose a two-stage framework to address this gap. First, we introduce HealthRubrics, a dataset of 7,034 physician-verified preference examples in which clinicians refine LLM-drafted rubrics to meet rigorous medical standards. Second, we distill these rubrics into HealthPrinciples: 119 broadly reusable, clinically grounded principles organized by clinical dimensions, enabling scalable supervision beyond manual annotation. We use HealthPrinciples for (1) offline alignment by synthesizing rubrics for unlabeled queries and (2) an inference-time tool for guided self-revision. A 30B parameter model that activates only 3B parameters at inference trained with our framework achieves 33.4

With the growing use of Retrieval-Augmented Generation (RAG), training large language models (LLMs) for context-sensitive reasoning and faithfulness is increasingly important. Existing RAG-oriented reinforcement learning (RL) methods rely on external rewards that often fail to evaluate document faithfulness, and may misjudge similar answers in open-domain settings. In addition, there is no RAG-based selfreward mechanism. Moreover, although such a mechanism could in principle estimate answer confidence given documents, the absence of objective feedback in a self-judgment can cause hallucination accumulation and eventual model collapse. To tackle these issues, we propose a novel "internal-external" hybrid reward framework centered on a Contrastive Likelihood Reward (CLR). CLR directly optimizes the log-likelihood gap between responses conditioned on prompts with and without supporting evidence. This encourages the model to extract relevant evidence and increases its confidence when grounded in a specific context. Experiments show that our method (used alone or combined with external correctness rewards) achieves strong performance on singlehop, multi-hop, vertical-domain, and faithfulness benchmarks. Our training code and models are coming soon.

We present SimpleSearch-VL, an efficient, reliable, and practical framework for multimodal agentic search. Its core idea is to improve the agent's own search-and-verification process rather than scaling data, tools, or auxiliary model components. For efficiency, Factorized Adaptive Rollout (FAR) improves sampling efficiency by forming more informative training groups while using redundant samples to mitigate long-tail latency and expose hard samples. For reliability, SimpleSearch-VL performs evidence-verified reasoning, explicitly using chain-of-thought verification to assess the relevance of retrieved visual and textual cues to the original context. For practicality, SimpleSearch-VL keeps a lightweight tool interface and performs webpage self-summary within the agent, requiring no additional external dependencies. With only 5K supervised tool-interleaved trajectories and 2K RL data, SimpleSearch-VL improves Qwen3-VL agentic baselines by 15.8 and 16.0 average points for the 8B and 30B-A3B variants, respectively. The SimpleSearch-VL-30B-A3B model further achieves performance competitive with agentic Gemini-3-Pro.

Deep Research systems based on web agents have shown strong potential in solving complex information-seeking tasks, yet their search efficiency remains underexplored. We observe that many state-of-the-art open-source web agents rely on long tool-call trajectories with cyclic reasoning loops and exploration of unproductive branches. To address this, we propose WebClipper, a framework that compresses web agent trajectories via graph-based pruning. Concretely, we model the agent’s search process as a state graph and cast trajectory optimization as a minimum-necessary Directed Acyclic Graph (DAG) mining problem, yielding pruned trajectories that preserve essential reasoning while eliminating redundant steps. Continued training on these refined trajectories enables the agent to evolve toward more efficient search patterns and reduces tool-call rounds by about 20% while improving accuracy. Furthermore, we introduce a new metric called F-AE Score to measure the model’s overall performance in balancing accuracy and efficiency. Experiments demonstrate that WebClipper compresses tool-call rounds under excellent performance, providing practical insight into balancing effectiveness and efficiency in web agent design.

Remote sensing foundation models, pretrained on massive remote sensing data, have shown impressive performance in several Earth observation (EO) tasks. These models usually use single-modal temporal data for pretraining, which is insufficient for multi-modal applications. Moreover, these models require a considerable number of samples for fine-tuning in downstream tasks, posing challenges in time-sensitive scenarios, such as rapid flood mapping. We present SkySense++, a multi-modal remote sensing foundation model for diverse EO tasks. SkySense++ has a factorized architecture to accommodate multi-modal images acquired by diverse sensors. We adopt progressive pretraining, which involves two stages, on meticulously curated datasets of 27 million multi-modal remote sensing images. The first representation-enhanced pretraining stage uses multi-granularity contrastive learning to obtain general representations. The second semantic-enhanced pretraining stage leverages masked semantic learning to learn semantically enriched representations, enabling few-shot capabilities. This ability allows the model to handle unseen tasks with minimal labelled data, alleviating the need for fine-tuning on extensive annotated data. SkySense++ demonstrates consistent improvements in classification, detection and segmentation over previous state-of-the-art models across 12 EO tasks in 7 domains: agriculture, forestry, oceanography, atmosphere, biology, land surveying and disaster management. This generalizability may lead to a new chapter of remote sensing foundation model applications for EO tasks at scale. Wu et al. developed SkySense++, a multi-modal remote sensing foundation model pretrained on 27 million multi-modal images, which achieved robust generalization and few-shot capabilities across several Earth observation tasks and domains, including agriculture and disaster management.