# Research Papers — 2025-12 · bundle page 104 of 175 > Full-text markdown of papers #5151–5200 of 8703 published in 2025-12, newest first. AI-readable mirrors of the full 8703-paper 2025-12 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2025-12/page/104 [← Previous bundle](https://aipod.dev/papers/archive/2025-12/page/103.md) [Next bundle →](https://aipod.dev/papers/archive/2025-12/page/105.md) ### Quantum Approaches to Urban Logistics: From Core QAOA to Clustered Scalability Authors: F. Picariello, G. Turati, R. Antonelli, I. Bailo, S. Bonura, G. Ciarfaglia, S. Cipolla, P. Cremonesi, M. Ferrari Dacrema, M. Gabusi, I. Gentile, V. Morreale et al. Published: 2025-12-11 Categories: quant-ph, cs.LG arXiv: https://arxiv.org/abs/2512.10813v1 The Traveling Salesman Problem (TSP) is a fundamental challenge in combinatorial optimization, widely applied in logistics and transportation. As the size of TSP instances grows, traditional algorithms often struggle to produce high-quality solutions within reasonable timeframes. This study investigates the potential of the Quantum Approximate Optimization Algorithm (QAOA), a hybrid quantum-classical method, to solve TSP under realistic constraints. We adopt a QUBO-based formulation of TSP that integrates real-world logistical constraints reflecting operational conditions, such as vehicle capacity, road accessibility, and time windows, while ensuring compatibility with the limitations of current quantum hardware. Our experiments are conducted in a simulated environment using high-performance computing (HPC) resources to assess QAOA's performance across different problem sizes and quantum circuit depths. In order to improve scalability, we propose clustering QAOA (Cl-QAOA), a hybrid approach combining classical machine learning with QAOA. This method decomposes large TSP instances into smaller sub-problems, making quantum optimization feasible even on devices with a limited number of qubits. The results offer a comprehensive evaluation of QAOA's strengths and limitations in solving constrained TSP scenarios. This study advances quantum optimization and lays groundwork for future large-scale applications. --- ### Graph Laplacian Transformer with Progressive Sampling for Prostate Cancer Grading Authors: Masum Shah Junayed, John Derek Van Vessem, Qian Wan, Gahie Nam, Sheida Nabavi Published: 2025-12-11 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.10808v1 Prostate cancer grading from whole-slide images (WSIs) remains a challenging task due to the large-scale nature of WSIs, the presence of heterogeneous tissue structures, and difficulty of selecting diagnostically relevant regions. Existing approaches often rely on random or static patch selection, leading to the inclusion of redundant or non-informative regions that degrade performance. To address this, we propose a Graph Laplacian Attention-Based Transformer (GLAT) integrated with an Iterative Refinement Module (IRM) to enhance both feature learning and spatial consistency. The IRM iteratively refines patch selection by leveraging a pretrained ResNet50 for local feature extraction and a foundation model in no-gradient mode for importance scoring, ensuring only the most relevant tissue regions are preserved. The GLAT models tissue-level connectivity by constructing a graph where patches serve as nodes, ensuring spatial consistency through graph Laplacian constraints and refining feature representations via a learnable filtering mechanism that enhances discriminative histological structures. Additionally, a convex aggregation mechanism dynamically adjusts patch importance to generate a robust WSI-level representation. Extensive experiments on five public and one private dataset demonstrate that our model outperforms state-of-the-art methods, achieving higher performance and spatial consistency while maintaining computational efficiency. --- ### HAROOD: A Benchmark for Out-of-distribution Generalization in Sensor-based Human Activity Recognition Authors: Wang Lu, Yao Zhu, Jindong Wang Published: 2025-12-11 Categories: cs.AI arXiv: https://arxiv.org/abs/2512.10807v3 Sensor-based human activity recognition (HAR) mines activity patterns from the time-series sensory data. In realistic scenarios, variations across individuals, devices, environments, and time introduce significant distributional shifts for the same activities. Recent efforts attempt to solve this challenge by applying or adapting existing out-of-distribution (OOD) algorithms, but only in certain distribution shift scenarios (e.g., cross-device or cross-position), lacking comprehensive insights on the effectiveness of these algorithms. For instance, is OOD necessary to HAR? Which OOD algorithm performs the best? In this paper, we fill this gap by proposing HAROOD, a comprehensive benchmark for HAR in OOD settings. We define 4 OOD scenarios: cross-person, cross-position, cross-dataset, and cross-time, and build a testbed covering 6 datasets, 16 comparative methods (implemented with CNN-based and Transformer-based architectures), and two model selection protocols. Then, we conduct extensive experiments and present several findings for future research, e.g., no single method consistently outperforms others, highlighting substantial opportunity for advancement. Our codebase is highly modular and easy to extend for new datasets, algorithms, comparisons, and analysis, with the hope to facilitate the research in OOD-based HAR. Our implementation is released and can be found at https://github.com/AIFrontierLab/HAROOD. --- ### HAROOD: A Benchmark for Out-of-distribution Generalization in Sensor-based Human Activity Recognition Authors: Wang Lu, Yao Zhu, Jindong Wang Published: 2025-12-11 Categories: cs.AI arXiv: https://arxiv.org/abs/2512.10807v1 Sensor-based human activity recognition (HAR) mines activity patterns from the time-series sensory data. In realistic scenarios, variations across individuals, devices, environments, and time introduce significant distributional shifts for the same activities. Recent efforts attempt to solve this challenge by applying or adapting existing out-of-distribution (OOD) algorithms, but only in certain distribution shift scenarios (e.g., cross-device or cross-position), lacking comprehensive insights on the effectiveness of these algorithms. For instance, is OOD necessary to HAR? Which OOD algorithm performs the best? In this paper, we fill this gap by proposing HAROOD, a comprehensive benchmark for HAR in OOD settings. We define 4 OOD scenarios: cross-person, cross-position, cross-dataset, and cross-time, and build a testbed covering 6 datasets, 16 comparative methods (implemented with CNN-based and Transformer-based architectures), and two model selection protocols. Then, we conduct extensive experiments and present several findings for future research, e.g., no single method consistently outperforms others, highlighting substantial opportunity for advancement. Our codebase is highly modular and easy to extend for new datasets, algorithms, comparisons, and analysis, with the hope to facilitate the research in OOD-based HAR. Our implementation is released and can be found at https://github.com/AIFrontierLab/HAROOD. --- ### Interpretable and Steerable Concept Bottleneck Sparse Autoencoders Authors: Akshay Kulkarni, Tsui-Wei Weng, Vivek Narayanaswamy, Shusen Liu, Wesam A. Sakla, Kowshik Thopalli Published: 2025-12-11 Categories: cs.LG, cs.CV arXiv: https://arxiv.org/abs/2512.10805v2 Sparse autoencoders (SAEs) promise a unified approach for mechanistic interpretability, concept discovery, and model steering in LLMs and LVLMs. However, realizing this potential requires learned features to be both interpretable and steerable. To that end, we introduce two new computationally inexpensive interpretability and steerability metrics for a systematic analysis of LVLM SAEs. This uncovers two observations; (i) a majority of SAE neurons exhibit either low interpretability or low steerability or both, rendering them ineffective for downstream use; and (ii) user-desired concepts are often absent in the SAE, thus limiting their practical utility. To address these limitations, we propose Concept Bottleneck Sparse Autoencoders (CB-SAE) - a novel post-hoc framework that prunes low-utility neurons and augments the latent space with a lightweight concept bottleneck aligned to a user-defined concept set. The resulting CB-SAE improves interpretability by +32.1% and steerability by +14.5% across LVLMs and image generation tasks. --- ### Interpretable and Steerable Concept Bottleneck Sparse Autoencoders Authors: Akshay Kulkarni, Tsui-Wei Weng, Vivek Narayanaswamy, Shusen Liu, Wesam A. Sakla, Kowshik Thopalli Published: 2025-12-11 Categories: cs.LG, cs.CV arXiv: https://arxiv.org/abs/2512.10805v1 Sparse autoencoders (SAEs) promise a unified approach for mechanistic interpretability, concept discovery, and model steering in LLMs and LVLMs. However, realizing this potential requires that the learned features be both interpretable and steerable. To that end, we introduce two new computationally inexpensive interpretability and steerability metrics and conduct a systematic analysis on LVLMs. Our analysis uncovers two observations; (i) a majority of SAE neurons exhibit either low interpretability or low steerability or both, rendering them ineffective for downstream use; and (ii) due to the unsupervised nature of SAEs, user-desired concepts are often absent in the learned dictionary, thus limiting their practical utility. To address these limitations, we propose Concept Bottleneck Sparse Autoencoders (CB-SAE) - a novel post-hoc framework that prunes low-utility neurons and augments the latent space with a lightweight concept bottleneck aligned to a user-defined concept set. The resulting CB-SAE improves interpretability by +32.1% and steerability by +14.5% across LVLMs and image generation tasks. We will make our code and model weights available. --- ### What matters for Representation Alignment: Global Information or Spatial Structure? Authors: Jaskirat Singh, Xingjian Leng, Zongze Wu, Liang Zheng, Richard Zhang, Eli Shechtman, Saining Xie Published: 2025-12-11 Categories: cs.CV, cs.AI, cs.GR, cs.LG, stat.ML arXiv: https://arxiv.org/abs/2512.10794v1 Representation alignment (REPA) guides generative training by distilling representations from a strong, pretrained vision encoder to intermediate diffusion features. We investigate a fundamental question: what aspect of the target representation matters for generation, its \textit{global} \revision{semantic} information (e.g., measured by ImageNet-1K accuracy) or its spatial structure (i.e. pairwise cosine similarity between patch tokens)? Prevalent wisdom holds that stronger global semantic performance leads to better generation as a target representation. To study this, we first perform a large-scale empirical analysis across 27 different vision encoders and different model scales. The results are surprising; spatial structure, rather than global performance, drives the generation performance of a target representation. To further study this, we introduce two straightforward modifications, which specifically accentuate the transfer of \emph{spatial} information. We replace the standard MLP projection layer in REPA with a simple convolution layer and introduce a spatial normalization layer for the external representation. Surprisingly, our simple method (implemented in $<$4 lines of code), termed iREPA, consistently improves convergence speed of REPA, across a diverse set of vision encoders, model sizes, and training variants (such as REPA, REPA-E, Meanflow, JiT etc). %, etc. Our work motivates revisiting the fundamental working mechanism of representational alignment and how it can be leveraged for improved training of generative models. The code and project page are available at https://end2end-diffusion.github.io/irepa --- ### LabelFusion: Fusing Large Language Models with Transformer Encoders for Robust Financial News Classification Authors: Michael Schlee, Christoph Weisser, Timo Kivimäki, Melchizedek Mashiku, Benjamin Saefken Published: 2025-12-11 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2512.10793v2 Financial news plays a central role in shaping investor sentiment and short-term dynamics in commodity markets. Many downstream financial applications, such as commodity price prediction or sentiment modeling, therefore rely on the ability to automatically identify news articles relevant to specific assets. However, obtaining large labeled corpora for financial text classification is costly, and transformer-based classifiers such as RoBERTa often degrade significantly in low-data regimes. Our results show that appropriately prompted out-of-the-box Large Language Models (LLMs) achieve strong performance even in such settings. Furthermore, we propose LabelFusion, a hybrid architecture that combines the output of a prompt-engineered LLM with contextual embeddings produced by a fine-tuned RoBERTa encoder through a lightweight Multilayer Perceptron (MLP) voting layer. Evaluated on a ten-class multi-label subset of the Reuters-21578 corpus, LabelFusion achieves a macro F1 score of 96.0% and an accuracy of 92.3% when trained on the full dataset, outperforming both standalone RoBERTa (F1 94.6%) and the standalone LLM (F1 93.9%). In low- to mid-data regimes, however, the LLM alone proves surprisingly competitive, achieving an F1 score of 75.9% even in a zero-shot setting and consistently outperforming LabelFusion until approximately 80% of the training data is available. These results suggest that LLM-only prompting is the preferred strategy under annotation constraints, whereas LabelFusion becomes the most effective solution once sufficient labeled data is available to train the encoder component. The code is available in an anonymized repository. --- ### LabelFusion: Learning to Fuse LLMs and Transformer Classifiers for Robust Text Classification Authors: Michael Schlee, Christoph Weisser, Timo Kivimäki, Melchizedek Mashiku, Benjamin Saefken Published: 2025-12-11 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2512.10793v1 LabelFusion is a fusion ensemble for text classification that learns to combine a traditional transformer-based classifier (e.g., RoBERTa) with one or more Large Language Models (LLMs such as OpenAI GPT, Google Gemini, or DeepSeek) to deliver accurate and cost-aware predictions across multi-class and multi-label tasks. The package provides a simple high-level interface (AutoFusionClassifier) that trains the full pipeline end-to-end with minimal configuration, and a flexible API for advanced users. Under the hood, LabelFusion integrates vector signals from both sources by concatenating the ML backbone's embeddings with the LLM-derived per-class scores -- obtained through structured prompt-engineering strategies -- and feeds this joint representation into a compact multi-layer perceptron (FusionMLP) that produces the final prediction. This learned fusion approach captures complementary strengths of LLM reasoning and traditional transformer-based classifiers, yielding robust performance across domains -- achieving 92.4% accuracy on AG News and 92.3% on 10-class Reuters 21578 topic classification -- while enabling practical trade-offs between accuracy, latency, and cost. --- ### The FACTS Leaderboard: A Comprehensive Benchmark for Large Language Model Factuality Authors: Aileen Cheng, Alon Jacovi, Amir Globerson, Ben Golan, Charles Kwong, Chris Alberti, Connie Tao, Eyal Ben-David, Gaurav Singh Tomar, Lukas Haas, Yonatan Bitton, Adam Bloniarz et al. Published: 2025-12-11 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2512.10791v1 We introduce The FACTS Leaderboard, an online leaderboard suite and associated set of benchmarks that comprehensively evaluates the ability of language models to generate factually accurate text across diverse scenarios. The suite provides a holistic measure of factuality by aggregating the performance of models on four distinct sub-leaderboards: (1) FACTS Multimodal, which measures the factuality of responses to image-based questions; (2) FACTS Parametric, which assesses models' world knowledge by answering closed-book factoid questions from internal parameters; (3) FACTS Search, which evaluates factuality in information-seeking scenarios, where the model must use a search API; and (4) FACTS Grounding (v2), which evaluates whether long-form responses are grounded in provided documents, featuring significantly improved judge models. Each sub-leaderboard employs automated judge models to score model responses, and the final suite score is an average of the four components, designed to provide a robust and balanced assessment of a model's overall factuality. The FACTS Leaderboard Suite will be actively maintained, containing both public and private splits to allow for external participation while guarding its integrity. It can be found at https://www.kaggle.com/benchmarks/google/facts . --- ### Natural Language Interface for Firewall Configuration Authors: F. Taghiyev, A. Aslanbayli Published: 2025-12-11 Categories: cs.NI, cs.AI arXiv: https://arxiv.org/abs/2512.10789v1 This paper presents the design and prototype implementation of a natural language interface for configuring enterprise firewalls. The framework allows administrators to express access control policies in plain language, which are then translated into vendor specific configurations. A compact schema bound intermediate representation separates human intent from device syntax and in the current prototype compiles to Palo Alto PAN OS command line configuration while remaining extensible to other platforms. Large language models are used only as assistive parsers that generate typed intermediate representation objects, while compilation and enforcement remain deterministic. The prototype integrates three validation layers, namely a static linter that checks structural and vendor specific constraints, a safety gate that blocks overly permissive rules such as any to any allows, and a Batfish based simulator that validates configuration syntax and referential integrity against a synthetic device model. The paper describes the architecture, implementation, and test methodology on synthetic network context datasets and discusses how this approach can evolve into a scalable auditable and human centered workflow for firewall policy management. --- ### Replace, Don't Expand: Mitigating Context Dilution in Multi-Hop RAG via Fixed-Budget Evidence Assembly Authors: Moshe Lahmy, Roi Yozevitch Published: 2025-12-11 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2512.10787v2 Retrieval-Augmented Generation (RAG) systems often fail on multi-hop queries when the initial retrieval misses a bridge fact. Prior corrective approaches, such as Self-RAG, CRAG, and Adaptive-$k$, typically address this by \textit{adding} more context or pruning existing lists. However, simply expanding the context window often leads to \textbf{context dilution}, where distractors crowd out relevant information. We propose \textbf{SEAL-RAG}, a training-free controller that adopts a \textbf{``replace, don't expand''} strategy to fight context dilution under a fixed retrieval depth $k$. SEAL executes a (\textbf{S}earch $\rightarrow$ \textbf{E}xtract $\rightarrow$ \textbf{A}ssess $\rightarrow$ \textbf{L}oop) cycle: it performs on-the-fly, entity-anchored extraction to build a live \textit{gap specification} (missing entities/relations), triggers targeted micro-queries, and uses \textit{entity-first ranking} to actively swap out distractors for gap-closing evidence. We evaluate SEAL-RAG against faithful re-implementations of Basic RAG, CRAG, Self-RAG, and Adaptive-$k$ in a shared environment on \textbf{HotpotQA} and \textbf{2WikiMultiHopQA}. On HotpotQA ($k=3$), SEAL improves answer correctness by \textbf{+3--13 pp} and evidence precision by \textbf{+12--18 pp} over Self-RAG. On 2WikiMultiHopQA ($k=5$), it outperforms Adaptive-$k$ by \textbf{+8.0 pp} in accuracy and maintains \textbf{96\%} evidence precision compared to 22\% for CRAG. These gains are statistically significant ($p<0.001$). By enforcing fixed-$k$ replacement, SEAL yields a predictable cost profile while ensuring the top-$k$ slots are optimized for precision rather than mere breadth. We release our code and data at https://github.com/mosherino/SEAL-RAG. --- ### Replace, Don't Expand: Mitigating Context Dilution in Multi-Hop RAG via Fixed-Budget Evidence Assembly Authors: Moshe Lahmy, Roi Yozevitch Published: 2025-12-11 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2512.10787v1 Retrieval-Augmented Generation (RAG) systems often fail on multi-hop queries when the initial retrieval misses a bridge fact. Prior corrective approaches, such as Self-RAG, CRAG, and Adaptive-$k$, typically address this by \textit{adding} more context or pruning existing lists. However, simply expanding the context window often leads to \textbf{context dilution}, where distractors crowd out relevant information. We propose \textbf{SEAL-RAG}, a training-free controller that adopts a \textbf{``replace, don't expand''} strategy to fight context dilution under a fixed retrieval depth $k$. SEAL executes a (\textbf{S}earch $\rightarrow$ \textbf{E}xtract $\rightarrow$ \textbf{A}ssess $\rightarrow$ \textbf{L}oop) cycle: it performs on-the-fly, entity-anchored extraction to build a live \textit{gap specification} (missing entities/relations), triggers targeted micro-queries, and uses \textit{entity-first ranking} to actively swap out distractors for gap-closing evidence. We evaluate SEAL-RAG against faithful re-implementations of Basic RAG, CRAG, Self-RAG, and Adaptive-$k$ in a shared environment on \textbf{HotpotQA} and \textbf{2WikiMultiHopQA}. On HotpotQA ($k=3$), SEAL improves answer correctness by \textbf{+3--13 pp} and evidence precision by \textbf{+12--18 pp} over Self-RAG. On 2WikiMultiHopQA ($k=5$), it outperforms Adaptive-$k$ by \textbf{+8.0 pp} in accuracy and maintains \textbf{96\%} evidence precision compared to 22\% for CRAG. These gains are statistically significant ($p<0.001$). By enforcing fixed-$k$ replacement, SEAL yields a predictable cost profile while ensuring the top-$k$ slots are optimized for precision rather than mere breadth. We release our code and data at https://github.com/mosherino/SEAL-RAG. --- ### Developing an LLM-Based Feedback System Grounded in Evidence-Centered Design to Support Physics Problem Solving Authors: Holger Maus, Fabian Kieser, Stefan Petersen, Peter Wulff, Paul Tschisgale Published: 2025-12-11 Categories: physics.ed-ph, cs.AI, cs.HC arXiv: https://arxiv.org/abs/2512.10785v3 Generative AI offers new opportunities for individualized and adaptive learning, e.g., through large language model (LLM)-based feedback systems. While LLMs can produce factually correct feedback for relatively straightforward conceptual tasks, delivering high-quality feedback for tasks that require advanced domain expertise, such as physics problem solving, remains a substantial challenge. This study presents the design and implementation of an LLM-based feedback system for physics problem solving grounded in evidence-centered design and reports a first evaluation within the German Physics Olympiad. Participants rated the usefulness and correctness of the generated feedback for each implemented problem. The collected ratings indicate that the feedback was generally perceived as useful and highly correct. However, an in-depth analysis revealed that the feedback contained errors in 20% of cases; errors that often went unnoticed by the students. We discuss the risks associated with uncritical reliance on LLM-based feedback and outline potential directions for generating more adaptive and reliable LLM-based feedback in the future. --- ### Developing and Evaluating a Large Language Model-Based Automated Feedback System Grounded in Evidence-Centered Design for Supporting Physics Problem Solving Authors: Holger Maus, Paul Tschisgale, Fabian Kieser, Stefan Petersen, Peter Wulff Published: 2025-12-11 Categories: physics.ed-ph, cs.AI, cs.HC arXiv: https://arxiv.org/abs/2512.10785v1 Generative AI offers new opportunities for individualized and adaptive learning, particularly through large language model (LLM)-based feedback systems. While LLMs can produce effective feedback for relatively straightforward conceptual tasks, delivering high-quality feedback for tasks that require advanced domain expertise, such as physics problem solving, remains a substantial challenge. This study presents the design of an LLM-based feedback system for physics problem solving grounded in evidence-centered design (ECD) and evaluates its performance within the German Physics Olympiad. Participants assessed the usefulness and accuracy of the generated feedback, which was generally perceived as useful and highly accurate. However, an in-depth analysis revealed that the feedback contained factual errors in 20% of cases; errors that often went unnoticed by the students. We discuss the risks associated with uncritical reliance on LLM-based feedback systems and outline potential directions for generating more adaptive and reliable LLM-based feedback in the future. --- ### Script Gap: Evaluating LLM Triage on Indian Languages in Native vs Romanized Scripts in a Real World Setting Authors: Manurag Khullar, Utkarsh Desai, Poorva Malviya, Aman Dalmia, Zheyuan Ryan Shi Published: 2025-12-11 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2512.10780v2 Large Language Models (LLMs) are increasingly deployed in high-stakes clinical applications in India. Speakers of Indian languages frequently communicate using romanized text rather than native scripts, yet existing research rarely quantifies or evaluates this orthographic variation in real world applications. We investigate how romanization impacts the reliability of LLMs in a critical domain: maternal and newborn healthcare triage. We benchmark leading LLMs on a real world dataset of user-generated health queries spanning five Indian languages and Nepali. Our results reveal consistent degradation in performance for romanized messages, with gap reaching up to 24 points across languages and models. We propose and evaluate an Uncertainty-based Selective Routing method to close this script gap. At our partner maternal health organization alone, this gap could cause nearly 2 million excess errors in triage. Our findings highlight a critical safety blind spot in LLM-based health systems: models that appear to understand romanized input may still fail to act on it reliably. --- ### Script Gap: Evaluating LLM Triage on Indian Languages in Native vs Roman Scripts in a Real World Setting Authors: Manurag Khullar, Utkarsh Desai, Poorva Malviya, Aman Dalmia, Zheyuan Ryan Shi Published: 2025-12-11 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2512.10780v1 Large Language Models (LLMs) are increasingly deployed in high-stakes clinical applications in India. In many such settings, speakers of Indian languages frequently communicate using romanized text rather than native scripts, yet existing research rarely evaluates this orthographic variation using real-world data. We investigate how romanization impacts the reliability of LLMs in a critical domain: maternal and newborn healthcare triage. We benchmark leading LLMs on a real-world dataset of user-generated queries spanning five Indian languages and Nepali. Our results reveal consistent degradation in performance for romanized messages, with F1 scores trailing those of native scripts by 5-12 points. At our partner maternal health organization in India, this gap could cause nearly 2 million excess errors in triage. Crucially, this performance gap by scripts is not due to a failure in clinical reasoning. We demonstrate that LLMs often correctly infer the semantic intent of romanized queries. Nevertheless, their final classification outputs remain brittle in the presence of orthographic noise in romanized inputs. Our findings highlight a critical safety blind spot in LLM-based health systems: models that appear to understand romanized input may still fail to act on it reliably. --- ### Grow Up and Merge: Scaling Strategies for Efficient Language Adaptation Authors: Kevin Glocker, Kätriin Kukk, Romina Oji, Marcel Bollmann, Marco Kuhlmann, Jenny Kunz Published: 2025-12-11 Categories: cs.CL, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2512.10772v1 Achieving high-performing language models which include medium- and lower-resource languages remains a challenge. Massively multilingual models still underperform compared to language-specific adaptations, especially at smaller model scales. In this work, we investigate scaling as an efficient strategy for adapting pretrained models to new target languages. Through comprehensive scaling ablations with approximately FLOP-matched models, we test whether upscaling an English base model enables more effective and resource-efficient adaptation than standard continued pretraining. We find that, once exposed to sufficient target-language data, larger upscaled models can match or surpass the performance of smaller models continually pretrained on much more data, demonstrating the benefits of scaling for data efficiency. Scaling also helps preserve the base model's capabilities in English, thus reducing catastrophic forgetting. Finally, we explore whether such scaled, language-specific models can be merged to construct modular and flexible multilingual systems. We find that while merging remains less effective than joint multilingual training, upscaled merges perform better than smaller ones. We observe large performance differences across merging methods, suggesting potential for improvement through merging approaches specialized for language-level integration. --- ### Template-Free Retrosynthesis with Graph-Prior Augmented Transformers Authors: Youjun Zhao Published: 2025-12-11 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.10770v2 Retrosynthesis reaction prediction aims to infer plausible reactant molecules for a given product and is a important problem in computer-aided organic synthesis. Despite recent progress, many existing models still fall short of the accuracy and robustness required for practical deployment. In this paper, we present a template-free, Transformer-based framework that removes the need for handcrafted reaction templates or additional chemical rule engines. Our model injects molecular graph information into the attention mechanism to jointly exploit SMILES sequences and structural cues, and further applies a paired data augmentation strategy to enhance training diversity and scale. Extensive experiments on the USPTO-50K benchmark demonstrate that our approach achieves state-of-the-art performance among template-free methods and substantially outperforms a vanilla Transformer baseline. --- ### Template-Free Retrosynthesis with Graph-Prior Augmented Transformers Authors: Youjun Zhao Published: 2025-12-11 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.10770v1 Retrosynthesis reaction prediction seeks to infer plausible reactant molecules for a given product and is a central problem in computer-aided organic synthesis. Despite recent progress, many existing models still fall short of the accuracy and robustness required for practical deployment. This work studies a template-free, Transformer-based framework that eliminates reliance on handcrafted reaction templates or additional chemical rule engines. The model injects molecular graph information into the attention mechanism to jointly exploit \SMILES\ sequences and structural cues, and further applies a paired data augmentation strategy to enhance training diversity and scale. On the USPTO-50K benchmark, our proposed approach achieves state-of-the-art performance among template-free methods and substantially outperforming a vanilla Transformer baseline. --- ### Blood Pressure Prediction for Coronary Artery Disease Diagnosis using Coronary Computed Tomography Angiography Authors: Rene Lisasi, Michele Esposito, Chen Zhao Published: 2025-12-11 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.10765v1 Computational fluid dynamics (CFD) based simulation of coronary blood flow provides valuable hemodynamic markers, such as pressure gradients, for diagnosing coronary artery disease (CAD). However, CFD is computationally expensive, time-consuming, and difficult to integrate into large-scale clinical workflows. These limitations restrict the availability of labeled hemodynamic data for training AI models and hinder broad adoption of non-invasive, physiology based CAD assessment. To address these challenges, we develop an end to end pipeline that automates coronary geometry extraction from coronary computed tomography angiography (CCTA), streamlines simulation data generation, and enables efficient learning of coronary blood pressure distributions. The pipeline reduces the manual burden associated with traditional CFD workflows while producing consistent training data. We further introduce a diffusion-based regression model designed to predict coronary blood pressure directly from CCTA derived features, bypassing the need for slow CFD computation during inference. Evaluated on a dataset of simulated coronary hemodynamics, the proposed model achieves state of the art performance, with an R2 of 64.42%, a root mean squared error of 0.0974, and a normalized RMSE of 0.154, outperforming several baseline approaches. This work provides a scalable and accessible framework for rapid, non-invasive blood pressure prediction to support CAD diagnosis. --- ### Designing AI-Resilient Assessments Using Interconnected Problems: A Theoretically Grounded and Empirically Validated Framework Authors: Kaihua Ding Published: 2025-12-11 Categories: cs.CY, cs.AI arXiv: https://arxiv.org/abs/2512.10758v3 The proliferation of generative AI tools has rendered traditional modular assessments in computing and data-centric education increasingly ineffective, creating a disconnect between academic evaluation and authentic skill measurement. This paper presents a theoretically grounded framework for designing AI-resilient assessments, supported by formal analysis and empirical validation. We make three primary contributions. First, we establish two formal propositions. (1) Assessments composed of interconnected problems, in which outputs serve as inputs to subsequent tasks, are inherently more AI-resilient than modular assessments due to their reliance on multi-step reasoning and sustained context. (2) Semi-structured problems with deterministic success criteria provide more reliable measures of student competency than fully open-ended projects, which allow AI systems to default to familiar solution templates. These results challenge widely cited recommendations in recent institutional and policy guidance that promote open-ended assessments as inherently more robust to AI assistance. Second, we validate these propositions through empirical analysis of three university data science courses (N = 117). We observe a substantial AI inflation effect: students achieve near-perfect scores on AI-assisted modular homework, while performance drops by approximately 30 percentage points on proctored exams (Cohen d = 1.51). In contrast, interconnected projects remain strongly aligned with modular assessments (r = 0.954, p < 0.001) while maintaining AI resistance, whereas proctored exams show weaker alignment (r = 0.726, p < 0.001). Third, we translate these findings into a practical assessment design procedure that enables educators to construct evaluations that promote deeper engagement, reflect industry practice, and resist trivial AI delegation. --- ### Designing AI-Resilient Assessments Using Interconnected Problems: A Theoretically Grounded and Empirically Validated Framework Authors: Kaihua Ding Published: 2025-12-11 Categories: cs.CY, cs.AI arXiv: https://arxiv.org/abs/2512.10758v1 The rapid adoption of generative AI has undermined traditional modular assessments in computing education, creating a disconnect between academic evaluation and industry practice. This paper presents a theoretically grounded framework for designing AI-resilient assessments, supported by formal analysis and multi-year empirical validation. We make three contributions. First, we establish two theoretical results: (1) assessments composed of interconnected problems, where outputs feed into subsequent stages, are more AI-resilient than modular assessments because current language models struggle with sustained multi-step reasoning and context; and (2) semi-structured problems with deterministic success criteria provide more reliable measures of student competency than fully open-ended projects, which allow AI systems to default to familiar solution patterns. These results challenge common policy and institutional guidance that promotes open-ended assessments as the primary safeguard for academic integrity. Second, we validate these results using data from four university data science courses (N = 138). While students achieve near-perfect scores on AI-assisted modular homework, performance drops by roughly 30 percentage points on proctored exams, indicating substantial AI score inflation. Interconnected projects remain strongly correlated with modular assessments, suggesting they measure the same underlying skills while resisting AI misuse. Proctored exams show weaker alignment, implying they may assess test-taking ability rather than intended learning outcomes. Third, we translate these findings into a practical assessment design framework. The proposed approach enables educators to create assessments that promote integrative thinking, reflect real-world AI-augmented workflows, and naturally resist trivial delegation to generative AI, thereby helping restore academic integrity. --- ### OPV: Outcome-based Process Verifier for Efficient Long Chain-of-Thought Verification Authors: Zijian Wu, Lingkai Kong, Wenwei Zhang, Songyang Gao, Yuzhe Gu, Zhongrui Cai, Tianyou Ma, Yuhong Liu, Zhi Wang, Runyuan Ma, Guangyu Wang, Wei Li et al. Published: 2025-12-11 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2512.10756v1 Large language models (LLMs) have achieved significant progress in solving complex reasoning tasks by Reinforcement Learning with Verifiable Rewards (RLVR). This advancement is also inseparable from the oversight automated by reliable verifiers. However, current outcome-based verifiers (OVs) are unable to inspect the unreliable intermediate steps in the long reasoning chains of thought (CoTs). Meanwhile, current process-based verifiers (PVs) have difficulties in reliably detecting errors in the complex long CoTs, limited by the scarcity of high-quality annotations due to the prohibitive costs of human annotations. Therefore, we propose the Outcome-based Process Verifier (OPV), which verifies the rationale process of summarized outcomes from long CoTs to achieve both accurate and efficient verification and enable large-scale annotation. To empower the proposed verifier, we adopt an iterative active learning framework with expert annotations to progressively improve the verification capability of OPV with fewer annotation costs. Specifically, in each iteration, the most uncertain cases of the current best OPV are annotated and then subsequently used to train a new OPV through Rejection Fine-Tuning (RFT) and RLVR for the next round. Extensive experiments demonstrate OPV's superior performance and broad applicability. It achieves new state-of-the-art results on our held-out OPV-Bench, outperforming much larger open-source models such as Qwen3-Max-Preview with an F1 score of 83.1 compared to 76.3. Furthermore, OPV effectively detects false positives within synthetic dataset, closely align with expert assessment. When collaborating with policy models, OPV consistently yields performance gains, e.g., raising the accuracy of DeepSeek-R1-Distill-Qwen-32B from 55.2% to 73.3% on AIME2025 as the compute budget scales. --- ### LDP: Parameter-Efficient Fine-Tuning of Multimodal LLM for Medical Report Generation Authors: Tianyu Zhou, Junyi Tang, Zehui Li, Dahong Qian, Suncheng Xiang Published: 2025-12-11 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.10750v1 Colonoscopic polyp diagnosis is pivotal for early colorectal cancer detection, yet traditional automated reporting suffers from inconsistencies and hallucinations due to the scarcity of high-quality multimodal medical data. To bridge this gap, we propose LDP, a novel framework leveraging multimodal large language models (MLLMs) for professional polyp diagnosis report generation. Specifically, we curate MMEndo, a multimodal endoscopic dataset comprising expert-annotated colonoscopy image-text pairs. We fine-tune the Qwen2-VL-7B backbone using Parameter-Efficient Fine-Tuning (LoRA) and align it with clinical standards via Direct Preference Optimization (DPO). Extensive experiments show that our LDP outperforms existing baselines on both automated metrics and rigorous clinical expert evaluations (achieving a Physician Score of 7.2/10), significantly reducing training computational costs by 833x compared to full fine-tuning. The proposed solution offers a scalable, clinically viable path for primary healthcare, with additional validation on the IU-XRay dataset confirming its robustness. --- ### PMB-NN: Physiology-Centred Hybrid AI for Personalized Hemodynamic Monitoring from Photoplethysmography Authors: Yaowen Zhang, Libera Fresiello, Peter H. Veltink, Dirk W. Donker, Ying Wang Published: 2025-12-11 Categories: physics.med-ph, cs.LG arXiv: https://arxiv.org/abs/2512.10745v1 Continuous monitoring of blood pressure (BP) and hemodynamic parameters such as peripheral resistance (R) and arterial compliance (C) are critical for early vascular dysfunction detection. While photoplethysmography (PPG) wearables has gained popularity, existing data-driven methods for BP estimation lack interpretability. We advanced our previously proposed physiology-centered hybrid AI method-Physiological Model-Based Neural Network (PMB-NN)-in blood pressure estimation, that unifies deep learning with a 2-element Windkessel based model parameterized by R and C acting as physics constraints. The PMB-NN model was trained in a subject-specific manner using PPG-derived timing features, while demographic information was used to infer an intermediate variable: cardiac output. We validated the model on 10 healthy adults performing static and cycling activities across two days for model's day-to-day robustness, benchmarked against deep learning (DL) models (FCNN, CNN-LSTM, Transformer) and standalone Windkessel based physiological model (PM). Validation was conducted on three perspectives: accuracy, interpretability and plausibility. PMB-NN achieved systolic BP accuracy (MAE: 7.2 mmHg) comparable to DL benchmarks, diastolic performance (MAE: 3.9 mmHg) lower than DL models. However, PMB-NN exhibited higher physiological plausibility than both DL baselines and PM, suggesting that the hybrid architecture unifies and enhances the respective merits of physiological principles and data-driven techniques. Beyond BP, PMB-NN identified R (ME: 0.15 mmHg$\cdot$s/ml) and C (ME: -0.35 ml/mmHg) during training with accuracy similar to PM, demonstrating that the embedded physiological constraints confer interpretability to the hybrid AI framework. These results position PMB-NN as a balanced, physiologically grounded alternative to purely data-driven approaches for daily hemodynamic monitoring. --- ### TRIDENT: A Redundant Architecture for Caribbean-Accented Emergency Speech Triage Authors: Elroy Galbraith, Chadwick Sutherland, Donahue Morgan Published: 2025-12-11 Categories: cs.CL arXiv: https://arxiv.org/abs/2512.10741v1 Emergency speech recognition systems exhibit systematic performance degradation on non-standard English varieties, creating a critical gap in services for Caribbean populations. We present TRIDENT (Transcription and Routing Intelligence for Dispatcher-Empowered National Triage), a three-layer dispatcher-support architecture designed to structure emergency call inputs for human application of established triage protocols (the ESI for routine operations and START for mass casualty events), even when automatic speech recognition fails. The system combines Caribbean-accent-tuned ASR, local entity extraction via large language models, and bio-acoustic distress detection to provide dispatchers with three complementary signals: transcription confidence, structured clinical entities, and vocal stress indicators. Our key insight is that low ASR confidence, rather than representing system failure, serves as a valuable queue prioritization signal -- particularly when combined with elevated vocal distress markers indicating a caller in crisis whose speech may have shifted toward basilectal registers. A complementary insight drives the entity extraction layer: trained responders and composed bystanders may report life-threatening emergencies without elevated vocal stress, requiring semantic analysis to capture clinical indicators that paralinguistic features miss. We describe the architectural design, theoretical grounding in psycholinguistic research on stress-induced code-switching, and deployment considerations for offline operation during disaster scenarios. This work establishes a framework for accent-resilient emergency AI that ensures Caribbean voices receive equitable access to established national triage protocols. Empirical validation on Caribbean emergency calls remains future work. --- ### Intern-S1-MO: Long-horizon Reasoning Agent for Olympiad?Level Mathematical Problem Solving Authors: Yuzhe Gu, Songyang Gao, Zijian Wu, Lingkai Kong, Wenwei Zhang, Zhongrui Cai, Fan Zheng, Tianyou Ma, Junhao Shen, Haiteng Zhao, Duanyang Zhang, Huilun Zhang et al. Published: 2025-12-11 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2512.10739v3 Large Reasoning Models (LRMs) have expanded the mathematical reasoning frontier through Chain-of-Thought (CoT) techniques and Reinforcement Learning with Verifiable Rewards (RLVR), capable of solving AIME-level problems. However, the performance of LRMs is heavily dependent on the extended reasoning context length. For solving ultra-hard problems like those in the International Mathematical Olympiad (IMO), the required reasoning complexity surpasses the space that an LRM can explore in a single round. Previous works attempt to extend the reasoning context of LRMs but remain prompt-based and built upon proprietary models, lacking systematic structures and training pipelines. Therefore, this paper introduces Intern-S1-MO, a long-horizon math agent that conducts multi-round hierarchical reasoning, composed of an LRM-based multi-agent system including reasoning, summary, and verification. By maintaining a compact memory in the form of lemmas, Intern-S1-MO can more freely explore the lemma-rich reasoning spaces in multiple reasoning stages, thereby breaking through the context constraints for IMO-level math problems. Furthermore, we propose OREAL-H, an RL framework for training the LRM using the online explored trajectories to simultaneously bootstrap the reasoning ability of LRM and elevate the overall performance of Intern-S1-MO. Experiments show that Intern-S1-MO can obtain 26 out of 35 points on the non-geometry problems of IMO2025, matching the performance of silver medalists. It also surpasses the current advanced LRMs on inference benchmarks such as HMMT2025, AIME2025, and CNMO2025. In addition, our agent officially participates in CMO2025 and achieves a score of 102/126 under the judgment of human experts, reaching the gold medal level. --- ### Long-horizon Reasoning Agent for Olympiad-Level Mathematical Problem Solving Authors: Songyang Gao, Yuzhe Gu, Zijian Wu, Lingkai Kong, Wenwei Zhang, Zhongrui Cai, Fan Zheng, Tianyou Ma, Junhao Shen, Haiteng Zhao, Duanyang Zhang, Huilun Zhang et al. Published: 2025-12-11 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2512.10739v1 Large language models (LLMs) have achieved significant progress in solving complex reasoning tasks by Reinforcement Learning with Verifiable Rewards (RLVR). This advancement is also inseparable from the oversight automated by reliable verifiers. However, current outcome-based verifiers (OVs) are unable to inspect the unreliable intermediate steps in the long reasoning chains of thought (CoTs). Meanwhile, current process-based verifiers (PVs) have difficulties in reliably detecting errors in the complex long CoTs, limited by the scarcity of high-quality annotations due to the prohibitive costs of human annotations. Therefore, we propose the \textbf{O}utcome-based \textbf{P}rocess \textbf{V}erifier (OPV), which verifies the rationale process of summarized outcomes from long CoTs to achieve both accurate and efficient verification and enable large-scale annotation. To empower the proposed verifier, we adopt an iterative active learning framework with expert annotations to progressively improve the verification capability of OPV with fewer annotation costs. Specifically, in each iteration, the most uncertain cases of the current best OPV are annotated and then subsequently used to train a new OPV through Rejection Fine-Tuning (RFT) and RLVR for the next round. Extensive experiments demonstrate OPV's superior performance and broad applicability. It achieves new state-of-the-art results on our held-out \textsc{\thisbench}, outperforming much larger open-source models such as Qwen3-Max-Preview with an F1 score of 83.1 compared to 76.3. Furthermore, OPV effectively detects false positives within synthetic dataset, closely align with expert assessment. When collaborating with policy models, OPV consistently yields performance gains, e.g., raising the accuracy of DeepSeek-R1-Distill-Qwen-32B from 55.2\% to 73.3\% on AIME2025 as the compute budget scales. --- ### LGAN: An Efficient High-Order Graph Neural Network via the Line Graph Aggregation Authors: Lin Du, Lu Bai, Jincheng Li, Lixin Cui, Hangyuan Du, Lichi Zhang, Yuting Chen, Zhao Li Published: 2025-12-11 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2512.10735v1 Graph Neural Networks (GNNs) have emerged as a dominant paradigm for graph classification. Specifically, most existing GNNs mainly rely on the message passing strategy between neighbor nodes, where the expressivity is limited by the 1-dimensional Weisfeiler-Lehman (1-WL) test. Although a number of k-WL-based GNNs have been proposed to overcome this limitation, their computational cost increases rapidly with k, significantly restricting the practical applicability. Moreover, since the k-WL models mainly operate on node tuples, these k-WL-based GNNs cannot retain fine-grained node- or edge-level semantics required by attribution methods (e.g., Integrated Gradients), leading to the less interpretable problem. To overcome the above shortcomings, in this paper, we propose a novel Line Graph Aggregation Network (LGAN), that constructs a line graph from the induced subgraph centered at each node to perform the higher-order aggregation. We theoretically prove that the LGAN not only possesses the greater expressive power than the 2-WL under injective aggregation assumptions, but also has lower time complexity. Empirical evaluations on benchmarks demonstrate that the LGAN outperforms state-of-the-art k-WL-based GNNs, while offering better interpretability. --- ### Textual Data Bias Detection and Mitigation -- An Extensible Pipeline with Experimental Evaluation Authors: Rebekka Görge, Sujan Sai Gannamaneni, Tabea Naeven, Hammam Abdelwahab, Héctor Allende-Cid, Armin B. Cremers, Lennard Helmer, Michael Mock, Anna Schmitz, Songkai Xue, Elif Yildirir, Maximilian Poretschkin et al. Published: 2025-12-11 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2512.10734v2 Textual data used to train large language models (LLMs) exhibits multifaceted bias manifestations encompassing harmful language and skewed demographic distributions. Regulations such as the European AI Act require identifying and mitigating biases against protected groups in data, with the ultimate goal of preventing unfair model outputs. However, practical guidance and operationalization are lacking. We propose a comprehensive data bias detection and mitigation pipeline comprising four components that address two data bias types, namely representation bias and (explicit) stereotypes for a configurable sensitive attribute. First, we leverage LLM-generated word lists created based on quality criteria to detect relevant group labels. Second, representation bias is quantified using the Demographic Representation Score. Third, we detect and mitigate stereotypes using sociolinguistically informed filtering. Finally, we compensate representation bias through Grammar- and Context-Aware Counterfactual Data Augmentation. We conduct a two-fold evaluation using the examples of gender, religion and age. First, the effectiveness of each individual component on data debiasing is evaluated through human validation and baseline comparison. The findings demonstrate that we successfully reduce representation bias and (explicit) stereotypes in a text dataset. Second, the effect of data debiasing on model bias reduction is evaluated by bias benchmarking of several models (0.6B-8B parameters), fine-tuned on the debiased text dataset. This evaluation reveals that LLMs fine-tuned on debiased data do not consistently show improved performance on bias benchmarks, exposing critical gaps in current evaluation methodologies and highlighting the need for targeted data manipulation to address manifested model bias. --- ### Textual Data Bias Detection and Mitigation - An Extensible Pipeline with Experimental Evaluation Authors: Rebekka Görge, Sujan Sai Gannamaneni, Tabea Naeven, Hammam Abdelwahab, Héctor Allende-Cid, Armin B. Cremers, Lennard Helmer, Michael Mock, Anna Schmitz, Songkai Xue, Elif Yildirir, Maximilian Poretschkin et al. Published: 2025-12-11 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2512.10734v1 Textual data used to train large language models (LLMs) exhibits multifaceted bias manifestations encompassing harmful language and skewed demographic distributions. Regulations such as the European AI Act require identifying and mitigating biases against protected groups in data, with the ultimate goal of preventing unfair model outputs. However, practical guidance and operationalization are lacking. We propose a comprehensive data bias detection and mitigation pipeline comprising four components that address two data bias types, namely representation bias and (explicit) stereotypes for a configurable sensitive attribute. First, we leverage LLM-generated word lists created based on quality criteria to detect relevant group labels. Second, representation bias is quantified using the Demographic Representation Score. Third, we detect and mitigate stereotypes using sociolinguistically informed filtering. Finally, we compensate representation bias through Grammar- and Context-Aware Counterfactual Data Augmentation. We conduct a two-fold evaluation using the examples of gender, religion and age. First, the effectiveness of each individual component on data debiasing is evaluated through human validation and baseline comparison. The findings demonstrate that we successfully reduce representation bias and (explicit) stereotypes in a text dataset. Second, the effect of data debiasing on model bias reduction is evaluated by bias benchmarking of several models (0.6B-8B parameters), fine-tuned on the debiased text dataset. This evaluation reveals that LLMs fine-tuned on debiased data do not consistently show improved performance on bias benchmarks, exposing critical gaps in current evaluation methodologies and highlighting the need for targeted data manipulation to address manifested model bias. --- ### IRG-MotionLLM: Interleaving Motion Generation, Assessment and Refinement for Text-to-Motion Generation Authors: Yuan-Ming Li, Qize Yang, Nan Lei, Shenghao Fu, Ling-An Zeng, Jian-Fang Hu, Xihan Wei, Wei-Shi Zheng Published: 2025-12-11 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.10730v2 Recent advances in motion-aware large language models have shown remarkable promise for jointly learning motion understanding and generation knowledge. However, these models typically treat understanding and generation separately, limiting the mutual benefits that could arise from interactive feedback between tasks. In this work, we reveal that motion assessment and refinement tasks can act as crucial bridges to enable knowledge flow from motion understanding to generation. Specifically, we propose Interleaved Reasoning for Motion Generation (IRMoGen), a novel paradigm that tightly couples motion generation with assessment and refinement through iterative text-motion dialogue. To realize this, we introduce IRG-MotionLLM, the first model that seamlessly interleaves motion generation, assessment, and refinement to improve the alignment between generated motion and goal text. IRG-MotionLLM is developed progressively with a novel three-stage training scheme, initializing and subsequently enhancing native IRMoGen capabilities. To facilitate this development, we construct an automated data engine to synthesize interleaved reasoning annotations from existing text-motion datasets. Extensive experiments demonstrate the properties brought by IRMoGen training, and the advanced cross-benchmark and cross-evaluator performance of IRG-MotionLLM. Code and models are available at https://github.com/HumanMLLM/IRG-MotionLLM. --- ### IRG-MotionLLM: Interleaving Motion Generation, Assessment and Refinement for Text-to-Motion Generation Authors: Yuan-Ming Li, Qize Yang, Nan Lei, Shenghao Fu, Ling-An Zeng, Jian-Fang Hu, Xihan Wei, Wei-Shi Zheng Published: 2025-12-11 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.10730v1 Recent advances in motion-aware large language models have shown remarkable promise for unifying motion understanding and generation tasks. However, these models typically treat understanding and generation separately, limiting the mutual benefits that could arise from interactive feedback between tasks. In this work, we reveal that motion assessment and refinement tasks act as crucial bridges to enable bidirectional knowledge flow between understanding and generation. Leveraging this insight, we propose Interleaved Reasoning for Motion Generation (IRMoGen), a novel paradigm that tightly couples motion generation with assessment and refinement through iterative text-motion dialogue. To realize this, we introduce IRG-MotionLLM, the first model that seamlessly interleaves motion generation, assessment, and refinement to improve generation performance. IRG-MotionLLM is developed progressively with a novel three-stage training scheme, initializing and subsequently enhancing native IRMoGen capabilities. To facilitate this development, we construct an automated data engine to synthesize interleaved reasoning annotations from existing text-motion datasets. Extensive experiments demonstrate that: (i) Assessment and refinement tasks significantly improve text-motion alignment; (ii) Interleaving motion generation, assessment, and refinement steps yields consistent performance gains across training stages; and (iii) IRG-MotionLLM clearly outperforms the baseline model and achieves advanced performance on standard text-to-motion generation benchmarks. Cross-evaluator testing further validates its effectiveness. Code & Data: https://github.com/HumanMLLM/IRG-MotionLLM/tree/main. --- ### Video Depth Propagation Authors: Luigi Piccinelli, Thiemo Wandel, Christos Sakaridis, Wim Abbeloos, Luc Van Gool Published: 2025-12-11 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.10725v1 Depth estimation in videos is essential for visual perception in real-world applications. However, existing methods either rely on simple frame-by-frame monocular models, leading to temporal inconsistencies and inaccuracies, or use computationally demanding temporal modeling, unsuitable for real-time applications. These limitations significantly restrict general applicability and performance in practical settings. To address this, we propose VeloDepth, an efficient and robust online video depth estimation pipeline that effectively leverages spatiotemporal priors from previous depth predictions and performs deep feature propagation. Our method introduces a novel Propagation Module that refines and propagates depth features and predictions using flow-based warping coupled with learned residual corrections. In addition, our design structurally enforces temporal consistency, resulting in stable depth predictions across consecutive frames with improved efficiency. Comprehensive zero-shot evaluation on multiple benchmarks demonstrates the state-of-the-art temporal consistency and competitive accuracy of VeloDepth, alongside its significantly faster inference compared to existing video-based depth estimators. VeloDepth thus provides a practical, efficient, and accurate solution for real-time depth estimation suitable for diverse perception tasks. Code and models are available at https://github.com/lpiccinelli-eth/velodepth --- ### Generalized Spherical Neural Operators: Green's Function Formulation Authors: Hao Tang, Hao Chen, Chao Li Published: 2025-12-11 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.10723v3 Neural operators offer powerful approaches for solving parametric partial differential equations, but extending them to spherical domains remains challenging due to the need to preserve intrinsic geometry while avoiding distortions that break rotational consistency. Existing spherical operators rely on rotational equivariance but often lack the flexibility for real-world complexity. We propose a generalized operator-design framework based on the designable spherical Green's function and its harmonic expansion, establishing a solid operator-theoretic foundation for spherical learning. Based on this, we propose an absolute and relative position-dependent Green's function that enables flexible balance of equivariance and invariance for real-world modeling. The resulting operator, Green's-function Spherical Neural Operator (GSNO) with a novel spectral learning method, can adapt to non-equivariant systems while retaining spectral efficiency and grid invariance. To exploit GSNO, we develop SHNet, a hierarchical architecture that combines multi-scale spectral modeling with spherical up-down sampling, enhancing global feature representation. Evaluations on diffusion MRI, shallow water dynamics, and global weather forecasting, GSNO and SHNet consistently outperform state-of-the-art methods. The theoretical and experimental results position GSNO as a principled and generalized framework for spherical operator design and learning, bridging rigorous theory with real-world complexity. The code is available at: https://github.com/haot2025/GSNO. --- ### Generalized Spherical Neural Operators: Green's Function Formulation Authors: Hao Tang, Hao Chen, Chao Li Published: 2025-12-11 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.10723v1 Neural operators offer powerful approaches for solving parametric partial differential equations, but extending them to spherical domains remains challenging due to the need to preserve intrinsic geometry while avoiding distortions that break rotational consistency. Existing spherical operators rely on rotational equivariance but often lack the flexibility for real-world complexity. We propose a general operator-design framework based on the designable spherical Green's function and its harmonic expansion, establishing a solid operator-theoretic foundation for spherical learning. Based on this, we propose an absolute and relative position-dependent Green's function that enables flexible balance of equivariance and invariance for real-world modeling. The resulting operator, Green's-function Spherical Neural Operator (GSNO) with a novel spectral learning method, can adapt to anisotropic, constraint-rich systems while retaining spectral efficiency. To exploit GSNO, we develop GSHNet, a hierarchical architecture that combines multi-scale spectral modeling with spherical up-down sampling, enhancing global feature representation. Evaluations on diffusion MRI, shallow water dynamics, and global weather forecasting, GSNO and GSHNet consistently outperform state-of-the-art methods. Our results position GSNO as a principled and general framework for spherical operator learning, bridging rigorous theory with real-world complexity. --- ### Beyond the Black Box: Identifiable Interpretation and Control in Generative Models via Causal Minimality Authors: Lingjing Kong, Shaoan Xie, Guangyi Chen, Yuewen Sun, Xiangchen Song, Eric P. Xing, Kun Zhang Published: 2025-12-11 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.10720v2 Deep generative models, while revolutionizing fields like image and text generation, largely operate as opaque ``black boxes'', hindering human understanding, control, and alignment. While methods like sparse autoencoders (SAEs) show remarkable empirical success, they often lack theoretical guarantees, risking subjective insights. Our primary objective is to establish a principled foundation for interpretable generative models. We demonstrate that the principle of causal minimality -- favoring the simplest causal explanation -- can endow the latent representations of modern generative models with clear causal interpretation and robust, component-wise identifiable control. We introduce a novel theoretical framework for hierarchical selection models, where higher-level concepts emerge from the constrained composition of lower-level variables, better capturing the complex dependencies in data generation. Under theoretically derived minimality conditions, we show that learned representations can be equivalent to the true latent variables of the data-generating process. Empirically, applying these constraints to leading text-to-image diffusion models allows us to extract their innate hierarchical concept graphs, offering fresh insights into their internal knowledge organization. Furthermore, these causally grounded concepts serve as levers for fine-grained model steering, paving the way for transparent, reliable systems. --- ### Beyond the Black Box: Identifiable Interpretation and Control in Generative Models via Causal Minimality Authors: Lingjing Kong, Shaoan Xie, Guangyi Chen, Yuewen Sun, Xiangchen Song, Eric P. Xing, Kun Zhang Published: 2025-12-11 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.10720v1 Deep generative models, while revolutionizing fields like image and text generation, largely operate as opaque black boxes, hindering human understanding, control, and alignment. While methods like sparse autoencoders (SAEs) show remarkable empirical success, they often lack theoretical guarantees, risking subjective insights. Our primary objective is to establish a principled foundation for interpretable generative models. We demonstrate that the principle of causal minimality -- favoring the simplest causal explanation -- can endow the latent representations of diffusion vision and autoregressive language models with clear causal interpretation and robust, component-wise identifiable control. We introduce a novel theoretical framework for hierarchical selection models, where higher-level concepts emerge from the constrained composition of lower-level variables, better capturing the complex dependencies in data generation. Under theoretically derived minimality conditions (manifesting as sparsity or compression constraints), we show that learned representations can be equivalent to the true latent variables of the data-generating process. Empirically, applying these constraints to leading generative models allows us to extract their innate hierarchical concept graphs, offering fresh insights into their internal knowledge organization. Furthermore, these causally grounded concepts serve as levers for fine-grained model steering, paving the way for transparent, reliable systems. --- ### SpaceDrive: Infusing Spatial Awareness into VLM-based Autonomous Driving Authors: Peizheng Li, Zhenghao Zhang, David Holtz, Hang Yu, Yutong Yang, Yuzhi Lai, Rui Song, Andreas Geiger, Andreas Zell Published: 2025-12-11 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.10719v3 End-to-end autonomous driving methods built on vision language models (VLMs) have undergone rapid development driven by their universal visual understanding and strong reasoning capabilities obtained from the large-scale pretraining. However, we find that current VLMs struggle to understand fine-grained 3D spatial relationships which is a fundamental requirement for systems interacting with the physical world. To address this issue, we propose SpaceDrive, a spatial-aware VLM-based driving framework that treats spatial information as explicit positional encodings (PEs) instead of textual digit tokens, enabling joint reasoning over semantic and spatial representations. SpaceDrive employs a universal positional encoder to all 3D coordinates derived from multi-view depth estimation, historical ego-states, and text prompts. These 3D PEs are first superimposed to augment the corresponding 2D visual tokens. Meanwhile, they serve as a task-agnostic coordinate representation, replacing the digit-wise numerical tokens as both inputs and outputs for the VLM. This mechanism enables the model to better index specific visual semantics in spatial reasoning and directly regress trajectory coordinates rather than generating digit-by-digit, thereby enhancing planning accuracy. Extensive experiments validate that SpaceDrive achieves state-of-the-art open-loop performance on the nuScenes dataset and the second-best Driving Score of 78.02 on the Bench2Drive closed-loop benchmark over existing VLM-based methods. Code is available at: https://github.com/zhenghao2519/SpaceDrive. --- ### SpaceDrive: Infusing Spatial Awareness into VLM-based Autonomous Driving Authors: Peizheng Li, Zhenghao Zhang, David Holtz, Hang Yu, Yutong Yang, Yuzhi Lai, Rui Song, Andreas Geiger, Andreas Zell Published: 2025-12-11 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.10719v1 End-to-end autonomous driving methods built on vision language models (VLMs) have undergone rapid development driven by their universal visual understanding and strong reasoning capabilities obtained from the large-scale pretraining. However, we find that current VLMs struggle to understand fine-grained 3D spatial relationships which is a fundamental requirement for systems interacting with the physical world. To address this issue, we propose SpaceDrive, a spatial-aware VLM-based driving framework that treats spatial information as explicit positional encodings (PEs) instead of textual digit tokens, enabling joint reasoning over semantic and spatial representations. SpaceDrive employs a universal positional encoder to all 3D coordinates derived from multi-view depth estimation, historical ego-states, and text prompts. These 3D PEs are first superimposed to augment the corresponding 2D visual tokens. Meanwhile, they serve as a task-agnostic coordinate representation, replacing the digit-wise numerical tokens as both inputs and outputs for the VLM. This mechanism enables the model to better index specific visual semantics in spatial reasoning and directly regress trajectory coordinates rather than generating digit-by-digit, thereby enhancing planning accuracy. Extensive experiments validate that SpaceDrive achieves state-of-the-art open-loop performance on the nuScenes dataset and the second-best Driving Score of 78.02 on the Bench2Drive closed-loop benchmark over existing VLM-based methods. --- ### CheXmask-U: Quantifying uncertainty in landmark-based anatomical segmentation for X-ray images Authors: Matias Cosarinsky, Nicolas Gaggion, Rodrigo Echeveste, Enzo Ferrante Published: 2025-12-11 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.10715v2 In this work, we study uncertainty estimation for anatomical landmark-based segmentation on chest X-rays. Inspired by hybrid neural network architectures that combine standard image convolutional encoders with graph-based generative decoders, and leveraging their variational latent space, we derive two complementary measures: (i) latent uncertainty, captured directly from the learned distribution parameters, and (ii) predictive uncertainty, obtained by generating multiple stochastic output predictions from latent samples. Through controlled corruption experiments we show that both uncertainty measures increase with perturbation severity, reflecting both global and local degradation. We demonstrate that these uncertainty signals can identify unreliable predictions by comparing with manual ground-truth, and support out-of-distribution detection on the CheXmask dataset. More importantly, we release CheXmask-U (huggingface.co/datasets/mcosarinsky/CheXmask-U), a large scale dataset of 657,566 chest X-ray landmark segmentations with per-node uncertainty estimates, enabling researchers to account for spatial variations in segmentation quality when using these anatomical masks. Our findings establish uncertainty estimation as a promising direction to enhance robustness and safe deployment of landmark-based anatomical segmentation methods in chest X-ray. A fully working interactive demo of the method is available at huggingface.co/spaces/matiasky/CheXmask-U and the source code at github.com/mcosarinsky/CheXmask-U. --- ### CheXmask-U: Quantifying uncertainty in landmark-based anatomical segmentation for X-ray images Authors: Matias Cosarinsky, Nicolas Gaggion, Rodrigo Echeveste, Enzo Ferrante Published: 2025-12-11 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.10715v1 Uncertainty estimation is essential for the safe clinical deployment of medical image segmentation systems, enabling the identification of unreliable predictions and supporting human oversight. While prior work has largely focused on pixel-level uncertainty, landmark-based segmentation offers inherent topological guarantees yet remains underexplored from an uncertainty perspective. In this work, we study uncertainty estimation for anatomical landmark-based segmentation on chest X-rays. Inspired by hybrid neural network architectures that combine standard image convolutional encoders with graph-based generative decoders, and leveraging their variational latent space, we derive two complementary measures: (i) latent uncertainty, captured directly from the learned distribution parameters, and (ii) predictive uncertainty, obtained by generating multiple stochastic output predictions from latent samples. Through controlled corruption experiments we show that both uncertainty measures increase with perturbation severity, reflecting both global and local degradation. We demonstrate that these uncertainty signals can identify unreliable predictions by comparing with manual ground-truth, and support out-of-distribution detection on the CheXmask dataset. More importantly, we release CheXmask-U (huggingface.co/datasets/mcosarinsky/CheXmask-U), a large scale dataset of 657,566 chest X-ray landmark segmentations with per-node uncertainty estimates, enabling researchers to account for spatial variations in segmentation quality when using these anatomical masks. Our findings establish uncertainty estimation as a promising direction to enhance robustness and safe deployment of landmark-based anatomical segmentation methods in chest X-ray. A fully working interactive demo of the method is available at huggingface.co/spaces/matiasky/CheXmask-U and the source code at github.com/mcosarinsky/CheXmask-U. --- ### PACIFIC: a framework for generating benchmarks to check Precise Automatically Checked Instruction Following In Code Authors: Itay Dreyfuss, Antonio Abu Nassar, Samuel Ackerman, Axel Ben David, Eitan Farchi, Rami Katan, Orna Raz, Marcel Zalmanovici Published: 2025-12-11 Categories: cs.SE, cs.AI arXiv: https://arxiv.org/abs/2512.10713v2 Large Language Model (LLM)-based code assistants have emerged as a powerful application of generative AI, demonstrating impressive capabilities in code generation and comprehension. A key requirement for these systems is their ability to accurately follow user instructions. We present Precise Automatically Checked Instruction Following In Code (PACIFIC), a novel framework designed to automatically generate benchmarks that rigorously assess sequential instruction-following and code dry-running capabilities in LLMs, while allowing control over benchmark difficulty. PACIFIC produces benchmark variants with clearly defined expected outputs, enabling straightforward and reliable evaluation through simple output comparisons. In contrast to existing approaches that often rely on tool usage or agentic behavior, our work isolates and evaluates the LLM's intrinsic ability to reason through code behavior step-by-step without execution (dry running) and to follow instructions. Furthermore, our framework mitigates training data contamination by facilitating effortless generation of novel benchmark variations. We validate our framework by generating a suite of benchmarks spanning a range of difficulty levels and evaluating multiple state-of-the-art LLMs. Our results demonstrate that PACIFIC can produce increasingly challenging benchmarks that effectively differentiate instruction-following and dry running capabilities, even among advanced models. Overall, our framework offers a scalable, contamination-resilient methodology for assessing core competencies of LLMs in code-related tasks. --- ### PACIFIC: a framework for generating benchmarks to check Precise Automatically Checked Instruction Following In Code Authors: Itay Dreyfuss, Antonio Abu Nassar, Samuel Ackerman, Axel Ben David, Rami Katan, Orna Raz, Marcel Zalmanovici Published: 2025-12-11 Categories: cs.SE, cs.AI arXiv: https://arxiv.org/abs/2512.10713v1 Large Language Model (LLM)-based code assistants have emerged as a powerful application of generative AI, demonstrating impressive capabilities in code generation and comprehension. A key requirement for these systems is their ability to accurately follow user instructions. We present Precise Automatically Checked Instruction Following In Code (PACIFIC), a novel framework designed to automatically generate benchmarks that rigorously assess sequential instruction-following and code dry-running capabilities in LLMs, while allowing control over benchmark difficulty. PACIFIC produces benchmark variants with clearly defined expected outputs, enabling straightforward and reliable evaluation through simple output comparisons. In contrast to existing approaches that often rely on tool usage or agentic behavior, our work isolates and evaluates the LLM's intrinsic ability to reason through code behavior step-by-step without execution (dry running) and to follow instructions. Furthermore, our framework mitigates training data contamination by facilitating effortless generation of novel benchmark variations. We validate our framework by generating a suite of benchmarks spanning a range of difficulty levels and evaluating multiple state-of-the-art LLMs. Our results demonstrate that PACIFIC can produce increasingly challenging benchmarks that effectively differentiate instruction-following and dry running capabilities, even among advanced models. Overall, our framework offers a scalable, contamination-resilient methodology for assessing core competencies of LLMs in code-related tasks. --- ### COMPARE: Clinical Optimization with Modular Planning and Assessment via RAG-Enhanced AI-OCT: Superior Decision Support for Percutaneous Coronary Intervention Compared to ChatGPT-5 and Junior Operators Authors: Wei Fang, Chiyao Wang, Wenshuai Ma, Hui Liu, Jianqiang Hu, Xiaona Niu, Yi Chu, Mingming Zhang, Jingxiao Yang, Dongwei Zhang, Zelin Li, Pengyun Liu et al. Published: 2025-12-11 Categories: cs.AI arXiv: https://arxiv.org/abs/2512.10702v1 Background: While intravascular imaging, particularly optical coherence tomography (OCT), improves percutaneous coronary intervention (PCI) outcomes, its interpretation is operator-dependent. General-purpose artificial intelligence (AI) shows promise but lacks domain-specific reliability. We evaluated the performance of CA-GPT, a novel large model deployed on an AI-OCT system, against that of the general-purpose ChatGPT-5 and junior physicians for OCT-guided PCI planning and assessment. Methods: In this single-center analysis of 96 patients who underwent OCT-guided PCI, the procedural decisions generated by the CA-GPT, ChatGPT-5, and junior physicians were compared with an expert-derived procedural record. Agreement was assessed using ten pre-specified metrics across pre-PCI and post-PCI phases. Results: For pre-PCI planning, CA-GPT demonstrated significantly higher median agreement scores (5(IQR 3.75-5)) compared to both ChatGPT-5 (3(2-4), P<0.001) and junior physicians (4(3-4), P<0.001). CA-GPT significantly outperformed ChatGPT-5 across all individual pre-PCI metrics and showed superior performance to junior physicians in stent diameter (90.3% vs. 72.2%, P<0.05) and length selection (80.6% vs. 52.8%, P<0.01). In post-PCI assessment, CA-GPT maintained excellent overall agreement (5(4.75-5)), significantly higher than both ChatGPT-5 (4(4-5), P<0.001) and junior physicians (5(4-5), P<0.05). Subgroup analysis confirmed CA-GPT's robust performance advantage in complex scenarios. Conclusion: The CA-GPT-based AI-OCT system achieved superior decision-making agreement versus a general-purpose large language model and junior physicians across both PCI planning and assessment phases. This approach provides a standardized and reliable method for intravascular imaging interpretation, demonstrating significant potential to augment operator expertise and optimize OCT-guided PCI. --- ### HybridVFL: Disentangled Feature Learning for Edge-Enabled Vertical Federated Multimodal Classification Authors: Mostafa Anoosha, Zeinab Dehghani, Kuniko Paxton, Koorosh Aslansefat, Dhavalkumar Thakker Published: 2025-12-11 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.10701v1 Vertical Federated Learning (VFL) offers a privacy-preserving paradigm for Edge AI scenarios like mobile health diagnostics, where sensitive multimodal data reside on distributed, resource-constrained devices. Yet, standard VFL systems often suffer performance limitations due to simplistic feature fusion. This paper introduces HybridVFL, a novel framework designed to overcome this bottleneck by employing client-side feature disentanglement paired with a server-side cross-modal transformer for context-aware fusion. Through systematic evaluation on the multimodal HAM10000 skin lesion dataset, we demonstrate that HybridVFL significantly outperforms standard federated baselines, validating the criticality of advanced fusion mechanisms in robust, privacy-preserving systems. --- ### How to Brake? Ethical Emergency Braking with Deep Reinforcement Learning Authors: Jianbo Wang, Galina Sidorenko, Johan Thunberg Published: 2025-12-11 Categories: cs.RO, cs.AI arXiv: https://arxiv.org/abs/2512.10698v1 Connected and automated vehicles (CAVs) have the potential to enhance driving safety, for example by enabling safe vehicle following and more efficient traffic scheduling. For such future deployments, safety requirements should be addressed, where the primary such are avoidance of vehicle collisions and substantial mitigating of harm when collisions are unavoidable. However, conservative worst-case-based control strategies come at the price of reduced flexibility and may compromise overall performance. In light of this, we investigate how Deep Reinforcement Learning (DRL) can be leveraged to improve safety in multi-vehicle-following scenarios involving emergency braking. Specifically, we investigate how DRL with vehicle-to-vehicle communication can be used to ethically select an emergency breaking profile in scenarios where overall, or collective, three-vehicle harm reduction or collision avoidance shall be obtained instead of single-vehicle such. As an algorithm, we provide a hybrid approach that combines DRL with a previously published method based on analytical expressions for selecting optimal constant deceleration. By combining DRL with the previous method, the proposed hybrid approach increases the reliability compared to standalone DRL, while achieving superior performance in terms of overall harm reduction and collision avoidance. --- ### Remember Me, Refine Me: A Dynamic Procedural Memory Framework for Experience-Driven Agent Evolution Authors: Zouying Cao, Jiaji Deng, Li Yu, Weikang Zhou, Zhaoyang Liu, Bolin Ding, Hai Zhao Published: 2025-12-11 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2512.10696v2 Procedural memory enables large language model (LLM) agents to internalize "how-to" knowledge, theoretically reducing redundant trial-and-error. However, existing frameworks predominantly suffer from a "passive accumulation" paradigm, treating memory as a static append-only archive. To bridge the gap between static storage and dynamic reasoning, we propose $\textbf{ReMe}$ ($\textit{Remember Me, Refine Me}$), a comprehensive framework for experience-driven agent evolution. ReMe innovates across the memory lifecycle via three mechanisms: 1) $\textit{multi-faceted distillation}$, which extracts fine-grained experiences by recognizing success patterns, analyzing failure triggers and generating comparative insights; 2) $\textit{context-adaptive reuse}$, which tailors historical insights to new contexts via scenario-aware indexing; and 3) $\textit{utility-based refinement}$, which autonomously adds valid memories and prunes outdated ones to maintain a compact, high-quality experience pool. Extensive experiments on BFCL-V3 and AppWorld demonstrate that ReMe establishes a new state-of-the-art in agent memory system. Crucially, we observe a significant memory-scaling effect: Qwen3-8B equipped with ReMe outperforms larger, memoryless Qwen3-14B, suggesting that self-evolving memory provides a computation-efficient pathway for lifelong learning. We release our code and the $\texttt{reme.library}$ dataset to facilitate further research. --- ### Remember Me, Refine Me: A Dynamic Procedural Memory Framework for Experience-Driven Agent Evolution Authors: Zouying Cao, Jiaji Deng, Li Yu, Weikang Zhou, Zhaoyang Liu, Bolin Ding, Hai Zhao Published: 2025-12-11 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2512.10696v1 Procedural memory enables large language model (LLM) agents to internalize "how-to" knowledge, theoretically reducing redundant trial-and-error. However, existing frameworks predominantly suffer from a "passive accumulation" paradigm, treating memory as a static append-only archive. To bridge the gap between static storage and dynamic reasoning, we propose $\textbf{ReMe}$ ($\textit{Remember Me, Refine Me}$), a comprehensive framework for experience-driven agent evolution. ReMe innovates across the memory lifecycle via three mechanisms: 1) $\textit{multi-faceted distillation}$, which extracts fine-grained experiences by recognizing success patterns, analyzing failure triggers and generating comparative insights; 2) $\textit{context-adaptive reuse}$, which tailors historical insights to new contexts via scenario-aware indexing; and 3) $\textit{utility-based refinement}$, which autonomously adds valid memories and prunes outdated ones to maintain a compact, high-quality experience pool. Extensive experiments on BFCL-V3 and AppWorld demonstrate that ReMe establishes a new state-of-the-art in agent memory system. Crucially, we observe a significant memory-scaling effect: Qwen3-8B equipped with ReMe outperforms larger, memoryless Qwen3-14B, suggesting that self-evolving memory provides a computation-efficient pathway for lifelong learning. We release our code and the $\texttt{reme.library}$ dataset to facilitate further research. ---