# Research Papers — 2026-02 · bundle page 102 of 210 > Full-text markdown of papers #5051–5100 of 10493 published in 2026-02, newest first. AI-readable mirrors of the full 10493-paper 2026-02 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2026-02/page/102 [← Previous bundle](https://aipod.dev/papers/archive/2026-02/page/101.md) [Next bundle →](https://aipod.dev/papers/archive/2026-02/page/103.md) ### The Appeal and Reality of Recycling LoRAs with Adaptive Merging Authors: Haokun Liu, Gyung Hyun Je, Marco Ciccone, Zhenlin Xu, Prasanth YSS, Colin Raffel Published: 2026-02-12 Categories: cs.LG, cs.SE arXiv: https://arxiv.org/abs/2602.12323v2 The widespread availability of fine-tuned LoRA modules for open pre-trained models has led to an interest in methods that can adaptively merge LoRAs to improve performance. These methods typically include some way of selecting LoRAs from a pool and tune merging coefficients based on a task-specific dataset. While adaptive merging methods have demonstrated improvements in some settings, no past work has attempted to recycle LoRAs found "in the wild" on model repositories like the Hugging Face Hub. To address this gap, we consider recycling from a pool of nearly 1,000 user-contributed LoRAs trained from the Llama 3.1 8B-Instruct language model. Our empirical study includes a range of adaptive and non-adaptive merging methods in addition to a new method designed via a wide search over the methodological design space. We demonstrate that adaptive merging methods can improve performance over the base model but provide limited benefit over training a new LoRA on the same data used to set merging coefficients. We additionally find not only that the specific choice of LoRAs to merge has little importance, but that using LoRAs with randomly initialized parameter values yields similar performance. This raises the possibility that adaptive merging from recycled LoRAs primarily works via some kind of regularization effect, rather than by enabling positive cross-task transfer. To better understand why past work has proven successful, we confirm that positive transfer is indeed possible when there are highly relevant LoRAs in the pool. We release the model checkpoints and code online. --- ### ForeAct: Steering Your VLA with Efficient Visual Foresight Planning Authors: Zhuoyang Zhang, Shang Yang, Qinghao Hu, Luke J. Huang, James Hou, Yufei Sun, Yao Lu, Song Han Published: 2026-02-12 Categories: cs.RO, cs.AI arXiv: https://arxiv.org/abs/2602.12322v1 Vision-Language-Action (VLA) models convert high-level language instructions into concrete, executable actions, a task that is especially challenging in open-world environments. We present Visual Foresight Planning (ForeAct), a general and efficient planner that guides a VLA step-by-step using imagined future observations and subtask descriptions. With an imagined future observation, the VLA can focus on visuo-motor inference rather than high-level semantic reasoning, leading to improved accuracy and generalization. Our planner comprises a highly efficient foresight image generation module that predicts a high-quality 640$\times$480 future observation from the current visual input and language instruction within only 0.33s on an H100 GPU, together with a vision-language model that reasons over the task and produces subtask descriptions for both the generator and the VLA. Importantly, state-of-the-art VLAs can integrate our planner seamlessly by simply augmenting their visual inputs, without any architectural modification. The foresight generator is pretrained on over 1 million multi-task, cross-embodiment episodes, enabling it to learn robust embodied dynamics. We evaluate our framework on a benchmark that consists of 11 diverse, multi-step real-world tasks. It achieves an average success rate of 87.4%, demonstrating a +40.9% absolute improvement over the $π_0$ baseline (46.5%) and a +30.3% absolute improvement over $π_0$ augmented with textual subtask guidance (57.1%). --- ### Abstractive Red-Teaming of Language Model Character Authors: Nate Rahn, Allison Qi, Avery Griffin, Jonathan Michala, Henry Sleight, Erik Jones Published: 2026-02-12 Categories: cs.LG arXiv: https://arxiv.org/abs/2602.12318v1 We want language model assistants to conform to a character specification, which asserts how the model should act across diverse user interactions. While models typically follow these character specifications, they can occasionally violate them in large-scale deployments. In this work, we aim to identify types of queries that are likely to produce such character violations at deployment, using much less than deployment-level compute. To do this, we introduce abstractive red-teaming, where we search for natural-language query categories, e.g. "The query is in Chinese. The query asks about family roles," that routinely elicit violations. These categories abstract over the many possible variants of a query which could appear in the wild. We introduce two algorithms for efficient category search against a character-trait-specific reward model: one based on reinforcement learning on a category generator LLM, and another which leverages a strong LLM to iteratively synthesize categories from high-scoring queries. Across a 12-principle character specification and 7 target models, we find that our algorithms consistently outperform baselines, and generate qualitatively interesting categories; for example, queries which ask Llama-3.1-8B-Instruct to predict the future lead to responses saying that AI will dominate humanity, and queries that ask GPT-4.1-Mini for essential prison survival items lead to enthusiastic recommendation of illegal weapons. Overall, we believe our results represent an important step towards realistic pre-deployment auditing of language model character. --- ### Free Lunch in Medical Image Foundation Model Pre-training via Randomized Synthesis and Disentanglement Authors: Yuhan Wei, Yuting He, Linshan Wu, Fuxiang Huang, Junlin Hou, Hao Chen Published: 2026-02-12 Categories: q-bio.QM, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2602.12317v1 Medical image foundation models (MIFMs) have demonstrated remarkable potential for a wide range of clinical tasks, yet their development is constrained by the scarcity, heterogeneity, and high cost of large-scale annotated datasets. Here, we propose RaSD (Randomized Synthesis and Disentanglement), a scalable framework for pre-training MIFMs entirely on synthetic data. By modeling anatomical structures and appearance variations with randomized Gaussian distributions, RaSD exposes models to sufficient multi-scale structural and appearance perturbations, forcing them to rely on invariant and task-relevant anatomical cues rather than dataset-specific textures, thereby enabling robust and transferable representation learning. We pre-trained RaSD on 1.2 million 3D volumes and 9.6 million 2D images, and extensively evaluated the resulting models across 6 imaging modalities, 48 datasets, and 56 downstream tasks. Across all evaluated downstream tasks, RaSD consistently outperforms training-from-scratch models, achieves the best performance on 17 tasks, and remains comparable to models pre-trained on large real datasets in most others. These results demonstrate that the capacity of synthetic data alone to drive robust representation learning. Our findings establish a paradigm shift in medical AI, demonstrating that synthetic data can serve as a "free lunch" for scalable, privacy-preserving, and clinically generalizable foundation models. --- ### GT-HarmBench: Benchmarking AI Safety Risks Through the Lens of Game Theory Authors: Pepijn Cobben, Xuanqiang Angelo Huang, Thao Amelia Pham, Isabel Dahlgren, Terry Jingchen Zhang, Zhijing Jin Published: 2026-02-12 Categories: cs.AI, cs.CL, cs.CY, cs.GT, cs.MA arXiv: https://arxiv.org/abs/2602.12316v3 Frontier AI systems are increasingly capable and deployed in high-stakes multi-agent environments. However, existing AI safety benchmarks largely evaluate single agents, leaving multi-agent risks such as coordination failure and conflict poorly understood. We introduce GT-HarmBench, a benchmark of 1,535 high-stakes scenarios spanning game-theoretic structures such as the Prisoner's Dilemma, Stag Hunt and Chicken. Scenarios are drawn from realistic AI risk contexts in the MIT AI Risk Repository. Across 15 frontier models, agents fail to choose socially beneficial actions in 38% of high-stakes cases, such as military escalation, election manipulation, and medical malpractice. We measure sensitivity to game-theoretic prompt framing and ordering, and analyze reasoning patterns driving failures. We further show that game-theoretic interventions improve socially beneficial outcomes by up to 18%. Our results highlight substantial reliability gaps and provide a broad standardized testbed for studying alignment in multi-agent environments. The benchmark and code are available at https://github.com/causalNLP/gt-harmbench. --- ### GT-HarmBench: Benchmarking AI Safety Risks Through the Lens of Game Theory Authors: Pepijn Cobben, Xuanqiang Angelo Huang, Thao Amelia Pham, Isabel Dahlgren, Terry Jingchen Zhang, Zhijing Jin Published: 2026-02-12 Categories: cs.AI, cs.CL, cs.CY, cs.GT, cs.MA arXiv: https://arxiv.org/abs/2602.12316v2 Frontier AI systems are increasingly capable and deployed in high-stakes multi-agent environments. However, existing AI safety benchmarks largely evaluate single agents, leaving multi-agent risks such as coordination failure and conflict poorly understood. We introduce GT-HarmBench, a benchmark of 1,535 high-stakes scenarios spanning game-theoretic structures such as the Prisoner's Dilemma, Stag Hunt and Chicken. Scenarios are drawn from realistic AI risk contexts in the MIT AI Risk Repository. Across 15 frontier models, agents fail to choose socially beneficial actions in 38% of high-stakes cases, such as military escalation, election manipulation, and medical malpractice. We measure sensitivity to game-theoretic prompt framing and ordering, and analyze reasoning patterns driving failures. We further show that game-theoretic interventions improve socially beneficial outcomes by up to 18%. Our results highlight substantial reliability gaps and provide a broad standardized testbed for studying alignment in multi-agent environments. The benchmark and code are available at https://github.com/causalNLP/gt-harmbench. --- ### AgenticShop: Benchmarking Agentic Product Curation for Personalized Web Shopping Authors: Sunghwan Kim, Ryang Heo, Yongsik Seo, Jinyoung Yeo, Dongha Lee Published: 2026-02-12 Categories: cs.IR, cs.AI arXiv: https://arxiv.org/abs/2602.12315v1 The proliferation of e-commerce has made web shopping platforms key gateways for customers navigating the vast digital marketplace. Yet this rapid expansion has led to a noisy and fragmented information environment, increasing cognitive burden as shoppers explore and purchase products online. With promising potential to alleviate this challenge, agentic systems have garnered growing attention for automating user-side tasks in web shopping. Despite significant advancements, existing benchmarks fail to comprehensively evaluate how well agentic systems can curate products in open-web settings. Specifically, they have limited coverage of shopping scenarios, focusing only on simplified single-platform lookups rather than exploratory search. Moreover, they overlook personalization in evaluation, leaving unclear whether agents can adapt to diverse user preferences in realistic shopping contexts. To address this gap, we present AgenticShop, the first benchmark for evaluating agentic systems on personalized product curation in open-web environment. Crucially, our approach features realistic shopping scenarios, diverse user profiles, and a verifiable, checklist-driven personalization evaluation framework. Through extensive experiments, we demonstrate that current agentic systems remain largely insufficient, emphasizing the need for user-side systems that effectively curate tailored products across the modern web. --- ### LatentAM: Real-Time, Large-Scale Latent Gaussian Attention Mapping via Online Dictionary Learning Authors: Junwoon Lee, Yulun Tian Published: 2026-02-12 Categories: cs.RO, cs.CV arXiv: https://arxiv.org/abs/2602.12314v2 We present LatentAM, an online 3D Gaussian Splatting (3DGS) mapping framework that builds scalable latent feature maps from streaming RGB-D observations for open-vocabulary robotic perception. Instead of distilling high-dimensional Vision-Language Model (VLM) embeddings using model-specific decoders, LatentAM proposes an online dictionary learning approach that is both model-agnostic and pretraining-free, enabling plug-and-play integration with different VLMs at test time. Specifically, our approach associates each Gaussian primitive with a compact query vector that can be converted into approximate VLM embeddings using an attention mechanism with a learnable dictionary. The dictionary is initialized efficiently from streaming observations and optimized online to adapt to evolving scene semantics under trust-region regularization. To scale to long trajectories and large environments, we further propose an efficient map management strategy based on voxel hashing, where optimization is restricted to an active local map on the GPU, while the global map is stored and indexed on the CPU to maintain bounded GPU memory usage. Experiments on public benchmarks and a large-scale custom dataset demonstrate that LatentAM attains significantly better feature reconstruction fidelity compared to state-of-the-art methods, while achieving near-real-time speed (12-35 FPS) on the evaluated datasets. Our project page, including the code, is available at: https://junwoonlee.github.io/projects/LatentAM --- ### LatentAM: Real-Time, Large-Scale Latent Gaussian Attention Mapping via Online Dictionary Learning Authors: Junwoon Lee, Yulun Tian Published: 2026-02-12 Categories: cs.RO, cs.CV arXiv: https://arxiv.org/abs/2602.12314v1 We present LatentAM, an online 3D Gaussian Splatting (3DGS) mapping framework that builds scalable latent feature maps from streaming RGB-D observations for open-vocabulary robotic perception. Instead of distilling high-dimensional Vision-Language Model (VLM) embeddings using model-specific decoders, LatentAM proposes an online dictionary learning approach that is both model-agnostic and pretraining-free, enabling plug-and-play integration with different VLMs at test time. Specifically, our approach associates each Gaussian primitive with a compact query vector that can be converted into approximate VLM embeddings using an attention mechanism with a learnable dictionary. The dictionary is initialized efficiently from streaming observations and optimized online to adapt to evolving scene semantics under trust-region regularization. To scale to long trajectories and large environments, we further propose an efficient map management strategy based on voxel hashing, where optimization is restricted to an active local map on the GPU, while the global map is stored and indexed on the CPU to maintain bounded GPU memory usage. Experiments on public benchmarks and a large-scale custom dataset demonstrate that LatentAM attains significantly better feature reconstruction fidelity compared to state-of-the-art methods, while achieving near-real-time speed (12-35 FPS) on the evaluated datasets. Our project page is at: https://junwoonlee.github.io/projects/LatentAM --- ### Visible and Hyperspectral Imaging for Quality Assessment of Milk: Property Characterisation and Identification Authors: Massimo Martinelli, Elena Tomassi, Nafiou Arouna, Morena Gabriele, Laryssa Perez Fabbri, Luisa Pozzo, Bianca Castiglioni, Paola Cremonesi, Giuseppe Conte, Davide Moroni, Laura Pucci Published: 2026-02-12 Categories: eess.IV, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2602.12313v2 Rapid and non-destructive assessment of milk quality is crucial to ensuring both nutritional value and food safety. In this study, we investigated the potential of visible and hyperspectral imaging as cost-effective and quick-response alternatives to conventional chemical analyses for characterizing key properties of cowś milk. A total of 52 milk samples were analysed to determine their biochemical composition (polyphenols, antioxidant capacity, and fatty acids) using spectrophotometer methods and standard gas-liquid and high-performance liquid chromatography (GLC/HPLC). Concurrently, visible (RGB) images were captured using a standard smartphone, and hyperspectral data were acquired in the near-infrared range. A comprehensive analytical framework, including eleven different machine learning algorithms, was employed to correlate imaging features with biochemical measurements. Analysis of visible images accurately distinguished between fresh samples and those stored for 12 days (100 percent accuracy) and achieved perfect discrimination between antibiotic-treated and untreated groups (100 percent accuracy). Moreover, image-derived features enabled perfect prediction of the polyphenols content and the antioxidant capacity using an XGBoost model. Hyperspectral imaging further achieved classification accuracies exceeding 95 percent for several individual fatty acids and 94.8 percent for treatment groups using a Random Forest model. These findings demonstrate that both visible and hyperspectral imaging, when coupled with machine learning, are powerful, non-invasive tools for the rapid assessment of milkś chemical and nutritional profiles, highlighting the strong potential of imaging-based approaches for milk quality assessment. --- ### Perceptual Self-Reflection in Agentic Physics Simulation Code Generation Authors: Prashant Shende, Bradley Camburn Published: 2026-02-12 Categories: cs.SE, cs.AI arXiv: https://arxiv.org/abs/2602.12311v1 We present a multi-agent framework for generating physics simulation code from natural language descriptions, featuring a novel perceptual self-reflection mechanism for validation. The system employs four specialized agents: a natural language interpreter that converts user requests into physics-based descriptions; a technical requirements generator that produces scaled simulation parameters; a physics code generator with automated self-correction; and a physics validator that implements perceptual self-reflection. The key innovation is perceptual validation, which analyzes rendered animation frames using a vision-capable language model rather than inspecting code structure directly. This approach addresses the ``oracle gap'' where syntactically correct code produces physically incorrect behavior--a limitation that conventional testing cannot detect. We evaluate the system across seven domains including classical mechanics, fluid dynamics, thermodynamics, electromagnetics, wave physics, reaction-diffusion systems, and non-physics data visualization. The perceptual self-reflection architecture demonstrates substantial improvement over single-shot generation baselines, with the majority of tested scenarios achieving target physics accuracy thresholds. The system exhibits robust pipeline stability with consistent code self-correction capability, operating at approximately \$0.20 per animation. These results validate our hypothesis that feeding visual simulation outputs back to a vision-language model for iterative refinement significantly outperforms single-shot code generation for physics simulation tasks and highlights the potential of agentic AI to support engineering workflows and physics data generation pipelines. --- ### Quantum walk inspired JPEG compression of images Authors: Abhishek Verma, Sahil Tomar, Sandeep Kumar Published: 2026-02-12 Categories: eess.IV, cs.AI, cs.CV, cs.ET, cs.IT arXiv: https://arxiv.org/abs/2602.12306v1 This work proposes a quantum inspired adaptive quantization framework that enhances the classical JPEG compression by introducing a learned, optimized Qtable derived using a Quantum Walk Inspired Optimization (QWIO) search strategy. The optimizer searches a continuous parameter space of frequency band scaling factors under a unified rate distortion objective that jointly considers reconstruction fidelity and compression efficiency. The proposed framework is evaluated on MNIST, CIFAR10, and ImageNet subsets, using Peak Signal to Noise Ratio (PSNR), Structural Similarity Index (SSIM), Bits Per Pixel (BPP), and error heatmap visual analysis as evaluation metrics. Experimental results show average gains ranging from 3 to 6 dB PSNR, along with better structural preservation of edges, contours, and luminance transitions, without modifying decoder compatibility. The structure remains JPEG compliant and can be implemented using accessible scientific packages making it ideal for deployment and practical research use. --- ### OptiML: An End-to-End Framework for Program Synthesis and CUDA Kernel Optimization Authors: Arijit Bhattacharjee, Heng Ping, Son Vu Le, Paul Bogdan, Nesreen K. Ahmed, Ali Jannesari Published: 2026-02-12 Categories: cs.LG, cs.AI, cs.DC, cs.MA, cs.SE arXiv: https://arxiv.org/abs/2602.12305v1 Generating high-performance CUDA kernels remains challenging due to the need to navigate a combinatorial space of low-level transformations under noisy and expensive hardware feedback. Although large language models can synthesize functionally correct CUDA code, achieving competitive performance requires systematic exploration and verification of optimization choices. We present OptiML, an end-to-end framework that maps either natural-language intent or input CUDA code to performance-optimized CUDA kernels by formulating kernel optimization as search under verification. OptiML consists of two decoupled stages. When the input is natural language, a Mixture-of-Thoughts generator (OptiML-G) acts as a proposal policy over kernel implementation strategies, producing an initial executable program. A search-based optimizer (OptiML-X) then refines either synthesized or user-provided kernels using Monte Carlo Tree Search over LLM-driven edits, guided by a hardware-aware reward derived from profiler feedback. Each candidate transformation is compiled, verified, and profiled with Nsight Compute, and evaluated by a composite objective that combines runtime with hardware bottleneck proxies and guardrails against regressions. We evaluate OptiML in both synthesis-and-optimize and optimization-only settings on a diverse suite of CUDA kernels. Results show that OptiML consistently discovers verified performance improvements over strong LLM baselines and produces interpretable optimization trajectories grounded in profiler evidence. --- ### OmniCustom: Sync Audio-Video Customization Via Joint Audio-Video Generation Model Authors: Maomao Li, Zhen Li, Kaipeng Zhang, Guosheng Yin, Zhifeng Li, Dong Xu Published: 2026-02-12 Categories: cs.SD, cs.AI, cs.MM, eess.AS arXiv: https://arxiv.org/abs/2602.12304v5 Existing mainstream video customization methods focus on generating identity-consistent videos based on given reference images and textual prompts. Benefiting from the rapid advancement of joint audio-video generation, this paper proposes a more compelling new task: sync audio-video customization, which aims to synchronously customize both video identity and audio timbre. Specifically, given a reference image $I^{r}$ and a reference audio $A^{r}$, this novel task requires generating videos that maintain the identity of the reference image while imitating the timbre of the reference audio, with spoken content freely specifiable through user-provided textual prompts. To this end, we propose OmniCustom, a powerful DiT-based audio-video customization framework that can synthesize a video following reference image identity, audio timbre, and text prompts all at once in a zero-shot manner. Our framework is built on three key contributions. First, identity and audio timbre control are achieved through separate reference identity and audio LoRA modules that operate through self-attention layers within the base audio-video generation model. Second, we introduce a contrastive learning objective alongside the standard flow matching objective. It uses predicted flows conditioned on reference inputs as positive examples and those without reference conditions as negative examples, thereby enhancing the model ability to preserve identity and timbre. Third, we train OmniCustom on our constructed large-scale, high-quality audio-visual human dataset. Extensive experiments demonstrate that OmniCustom outperforms existing methods in generating audio-video content with consistent identity and timbre fidelity. Project page: https://omnicustom-project.github.io/page/. --- ### Scaling Verification Can Be More Effective than Scaling Policy Learning for Vision-Language-Action Alignment Authors: Jacky Kwok, Xilun Zhang, Mengdi Xu, Yuejiang Liu, Azalia Mirhoseini, Chelsea Finn, Marco Pavone Published: 2026-02-12 Categories: cs.RO, cs.AI, eess.SY arXiv: https://arxiv.org/abs/2602.12281v2 The long-standing vision of general-purpose robots hinges on their ability to understand and act upon natural language instructions. Vision-Language-Action (VLA) models have made remarkable progress toward this goal, yet their generated actions can still misalign with the given instructions. In this paper, we investigate test-time verification as a means to shrink the "intention-action gap." We first characterize the test-time scaling laws for embodied instruction following and demonstrate that jointly scaling the number of rephrased instructions and generated actions greatly increases test-time sample diversity, often recovering correct actions more efficiently than scaling each dimension independently. To capitalize on these scaling laws, we present CoVer, a contrastive verifier for vision-language-action alignment, and show that our architecture scales gracefully with additional computational resources and data. We then introduce CoVer-VLA, a hierarchical test-time verification pipeline using the trained verifier. At deployment, our framework precomputes a diverse set of rephrased instructions from a Vision-Language-Model (VLM), repeatedly generates action candidates for each instruction, and then uses the verifier to select the optimal high-level prompt and low-level action chunks. Compared to scaling policy pre-training on the same data, our verification approach yields 22% gains in-distribution and 13% out-of-distribution on the SIMPLER benchmark, with a further 45% improvement in real-world experiments. On the PolaRiS benchmark, CoVer-VLA achieves 14% gains in task progress and 9% in success rate. --- ### Stroke of Surprise: Progressive Semantic Illusions in Vector Sketching Authors: Huai-Hsun Cheng, Siang-Ling Zhang, Yu-Lun Liu Published: 2026-02-12 Categories: cs.CV arXiv: https://arxiv.org/abs/2602.12280v2 Visual illusions traditionally rely on spatial manipulations such as multi-view consistency. In this work, we introduce Progressive Semantic Illusions, a novel vector sketching task where a single sketch undergoes a dramatic semantic transformation through the sequential addition of strokes. We present Stroke of Surprise, a generative framework that optimizes vector strokes to satisfy distinct semantic interpretations at different drawing stages. The core challenge lies in the "dual-constraint": initial prefix strokes must form a coherent object (e.g., a duck) while simultaneously serving as the structural foundation for a second concept (e.g., a sheep) upon adding delta strokes. To address this, we propose a sequence-aware joint optimization framework driven by a dual-branch Score Distillation Sampling (SDS) mechanism. Unlike sequential approaches that freeze the initial state, our method dynamically adjusts prefix strokes to discover a "common structural subspace" valid for both targets. Furthermore, we introduce a novel Overlay Loss that enforces spatial complementarity, ensuring structural integration rather than occlusion. Extensive experiments demonstrate that our method significantly outperforms state-of-the-art baselines in recognizability and illusion strength, successfully expanding visual anagrams from the spatial to the temporal dimension. Project page: https://stroke-of-surprise.github.io/ --- ### UniT: Unified Multimodal Chain-of-Thought Test-time Scaling Authors: Leon Liangyu Chen, Haoyu Ma, Zhipeng Fan, Ziqi Huang, Animesh Sinha, Xiaoliang Dai, Jialiang Wang, Zecheng He, Jianwei Yang, Chunyuan Li, Junzhe Sun, Chu Wang et al. Published: 2026-02-12 Categories: cs.CV, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2602.12279v2 Unified models can handle both multimodal understanding and generation within a single architecture, yet they typically operate in a single pass without iteratively refining their outputs. Many multimodal tasks, especially those involving complex spatial compositions, multiple interacting objects, or evolving instructions, require decomposing instructions, verifying intermediate results, and making iterative corrections. While test-time scaling (TTS) has demonstrated that allocating additional inference compute for iterative reasoning substantially improves language model performance, extending this paradigm to unified multimodal models remains an open challenge. We introduce UniT, a framework for multimodal chain-of-thought test-time scaling that enables a single unified model to reason, verify, and refine across multiple rounds. UniT combines agentic data synthesis, unified model training, and flexible test-time inference to elicit cognitive behaviors including verification, subgoal decomposition, and content memory. Our key findings are: (1) unified models trained on short reasoning trajectories generalize to longer inference chains at test time; (2) sequential chain-of-thought reasoning provides a more scalable and compute-efficient TTS strategy than parallel sampling; (3) training on generation and editing trajectories improves out-of-distribution visual reasoning. These results establish multimodal test-time scaling as an effective paradigm for advancing both generation and understanding in unified models. --- ### AttentionRetriever: Attention Layers are Secretly Long Document Retrievers Authors: David Jiahao Fu, Lam Thanh Do, Jiayu Li, Kevin Chen-Chuan Chang Published: 2026-02-12 Categories: cs.IR, cs.AI arXiv: https://arxiv.org/abs/2602.12278v1 Retrieval augmented generation (RAG) has been widely adopted to help Large Language Models (LLMs) to process tasks involving long documents. However, existing retrieval models are not designed for long document retrieval and fail to address several key challenges of long document retrieval, including context-awareness, causal dependence, and scope of retrieval. In this paper, we proposed AttentionRetriever, a novel long document retrieval model that leverages attention mechanism and entity-based retrieval to build context-aware embeddings for long document and determine the scope of retrieval. With extensive experiments, we found AttentionRetriever is able to outperform existing retrieval models on long document retrieval datasets by a large margin while remaining as efficient as dense retrieval models. --- ### Agentic Test-Time Scaling for WebAgents Authors: Nicholas Lee, Lutfi Eren Erdogan, Chris Joseph John, Surya Krishnapillai, Michael W. Mahoney, Kurt Keutzer, Amir Gholami Published: 2026-02-12 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2602.12276v2 Test-time scaling has become a standard way to improve performance and boost reliability of neural network models. However, its behavior on agentic, multi-step tasks remains less well-understood: small per-step errors can compound over long horizons; and we find that naive policies that uniformly increase sampling show diminishing returns. In this work, we present CATTS, a simple technique for dynamically allocating compute for multi-step agents. We first conduct an empirical study of inference-time scaling for web agents. We find that uniformly increasing per-step compute quickly saturates in long-horizon environments. We then investigate stronger aggregation strategies, including an LLM-based Arbiter that can outperform naive voting, but that can overrule high-consensus decisions. We show that uncertainty statistics derived from the agent's own vote distribution (entropy and top-1/top-2 margin) correlate with downstream success and provide a practical signal for dynamic compute allocation. Based on these findings, we introduce Confidence-Aware Test-Time Scaling (CATTS), which uses vote-derived uncertainty to allocate compute only when decisions are genuinely contentious. CATTS improves performance on WebArena-Lite, Online-Mind2Web, and GoBrowse by up to 11.8% over majority voting while using fewer tokens than uniform scaling, providing both efficiency gains and an interpretable decision rule. --- ### Agentic Test-Time Scaling for WebAgents Authors: Nicholas Lee, Lutfi Eren Erdogan, Chris Joseph John, Surya Krishnapillai, Michael W. Mahoney, Kurt Keutzer, Amir Gholami Published: 2026-02-12 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2602.12276v1 Test-time scaling has become a standard way to improve performance and boost reliability of neural network models. However, its behavior on agentic, multi-step tasks remains less well-understood: small per-step errors can compound over long horizons; and we find that naive policies that uniformly increase sampling show diminishing returns. In this work, we present CATTS, a simple technique for dynamically allocating compute for multi-step agents. We first conduct an empirical study of inference-time scaling for web agents. We find that uniformly increasing per-step compute quickly saturates in long-horizon environments. We then investigate stronger aggregation strategies, including an LLM-based Arbiter that can outperform naive voting, but that can overrule high-consensus decisions. We show that uncertainty statistics derived from the agent's own vote distribution (entropy and top-1/top-2 margin) correlate with downstream success and provide a practical signal for dynamic compute allocation. Based on these findings, we introduce Confidence-Aware Test-Time Scaling (CATTS), which uses vote-derived uncertainty to allocate compute only when decisions are genuinely contentious. CATTS improves performance on WebArena-Lite and GoBrowse by up to 9.1% over React while using up to 2.3x fewer tokens than uniform scaling, providing both efficiency gains and an interpretable decision rule. --- ### On-Policy Context Distillation for Language Models Authors: Tianzhu Ye, Li Dong, Xun Wu, Shaohan Huang, Furu Wei Published: 2026-02-12 Categories: cs.CL arXiv: https://arxiv.org/abs/2602.12275v2 Context distillation enables language models to internalize in-context knowledge into their parameters. In our work, we propose On-Policy Context Distillation (OPCD), a framework that bridges on-policy distillation with context distillation by training a student model on its own generated trajectories while minimizing reverse Kullback-Leibler divergence against a context-conditioned teacher. We demonstrate the effectiveness of OPCD on two important applications: experiential knowledge distillation, where models extract and consolidate transferable knowledge from their historical solution traces, and system prompt distillation, where models internalize beneficial behaviors encoded in optimized prompts. Across mathematical reasoning, text-based games, and domain-specific tasks, OPCD consistently outperforms baseline methods, achieving higher task accuracy while better preserving out-of-distribution capabilities. We further show that OPCD enables effective cross-size distillation, where smaller student models can internalize experiential knowledge from larger teachers. --- ### Function-Space Decoupled Diffusion for Forward and Inverse Modeling in Carbon Capture and Storage Authors: Xin Ju, Jiachen Yao, Anima Anandkumar, Sally M. Benson, Gege Wen Published: 2026-02-12 Categories: cs.LG, physics.geo-ph arXiv: https://arxiv.org/abs/2602.12274v2 Accurate characterization of subsurface flow is critical for Carbon Capture and Storage (CCS) but remains challenged by the ill-posed nature of inverse problems with sparse observations. We present Function-space Decoupled Diffusion Posterior Sampling (Fun-DDPS), a generative framework that combines function-space diffusion models with differentiable neural operator surrogates for both forward and inverse modeling. Our approach learns a prior distribution over geological parameters (geomodel) using a single-channel diffusion model, then leverages a Local Neural Operator (LNO) surrogate to provide physics-consistent guidance for cross-field conditioning on the dynamics field. This decoupling allows the diffusion prior to robustly recover missing information in parameter space, while the surrogate provides efficient gradient-based guidance for data assimilation. We demonstrate Fun-DDPS on synthetic CCS modeling datasets, achieving two key results: (1) For forward modeling with only 25% observations, Fun-DDPS achieves 7.7% relative error compared to 86.9% for standard surrogates (an 11x improvement), proving its capability to handle extreme data sparsity where deterministic methods fail. (2) We provide the first rigorous validation of diffusion-based inverse solvers against asymptotically exact Rejection Sampling (RS) posteriors. Both Fun-DDPS and the joint-state baseline (Fun-DPS) achieve Jensen-Shannon divergence less than 0.06 against the ground truth. Crucially, Fun-DDPS produces physically consistent realizations free from the high-frequency artifacts observed in joint-state baselines, achieving this with 4x improved sample efficiency compared to rejection sampling. --- ### Learning to Control: The iUzawa-Net for Nonsmooth Optimal Control of Linear PDEs Authors: Yongcun Song, Xiaoming Yuan, Hangrui Yue, Tianyou Zeng Published: 2026-02-12 Categories: math.OC, cs.LG, math.NA arXiv: https://arxiv.org/abs/2602.12273v1 We propose an optimization-informed deep neural network approach, named iUzawa-Net, aiming for the first solver that enables real-time solutions for a class of nonsmooth optimal control problems of linear partial differential equations (PDEs). The iUzawa-Net unrolls an inexact Uzawa method for saddle point problems, replacing classical preconditioners and PDE solvers with specifically designed learnable neural networks. We prove universal approximation properties and establish the asymptotic $\varepsilon$-optimality for the iUzawa-Net, and validate its promising numerical efficiency through nonsmooth elliptic and parabolic optimal control problems. Our techniques offer a versatile framework for designing and analyzing various optimization-informed deep learning approaches to optimal control and other PDE-constrained optimization problems. The proposed learning-to-control approach synergizes model-based optimization algorithms and data-driven deep learning techniques, inheriting the merits of both methodologies. --- ### MonarchRT: Efficient Attention for Real-Time Video Generation Authors: Krish Agarwal, Zhuoming Chen, Cheng Luo, Yongqi Chen, Haizhong Zheng, Xun Huang, Atri Rudra, Beidi Chen Published: 2026-02-12 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2602.12271v1 Real-time video generation with Diffusion Transformers is bottlenecked by the quadratic cost of 3D self-attention, especially in real-time regimes that are both few-step and autoregressive, where errors compound across time and each denoising step must carry substantially more information. In this setting, we find that prior sparse-attention approximations break down, despite showing strong results for bidirectional, many-step diffusion. Specifically, we observe that video attention is not reliably sparse, but instead combines pronounced periodic structure driven by spatiotemporal position with dynamic, sparse semantic correspondences and dense mixing, exceeding the representational capacity of even oracle top-k attention. Building on this insight, we propose Monarch-RT, a structured attention parameterization for video diffusion models that factorizes attention using Monarch matrices. Through appropriately aligned block structure and our extended tiled Monarch parameterization, we achieve high expressivity while preserving computational efficiency. We further overcome the overhead of parameterization through finetuning, with custom Triton kernels. We first validate the high efficacy of Monarch-RT over existing sparse baselines designed only for bidirectional models. We further observe that Monarch-RT attains up to 95% attention sparsity with no loss in quality when applied to the state-of-the-art model Self-Forcing, making Monarch-RT a pioneering work on highly-capable sparse attention parameterization for real-time video generation. Our optimized implementation outperforms FlashAttention-2, FlashAttention-3, and FlashAttention-4 kernels on Nvidia RTX 5090, H100, and B200 GPUs respectively, providing kernel speedups in the range of 1.4-11.8X. This enables us, for the first time, to achieve true real-time video generation with Self-Forcing at 16 FPS on a single RTX 5090. --- ### Creative Ownership in the Age of AI Authors: Annie Liang, Jay Lu Published: 2026-02-12 Categories: econ.TH, cs.AI, cs.GT arXiv: https://arxiv.org/abs/2602.12270v1 Copyright law focuses on whether a new work is "substantially similar" to an existing one, but generative AI can closely imitate style without copying content, a capability now central to ongoing litigation. We argue that existing definitions of infringement are ill-suited to this setting and propose a new criterion: a generative AI output infringes on an existing work if it could not have been generated without that work in its training corpus. To operationalize this definition, we model generative systems as closure operators mapping a corpus of existing works to an output of new works. AI generated outputs are \emph{permissible} if they do not infringe on any existing work according to our criterion. Our results characterize structural properties of permissible generation and reveal a sharp asymptotic dichotomy: when the process of organic creations is light-tailed, dependence on individual works eventually vanishes, so that regulation imposes no limits on AI generation; with heavy-tailed creations, regulation can be persistently constraining. --- ### CM2: Reinforcement Learning with Checklist Rewards for Multi-Turn and Multi-Step Agentic Tool Use Authors: Zhen Zhang, Kaiqiang Song, Xun Wang, Yebowen Hu, Weixiang Yan, Chenyang Zhao, Henry Peng Zou, Haoyun Deng, Sathish Reddy Indurthi, Shujian Liu, Simin Ma, Xiaoyang Wang et al. Published: 2026-02-12 Categories: cs.AI arXiv: https://arxiv.org/abs/2602.12268v2 AI agents are increasingly used to solve real-world tasks by reasoning over multi-turn user interactions and invoking external tools. However, applying reinforcement learning to such settings remains difficult: realistic objectives often lack verifiable rewards and instead emphasize open-ended behaviors; moreover, RL for multi-turn, multi-step agentic tool use is still underexplored; and building and maintaining executable tool environments is costly, limiting scale and coverage. We propose CM2, an RL framework that replaces verifiable outcome rewards with checklist rewards. CM2 decomposes each turn's intended behavior into fine-grained binary criteria with explicit evidence grounding and structured metadata, turning open-ended judging into more stable classification-style decisions. To balance stability and informativeness, our method adopts a strategy of sparse reward assignment but dense evaluation criteria. Training is performed in a scalable LLM-simulated tool environment, avoiding heavy engineering for large tool sets. Experiments show that CM2 consistently improves over supervised fine-tuning. Starting from an 8B Base model and training on an 8k-example RL dataset, CM2 improves over the SFT counterpart by 8 points on tau^-Bench, by 10 points on BFCL-V4, and by 12 points on ToolSandbox. The results match or even outperform similarly sized open-source baselines, including the judging model. CM2 thus provides a scalable recipe for optimizing multi-turn, multi-step tool-using agents without relying on verifiable rewards. Code provided by the open-source community: https://github.com/namezhenzhang/CM2-RLCR-Tool-Agent. --- ### Self-Supervised Learning via Flow-Guided Neural Operator on Time-Series Data Authors: Duy Nguyen, Jiachen Yao, Jiayun Wang, Julius Berner, Animashree Anandkumar Published: 2026-02-12 Categories: cs.LG arXiv: https://arxiv.org/abs/2602.12267v2 Self-supervised learning (SSL) is a powerful paradigm for learning from unlabeled time-series data. However, popular methods such as masked autoencoders (MAEs) rely on reconstructing inputs from a fixed, predetermined masking ratio. Instead of this static design, we propose treating the corruption level as a new degree of freedom for representation learning, enhancing flexibility and performance. To achieve this, we introduce the Flow-Guided Neural Operator (FGNO), a novel framework combining operator learning with flow matching for SSL training. FGNO learns mappings in functional spaces by using Short-Time Fourier Transform to unify different time resolutions. We extract a rich hierarchy of features by tapping into different network layers and flow times that apply varying strengths of noise to the input data. This enables the extraction of versatile representations, from low-level patterns to high-level global features, using a single model adaptable to specific tasks. Unlike prior generative SSL methods that use noisy inputs during inference, we propose using clean inputs for representation extraction while learning representations with noise; this eliminates randomness and boosts accuracy. We evaluate FGNO across three biomedical domains, where it consistently outperforms established baselines. Our method yields up to 35% AUROC gains in neural signal decoding (BrainTreeBank), 16% RMSE reductions in skin temperature prediction (DREAMT), and over 20% improvement in accuracy and macro-F1 on SleepEDF under low-data regimes. These results highlight FGNO's robustness to data scarcity and its superior capacity to learn expressive representations for diverse time series. --- ### Few-Step Diffusion Language Models via Trajectory Self-Distillation Authors: Tunyu Zhang, Xinxi Zhang, Ligong Han, Haizhou Shi, Xiaoxiao He, Zhuowei Li, Hao Wang, Kai Xu, Akash Srivastava, Chengzhi Mao, Hao Wang, Vladimir Pavlovic et al. Published: 2026-02-12 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2602.12262v4 Diffusion large language models (DLLMs) have emerged as powerful generative models with the promise of fast text generation through parallel decoding. However, realizing this potential in practice remains challenging: reducing the number of decoding steps, typically causes a substantial degradation in output quality due to token factorization error. To alleviate this, we propose a self-distillation framework that trains a few-step student to match the generative trajectory of a full-step teacher. We theoretically and empirically show that trajectory-level supervision mitigates this factorization error, thereby enabling effective few-step decoding. We further incorporate Direct Discriminative Optimization (DDO), a reverse-KL objective that encourages mode-seeking toward the teacher's modes, yielding stronger performance on challenging reasoning tasks. Across reasoning and code-generation benchmarks, our method substantially narrows the gap between few-step and full-step decoding. The source code is available at https://github.com/Tyrion58/T3D. --- ### Think like a Scientist: Physics-guided LLM Agent for Equation Discovery Authors: Jianke Yang, Ohm Venkatachalam, Mohammad Kianezhad, Sharvaree Vadgama, Rose Yu Published: 2026-02-12 Categories: cs.AI, cs.LG arXiv: https://arxiv.org/abs/2602.12259v2 Explaining observed phenomena through symbolic, interpretable formulas is a fundamental goal of science. Recently, large language models (LLMs) have emerged as promising tools for symbolic equation discovery, owing to their broad domain knowledge and strong reasoning capabilities. However, most existing LLM-based systems try to guess equations directly from data, without modeling the multi-step reasoning process that scientists often follow: first inferring physical properties such as symmetries, then using these as priors to restrict the space of candidate equations. We introduce KeplerAgent, an agentic framework that explicitly follows this scientific reasoning process. The agent coordinates physics-based tools to extract intermediate structure and uses these results to configure symbolic regression engines such as PySINDy and PySR, including their function libraries and structural constraints. Across a suite of physical equation benchmarks, KeplerAgent achieves substantially higher symbolic accuracy and greater robustness to noisy data than both LLM and traditional baselines. --- ### On the implicit regularization of Langevin dynamics with projected noise Authors: Govind Menon, Austin J. Stromme, Adrien Vacher Published: 2026-02-12 Categories: math.PR, cs.AI arXiv: https://arxiv.org/abs/2602.12257v1 We study Langevin dynamics with noise projected onto the directions orthogonal to an isometric group action. This mathematical model is introduced to shed new light on the effects of symmetry on stochastic gradient descent for over-parametrized models. Our main result identifies a novel form of implicit regularization: when the initial and target density are both invariant under the group action, Langevin dynamics with projected noise is equivalent in law to Langevin dynamics with isotropic diffusion but with an additional drift term proportional to the negative log volume of the group orbit. We prove this result by constructing a coupling of the two processes via a third process on the group itself, and identify the additional drift as the mean curvature of the orbits. --- ### Is Online Linear Optimization Sufficient for Strategic Robustness? Authors: Yang Cai, Haipeng Luo, Chen-Yu Wei, Weiqiang Zheng Published: 2026-02-12 Categories: cs.GT, cs.LG arXiv: https://arxiv.org/abs/2602.12253v1 We consider bidding in repeated Bayesian first-price auctions. Bidding algorithms that achieve optimal regret have been extensively studied, but their strategic robustness to the seller's manipulation remains relatively underexplored. Bidding algorithms based on no-swap-regret algorithms achieve both desirable properties, but are suboptimal in terms of statistical and computational efficiency. In contrast, online gradient ascent is the only algorithm that achieves $O(\sqrt{TK})$ regret and strategic robustness (KSS24), where $T$ denotes the number of auctions and $K$ the number of bids. In this paper, we explore whether simple online linear optimization (OLO) algorithms suffice for bidding algorithms with both desirable properties. Our main result shows that sublinear linearized regret is sufficient for strategic robustness. Specifically, we construct simple black-box reductions that convert any OLO algorithm into a strategically robust no-regret bidding algorithm, in both known and unknown value distribution settings. For the known value distribution case, our reduction yields a bidding algorithm that achieves $O(\sqrt{T \log K})$ regret and strategic robustness (with exponential improvement on the $K$-dependence compared to (KSS24)). For the unknown value distribution case, our reduction gives a bidding algorithm with high-probability $O(\sqrt{T (\log K+\log(T/δ)})$ regret and strategic robustness, while removing the bounded density assumption made in (KSS24). --- ### A technical curriculum on language-oriented artificial intelligence in translation and specialised communication Authors: Ralph Krüger Published: 2026-02-12 Categories: cs.CL, cs.AI, cs.HC arXiv: https://arxiv.org/abs/2602.12251v1 This paper presents a technical curriculum on language-oriented artificial intelligence (AI) in the language and translation (L&T) industry. The curriculum aims to foster domain-specific technical AI literacy among stakeholders in the fields of translation and specialised communication by exposing them to the conceptual and technical/algorithmic foundations of modern language-oriented AI in an accessible way. The core curriculum focuses on 1) vector embeddings, 2) the technical foundations of neural networks, 3) tokenization and 4) transformer neural networks. It is intended to help users develop computational thinking as well as algorithmic awareness and algorithmic agency, ultimately contributing to their digital resilience in AI-driven work environments. The didactic suitability of the curriculum was tested in an AI-focused MA course at the Institute of Translation and Multilingual Communication at TH Koeln. Results suggest the didactic effectiveness of the curriculum, but participant feedback indicates that it should be embedded into higher-level didactic scaffolding - e.g., in the form of lecturer support - in order to enable optimal learning conditions. --- ### Community Concealment from Graph Neural Networks Authors: Dalyapraz Manatova, Pablo Moriano, L. Jean Camp Published: 2026-02-12 Categories: cs.LG, cs.CR, cs.SI arXiv: https://arxiv.org/abs/2602.12250v2 Graph neural networks (GNNs) enable powerful unsupervised learning of communities. However, such inference may inadvertently expose sensitive group structures, critical clustered patterns, or collective behaviors, raising concerns about sensitive group-level privacy. In social and critical infrastructure networks, unauthorized community inference can reveal coordinated asset groups, operational hierarchies, and system dependencies that may be exploited for reconnaissance or profiling. We study a defensive setting in which a network (or defender) operator seeks to conceal a community of interest while making only small, utility-preserving modifications to the network. Our analysis shows that community concealment depends on two measurable factors: the connectivity at the community boundary and the feature similarity between the protected community and its neighbors. Guided by these observations, we introduce Feature-Community-guided DICE (FCom-DICE), a perturbation strategy built on DICE (Disconnect Internally Connect Externally) that rewires a set of structurally influential edges and adjusts node features to reduce the distinctiveness exploited by GNN message passing. Across synthetic benchmarks and real network graphs such as Facebook, Wikipedia, and Bitcoin Transactions, FCom-DICE consistently outperforms structure-only DICE under the same perturbation budgets. The largest improvements are observed for communities that are weakly connected to the rest of the network and well separated in feature space. These gains are achieved while preserving key structural and feature characteristics of the original network. These results demonstrate the effectiveness of feature-aware perturbations for reducing the recoverability of targeted communities under GNN-based community inference. --- ### Community Concealment from Unsupervised Graph Learning-Based Clustering Authors: Dalyapraz Manatova, Pablo Moriano, L. Jean Camp Published: 2026-02-12 Categories: cs.LG, cs.CR, cs.SI arXiv: https://arxiv.org/abs/2602.12250v1 Graph neural networks (GNNs) are designed to use attributed graphs to learn representations. Such representations are beneficial in the unsupervised learning of clusters and community detection. Nonetheless, such inference may reveal sensitive groups, clustered systems, or collective behaviors, raising concerns regarding group-level privacy. Community attribution in social and critical infrastructure networks, for example, can expose coordinated asset groups, operational hierarchies, and system dependencies that could be used for profiling or intelligence gathering. We study a defensive setting in which a data publisher (defender) seeks to conceal a community of interest while making limited, utility-aware changes in the network. Our analysis indicates that community concealment is strongly influenced by two quantifiable factors: connectivity at the community boundary and feature similarity between the protected community and adjacent communities. Informed by these findings, we present a perturbation strategy that rewires a set of selected edges and modifies node features to reduce the distinctiveness leveraged by GNN message passing. The proposed method outperforms DICE in our experiments on synthetic benchmarks and real network graphs under identical perturbation budgets. Overall, it achieves median relative concealment improvements of approximately 20-45% across the evaluated settings. These findings demonstrate a mitigation strategy against GNN-based community learning and highlight group-level privacy risks intrinsic to graph learning. --- ### "Sorry, I Didn't Catch That": How Speech Models Miss What Matters Most Authors: Kaitlyn Zhou, Martijn Bartelds, Federico Bianchi, James Zou Published: 2026-02-12 Categories: cs.AI, cs.CL, cs.CY arXiv: https://arxiv.org/abs/2602.12249v2 Despite speech recognition systems achieving low word error rates on standard benchmarks, they often fail on short, high-stakes utterances in real-world deployments. Here, we study this failure mode in a high-stakes task: the transcription of U.S. street names as spoken by U.S. participants. We evaluate 15 models from OpenAI, Deepgram, Google, and Microsoft on recordings from linguistically diverse U.S. speakers and find an average transcription error rate of 44%. We quantify the downstream impact of failed transcriptions by geographic locations and show that mis-transcriptions systematically cause errors for all speakers, but that routing distance errors are twice as large for non-English primary speakers compared to English primary speakers. To mitigate this harm, we introduce a synthetic data generation approach that produces diverse pronunciations of named entities using open-source text-to-speech models. Fine-tuning with less than 1,000 synthetic samples improves street name transcription accuracy by nearly 60% (relative to base models) for non-English primary speakers. Our results highlight a critical gap between benchmark performance and real-world reliability in speech systems and demonstrate a simple, scalable path to reducing high-stakes transcription errors. --- ### ExtractBench: A Benchmark and Evaluation Methodology for Complex Structured Extraction Authors: Nick Ferguson, Josh Pennington, Narek Beghian, Aravind Mohan, Douwe Kiela, Sheshansh Agrawal, Thien Hang Nguyen Published: 2026-02-12 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2602.12247v2 Unstructured documents like PDFs contain valuable structured information, but downstream systems require this data in reliable, standardized formats. LLMs are increasingly deployed to automate this extraction, making accuracy and reliability paramount. However, progress is bottlenecked by two gaps. First, no end-to-end benchmark evaluates PDF-to-JSON extraction under enterprise-scale schema breadth. Second, no principled methodology captures the semantics of nested extraction, where fields demand different notions of correctness (exact match for identifiers, tolerance for quantities, semantic equivalence for names), arrays require alignment, and omission must be distinguished from hallucination. We address both gaps with ExtractBench, an open-source benchmark and evaluation framework for PDF-to-JSON structured extraction. The benchmark pairs 35 PDF documents with JSON Schemas and human-annotated gold labels across economically valuable domains, yielding 12,867 evaluatable fields spanning schema complexities from tens to hundreds of fields. The evaluation framework treats the schema as an executable specification: each field declares its scoring metric. Baseline evaluations reveal that frontier models (GPT-5/5.2, Gemini-3 Flash/Pro, Claude 4.5 Opus/Sonnet) remain unreliable on realistic schemas. Performance degrades sharply with schema breadth, culminating in 0% valid output on a 369-field financial reporting schema across all tested models. We release ExtractBench at https://github.com/ContextualAI/extract-bench. --- ### Intrinsic-Energy Joint Embedding Predictive Architectures Induce Quasimetric Spaces Authors: Anthony Kobanda, Waris Radji Published: 2026-02-12 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2602.12245v1 Joint-Embedding Predictive Architectures (JEPAs) aim to learn representations by predicting target embeddings from context embeddings, inducing a scalar compatibility energy in a latent space. In contrast, Quasimetric Reinforcement Learning (QRL) studies goal-conditioned control through directed distance values (cost-to-go) that support reaching goals under asymmetric dynamics. In this short article, we connect these viewpoints by restricting attention to a principled class of JEPA energy functions : intrinsic (least-action) energies, defined as infima of accumulated local effort over admissible trajectories between two states. Under mild closure and additivity assumptions, any intrinsic energy is a quasimetric. In goal-reaching control, optimal cost-to-go functions admit exactly this intrinsic form ; inversely, JEPAs trained to model intrinsic energies lie in the quasimetric value class targeted by QRL. Moreover, we observe why symmetric finite energies are structurally mismatched with one-way reachability, motivating asymmetric (quasimetric) energies when directionality matters. --- ### Moonshine v2: Ergodic Streaming Encoder ASR for Latency-Critical Speech Applications Authors: Manjunath Kudlur, Evan King, James Wang, Pete Warden Published: 2026-02-12 Categories: cs.CL, cs.LG, cs.SD arXiv: https://arxiv.org/abs/2602.12241v1 Latency-critical speech applications (e.g., live transcription, voice commands, and real-time translation) demand low time-to-first-token (TTFT) and high transcription accuracy, particularly on resource-constrained edge devices. Full-attention Transformer encoders remain a strong accuracy baseline for automatic speech recognition (ASR) because every frame can directly attend to every other frame, which resolves otherwise locally ambiguous acoustics using distant lexical context. However, this global dependency incurs quadratic complexity in sequence length, inducing an inherent "encode-the-whole-utterance" latency profile. For streaming use cases, this causes TTFT to grow linearly with utterance length as the encoder must process the entire prefix before any decoder token can be emitted. To better meet the needs of on-device, streaming ASR use cases we introduce Moonshine v2, an ergodic streaming-encoder ASR model that employs sliding-window self-attention to achieve bounded, low-latency inference while preserving strong local context. Our models achieve state of the art word error rates across standard benchmarks, attaining accuracy on-par with models 6x their size while running significantly faster. These results demonstrate that carefully designed local attention is competitive with the accuracy of full attention at a fraction of the size and latency cost, opening new possibilities for interactive speech interfaces on edge devices. --- ### Olmix: A Framework for Data Mixing Throughout LM Development Authors: Mayee F. Chen, Tyler Murray, David Heineman, Matt Jordan, Hannaneh Hajishirzi, Christopher Ré, Luca Soldaini, Kyle Lo Published: 2026-02-12 Categories: cs.LG, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2602.12237v1 Data mixing -- determining the ratios of data from different domains -- is a first-order concern for training language models (LMs). While existing mixing methods show promise, they fall short when applied during real-world LM development. We present Olmix, a framework that addresses two such challenges. First, the configuration space for developing a mixing method is not well understood -- design choices across existing methods lack justification or consensus and overlook practical issues like data constraints. We conduct a comprehensive empirical study of this space, identifying which design choices lead to a strong mixing method. Second, in practice, the domain set evolves throughout LM development as datasets are added, removed, partitioned, and revised -- a problem setting largely unaddressed by existing works, which assume fixed domains. We study how to efficiently recompute the mixture after the domain set is updated, leveraging information from past mixtures. We introduce mixture reuse, a mechanism that reuses existing ratios and recomputes ratios only for domains affected by the update. Over a sequence of five domain-set updates mirroring real-world LM development, mixture reuse matches the performance of fully recomputing the mix after each update with 74% less compute and improves over training without mixing by 11.6% on downstream tasks. --- ### Energy-Aware Spike Budgeting for Continual Learning in Spiking Neural Networks for Neuromorphic Vision Authors: Anika Tabassum Meem, Muntasir Hossain Nadid, Md Zesun Ahmed Mia Published: 2026-02-12 Categories: cs.NE, cs.AI, cs.CV arXiv: https://arxiv.org/abs/2602.12236v2 Neuromorphic vision systems based on spiking neural networks (SNNs) offer ultra-low-power perception for event-based and frame-based cameras, yet catastrophic forgetting remains a critical barrier to deployment in continually evolving environments. Existing continual learning methods, developed primarily for artificial neural networks, seldom jointly optimize accuracy and energy efficiency, with particularly limited exploration on event-based datasets. We propose an energy-aware spike budgeting framework for continual SNN learning that integrates experience replay, learnable leaky integrate-and-fire neuron parameters, and an adaptive spike scheduler to enforce dataset-specific energy constraints during training. Our approach exhibits modality-dependent behavior: on frame-based datasets (MNIST, CIFAR-10), spike budgeting acts as a sparsity-inducing regularizer, improving accuracy while reducing spike rates by up to 47\%; on event-based datasets (DVS-Gesture, N-MNIST, CIFAR-10-DVS), controlled budget relaxation enables accuracy gains up to 17.45 percentage points with minimal computational overhead. Across five benchmarks spanning both modalities, our method demonstrates consistent performance improvements while minimizing dynamic power consumption, advancing the practical viability of continual learning in neuromorphic vision systems. --- ### Detecting Overflow in Compressed Token Representations for Retrieval-Augmented Generation Authors: Julia Belikova, Danila Rozhevskii, Dennis Svirin, Konstantin Polev, Alexander Panchenko Published: 2026-02-12 Categories: cs.CL arXiv: https://arxiv.org/abs/2602.12235v2 Efficient long-context processing remains a crucial challenge for contemporary large language models (LLMs), especially in resource-constrained environments. Soft compression architectures promise to extend effective context length by replacing long token sequences with smaller sets of learned compressed tokens. Yet, the limits of compressibility -- and when compression begins to erase task-relevant content -- remain underexplored. In this paper, we define token overflow as a regime in which compressed representations no longer contain sufficient information to answer a given query, and propose a methodology to characterize and detect it. In the xRAG soft-compression setting, we find that query-agnostic saturation statistics reliably separate compressed from uncompressed token representations, providing a practical tool for identifying compressed tokens but showing limited overflow detection capability. Lightweight probing classifiers over both query and context xRAG representations detect overflow with 0.72 AUC-ROC on average on HotpotQA, SQuADv2, and TriviaQA datasets, demonstrating that incorporating query information improves detection performance. These results advance from query-independent diagnostics to query-aware detectors, enabling low-cost pre-LLM gating to mitigate compression-induced errors. --- ### Categorical Flow Maps Authors: Daan Roos, Oscar Davis, Floor Eijkelboom, Michael Bronstein, Max Welling, İsmail İlkan Ceylan, Luca Ambrogioni, Jan-Willem van de Meent Published: 2026-02-12 Categories: cs.LG arXiv: https://arxiv.org/abs/2602.12233v1 We introduce Categorical Flow Maps, a flow-matching method for accelerated few-step generation of categorical data via self-distillation. Building on recent variational formulations of flow matching and the broader trend towards accelerated inference in diffusion and flow-based models, we define a flow map towards the simplex that transports probability mass toward a predicted endpoint, yielding a parametrisation that naturally constrains model predictions. Since our trajectories are continuous rather than discrete, Categorical Flow Maps can be trained with existing distillation techniques, as well as a new objective based on endpoint consistency. This continuous formulation also automatically unlocks test-time inference: we can directly reuse existing guidance and reweighting techniques in the categorical setting to steer sampling toward downstream objectives. Empirically, we achieve state-of-the-art few-step results on images, molecular graphs, and text, with strong performance even in single-step generation. --- ### Diffusion Alignment Beyond KL: Variance Minimisation as Effective Policy Optimiser Authors: Zijing Ou, Jacob Si, Junyi Zhu, Ondrej Bohdal, Mete Ozay, Taha Ceritli, Yingzhen Li Published: 2026-02-12 Categories: cs.LG arXiv: https://arxiv.org/abs/2602.12229v1 Diffusion alignment adapts pretrained diffusion models to sample from reward-tilted distributions along the denoising trajectory. This process naturally admits a Sequential Monte Carlo (SMC) interpretation, where the denoising model acts as a proposal and reward guidance induces importance weights. Motivated by this view, we introduce Variance Minimisation Policy Optimisation (VMPO), which formulates diffusion alignment as minimising the variance of log importance weights rather than directly optimising a Kullback-Leibler (KL) based objective. We prove that the variance objective is minimised by the reward-tilted target distribution and that, under on-policy sampling, its gradient coincides with that of standard KL-based alignment. This perspective offers a common lens for understanding diffusion alignment. Under different choices of potential functions and variance minimisation strategies, VMPO recovers various existing methods, while also suggesting new design directions beyond KL. --- ### Two-Sided Time-Independent Regret for Matching Markets with Limited Interviews Authors: Amirmahdi Mirfakhar, Xuchuang Wang, Mengfan Xu, Hedyeh Beyhaghi, Mohammad Hajiesmaili Published: 2026-02-12 Categories: cs.GT, cs.AI, econ.TH arXiv: https://arxiv.org/abs/2602.12224v2 Two-sided matching platforms rely on preferences from both sides, yet participants can evaluate only a small fraction of potential partners. In practice, they use low-cost pre-match screening, e.g., interviews, profile views, or trial tasks, to form noisy impressions before committing to applications and offers. We study bandit learning in matching markets with interviews, modeling these interactions as queried \emph{hints}~\citep{DBLP:conf/innovations/BhaskaraGIKM23} that reveal partial preference information to both sides while constraining subsequent applications. Our framework also allows firm-side uncertainty: firms, like agents, learn their preferences and may make early hiring mistakes. To address this, we introduce strategic deferral, a firm-side action that permits temporary vacancy, corrects premature commitments, and enables decentralized learning under coarse anonymous feedback. We design algorithms for centralized and decentralized markets and show that a constant number of interviews per round suffices for horizon-independent regret, improving over the $O(\log T)$ guarantees known without interviews. Our bounds are near-optimal: the centralized guarantee is within a factor $m$ of an information-theoretic lower bound, while decentralized algorithms match it up to polynomial factors in structured markets and remain horizon-independent in general markets. --- ### Towards On-Policy SFT: Distribution Discriminant Theory and its Applications in LLM Training Authors: Miaosen Zhang, Yishan Liu, Shuxia Lin, Xu Yang, Qi Dai, Chong Luo, Weihao Jiang, Peng Hou, Anxiang Zeng, Xin Geng, Baining Guo Published: 2026-02-12 Categories: cs.LG, cs.AI, cs.CV arXiv: https://arxiv.org/abs/2602.12222v2 Supervised fine-tuning (SFT) is computationally efficient but often yields inferior generalization compared to reinforcement learning (RL). This gap is primarily driven by RL's use of on-policy data. We propose a framework to bridge this chasm by enabling On-Policy SFT. We first present \textbf{\textit{Distribution Discriminant Theory (DDT)}}, which explains and quantifies the alignment between data and the model-induced distribution. Leveraging DDT, we introduce two complementary techniques: (i) \textbf{\textit{In-Distribution Finetuning (IDFT)}}, a loss-level method to enhance generalization ability of SFT, and (ii) \textbf{\textit{Hinted Decoding}}, a data-level technique that can re-align the training corpus to the model's distribution. Extensive experiments demonstrate that our framework achieves generalization performance surpassing prominent offline RL algorithms, including DPO and SimPO, while maintaining the efficiency of an SFT pipeline. The proposed framework thus offers a practical alternative in domains where RL is infeasible. We open-source the code here: https://github.com/zhangmiaosen2000/Towards-On-Policy-SFT --- ### Best of Both Worlds: Multimodal Reasoning and Generation via Unified Discrete Flow Matching Authors: Onkar Susladkar, Tushar Prakash, Gayatri Deshmukh, Kiet A. Nguyen, Jiaxun Zhang, Adheesh Juvekar, Tianshu Bao, Lin Chai, Sparsh Mittal, Inderjit S Dhillon, Ismini Lourentzou Published: 2026-02-12 Categories: cs.CV arXiv: https://arxiv.org/abs/2602.12221v2 We propose UniDFlow, a unified discrete flow-matching framework for multimodal understanding, generation, and editing. It decouples understanding and generation via task-specific low-rank adapters, avoiding objective interference and representation entanglement, while a novel reference-based multimodal preference alignment optimizes relative outcomes under identical conditioning, improving faithfulness and controllability without large-scale retraining. UniDFlpw achieves SOTA performance across eight benchmarks and exhibits strong zero-shot generalization to tasks including inpainting, in-context image generation, reference-based editing, and compositional generation, despite no explicit task-specific training. --- ### The Observer Effect in World Models: Invasive Adaptation Corrupts Latent Physics Authors: Christian Internò, Jumpei Yamaguchi, Loren Amdahl-Culleton, Markus Olhofer, David Klindt, Barbara Hammer Published: 2026-02-12 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2602.12218v1 Determining whether neural models internalize physical laws as world models, rather than exploiting statistical shortcuts, remains challenging, especially under out-of-distribution (OOD) shifts. Standard evaluations often test latent capability via downstream adaptation (e.g., fine-tuning or high-capacity probes), but such interventions can change the representations being measured and thus confound what was learned during self-supervised learning (SSL). We propose a non-invasive evaluation protocol, PhyIP. We test whether physical quantities are linearly decodable from frozen representations, motivated by the linear representation hypothesis. Across fluid dynamics and orbital mechanics, we find that when SSL achieves low error, latent structure becomes linearly accessible. PhyIP recovers internal energy and Newtonian inverse-square scaling on OOD tests (e.g., $ρ> 0.90$). In contrast, adaptation-based evaluations can collapse this structure ($ρ\approx 0.05$). These findings suggest that adaptation-based evaluation can obscure latent structures and that low-capacity probes offer a more accurate evaluation of physical world models. --- ### VIRENA: Virtual Arena for Research, Education, and Democratic Innovation Authors: Emma Hoes, K. Jonathan Klueser, Fabrizio Gilardi Published: 2026-02-12 Categories: cs.HC, cs.AI, cs.SI arXiv: https://arxiv.org/abs/2602.12207v2 Digital platforms shape how people communicate, deliberate, and form opinions. Studying these dynamics has become increasingly difficult due to restricted data access, ethical constraints on real-world experiments, and limitations of existing research tools. VIRENA (Virtual Arena) is a platform that enables controlled experimentation in realistic social media environments. Multiple participants interact simultaneously in realistic replicas of feed-based platforms (Instagram, Facebook, Reddit) and messaging apps (WhatsApp, Messenger). Large language model-powered AI agents participate alongside humans with configurable personas and realistic behavior. Researchers can manipulate content moderation approaches, pre-schedule stimulus content, and run experiments across conditions through a visual interface requiring no programming skills. VIRENA makes possible research designs that were previously impractical: studying human--AI interaction in realistic social contexts, experimentally comparing moderation interventions, and observing group deliberation as it unfolds. Built on open-source technologies that ensure data remain under institutional control and comply with data protection requirements, VIRENA is currently in use at the University of Zurich and available for pilot collaborations. Designed for researchers, educators, and public organizations alike, VIRENA's no-code interface makes controlled social media simulation accessible across disciplines and sectors. This paper documents its design, architecture, and capabilities. --- ### DeepGen 1.0: A Lightweight Unified Multimodal Model for Advancing Image Generation and Editing Authors: Dianyi Wang, Ruihang Li, Feng Han, Chaofan Ma, Wei Song, Siyuan Wang, Yibin Wang, Yi Xin, Hongjian Liu, Zhixiong Zhang, Shengyuan Ding, Tianhang Wang et al. Published: 2026-02-12 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2602.12205v2 Current unified multimodal models for image generation and editing typically rely on massive parameter scales (e.g., >10B), entailing prohibitive training costs and deployment footprints. In this work, we present DeepGen 1.0, a lightweight 5B unified model that achieves comprehensive capabilities competitive with or surpassing much larger counterparts. To overcome the limitations of compact models in semantic understanding and fine-grained control, we introduce Stacked Channel Bridging (SCB), a deep alignment framework that extracts hierarchical features from multiple VLM layers and fuses them with learnable 'think tokens' to provide the generative backbone with structured, reasoning-rich guidance. We further design a data-centric training strategy spanning three progressive stages: (1) Alignment Pre-training on large-scale image-text pairs and editing triplets to synchronize VLM and DiT representations, (2) Joint Supervised Fine-tuning on a high-quality mixture of generation, editing, and reasoning tasks to foster omni-capabilities, and (3) Reinforcement Learning with MR-GRPO, which leverages a mixture of reward functions and supervision signals, resulting in substantial gains in generation quality and alignment with human preferences, while maintaining stable training progress and avoiding visual artifacts. Despite being trained on only ~50M samples, DeepGen 1.0 achieves leading performance across diverse benchmarks, surpassing the 80B HunyuanImage by 28% on WISE and the 27B Qwen-Image-Edit by 37% on UniREditBench. By open-sourcing our training code, weights, and datasets, we provide an efficient, high-performance alternative to democratize unified multimodal research. --- ### Learning to Forget Attention: Memory Consolidation for Adaptive Compute Reduction Authors: Ibne Farabi Shihab, Sanjeda Akter, Anuj Sharma Published: 2026-02-12 Categories: cs.LG arXiv: https://arxiv.org/abs/2602.12204v1 Hybrid architectures combining state-space models with attention have achieved strong efficiency-quality tradeoffs, yet existing approaches either apply attention uniformly or learn static sparse patterns. This misses a key opportunity: \emph{attention demand should decrease over time as recurring patterns become familiar}. We present a surprising finding from analyzing GPT-2 models: \textbf{88\%} of attention operations retrieve information already predictable from the model's hidden state, and this redundancy does \emph{not} decrease during training. Motivated by this observation, we introduce \textbf{\ours{}} (\textbf{C}onsolidation-based \textbf{R}outing for \textbf{A}daptive \textbf{M}emory), a biologically inspired memory consolidation mechanism that gradually distills episodic retrievals into parametric semantic memory. Unlike prior sparse attention methods, \ours{} exhibits \emph{decreasing attention utilization} over training, achieving a \textbf{37.8$\times$} reduction through a sharp phase transition at approximately 3K steps. We prove that this capability is \emph{impossible} without consolidation: any static routing scheme requires $Ω(f \cdot n)$ attention for tasks with recurring patterns of frequency $f$. On our proposed SRCD benchmark, \ours{} achieves \textbf{100\% retrieval accuracy} at 1.6\% attention compute (vs.\ 68\% for baselines), and consolidated patterns transfer to unseen tasks with \textbf{48--52\%} attention reduction without retraining. Remarkably, the learned consolidation dynamics quantitatively match human episodic-to-semantic memory transition curves from cognitive psychology ($γ= 0.43$ vs.\ $γ_{\text{human}} \approx 0.4$--$0.5$). Code and benchmarks are available at (anonymized). ---