Sky-image irradiance forecasting is dominated by a two-stage design: a video-prediction model forecasts future sky frames, and a separately trained regressor maps those frames to irradiance. We propose a single-stage alternative that co-denoises the future sky-image and clear-sky-index sequences in one diffusion process, with an irradiance head that reads a denoised preview of the future sky rather than the noised diffusion state. The forecaster needs no regression stage and runs from sky images and deterministic covariates alone, with no measured irradiance at its input --- the situation at a site with a camera but no pyranometer. On 2023 SIRTA data the model reaches a pooled RMSE of $104.1$~W\,m$^{-2}$ and a skill score of $+0.34$ on the continuous ranked probability score (CRPS) against smart persistence, higher than any of twelve baseline configurations, and retains that skill when retrained on 2021--2022. On the model's own generated frames, a separately trained regressor reaches $150.6$~W\,m$^{-2}$ against the joint head's $104.1$: roughly $46$~W\,m$^{-2}$ that joint training recovers and a cascade discards. Point accuracy is competitive but not best: a deterministic network that also reads measured irradiance attains $97.6$~W\,m$^{-2}$. The model's CRPS skill leads that network's in every sky-condition tertile, by $0.26$ to $0.30$.
Real-world environments are messy, with paths often blocked by everyday objects like chairs and toys, requiring robots to perform interactive navigation by manipulating obstacles to clear a path. Existing methods rely on extensive simulation-based learning or per-scene instructions specifying which objects to avoid or manipulate. They struggle to generalize across novel objects, diverse environments, and varied contexts such as a cart of fragile glassware that should be avoided while an empty one can be pushed. We formalize this problem as open-world interactive navigation and present MessyNav, a zero-shot framework based on vision-language models (VLMs). At its core, MessyNav performs explicit VLM-based affordance reasoning before task planning to decide whether to swipe, pick up, or avoid an object and produce a high-level plan. Evaluated on a mobile manipulator across real-world indoor and outdoor environments, MessyNav improves success rate by 27 percentage points (pp) over the strongest baseline in real-robot trials and achieves statistically significant gains in a larger-scale offline evaluation, without task-specific training.
Training compute is increasingly outpacing the availability of high-quality data. This shifts the central challenge from optimal compute allocation to extracting maximum value from limited data. The widely adopted Chinchilla scaling law assumes every training token is unique. This limits its ability to guide pretraining decisions in data-constrained regimes. We model the excess loss under repetition with a simple additive overfitting penalty and find that it accurately describes model behavior. Our scaling law yields qualitatively new compute-optimal allocation advice. Beyond a point, further repetition is counterproductive and compute is better spent on model capacity. We show that following our law's recommended configuration improves performance in data-constrained regimes. Finally, because our one-parameter form isolates overfitting in a single coefficient, it enables direct comparison across training configurations. As a case study, we show that strong weight decay ($λ=1.0$) reduces this coefficient by approximately 70\%, providing a scaling-law explanation for recent findings that optimal weight decay in data-constrained regimes is an order of magnitude larger than standard practice.
Michal Korniak, Marco Bagatella, Ralf Römer, Michal Nauman, Yarden As, Pranav Vignesh, Manan Agarwal, Leander Diaz-Bone 等 24 位作者
AI 速览：作者系统评估通用机器人策略能否在部署阶段通过在线强化学习实现多任务同时改进，并探索使该改进泛化到未训练任务的关键算法配方。
The introduction of generalist robot policies has enabled a single, large model to learn to perform a variety of tasks from a vast pre-training corpus. While these models often exhibit powerful zero-shot generalization capabilities, they typically require an adaptation phase to achieve sufficient performance in deployment. Iterated self-improvement via reinforcement learning (RL) has demonstrated promise in enabling this adaptation, yet such approaches have been largely explored in single-task settings - where the goal is to simply improve on a narrow deployment task - which lies in stark contrast to the multi-task capabilities of the base policy. In this work we ask whether generalist robot policies are capable of multi-task, generalist improvement in deployment - targeting many diverse tasks simultaneously - and what algorithmic recipe enables this. To answer this question, we first develop a set of structured evaluations to test generalist improvement capabilities, and systematically evaluate algorithmic and practical design choices regarding value estimation, policy extraction and resource allocation. We show that not only can generalist policies be improved online across diverse tasks, but that the right recipe can enable generalization and increase performance on tasks not considered during online training. We characterize when this generalization occurs and show that both the training task distribution and the use of value-based RL are critical. To the best of our knowledge, this is among the first evidence for online generalist improvement for robot foundation models.
Hehui Zheng, Bhavya Sukhija, Chenhao Li, Klemens Iten, Andreas Krause, Robert K. Katzschmann
Soft robots offer adaptability and safety in unstructured environments, yet their compliant, high-dimensional, and nonlinear dynamics make modeling for control difficult. Existing data-driven approaches often fail to generalize, constrained by narrowly focused task demonstrations or inefficient random exploration. We introduce an uncertainty-aware active exploration framework that autonomously learns task-agnostic and generalizable dynamics models of soft robotic systems. We evaluate this framework on three simulated platforms---a continuum arm, an articulated fish in fluid, and a musculoskeletal leg---and on a real-world pneumatically actuated continuum arm. Compared with random exploration and task-specific baseline, our method achieves higher modeling accuracy by actively collecting informative data. This improved model quality yields higher returns in zero-shot control on unseen tasks and robust performance under sensing noise, actuation delays, and material nonlinearities. These results demonstrate that uncertainty-driven active exploration enables data-efficient dynamics learning across soft robotic morphologies, supporting autonomous and adaptable control for compliant robots.
We evaluate, at implementation level and under an explicit threat model, the verification pipelines of three Verifiable Credential (VC) formats: SD-JWT VC; the W3C Verifiable Credentials Data Model 2.0 with the Data Integrity cryptosuite eddsa-rdfc-2022 (W3C VCDM); and mdoc (ISO/IEC 18013-5). For W3C VCDM we identify and exercise three structural attack surfaces from remote context resolution and RDFC-1.0 canonicalization: denial of service (DoS) via poison graphs, context injection through @context, and server-side request forgery (SSRF) reachable via malicious URLs, all at default settings. A poison graph, whose blank-node references are symmetric enough to force RDFC-1.0 into deep recursion, raises it to 278x the baseline from five blank nodes in 527 bytes, and to 13.5 s from eight. Every experiment ran in four environments sharing runtime and library versions; the multiple moves with the machine (251--424x for one input), not with its speed. Cost tracks that symmetry rather than the node count: an acyclic input with 1,500 blank nodes stays at 462--714x. A call limit separates the two: every attack family aborts at 16 or below, every legitimate credential completes at 1, but a shallow, wide recursion, such as the acyclic control, passes whatever the bound. DoS does not presuppose adversarial input: with production education schemas, canonicalization costs 23--34x a simple credential in one environment, 16--49x across four. Effective mitigations are a static document loader, which removes the retrieval latency and both remote-context surfaces; a context allowlist; and a call limit with a size bound. SD-JWT VC and mdoc need neither canonicalization nor external references, and our verifiers resisted every vector evaluated. These are not format-specific defects but properties of the pipeline and its configuration.
We compare the signing and verification performance of three Verifiable Credential (VC) formats--SD-JWT VC; the W3C Verifiable Credentials Data Model 2.0 with a Data Integrity cryptosuite (W3C VCDM); and mdoc (ISO/IEC 18013-5)--over one scope: issue, present with selective disclosure, verify. Node.js, Go and Python implementations ran in four environments, one a cloud VM (N=2,000 across five runs; medians of each statistic and the spread). For an end-to-end exchange disclosing 5 of 20 attributes, the total (issue, present, verify) on Node.js in the cloud VM is 0.265 ms for mdoc, 0.298 ms for SD-JWT VC and 0.534 ms for W3C VCDM. The four span a factor of five in speed; across them W3C VCDM takes 1.73--2.04x mdoc and 1.57--1.79x SD-JWT VC--at least five times our 10% band--and the ordering survives unifying the algorithm on Ed25519 and P-256. The mdoc--SD-JWT VC difference is not robust: the totals differ by 12%, 28%, 19% and 10% in the four environments, and at verification mdoc is the slower in one of them. RDFC-1.0 canonicalization produces the gap: the cryptosuite canonicalizes twice per operation, so one exchange canonicalizes six times. We attribute that cost to the specification rather than to a library mainly by replacing RDFC-1.0 with JCS, which brings W3C VCDM verification to within 5% of SD-JWT VC in three environments and within 12% in the fourth; three implementations in three languages agree, though they share a lineage. Because eddsa-rdfc-2022 has no derivation step, presenting a subset re-signs it; ecdsa-sd-2023, which preserves the issuer signature, costs 3.4 ms to issue and 1.5--2.4 ms to derive and verify there. With production education schemas, canonicalization costs 31--49x a simple credential across the four, and blank node labeling is not the driver. The measured path ends at the issuer signature check.
Recent progress has rapidly advanced our understanding of the mechanisms underlying in-context learning in modern attention-based neural networks. However, existing results focus exclusively on unimodal data; in contrast, the theoretical underpinnings of in-context learning for multi-modal data remain poorly understood. We introduce a mathematically tractable framework for studying multi-modal learning and explore when transformer-like architectures can recover Bayes-optimal performance in-context. To model multi-modal problems, we assume the observed data arises from a latent factor model. Our first result comprises a negative take on expressibility: we prove that single-layer self-attention models fail to recover the Bayes-optimal predictor uniformly over the task distribution. To address this limitation, we introduce a novel, linearized cross-attention mechanism, which we study in the regime where both the number of cross-attention layers and the context length are large. We show that this cross-attention mechanism is provably Bayes optimal when optimized using gradient flow. Our results underscore the benefits of depth for in-context learning and establish the provable utility of cross-attention for multi-modal distributions.
We introduce Nemotron 3.5 Super and Nemotron 3.5 Lightning, open and efficient Mixture-of-Experts Hybrid Mamba-Transformer models designed for reasoning, long-context understanding, and agentic workloads. Super is a 120B-parameter, 12B-active multimodal model, while Lightning is a smaller 30B-parameter, 3B-active text model; both designed for efficient deployment. Both models combine large-scale distillation, long-context training, and native speculative decoding, while Super further incorporates Indic continued pretraining and agentic software-engineering mid-training. We develop a comprehensive post-training pipeline spanning supervised fine-tuning and reinforcement learning (RLVR and RLHF), complemented in Super by specialized teacher training and multi- teacher on-policy distillation (MOPD). This transforms Super from a text-only base model into a multimodal agent capable of reasoning over text, images, video, and interactive tool environments. We further explore test-time scaling through parallel agent execution and self-verification. Across training and deployment, we jointly optimize capability and efficiency. Indic continued pretraining improves Indian-language benchmarks by up to 9.82 percentage points, while agentic mid-training strengthens software-engineering capabilities and robustness across agent harnesses. The post-trained Super model achieves 51.78% on SWE-bench Pro, surpassing Nemotron 3 Ultra despite using approximately 4.6x fewer active parameters. Post-training MTP boosting improves speculative- decoding acceptance length by up to 21.8% under probabilistic draft sampling, while NVFP4 quantization achieves 99.62% median accuracy recovery relative to BF16. Inference optimizations enable up to 4,856 output tokens/s/GPU on GB200 while sustaining at least 50 tokens/s/user.
My Chiffon Nguyen, Tianyi Hu, Vinod Anbalagan, Henry Gagnier, Pouya Sadeghi, Farah Abdou, Tidor-Vlad Pricope, Diego MMS Oliveira 等 164 位作者
As LLMs expand their language coverage, evaluating how well they understand and represent diverse cultures becomes increasingly important. However, many benchmarks remain English-centric, with limited cultural coverage and a focus on factual knowledge over its application in real-world cultural contexts. We propose CulturalRiddles, a benchmark of original, culturally grounded riddles created entirely by humans through a participatory community effort, spanning 62 cultures (45 countries, 51 languages). These riddles are deeply rooted in their respective cultures and require abductive and compositional reasoning, alongside detailed factual and social knowledge, language skills, and an understanding of culturally specific associations. We benchmark 28 LLM configurations using both automatic and large-scale human evaluation. Even frontier models achieve at most 22% accuracy in some communities. We publicly release all artifacts alongside an open contribution process to expand the benchmark's linguistic and cultural coverage.
Yuanbo Xie, Yingjie Zhang, Yulin Li, Shouyou Song, Xiaokun Chen Stanford University, Zhihan Liu, Liya Su, Tingwen Liu Institute of Information Engineering, Chinese Academy of Sciences
Retrieval-Augmented Generation (RAG) systems augment large language models with external knowledge, yet introduce a critical security vulnerability: RAG Knowledge Base Leakage, wherein adversarial prompts can induce the model to divulge retrieved proprietary content. Recent studies reveal that such leakage can be executed through adaptive and iterative attack strategies (named RAG extraction attack), while effective countermeasures remain notably lacking. To bridge this gap, we propose CanaryRAG, a runtime defense mechanism inspired by stack canaries in software security. CanaryRAG embeds carefully designed canary tokens into retrieved chunks and reformulates RAG extraction defense as a dual-path runtime integrity game. Leakage is detected in real time whenever either the target or oracle path violates its expected canary behavior, including under adaptive suppression and obfuscation. Extensive evaluations against existing attacks demonstrate that CanaryRAG provides robust defense, achieving substantially lower chunk recovery rates than state-of-the-art baselines while imposing negligible impact on task performance and inference latency. Moreover, as a plug-and-play solution, CanaryRAG can be seamlessly integrated into arbitrary RAG pipelines without requiring retraining or structural modifications, offering a practical and scalable safeguard for proprietary data.
尚无 AI 解读，展示原文摘要
cs.CL
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表
Ya Su, Hu Zhang Shanxi University, Dan Qiao, YuJie Wang, Yunxiao Zhao Shanxi University, Yue Fan Shanxi University, Shike Li Shanxi University, Ru Li Shanxi University 等 9 位作者
Document-level Event Causality Identification (DECI) aims to identify causal relations among multiple events within unstructured text. Existing methods predominantly rely on local semantic similarity for independent event-pair discrimination, thereby overlooking the influence of the overall narrative backbone in the propagation of causal dependencies and the role differentiation of events within multi-cause/multi-effect structures. Therefore, we propose a suggest-verify-revise approach for document-level Event Causality Identification with narrative consistency (SVRECI). In the suggest stage, we integrate multi-dimensional heuristic causal suggestions generated by an LLM with structural suggestions derived from hypergraph modeling to provide multi-source initial support for candidate event pairs. In the verify stage, we introduce a Topological Hawkes process to perform constrained verification of narrative propagation consistency among events. In the revise stage, we construct a dynamically evolving document-level causal graph and incorporate a structure-aware dual-level contrastive learning mechanism at both the event and event-pair levels, iteratively reducing noisy edges over multiple iterations. Experimental results on EventStoryLine and Causal-TimeBank datasets demonstrate that our approach outperforms previous methods.
尚无 AI 解读，展示原文摘要
cs.CL
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表
Angelina Tsai, Shreya Subramanian, Catherine Liu, Kimberly Lopez, Leif Zinn-Brooks, Alexia E. Schulz, Adaku Uchendu MIT Lincoln Laboratory
Adversarial perturbations in the context of large language models (LLMs) are subtle changes added to input data (i.e., images or text) that are designed to alter predictions or outputs of machine learning models. We introduce several novel visualizations using topological data analysis (TDA) (leveraging persistent homology) to characterize how adversarial perturbations act on text inputs, specifically, how sandbagging and code-injection attacksalter the geometric structure of attention heads in transformer models. By computing persistent homology metrics from attention maps across different model architectures (such as BERT, RoBERTa, ELECTRA, DistilGPT, etc.), we find that adversarial inputs alter higher-dimensional topological features (H_{1} loops and H_{2} voids) in ways that distinguish them from clean, non-adversarial inputs.
尚无 AI 解读，展示原文摘要
cs.CL
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 4: Student Research Workshop) · 2026 · 正式发表
Advances in large reasoning models have shown strong performance on complex reasoning tasks by scaling test-time compute through extended inference-time thinking. However, recent studies observe that in vision-dependent tasks, extended textual reasoning at inference time can often degrade performance as models progressively lose attention to visual tokens, increasingly relying on textual priors alone. To address this, prior works used reinforcement learning (RL)-based fine-tuning to route visual tokens or employ refocusing mechanisms during reasoning. While effective, these methods are computationally expensive, requiring large-scale data generation and policy optimization. To leverage the benefits of inference-time compute without additional RL fine-tuning, we propose VisRef, a visually grounded test-time scaling framework. Our key idea is to actively guide the reasoning process through re-injecting a coreset of visual tokens that are semantically relevant to the reasoning context yet diverse and globally representative of the image for more grounded multi-modal reasoning. Experiments on three visual reasoning benchmarks with state-of-the-art multi-modal large reasoning models demonstrate that under fixed inference-time compute budgets, VisRef consistently outperforms existing test-time scaling approaches by up to 6.4%.
尚无 AI 解读，展示原文摘要
cs.CV
Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition · 2026 · 正式发表
Mark Chen, Jerry Tworek, Heewoo Jun, Qiming Yuan, Henrique Ponde de Oliveira Pinto, Jared Kaplan, Harri Edwards, Yuri Burda 等 58 位作者
We introduce Codex, a GPT language model fine-tuned on publicly available code from GitHub, and study its Python code-writing capabilities. A distinct production version of Codex powers GitHub Copilot. On HumanEval, a new evaluation set we release to measure functional correctness for synthesizing programs from docstrings, our model solves 28.8% of the problems, while GPT-3 solves 0% and GPT-J solves 11.4%. Furthermore, we find that repeated sampling from the model is a surprisingly effective strategy for producing working solutions to difficult prompts. Using this method, we solve 70.2% of our problems with 100 samples per problem. Careful investigation of our model reveals its limitations, including difficulty with docstrings describing long chains of operations and with binding operations to variables. Finally, we discuss the potential broader impacts of deploying powerful code generation technologies, covering safety, security, and economics.
Tom B. Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam 等 31 位作者
Recent work has demonstrated substantial gains on many NLP tasks and benchmarks by pre-training on a large corpus of text followed by fine-tuning on a specific task. While typically task-agnostic in architecture, this method still requires task-specific fine-tuning datasets of thousands or tens of thousands of examples. By contrast, humans can generally perform a new language task from only a few examples or from simple instructions - something which current NLP systems still largely struggle to do. Here we show that scaling up language models greatly improves task-agnostic, few-shot performance, sometimes even reaching competitiveness with prior state-of-the-art fine-tuning approaches. Specifically, we train GPT-3, an autoregressive language model with 175 billion parameters, 10x more than any previous non-sparse language model, and test its performance in the few-shot setting. For all tasks, GPT-3 is applied without any gradient updates or fine-tuning, with tasks and few-shot demonstrations specified purely via text interaction with the model. GPT-3 achieves strong performance on many NLP datasets, including translation, question-answering, and cloze tasks, as well as several tasks that require on-the-fly reasoning or domain adaptation, such as unscrambling words, using a novel word in a sentence, or performing 3-digit arithmetic. At the same time, we also identify some datasets where GPT-3's few-shot learning still struggles, as well as some datasets where GPT-3 faces methodological issues related to training on large web corpora. Finally, we find that GPT-3 can generate samples of news articles which human evaluators have difficulty distinguishing from articles written by humans. We discuss broader societal impacts of this finding and of GPT-3 in general.
Christian Szegedy, Wojciech Zaremba, Ilya Sutskever, Joan Bruna, Dumitru Erhan, Ian Goodfellow, Rob Fergus
Deep neural networks are highly expressive models that have recently achieved state of the art performance on speech and visual recognition tasks. While their expressiveness is the reason they succeed, it also causes them to learn uninterpretable solutions that could have counter-intuitive properties. In this paper we report two such properties. First, we find that there is no distinction between individual high level units and random linear combinations of high level units, according to various methods of unit analysis. It suggests that it is the space, rather than the individual units, that contains of the semantic information in the high layers of neural networks. Second, we find that deep neural networks learn input-output mappings that are fairly discontinuous to a significant extend. We can cause the network to misclassify an image by applying a certain imperceptible perturbation, which is found by maximizing the network's prediction error. In addition, the specific nature of these perturbations is not a random artifact of learning: the same perturbation can cause a different network, that was trained on a different subset of the dataset, to misclassify the same input.