arXiv subjects
Publications and source records attributed to :.
In this talk, we use several examples to elaborate on how a recently proposed algorithm can turn non-trivial Feynman integrals into an $\varepsilon $-factorised manner, regardless of their hidden geometric essence. In particular, some extra details about three-loop banana integrals with unequal-mass configuration are provided.
The discovery of novel solid-state materials, such as electrocatalysts, super-ionic conductors, or photovoltaic materials, plays a critical role in addressing various global challenges. It has, for instance, the potential to significantly improve the efficiency of renewable energy production and storage, thereby making substantial contributions to climate crisis mitigation strategies. In this paper, we introduce Crystal-GFN, a generative model of crystal structures possessing desirable properties and constraints. Operating as a multi-environment, continuous-discrete GFlowNet, it sequentially samples structural attributes of crystalline materials, namely space group, composition and lattice parameters. This domain-inspired approach enables the flexible incorporation of physicochemical and geometric hard constraints. We demonstrate the capabilities of Crystal-GFN to efficiently discover diverse and valid crystals with various properties: low predicted formation energy (median -3.2 eV/atom), band gap close to a target value and high density. Overall, Crystal-GFN is a crystal generation method that addresses several existing challenges in the literature and opens promising paths for accelerating materials discovery with machine learning.
The ESSnuSB experiment aims to measure the leptonic CP phase $δ_{CP}$ with an unprecedented resolution by probing neutrino oscillations at the second oscillation maximum. In the present work, the complementarity between the long-baseline neutrino program and atmospheric neutrinos is investigated for ESSnuSB. By simulating atmospheric neutrino events equivalent of 5.4 Mt$\cdot$year exposure, the resolution for $δ_{\rm CP}^{}$ is found to improve from $7.5^\circ$ ($6.7^\circ$) to $7.1^\circ$ ($6.5^\circ$) at $1σ$~CL for $δ_{\rm CP}^{} = -90^\circ$ ($+90^\circ$) with respect to super-beam neutrinos, resolving also the degeneracies arising from neutrino mass ordering. These findings highlight the synergies that exist between super-beam neutrinos and atmospheric neutrinos in ESSnuSB.
Multilinguality is a core capability for modern foundation models, yet training high-quality multilingual models remains challenging due to uneven data availability across languages. A further challenge is the performance interference that can arise from joint multilingual training, commonly referred to as the "curse of multilinguality". We study multilingual data curation across thirteen languages and find that many reported regressions are not inherent to multilingual scaling but instead stem from correctable deficiencies in data quality and composition rather than fundamental capacity limits. In controlled bilingual experiments, improving data quality for any single language benefits others: curating English improves non-English performance in 12 of 13 languages, while curating non-English yields reciprocal improvements in English. Bespoke per-language curation produces substantially larger within-language improvements. Extending these findings to large-scale general-purpose training mixtures, we show that curated multilingual allocations comprising under 8% of total tokens remain remarkably effective. We operationalize this approach within an effort that produced a 20T-token pretraining corpus derived entirely from public sources. Models with 3B and 8B parameters trained on a 1T-token random subset achieve competitive multilingual accuracy with 4-10x fewer training FLOPs than strong public baselines, establishing a new Pareto frontier in multilingual performance versus compute. Moreover, these benefits extend to frontier model scale: the 20T-token corpus served as part of the pretraining dataset for Trinity Large (400B/A13B), which exhibits strong multilingual performance relative to its training FLOPs. These results show that targeted, per-language data curation mitigates multilingual interference and enables compute-efficient multilingual scaling.
Using data samples of 988.4 fb$^{-1}$ and 427.9 fb$^{-1}$ collected with the Belle and Belle II detectors, we present a study of the singly Cabibbo-suppressed decays $Ξ_c^{0} \to Λη$, $Λη'$, and $Λπ^0$. We observe the decay $Ξ_c^0 \to Λη$ and find evidence for the decay $Ξ_c^0 \to Λη'$, with corresponding branching ratios determined to be ${\mathcal{B}(Ξ_c^0 \to Λη)}/{\mathcal{B}(Ξ_c^0 \to Ξ^- π^+)}= (4.16 \pm 0.91 \pm {0.23})\%$ and ${\mathcal{B}(Ξ_c^0 \to Λη')}/{\mathcal{B}(Ξ_c^0 \to Ξ^- π^+)}= (2.48 \pm 0.82 \pm {0.12})\%$, respectively. We find no significant signal in the $Ξ_c^0 \to Λπ^0$ decay mode and set an upper limit at the 90% credibility level of ${\mathcal{B}(Ξ_c^0 \to Λπ^0)}/{\mathcal{B}(Ξ_c^0 \to Ξ^- π^+)}< {3.5\%}$. Multiplying these ratios by the world-average branching fraction of the normalization channel, $\mathcal{B}(Ξ_c^0 \to Ξ^- π^+)=(1.43 \pm 0.27)\%$, we obtain the absolute branching fractions of $\mathcal{B}(Ξ_c^0 \to Λη)= (5.95 \pm 1.30 \pm {0.32} \pm 1.13) \times 10^{-4}$, $\mathcal{B}(Ξ_c^0 \to Λη')= (3.55 \pm 1.17 \pm {0.17} \pm 0.68) \times 10^{-4}$, and an upper limit at the 90% credibility level on the absolute branching fraction of $\mathcal{B}(Ξ_c^0 \to Λπ^0)< {5.2} \times 10^{-4}$. The quoted first and second uncertainties are statistical and systematic, respectively, while the third uncertainties arise from the branching fraction of the normalization mode. These results are consistent with most theoretical predictions and further the understanding of the underlying decay mechanisms.
We introduce [Cosmos-Predict2.5], the latest generation of the Cosmos World Foundation Models for Physical AI. Built on a flow-based architecture, [Cosmos-Predict2.5] unifies Text2World, Image2World, and Video2World generation in a single model and leverages [Cosmos-Reason1], a Physical AI vision-language model, to provide richer text grounding and finer control of world simulation. Trained on 200M curated video clips and refined with reinforcement learning-based post-training, [Cosmos-Predict2.5] achieves substantial improvements over [Cosmos-Predict1] in video quality and instruction alignment, with models released at 2B and 14B scales. These capabilities enable more reliable synthetic data generation, policy evaluation, and closed-loop simulation for robotics and autonomous systems. We further extend the family with [Cosmos-Transfer2.5], a control-net style framework for Sim2Real and Real2Real world translation. Despite being 3.5$\times$ smaller than [Cosmos-Transfer1], it delivers higher fidelity and robust long-horizon video generation. Together, these advances establish [Cosmos-Predict2.5] and [Cosmos-Transfer2.5] as versatile tools for scaling embodied intelligence. To accelerate research and deployment in Physical AI, we release source code, pretrained checkpoints, and curated benchmarks under the NVIDIA Open Model License at https://github.com/nvidia-cosmos/cosmos-predict2.5 and https://github.com/nvidia-cosmos/cosmos-transfer2.5. We hope these open resources lower the barrier to adoption and foster innovation in building the next generation of embodied intelligence.
We present GLM-5, a next-generation foundation model designed to transition the paradigm of vibe coding to agentic engineering. Building upon the agentic, reasoning, and coding (ARC) capabilities of its predecessor, GLM-5 adopts DSA to significantly reduce training and inference costs while maintaining long-context fidelity. To advance model alignment and autonomy, we implement a new asynchronous reinforcement learning infrastructure that drastically improves post-training efficiency by decoupling generation from training. Furthermore, we propose novel asynchronous agent RL algorithms that further improve RL quality, enabling the model to learn from complex, long-horizon interactions more effectively. Through these innovations, GLM-5 achieves state-of-the-art performance on major open benchmarks. Most critically, GLM-5 demonstrates unprecedented capability in real-world coding tasks, surpassing previous baselines in handling end-to-end software engineering challenges. Code, models, and more information are available at https://github.com/zai-org/GLM-5.
We introduce Habilis-$β$, a fast-motion and long-lasting on-device vision-language-action (VLA) model designed for real-world deployment. Current VLA evaluation remains largely confined to single-trial success rates under curated resets, which fails to capture the fast-motion and long-lasting capabilities essential for practical operation. To address this, we introduce the Productivity-Reliability Plane (PRP), which evaluates performance through Tasks per Hour (TPH) and Mean Time Between Intervention (MTBI) under a continuous-run protocol that demands both high-speed execution and sustained robustness. Habilis-$β$ achieves high performance by integrating language-free pre-training on large-scale play data for robust interaction priors with post-training on cyclic task demonstrations that capture state drift across consecutive task iterations. The system further employs ESPADA for phase-adaptive motion shaping to accelerate free-space transit, utilizes rectified-flow distillation to enable high-frequency control on edge devices, and incorporates classifier-free guidance (CFG) as a deployment-time knob to dynamically balance instruction adherence and learned interaction priors. In 1-hour continuous-run evaluations, Habilis-$β$ achieves strong performance under the PRP metrics, compared to $π_{0.5}$ in both simulation and real-world environments. In simulation, Habilis-$β$ achieves 572.6 TPH and 39.2 s MTBI (vs. 120.5 TPH and 30.5 s for $π_{0.5}$), while in a real-world humanoid logistics workflow it achieves 124 TPH and 137.4 s MTBI (vs. 19 TPH and 46.1 s for $π_{0.5}$). Finally, Habilis-$β$ achieves the highest reported performance on the standard RoboTwin 2.0 leaderboard across representative tasks, validating its effectiveness in complex manipulation scenarios.
While recent advances in neural representations and generative models have revolutionized 3D content creation, the field remains constrained by significant data processing bottlenecks. To address this, we introduce HY3D-Bench, an open-source ecosystem designed to establish a unified, high-quality foundation for 3D generation. Our contributions are threefold: (1) We curate a library of 250k high-fidelity 3D objects distilled from large-scale repositories, employing a rigorous pipeline to deliver training-ready artifacts, including watertight meshes and multi-view renderings; (2) We introduce structured part-level decomposition, providing the granularity essential for fine-grained perception and controllable editing; and (3) We bridge real-world distribution gaps via a scalable AIGC synthesis pipeline, contributing 125k synthetic assets to enhance diversity in long-tail categories. Validated empirically through the training of Hunyuan3D-2.1-Small, HY3D-Bench democratizes access to robust data resources, aiming to catalyze innovation across 3D perception, robotics, and digital content creation.
We present a measurement of the time-dependent $CP$ asymmetry in $B^0 \to K_{\rm S}^0 \: π^{+} π^{-} γ$ decays using a data set of 365 fb$^{-1}$ recorded by the Belle II experiment and the final data set of 711 fb$^{-1}$ recorded by the Belle experiment at the ${\rm Υ(4S)}$ resonance. The direct and mixing-induced time-dependent $CP$ violation parameters $C$ and $S$ are determined along with two additional quantities, $S^{+}$ and $S^{-}$, defined in the two halves of the $m^2(K_{\rm S}^0 π^{+})-m^2(K_{\rm S}^0 π^{-})$ plane. The measured values are $C = -0.17 \pm 0.09 \pm 0.04$, $S = -0.29 \pm 0.11 \pm 0.05$, $S^{+} = -0.57 \pm 0.23 \pm 0.10$ and $S^{-} = 0.31 \pm 0.24 \pm 0.05$, where the first uncertainty is statistical and the second systematic.
Recent advances in learned control, large-scale simulation, and generative models have accelerated progress toward general-purpose robotic controllers, yet the field still lacks platforms suitable for safe, expressive, long-term deployment in human environments. Most existing humanoids are either closed industrial systems or academic prototypes that are difficult to deploy and operate around people, limiting progress in robotics. We introduce Sprout, a developer platform designed to address these limitations through an emphasis on safety, expressivity, and developer accessibility. Sprout adopts a lightweight form factor with compliant control, limited joint torques, and soft exteriors to support safe operation in shared human spaces. The platform integrates whole-body control, manipulation with integrated grippers, and virtual-reality-based teleoperation within a unified hardware-software stack. An expressive head further enables social interaction -- a domain that remains underexplored on most utilitarian humanoids. By lowering physical and technical barriers to deployment, Sprout expands access to capable humanoid platforms and provides a practical basis for developing embodied intelligence in real human environments.
In this proceedings we explore the physics potential of the ESSnuSBplus setup to study beam and non-beam based physics scenarios in both standard and new physics cases. The ESSnuSBplus setup consists of three neutrino sources: the main ESS linac, a low energy monitored neutrino beam and a low energy nuSTORM facility and three detectors: the main far detector and two near detectors. The goal of this facility is to measure the leptonic CP phase with extremely high precision and the neutrino nucleus cross-section in the few hundred MeV region.
Empirical evaluation serves as the primary compass guiding research progress in foundation models. Despite a large body of work focused on training frontier vision-language models (VLMs), approaches to their evaluation remain nascent. To guide their maturation, we propose three desiderata that evaluations should satisfy: (1) faithfulness to the modality and application, (2) discriminability between models of varying quality, and (3) efficiency in compute. Through this lens, we identify critical failure modes that violate faithfulness and discriminability, misrepresenting model capabilities: (i) multiple-choice formats reward guessing, poorly reflect downstream use cases, and saturate early as models improve; (ii) blindly solvable questions, which can be answered without images, constitute up to 70% of some evaluations; and (iii) mislabeled or ambiguous samples compromise up to 42% of examples in certain datasets. Regarding efficiency, the computational burden of evaluating frontier models has become prohibitive: by some accounts, nearly 20% of development compute is devoted to evaluation alone. Rather than discarding existing benchmarks, we curate them via transformation and filtering to maximize fidelity and discriminability. We find that converting multiple-choice questions to generative tasks reveals sharp capability drops of up to 35%. In addition, filtering blindly solvable and mislabeled samples improves discriminative power while simultaneously reducing computational cost. We release DatBench-Full, a cleaned evaluation suite of 33 datasets spanning nine VLM capabilities, and DatBench, a discriminative subset that achieves 13x average speedup (up to 50x) while closely matching the discriminative power of the original datasets. Our work outlines a path toward evaluation practices that are both rigorous and sustainable as VLMs continue to scale.
We present a search for an invisible hidden-sector particle $ X_{\rm{inv}}$, produced in $B^0 \to \bar{D}^0 X_{\rm{inv}}$ and $B^\pm\to h X_{\rm{inv}}$ decays, where $h = π^\pm$, $K^\pm$, $D_{s}^\pm$, $p^\pm$. The search is performed using $e^+ e^-$ collision data recorded with the Belle detector, corresponding to 711~fb$^{-1}$. No significant signal is observed. We set 90\% confidence level upper limits ranging between $10^{-4}$ and $10^{-6}$ on the branching fraction $\mathcal{B}(B \to h X_{\rm{inv}})$ as a function of $m_{ X_{\rm{inv}}}$. Corresponding limits are set on $\mathcal{B}(B \to hX)\times\mathcal{B}(X\toγγ)$ for lifetimes $cτ_X$ between $10~μ$m and 10 m. Many of these limits are the first direct constraints on their respective decays. Our results provide the most stringent exclusion limits to date on the branching fractions for all search channels, and exclude previously unexplored regions of parameter space relevant to several new physics models.
Large Language Models (LLMs) have demonstrated significant potential across various domains, particularly in banking and finance, where they can automate complex tasks and enhance decision-making at scale. Due to privacy, security, and regulatory concerns, organizations often prefer on-premise deployment of LLMs. The ThaiLLM initiative aims to enhance Thai language capabilities in open-LLMs, enabling Thai industry to leverage advanced language models. However, organizations often face a trade-off between deploying multiple specialized models versus the prohibitive expense of training a single multi-capability model. To address this, we explore model merging as a resource-efficient alternative for developing high-performance, multi-capability LLMs. We present results from two key experiments: first, merging Qwen-8B with ThaiLLM-8B demonstrates how ThaiLLM-8B enhances Thai general capabilities, showing an uplift of M3 and M6 O-NET exams over the general instruction-following Qwen-8B. Second, we merge Qwen-8B with both ThaiLLM-8B and THaLLE-CFA-8B. This combination results in further improvements in performance across both general and financial domains, by demonstrating an uplift in both M3 and M6 O-NET, Flare-CFA, and Thai-IC benchmarks. The report showcases the viability of model merging for efficiently creating multi-capability LLMs.
End-to-end architectures trained via imitation learning have advanced autonomous driving by scaling model size and data, yet performance remains brittle in safety-critical long-tail scenarios where supervision is sparse and causal understanding is limited. We introduce Alpamayo-R1 (AR1), a vision-language-action model (VLA) that integrates Chain of Causation reasoning with trajectory planning for complex driving scenarios. Our approach features three key innovations: (1) the Chain of Causation (CoC) dataset, built through a hybrid auto-labeling and human-in-the-loop pipeline producing decision-grounded, causally linked reasoning traces aligned with driving behaviors; (2) a modular VLA architecture combining Cosmos-Reason, a vision-language model pre-trained for Physical AI, with a diffusion-based trajectory decoder that generates dynamically feasible trajectories in real time; (3) a multi-stage training strategy using supervised fine-tuning to elicit reasoning and reinforcement learning (RL) to enforce reasoning-action consistency and optimize reasoning quality. AR1 achieves up to a 12% improvement in planning accuracy on challenging cases compared to a trajectory-only baseline, with a 35% reduction in close encounter rate in closed-loop simulation. RL post-training improves reasoning quality by 45% and reasoning-action consistency by 37%. Model scaling from 0.5B to 7B parameters shows consistent improvements. On-vehicle road tests confirm real-time performance (99 ms latency) and successful urban deployment. By bridging interpretable reasoning with precise control, AR1 demonstrates a practical path towards Level 4 autonomous driving. Model weights are available at https://huggingface.co/nvidia/Alpamayo-R1-10B with inference code at https://github.com/NVlabs/alpamayo.
We introduce Yuan3.0 Flash, an open-source Mixture-of-Experts (MoE) MultiModal Large Language Model featuring 3.7B activated parameters and 40B total parameters, specifically designed to enhance performance on enterprise-oriented tasks while maintaining competitive capabilities on general-purpose tasks. To address the overthinking phenomenon commonly observed in Large Reasoning Models (LRMs), we propose Reflection-aware Adaptive Policy Optimization (RAPO), a novel RL training algorithm that effectively regulates overthinking behaviors. In enterprise-oriented tasks such as retrieval-augmented generation (RAG), complex table understanding, and summarization, Yuan3.0 Flash consistently achieves superior performance. Moreover, it also demonstrates strong reasoning capabilities in domains such as mathematics, science, etc., attaining accuracy comparable to frontier model while requiring only approximately 1/4 to 1/2 of the average tokens. Yuan3.0 Flash has been fully open-sourced to facilitate further research and real-world deployment: https://github.com/Yuan-lab-LLM/Yuan3.0.
One-to-one tutoring is widely considered the gold standard for personalized education, yet it remains prohibitively expensive to scale. To evaluate whether generative AI might help expand access to this resource, we conducted an exploratory randomized controlled trial (RCT) with $N = 165$ students across five UK secondary schools. We integrated LearnLM -- a generative AI model fine-tuned for pedagogy -- into chat-based tutoring sessions on the Eedi mathematics platform. In the RCT, expert tutors directly supervised LearnLM, with the remit to revise each message it drafted until they would be satisfied sending it themselves. LearnLM proved to be a reliable source of pedagogical instruction, with supervising tutors approving 76.4% of its drafted messages making zero or minimal edits (i.e., changing only one or two characters). This translated into effective tutoring support: students guided by LearnLM performed at least as well as students chatting with human tutors on each learning outcome we measured. In fact, students who received support from LearnLM were 5.5 percentage points more likely to solve novel problems on subsequent topics (with a success rate of 66.2%) than those who received tutoring from human tutors alone (rate of 60.7%). In interviews, tutors highlighted LearnLM's strength at drafting Socratic questions that encouraged deeper reflection from students, with multiple tutors even reporting that they learned new pedagogical practices from the model. Overall, our results suggest that pedagogically fine-tuned AI tutoring systems may play a promising role in delivering effective, individualized learning support at scale.