Knowledge catalogue
arXiv:2609.32792v1 Announce Type: new Abstract: LLM post-training combines supervised fine-tuning (SFT), a mode-covering forward-KL objective, with reinforcement learning (RL), a mode-seeking reverse-
arXiv:2601.06597v3 Announce Type: replace Abstract: Can we design a model such that its stochastic training favours a desired class of solutions without enforcing an explicit penalty? Under suitable c
arXiv:2505.11335v4 Announce Type: replace Abstract: Graph Neural Networks (GNNs) have demonstrated remarkable effectiveness on graph-based tasks. However, their predictive confidence is often miscalib
arXiv:2609.34458v1 Announce Type: cross Abstract: Classical statistical theory is insufficient to explain the successes of general-purpose AI models, because it depends on handcrafted inductive biases
arXiv:2603.16731v2 Announce Type: replace Abstract: Quantizing optimizer states is becoming an important ingredient of memory-efficient large-scale pre-training, but the resulting optimizer dynamics r
arXiv:2609.31983v1 Announce Type: new Abstract: The phenomenon of the top subspace stabilization of the Hessian matrix is an surprising and critical aspect in study of the second-order information of
arXiv:2609.31900v1 Announce Type: new Abstract: Modern LLM post-training composes supervised fine-tuning (SFT), reinforcement learning with verifiable rewards (RLVR), and on-policy distillation (OPD)
arXiv:2609.33996v1 Announce Type: new Abstract: Learned mask refiners improve segmentation accuracy, but it is hard to tell how much of the improvement comes from explicit structure rather than from e
arXiv:2609.35450v1 Announce Type: new Abstract: Physical contact often determines how a humanoid should respond during loco-manipulation, yet vision and proprioception alone are often insufficient to
arXiv:2512.11609v3 Announce Type: replace Abstract: In embodied intelligence, the embodiment gap between robotic and human hands brings significant challenges for learning from human demonstrations. A
arXiv:2609.32831v1 Announce Type: new Abstract: Unified multimodal models combine understanding, generation, and editing within a single network, offering a promising foundation for versatile multimod
arXiv:2609.34688v1 Announce Type: new Abstract: Reward-maximizing reinforcement learning (RL) is widely used to post-train stochastic diffusion and flow policies for text-to-image (T2I) generation. Ho
arXiv:2609.34182v1 Announce Type: new Abstract: Dexterous manipulation requires tactile feedback.However, robot tactile demonstrations are difficult to scale,because dexterous-hand teleoperation provi
arXiv:2609.34639v1 Announce Type: new Abstract: We study approximate sampling: given N independent samples from a proposal distribution mu, the goal is to select one whose distribution is close to a t
arXiv:2512.17992v2 Announce Type: replace Abstract: Long-horizon robotic tasks are hard due to continuous state-action spaces and sparse feedback. Symbolic world models help by decomposing tasks into
arXiv:2609.35763v1 Announce Type: new Abstract: Distributional training provides collective supervision for one-step visual generation by matching real and generated features in frozen representation
arXiv:2605.06665v2 Announce Type: replace Abstract: Most Mixture-of-Experts (MoE) transformers preassign each expert to one layer for the whole of training. We study whether a brief global-access phas
arXiv:2609.35483v1 Announce Type: new Abstract: Many learning tasks map an input distribution to an output distribution. A natural way to model such an operator is to transform each input sample using
arXiv:2609.34643v1 Announce Type: new Abstract: Cover's Universal Portfolio (Cover, 1991) matches the performance of the best constant rebalanced portfolio in hindsight. We generalize this framework t
arXiv:2609.35055v1 Announce Type: new Abstract: Long contexts are central to modern transformer systems, but most expressivity results choose a different network for each fixed sequence length. We stu
arXiv:2609.34406v1 Announce Type: cross Abstract: Sampling latency compounds in diffusion workflows, where users generate and discard many candidates before keeping one. Surprisingly, the poor outputs
arXiv:2609.32824v1 Announce Type: new Abstract: extit{Gromov-Wasserstein Distances} (GWDs) provide quantitative ways of comparing probabilistic distributions defined on different metric spaces by appl
arXiv:2609.31766v1 Announce Type: new Abstract: Contextual image misuse pairs an image with a misleading claim. We study image-claim correspondence in fact-checked pairs containing out-of-context reus
arXiv:2609.32518v1 Announce Type: new Abstract: Distilling bidirectional multi-step video diffusion transformers into few-step causal models has become a common approach for streaming video generation
arXiv:2609.31671v1 Announce Type: new Abstract: Event cameras are well suited to pedestrian crossing detection, and spiking neural networks (SNNs) can process their output natively, but current SNN de
arXiv:2602.05548v4 Announce Type: replace Abstract: Reinforcement Learning with Verifiable Rewards (RLVR), particularly GRPO, has become the standard for eliciting LLM reasoning. However, its efficien
arXiv:2609.34036v1 Announce Type: new Abstract: On-policy distillation (OPD) trains a student on its own rollouts using dense supervision from a teacher. In multi-turn environments, a mistake at a cri
arXiv:2609.32813v1 Announce Type: new Abstract: Large vision-language models (VLMs) have emerged as a powerful paradigm for urban and spatial AI. However, current state-of-the-art large VLMs still str
arXiv:2609.32386v1 Announce Type: new Abstract: Impaired glucose regulation is a major contributor to metabolic dysfunction and type 2 diabetes. This study developed an interpretable machine-learning
arXiv:2609.34682v1 Announce Type: new Abstract: Advances in multimodal understanding, reasoning, and tool use enable agents to tackle increasingly complex visual reasoning tasks. By distilling past ex
arXiv:2609.33402v1 Announce Type: new Abstract: Universal multimodal retrieval requires compact embeddings that preserve task-relevant semantic information across diverse modalities. Prior works have
arXiv:2609.33888v1 Announce Type: new Abstract: Can vanilla policy optimization explore enough to achieve near-optimal regret in stochastic contextual bandits? We show that standard exponential policy
arXiv:2609.34390v1 Announce Type: new Abstract: Multi-animal tracking (MAT) supports the study of animal movement, behavior, and group interactions. However, general multi-object tracking (MOT) benchm
arXiv:2609.31999v1 Announce Type: new Abstract: Whereas previous results on abilities and limitations of transformers have restricted the definition of transformers in various ways, here we study soft
arXiv:2609.32840v1 Announce Type: new Abstract: Accurate assessment of Schistosoma japonicum-associated liver fibrosis is essential for disease management and long-term follow-up in endemic regions. U
arXiv:2609.33264v1 Announce Type: new Abstract: Video world models are emerging as data engines, action planners, and generative simulators for autonomous driving, but existing benchmarks primarily as
arXiv:2609.35641v1 Announce Type: cross Abstract: Precise instruction following in image generation, such as satisfying object counts and spatial relations, remains an open challenge at least in part
arXiv:2609.34553v1 Announce Type: new Abstract: Formal verification can play a key role in ensuring the reliability of Deep Neural Networks (DNNs) deployed in safety-critical systems. Modern DNN verif
arXiv:2609.35020v1 Announce Type: new Abstract: Vision Sparse Autoencoders (SAEs) have become a popular tool in Mechanistic Interpretability due to their presumed ability to disentangle complex featur
arXiv:2609.35028v1 Announce Type: new Abstract: Large language models (LLMs) have made misinformation inexpensive to produce but not to verify, creating a growing asymmetry in the information ecosyste
arXiv:2609.33253v1 Announce Type: new Abstract: We present VGGT-Diff, a geometry-routed multi-view diffusion model for sparse-view novel view synthesis. Existing novel view synthesis (NVS) methods fac
arXiv:2609.33844v1 Announce Type: cross Abstract: Modeling temporal dependence and uncertainty is central to forecasting with world models. The Visual Bayesian Regression World Model combines visual f
arXiv:2609.33603v1 Announce Type: new Abstract: Reliable optical chemical structure recognition (OCSR) is essential for building high-quality chemical data from scientific literature, yet even small r
arXiv:2609.33464v1 Announce Type: new Abstract: World models learn internal representations of environment dynamics to predict future states, enabling agents to optimize action plans without physical
arXiv:2609.31697v1 Announce Type: new Abstract: Low-light conditions can significantly degrade the ability of vision-language models (VLMs) to accurately describe human actions in videos. In this work
arXiv:2609.33935v1 Announce Type: new Abstract: Adapting video models to new tasks typically requires dedicated data curation and fine-tuning. While visual analogy provides a training-free alternative
arXiv:2609.31810v1 Announce Type: cross Abstract: Video-to-music models have advanced considerably in the last few years, particularly in film and music video applications. In this paper, we investiga
arXiv:2609.35134v1 Announce Type: new Abstract: Video editing has advanced substantially in recent years, with methods increasingly accounting for the visual consequences of edits, such as changes to
arXiv:2609.34041v1 Announce Type: new Abstract: Safe reinforcement learning seeks policies that maximise task performance while satisfying safety constraints. In driving benchmarks, however, collision
arXiv:2609.32918v1 Announce Type: cross Abstract: Vision-language models (VLMs) are increasingly being used in scientific workflows, but their ability as agents to directly control laboratory experime
arXiv:2609.33325v1 Announce Type: new Abstract: Visual backbones have evolved from Convolutional Neural Networks (CNNs) with local aggregation to Vision Transformers (ViTs) with global interactions, S
arXiv:2503.10547v3 Announce Type: replace Abstract: Concept-based explanations help users understand vision models through recognizable visual patterns. However, existing methods often rely on correla
arXiv:2609.31746v1 Announce Type: new Abstract: Sub-billion-parameter Vision-Language Models are increasingly viable for on-device deployment, yet compact model size alone does not guarantee usability
arXiv:2609.34472v1 Announce Type: new Abstract: Acne assessment is crucial for clinical decision-making, yet traditional grading and counting are subjective and fail to account for lesion size. While
arXiv:2602.01915v2 Announce Type: replace Abstract: Recent advances in Large Language Models (LLMs) and Vision-Language Models (VLMs) have enabled powerful semantic and multimodal reasoning capabiliti
arXiv:2609.33104v1 Announce Type: new Abstract: While action-conditioned video prediction provides an intuitive world model for robotics, purely data-driven predictors often suffer from compounding er
arXiv:2609.35464v1 Announce Type: new Abstract: Images capture the world at one moment, whereas video reveals how it changes. Image self-supervision learns spatial structure from a single moment, whil
arXiv:2607.11397v2 Announce Type: replace Abstract: Human videos provide rich information about object manipulation at a scale difficult to reproduce with robots, but often lack action annotations tha
arXiv:2609.34250v1 Announce Type: new Abstract: Pretrained world action models (WAMs) provide generalist capabilities across diverse robotic manipulation tasks, yet improving target-task performance t
arXiv:2606.15835v2 Announce Type: replace Abstract: Diffusion models have achieved impressive empirical success in generative tasks, and their convergence theory is now relatively well understood. Mot