Does self-distillation improve reasoning?
arXiv:2610.18613v1 Announce Type: new Abstract: Self-distillation is an emerging approach that trains a model to imitate its own predictions when conditioned on privileged information, such as correct
Knowledge catalogue
arXiv:2610.18613v1 Announce Type: new Abstract: Self-distillation is an emerging approach that trains a model to imitate its own predictions when conditioned on privileged information, such as correct
arXiv:2610.14112v1 Announce Type: cross Abstract: Calibrated typed-decision ('System One') models return a label and a probability in a single forward pass at a small fraction of the cost of large lan
arXiv:2610.18870v1 Announce Type: new Abstract: We generally perceive that training text-to-image diffusion models requires heavy compute: it is true if we aim at a comprehensive frontier model. This
arXiv:2610.13990v1 Announce Type: new Abstract: Sparse attention visits few keys, yet finding them can still require scoring every key with every indexer head. We introduce DoSA (Dots Sparse Attention
arXiv:2610.18862v1 Announce Type: cross Abstract: Reasoning models achieve strong performance through long chains of thought (CoT), but often spend more tokens than needed to solve a problem. Existing
arXiv:2610.14376v1 Announce Type: new Abstract: Test-time adaptation (TTA) enables time series forecasters to update after deployment as new observations become available. However, adaptation objectiv
arXiv:2610.18609v1 Announce Type: cross Abstract: Direct Preference Optimization (DPO) has become a widely adopted approach for aligning vision-language models (VLMs), yet the security of multimodal p
arXiv:2610.16618v1 Announce Type: new Abstract: Policy representations in reinforcement learning have expanded beyond Gaussian policies to diffusion and flow-matching based generative policies. Howeve
arXiv:2610.18373v1 Announce Type: new Abstract: Quadruped robots performing outdoor deliveries must interpret complex scenes, anticipate changes, and navigate in socially compliant ways. Robot demonst
arXiv:2610.16491v1 Announce Type: new Abstract: Improving an agent through successive revisions requires balancing exploration and evaluation under a limited budget. A narrow search can miss useful se
arXiv:2610.15696v1 Announce Type: new Abstract: We present DRanker (Disney+ Ranker), a transformer-based multitask ranking model serving personalized recommendations on the Disney+ homepage. We descri
arXiv:2610.16071v1 Announce Type: new Abstract: Reliable deployment of LLM-based multi-agent systems calls for runtime auditors that predict terminal outcomes from observed execution prefixes. These a
arXiv:2603.00466v2 Announce Type: replace Abstract: Despite impressive progress in video generation, existing models remain limited to surface-level plausibility, lacking a coherent and unified unders
arXiv:2610.15662v1 Announce Type: new Abstract: Generative policies offer expressive action distributions for online reinforcement learning (RL), but many diffusion- and flow-based policies require co
arXiv:2610.15112v1 Announce Type: new Abstract: High-fidelity closed-loop simulation for autonomous driving increasingly relies on dense multi-view 3D reconstruction. While recent feed-forward geometr
arXiv:2610.17903v1 Announce Type: new Abstract: Vision-Language Models (VLMs) and world models have demonstrated strong capabilities in semantic understanding and future scene evolution for autonomous
arXiv:2602.04692v3 Announce Type: replace Abstract: Referring Multi-Object Tracking (RMOT) aims to track specific targets based on language descriptions, playing a vital role in interactive AI systems
arXiv:2607.17291v2 Announce Type: replace Abstract: Deep research agents increasingly operate over the open web, where relevant records coexist with redundant summaries, outdated reports, and misleadi
arXiv:2610.14016v1 Announce Type: new Abstract: Most existing sports datasets are individual-centric or ball-centric, focusing on individual player actions while ignoring interactions among teammates.
arXiv:2610.14945v1 Announce Type: new Abstract: Visual localization enables unmanned aerial vehicles (UAVs) to maintain absolute positioning in GNSS-denied environments. A common approach matches UAV
arXiv:2610.14499v1 Announce Type: new Abstract: 3D human pose estimation requires accurate modeling of motion-dependent skeletal structures and inter-joint dependencies. Existing methods typically rel
arXiv:2610.17730v1 Announce Type: new Abstract: Low-rank adaptation (LoRA) often underperforms full fine-tuning. We investigate this performance gap through a tangent-space lens: factorized updates ar
arXiv:2610.16822v1 Announce Type: new Abstract: Architecture attribute predictors reduce the cost of neural architecture search by estimating accuracy or latency from a small set of evaluated candidat
arXiv:2610.18518v1 Announce Type: new Abstract: Four-dimensional cone-beam computed tomography (4D CBCT) provides phase-resolved volumetric images for image-guided radiotherapy. Yet good reconstructio
arXiv:2610.18819v1 Announce Type: new Abstract: Complex image editing requires faithful task interpretation and accurate execution, but a failed edit does not reveal which responsibility failed. We pr
arXiv:2610.14630v1 Announce Type: new Abstract: Dual-lens super-resolution (DuSR) aims to enhance a wide-angle low-resolution (LR) image by exploiting a telephoto reference (Ref) image. A central chal
arXiv:2610.17853v1 Announce Type: new Abstract: Multimodal large language models (MLLMs) have achieved strong performance on diverse vision-language tasks, but their computational demands limit deploy
arXiv:2610.16348v1 Announce Type: new Abstract: During adolescence, significant changes in brain development are influenced by hormonal changes that impact both the body and the brain. This study pres
arXiv:2603.16157v2 Announce Type: replace Abstract: While Reinforcement Learning (RL) enhances Large Language Model reasoning, on-policy algorithms like GRPO are sample-inefficient as they discard pas
arXiv:2610.18018v1 Announce Type: new Abstract: Low-light image enhancement (LLIE) is an important image/video processing problem for reliable visual systems under poor illumination, yet it remains ch
arXiv:2610.17679v1 Announce Type: new Abstract: Chain-of-thought (CoT) improves reasoning but incurs substantial autoregressive rationale cost. Continuous CoT moves intermediate computation into recur
arXiv:2510.27042v3 Announce Type: replace-cross Abstract: Increasing the thinking budget of AI models can significantly improve accuracy, but not all questions warrant the same amount of reasoning. Us
arXiv:2610.15384v1 Announce Type: new Abstract: Compressing the key-value (KV) cache is essential for serving long-context large language models (LLMs) under memory constraints. While current scalar q
arXiv:2610.14120v1 Announce Type: cross Abstract: A defined-benefit pension is only as safe as its sponsor, yet sponsor distress is under-instrumented relative to plan-side modelling. We present a tra
arXiv:2610.14624v1 Announce Type: new Abstract: Interpreting Earth system conditions is central to understanding environmental risks and informing decisions across weather- and climate-sensitive secto
arXiv:2610.16523v1 Announce Type: cross Abstract: Training LLM agents to interact with users over multiple turns is challenging because the learning signal depends not only on the agent, but also on t
arXiv:2610.17718v1 Announce Type: cross Abstract: Electrocardiogram (ECG) models need supervision that targets evolving weaknesses and remains faithful to waveform evidence. We introduce ECG-RSI, a re
arXiv:2610.16433v1 Announce Type: new Abstract: Time Series Foundation Models (TSFMs) provide general-purpose forecasting capabilities, yet their forecasts can remain misaligned with the temporal char
arXiv:2610.14277v1 Announce Type: cross Abstract: Tensor parallelism ends every layer with an all-reduce that sits between two dependent matrix multiplications, so its cost cannot be hidden and grows
arXiv:2601.06157v2 Announce Type: replace Abstract: Alignment in large language models (LLMs) is still largely static: after training, the policy is frozen. DPO, GRPO methods typically imprint one beh
arXiv:2610.14926v1 Announce Type: new Abstract: Multi-finger hands contact objects at different times, so a global close command can overload early-contact fingers while others are still moving. Dense
arXiv:2610.17453v1 Announce Type: new Abstract: We propose a continuous learning formulation for T non-forgetting tasks with the spherical parameter space S^{d-1}(sqrt{d}) in asymptotic regime drighta
arXiv:2610.17466v1 Announce Type: new Abstract: Frechet Distributional (FD) objectives provide an effective distribution-level signal for post-training generative models. For latent generative models,
arXiv:2610.15934v1 Announce Type: new Abstract: Motion planning for automated vehicles in interactive traffic has to simultaneously ensure safety, respect traffic rules, and maintain comfortable progr
arXiv:2610.16343v1 Announce Type: cross Abstract: Multi-agent debate has shown promise for improving the reasoning of large language models, yet recent theory suggests its benefits are highly regime-d
arXiv:2610.16792v1 Announce Type: new Abstract: Adaptive optimization methods have received considerable attention in recent years due to their strong empirical performance. However, they generally re
arXiv:2610.16866v1 Announce Type: new Abstract: Reinforcement learning (RL) post-training with generative reward models (GenRM) has become a central recipe for improving large language models (LLMs),
arXiv:2610.18284v1 Announce Type: new Abstract: Generated images often have coherent layouts and visual content, yet local errors in small text and complex characters still limit their readability and
arXiv:2510.11068v4 Announce Type: replace Abstract: Real-world deployment often exposes vision models to distribution shifts, making test-time adaptation (TTA) important for reliable prediction. Exist
arXiv:2610.16176v1 Announce Type: new Abstract: Situated 3D Question Answering requires models to interpret a 3D scene relative to an agent's pose. Existing methods commonly convey that pose through a
arXiv:2610.17572v1 Announce Type: new Abstract: While expert demonstrations on real robots are difficult to collect, human ego-centric videos are easy to collect and abundant on the internet. However,
arXiv:2610.18872v1 Announce Type: cross Abstract: Learning humanoid loco-manipulation from human demonstrations rests on observing how locomotion, posture, and object interaction evolve together as ta
arXiv:2610.18822v1 Announce Type: new Abstract: Egocentric human videos offer a scalable alternative to robot teleoperation for collecting dexterous manipulation data. However, two challenges limit th
arXiv:2610.18499v1 Announce Type: new Abstract: Egocentric video understanding spans diverse capabilities, modalities, and evaluation formats, yet these are often studied in isolation despite their ti
arXiv:2610.17128v1 Announce Type: new Abstract: Egocentric video provides a continuous first-person record of everyday activities, making it a natural foundation for wearable AI assistants. Unlike off
arXiv:2610.16256v1 Announce Type: new Abstract: Head-mounted body and bimanual motion capture supports self-avatars, teleoperation, and embodied demonstrations. However, a single egocentric camera is
OpenAI is facing a new wave of lawsuits from eight US states that accuse the company of harming children and misleading customers about the safety of ChatGPT, as reported earlier by Reuters. The lawsu
arXiv:2610.15555v1 Announce Type: new Abstract: Training-free story visualization without reference images must preserve recurring entities across frames while scaling to long narratives. Existing met
arXiv:2610.16118v1 Announce Type: new Abstract: Hour-long videos, streaming feeds, and movie-length content are pushing vision-language models (VLMs) past their breaking point -- the standard recipe o
arXiv:2610.15580v1 Announce Type: new Abstract: This paper presents an algorithm for detecting and fitting multiple ellipsoids in 2D, 3D and higher-dimensional data. The representation is based on tan