Search: “pretraining”

Spa-Bench: A Real-Robot Benchmark for Spatial Grounding in Vision-Language-Action Models

model-releasesarxiv-cs-ro

arXiv:2610.12696v1 Announce Type: new Abstract: Vision-language-action models (VLAs) offer a promising route toward general-purpose robotics by transferring semantic knowledge from large-scale vision-

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Safety

Splat2Act: From Captured Scenes to Personalized VLA Manipulation

DGX agent

arXiv:2610.13945v1 Announce Type: new Abstract: A user who brings a generalist manipulation policy into their home or office wants it to pick up their objects from their shelves, counters, and tables.

safetyarxiv-cs-ro

12 Oct 2026

Vision Transformers Learn Word Meaning from Images Alone

arXiv:2610.13252v1 Announce Type: cross Abstract: Can a vision model learn what words mean from pixels alone? We show that the two largest DINOv3 vision transformers organize words rendered in images

safetyarxiv-cs-cl

Vocal Intensity Conversion with Continuous SPL Control via Adversarial Training

arXiv:2610.13608v1 Announce Type: cross Abstract: Vocal intensity conversion renders a recorded utterance at a different vocal intensity, rather than simply scaling its amplitude, while preserving its

Weight Clustering Reveals the Importance of Relative Rank in Large Language Models

model-releasesarxiv-cs-cl

arXiv:2603.17917v2 Announce Type: replace-cross Abstract: Large language models (LLMs) contain billions of parameters, but many exact weight values are not essential. We apply scalar K-means clusterin

What Should SFT Learn? Conditional Information Gain for Generalizable Post-Training

arXiv:2610.12995v1 Announce Type: cross Abstract: Supervised fine-tuning (SFT) remains a widely used approach for post-training large language models, because fixed demonstrations can directly introdu

Zebra3D: Efficient Hybrid Foundation Models For Scalable 3D Reconstruction

arXiv:2610.12727v1 Announce Type: cross Abstract: Feed-forward 3D foundation models reconstruct scenes from unposed images, but the quadratic cost of global attention limits their scalability to large

zip2zip++: Towards Inference-Time Token Compression with Minimal Quality Loss

arXiv:2610.13605v1 Announce Type: new Abstract: Dynamic vocabularies built on-the-fly via Lempel-Ziv-Welch (LZW) compression let a language model represent the same text in fewer tokens, but prior wor

researcharxiv-cs-cl

AdaCast: Conditional Parameter Generation for Adaptive Time Series Forecasting

arXiv:2610.12240v1 Announce Type: cross Abstract: Time-series foundation models (TSFMs) have achieved strong forecasting performance across domains. However, most adaptation methods remain static. Exi

Addressing Overcommitment in the Reasoning of Gendered Economic Memes under Multimodal Ambiguity

arXiv:2610.11724v1 Announce Type: new Abstract: Multimodal meme understanding is increasingly used to analyze socially sensitive content, yet existing models often exhibit biased behavior when interpr

safetyarxiv-cs-lg

Tutorials

AgentGarten: Code Worlds for Evolving Agents

arXiv:2610.12374v1 Announce Type: new Abstract: Interactive virtual worlds allow agents to learn through exploration and interaction. What agents can learn is bounded by the environments they practice

tutorialsarxiv-cs-cv

Amortized Off-Policy Evaluation for LLMs

model-releasesarxiv-cs-lg

arXiv:2610.10848v1 Announce Type: new Abstract: Accurate evaluation is central to selecting which LLM to deploy, yet testing a candidate on live traffic exposes real users to an unvetted model. Teams

ARC: A Reasoning Recipe for Robot Foundation Models

arXiv:2610.12386v1 Announce Type: cross Abstract: The prevailing approach to improving robot foundation models (RFMs) relies on larger models, more robot demonstrations, and costly training at scale.

AuraLuxMuse: Adaptive Fusion Modeling for Aesthetic Stage Lighting Design with Music and Expert Guidance

arXiv:2610.11792v1 Announce Type: cross Abstract: We present AuraLuxMuse, a novel system for automated aesthetic stage lighting design that integrates expert knowledge, representation learning, and pr

Benchmark Designers Should 'Train on the Test Set' to Expose Exploitable Non-Visual Shortcuts

arXiv:2511.04655v2 Announce Type: replace Abstract: Multimodal LLMs can answer many questions in vision-centric benchmarks without looking at the image, using linguistic priors and skewed answer distr

Beyond Resolution: Object-to-Image Ratio Mismatch in Instance Retrieval

arXiv:2610.11489v1 Announce Type: new Abstract: Visual instance retrieval often fails when the same object appears at different apparent sizes in the query and gallery. We show that the dominant cause

Brain foundation model-guided source-selective domain adaptation for cross-subject EEG decoding

arXiv:2507.21037v3 Announce Type: replace Abstract: Cross-subject motor-imagery electroencephalography (MI-EEG) decoding remains challenging because substantial inter-subject variability can cause bot

safetyarxiv-cs-lg

BridgeGuard: Explicit Safety Drift for Diffusion-based Autonomous Driving

arXiv:2610.11483v1 Announce Type: new Abstract: Diffusion-based driving planners capture diverse behaviors but can generate unsafe trajectories under distribution shift. We propose BridgeGuard, a safe

CARE: Constrained Attention Refinement for Fine-Grained Visual Classification via Teacher-Student Distillation

arXiv:2610.11153v1 Announce Type: cross Abstract: Fine-grained visual classification requires models to recognize subtle local traits while exposing the visual evidence behind their predictions. Class

Conditional Residual Prediction: Improving Autoregressive Video Diffusion without a Bidirectional Teacher

arXiv:2610.11479v1 Announce Type: cross Abstract: Causal video diffusion models generate video autoregressively, which suits streaming, interactive, and long-video generation. Under standard training,

CRISP: Fixing Flying Pixels in Latent LiDAR Generation via Diffusion Decoding

arXiv:2610.11376v1 Announce Type: cross Abstract: Latent LiDAR pipelines suffer from flying pixels: convolutional VAEs blur sharp radial depth discontinuities, yielding edge depths that back-project t

CSF: Contextual Safety Filtering for Motion Generators

arXiv:2610.12467v1 Announce Type: cross Abstract: Text-conditioned motion generators produce trackable whole-body motion, but they have no notion of scene-dependent safety: the same action may target

safetyarxiv-cs-lg

Density Ratio Estimation with Stein Displacement Fields

arXiv:2610.12437v1 Announce Type: cross Abstract: Density ratios quantify distribution shift from a probability-mass point of view, whereas displacement fields describe, from a dynamical point of view

Diffu-LoRA: A Novel Low-Rank Adaptation for Personalized Diffusion Models

model-releasesarxiv-cs-cl

arXiv:2610.10550v1 Announce Type: new Abstract: Personalizing text-to-image diffusion models from a few reference images requires preserving subject identity while following prompts that describe new

Distillation for Incrimination and Distillation for Capabilities

arXiv:2610.11012v1 Announce Type: new Abstract: Powerful misaligned AI models might recognize alignment evaluations and strategically behave well on them, rendering direct audits uninformative. Howeve

Equal Path Cost, Unequal Output Effects: Understanding Perturbation Propagation in Diffusion Models

arXiv:2610.11380v1 Announce Type: new Abstract: Diffusion models have achieved remarkable success in generative modeling, with their sampling procedures routinely modified to control generation and im

Expression-Diverse References for Identity-Preserving Video Generation

arXiv:2610.11023v1 Announce Type: new Abstract: Identity-preserving video generation aims to maintain a subject's identity while synthesizing realistic videos. Yet a single reference portrait captures

Freeze the Decoder, Heal the Encoder: Parameter-Efficient Adaptation for SVD-Based KV-Cache Compression

arXiv:2610.10552v1 Announce Type: cross Abstract: Comparing parameter-efficient fine-tuning recipes under a single, shared learning rate is a common but flawed practice: when the arms being compared h

From Prompting to Composing: A Spatial Canvas Interface for Poster Generation

arXiv:2610.12230v1 Announce Type: cross Abstract: Text prompting is an indirect interface for poster generation, requiring users to encode inherently two-dimensional composition intent into a one-dime

GHARP: Real-time Gaussian Head Animation from Large-scale Reconstruction Prior

arXiv:2610.10945v1 Announce Type: new Abstract: We present GHARP (Real-time Gaussian Head Animation from Large-scale Reconstruction Prior), a method that animates 3D human heads in real time from a fe

Local Ai

Learning Kilometer-Scale Weather Prediction with Global-Regional Alignment

arXiv:2610.12401v1 Announce Type: new Abstract: Kilometer-scale regional weather forecasting is essential for local weather warnings and weather-sensitive decisions. Existing data-driven approaches of

local-aiarxiv-cs-lg

Learning Projection-Aware 360-Degree Image Rectification via Dual-Projection Fusion

arXiv:2512.00911v2 Announce Type: replace Abstract: Panoramic cameras provide a 360{eg} field of view and are widely used in panoramic vision, immersive visual computing, and robotic perception. Howev

researcharxiv-cs-cv

Tutorials

Mid-Training Language Models on Raw Video

arXiv:2610.11019v1 Announce Type: cross Abstract: Multimodal large language models learn mostly from paired image-text data or annotated video, and raw web video is rarely used to further train an exi

tutorialsarxiv-cs-ai

Agents

MiMo-V2.6: Scaling Reinforcement Learning Towards Self-Improvement

arXiv:2610.11959v1 Announce Type: new Abstract: Reinforcement learning (RL) is the central training paradigm for advancing large foundation models towards self-improvement. This report introduces the

agentsarxiv-cs-cl

MiniWAM: Learning Compact Future Targets for Efficient World-Action Modeling

arXiv:2610.12194v1 Announce Type: new Abstract: World modeling has emerged as an effective co-training objective for robot policies, giving rise to World Action Models (WAMs) that jointly predict acti

safetyarxiv-cs-ro

MotiveMob: Motivation as Semantic Action for Closed-Loop Human Mobility Generation

arXiv:2610.11442v1 Announce Type: new Abstract: Human mobility generation, an important task in urban research, synthesizes trajectory data for urban planning and transportation management. Human mobi

NanoProof: Open and Efficient Automated Theorem Proving in Lean 4

arXiv:2610.11605v1 Announce Type: cross Abstract: We introduce NanoProof, to our knowledge the first factorized execution-guided theorem prover in Lean 4 whose training data, extraction tooling, train

Obscuring Data Contamination Through Translation: Evidence from Arabic Corpora

arXiv:2601.14994v2 Announce Type: replace-cross Abstract: Data contamination can invalidate benchmark evaluation when a model benefits from memorized evaluation content rather than genuine generalizat

Optimizing Large Language Models with Chained LMOs

arXiv:2610.10975v1 Announce Type: cross Abstract: Muon has motivated a growing family of optimizers that compose multiple matrix normalizations, but these methods remain fragmented and lack a unified

PearlVLA: Progressive Embodied Action-Plan Refinement in Latent Space

arXiv:2606.17924v2 Announce Type: replace-cross Abstract: Current Vision-Language-Action (VLA) models face a trade-off between efficient action generation and explicit deliberation. Directly decoding

Phonologically Informed Tokenization for German Speech Recognition: A Cross-Domain Study

arXiv:2610.11646v1 Announce Type: new Abstract: German is a morphologically rich language whose syllable structure is exceptionally well-predicted by the Knuth--Liang hyphenation algorithm. We ask whe

researcharxiv-cs-cl

PLaW-VLA: Predictive Latent World Modeling for Vision-Language-Action Policies

arXiv:2610.12285v1 Announce Type: new Abstract: Learning to predict how the world evolves can provide vision-language-action (VLA) policies with predictive context for long-horizon control, but its ef

safetyarxiv-cs-ro

PulseBound: Future-Beat State Forecasting Under an Explicit Information Boundary

arXiv:2610.12010v1 Announce Type: new Abstract: Predictive representation learning from photoplethysmography (PPG) can violate causal information access even with causal attention, as normalization, n

READ-Bench: Benchmarking Historical Instance Retrieval for Time-Series Diagnosis

arXiv:2609.32123v2 Announce Type: replace Abstract: Time-series diagnostic systems rarely rely on retrieving relevant historical cases, and when they do, retrieval is evaluated only indirectly through

Rethinking Contrastive Loss in CLIP Post-training: A Complementary Framework with Frozen Text Encoder

arXiv:2610.11374v1 Announce Type: cross Abstract: CLIP serves as a foundational vision-language model and the de facto vision encoder for downstream VLMs such as LLaVA. Post-training offers a lightwei

Revisiting Identity and Spectra Dispersion in Media-Bridged Time Series Forecasting: Linking Multivariate Signals and Narrative Flows

arXiv:2610.11924v1 Announce Type: new Abstract: Media-bridged time series forecasting is expanding to encompass traditional 'multivariate' and emerging 'multimodal' (e.g., through textual assistance).

safetyarxiv-cs-cv

Applications

Rewiring Semantics, Dynamics, and Control: A Simple yet Effective Action-Centric Tri-Stream Transformer

arXiv:2610.11416v1 Announce Type: cross Abstract: Vision-Language-Action (VLA) models have emerged as a prominent framework for complex robotic manipulation, building on the strong semantic understand

applicationsarxiv-cs-ai

Rounding in Preconditioner Space: Redesigning 4-bit AdamW Optimizer-State Quantization

model-releasesarxiv-cs-lg

arXiv:2610.12444v1 Announce Type: new Abstract: Quantizing AdamW's optimizer states reduces persistent storage, but quantization errors propagate through the moment recurrences and perturb subsequent