Search: “inference”

LLM-Guided Future Hypotheses for Horizon-Aware Exploration in Multi-Step Robot Manipulation

arXiv:2605.29864v1 Announce Type: new Abstract: Multi-step robot manipulation requires acting under uncertainty about how the scene will evolve, making exploration and policy adaptation challenging. W

safetyarxiv-cs-ro

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Research

LLMBridge: An LLM Pipeline for End-to-end Referential Bridging Resolution in English

DGX agent

arXiv:2605.29048v1 Announce Type: new Abstract: In this paper, we introduce LLMBridge, a new LLM based system for the task of end-to-end referential bridging resolution in English. Our bridging resolu

researcharxiv-cs-cl

29 May 2026

LoopFM: Learning frOm HistOrical RePresentations of Foundation Model for Recommendation

arXiv:2605.29280v1 Announce Type: cross Abstract: Knowledge distillation (KD) transfers a single scalar prediction from a large foundation model (FM) to compact vertical models (VMs), suffering from d

Mask the Target: A Plug-and-Play Regularizer Against LoRA Forgetting

arXiv:2605.29498v1 Announce Type: new Abstract: Low-Rank Adaptation (LoRA) has become one of the most widely used fine-tuning mechanisms for adapting large language models to new domains, tasks, and u

safetyarxiv-cs-cl

Agents

MINDGAMES: A Live Arena for Evaluating Social and Strategic Reasoning in Multi-Agent LLMs

arXiv:2605.29512v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly deployed as interactive agents, yet their capacity for social and strategic reasoning over extended intera

agentsarxiv-cs-ai

Mitigating Hallucination in Vision-Language Models through Barrier-Regulated Adaptive Closed-form Steering

arXiv:2605.29881v1 Announce Type: cross Abstract: Large vision-language models (LVLMs) often hallucinate objects that are not present in the input image, largely because visual grounding weakens as de

Mitigating State Aliasing in Vision-Language-Action Models via Inverse Dynamics Learning

arXiv:2605.29577v1 Announce Type: new Abstract: Vision-Language-Action (VLA) models have emerged as a promising framework that unifies perception, reasoning, and control for robot manipulation by adap

safetyarxiv-cs-cv

Modeling Hierarchical Thinking in Large Reasoning Models

arXiv:2510.22437v2 Announce Type: replace Abstract: Large Reasoning Models (LRMs) solve complex tasks by generating long Chain-of-Thought (CoT) sequences; however, the emergent dynamics governing reas

safetyarxiv-cs-ai

Moment Matching Q-Learning

arXiv:2605.29033v1 Announce Type: new Abstract: Score-based and flow-based generative models exhibit remarkable expressive capacity in capturing complex distributions, and have been extensively deploy

MuPHI: Learning Implicit Multimodal Harm Reasoning via Semantically Grounded Reward Optimization

arXiv:2605.29951v1 Announce Type: new Abstract: Understanding how harm emerges from interaction between otherwise benign image-text pairs requires intent-aware cross-modal reasoning beyond surface-lev

Nine Judges, Two Effective Votes: Correlated Errors Undermine LLM Evaluation Panels

arXiv:2605.29800v1 Announce Type: new Abstract: LLM-as-a-judge panels aggregate votes from multiple models, with the expectation that diverse models yield more reliable evaluations. We develop a frame

researcharxiv-cs-cl

Non-Forgetting Knowledge Allocation with Bi-level Competition for Class-Incremental Learning

arXiv:2605.29592v1 Announce Type: new Abstract: Class-Incremental Learning (CIL) with pre-trained models (PTMs) aims to sequentially adapt PTMs to new categories without forgetting old knowledge. Buil

researcharxiv-cs-cv

Order-Agnostic Autoregressive Modelling with Missing Data

arXiv:2605.06355v2 Announce Type: replace Abstract: Order-Agnostic autoregressive models have demonstrated strong performance in deep generative modeling, yet their use in settings with incomplete dat

applicationsarxiv-cs-lg

Overcoming Forgetting in LLM Fine-Tuning with Evolution Strategies

arXiv:2605.30148v1 Announce Type: cross Abstract: Evolution Strategies (ES) has recently emerged as a competitive alternative to reinforcement learning (RL) for large language model (LLM) fine-tuning,

Phase-Conditioned Imitation Learning with Autonomous Failure Recovery for Robust Deformable Object Manipulation

arXiv:2605.29407v1 Announce Type: new Abstract: This paper presents a phase-conditioned, force-aware framework for robust deformable object manipulation. Standard imitation learning policies such as A

safetyarxiv-cs-ro

Pocket-Dentist: On-Device Dental Image Understanding via Efficient Multimodal Large Language Models

arXiv:2605.29299v1 Announce Type: cross Abstract: Evaluations of dental vision-language models remain fragmented across datasets, task definitions and metrics, and often ignore their computational cos

Prioritize the Process, Not Just the Outcome: Rewarding Latent Thought Trajectories Improves Reasoning in Looped Language Models

model-releasesarxiv-cs-lg

arXiv:2602.10520v3 Announce Type: replace Abstract: Looped Language Models (LoopLMs) perform multi-step latent reasoning prior to token generation and outperform conventional LLMs on reasoning benchma

Pushing the Limits of Block Rotations in Post-Training Quantization

arXiv:2601.22347v2 Announce Type: replace-cross Abstract: Recent post-training quantization (PTQ) methods have adopted block rotations to diffuse outliers prior to rounding. While this reduces the ove

Local Ai

RadioFormer3D: Weakly Supervised 3D Radio Map Estimation in Low-Altitude Airspace via Generative Modeling

arXiv:2605.29538v1 Announce Type: new Abstract: With the emergence of wireless applications in three-dimensional environments, such as the low-altitude airspace and 3D heterogeneous networks, radio ma

local-aiarxiv-cs-cv

Ray Data LLM enables 2x throughput over vLLM’s synchronous LLM engine at production-scale

Ray Data LLM, developed by Anyscale, achieves 2x higher throughput compared to vLLM's synchronous LLM engine when operating at production scale. This performance improvement likely stems from Ray Data

applicationsanyscale-ray

ReasonOps: Operator Segmentation for LLM Reasoning Traces

arXiv:2605.29192v1 Announce Type: new Abstract: Chain-of-thought traces from large reasoning models can span tens of thousands of tokens, yet we lack a vocabulary for describing their internal structu

Reinforcing Few-step Generators via Reward-Tilted Distribution Matching

arXiv:2605.26108v2 Announce Type: replace Abstract: Recent advances in few-step diffusion distillation have enabled efficient image generation, yet aligning these models with human preferences remains

safetyarxiv-cs-cv

Agents

Revisiting Observation Reduction for Web Agents: Comprehensive Evaluation with a Lightweight Framework

arXiv:2605.29397v1 Announce Type: new Abstract: HTML observations in LLM-based web agents are extremely long, and while many reduction methods have been proposed, it remains unclear which methods redu

agentsarxiv-cs-cl

Riemannian AmbientFlow: Towards Simultaneous Manifold Learning and Generative Modeling from Corrupted Data

arXiv:2601.18728v2 Announce Type: replace Abstract: Modern generative modeling methods have demonstrated strong performance in learning complex data distributions from clean samples. In many scientifi

Robust and Generalizable Safety Steering for Text-to-Image Diffusion Transformers

arXiv:2605.30049v1 Announce Type: new Abstract: Diffusion Transformers have become a powerful backbone for text-to-image generation, but their layered and cross-modal generation process makes safety c

safetyarxiv-cs-ai

Tutorials

Routing by Reaching: Composition of Pre-trained GFlowNets for Multi-Objective Generation

arXiv:2602.21565v2 Announce Type: replace Abstract: Generative Flow Networks (GFlowNets) learn to sample diverse candidates in proportion to a reward function, making them well-suited for scientific d

tutorialsarxiv-cs-lg

Hardware

Run Step 3.7 Flash on NVIDIA GPUs with Enterprise-Ready Multimodal AI

Step 3.7 Flash is a 198B-parameter Mixture-of-Experts vision-language model designed for enterprise-scale production workloads, featuring native image and video input, a 256k context window, and confi

hardwarenvidia-developer

Local Ai

S3Mem: Structured Spatiotemporal Scene-Event Memory for Long-Horizon Interactive Question Answering

arXiv:2605.28831v1 Announce Type: cross Abstract: Long-horizon interactive agents often accumulate large trajectory histories yet still fail to answer questions about earlier events reliably. We argue

local-aiarxiv-cs-ai

SAAS: Self-Aware Reinforcement Learning for Over-Search Mitigation in Agentic Search

arXiv:2605.29796v1 Announce Type: new Abstract: Agentic search enables LLMs to solve complex multi-hop questions through iterative reasoning and external search. Despite the effectiveness, these syste

SADA: Safe and Adaptive Aggregation of Multiple Black-Box Predictions in Semi-Supervised Learning

arXiv:2509.21707v3 Announce Type: replace-cross Abstract: Semi-supervised learning (SSL) arises in practice when labeled data are scarce or expensive to obtain, while large quantities of unlabeled dat

SAM3D-Phys: Towards Multi-Object Interactive Simulation in Real World

arXiv:2605.30239v1 Announce Type: new Abstract: This work addresses the problem of recovering complete, simulatable object geometry from reconstructed real-world scenes, enabling physics-based interac

applicationsarxiv-cs-cv

SCoOP: Semantic Consistent Opinion Pooling for Uncertainty Quantification in Multiple Vision-Language Model Systems

arXiv:2603.23853v3 Announce Type: replace Abstract: Combining multiple Vision-Language Models (VLMs) can enhance multimodal reasoning and robustness, but aggregating heterogeneous models' outputs ampl

Semantic and Visual Evidence for Efficient Long-Video Reasoning: A Solution for the HD-EPIC VQA Challenge

arXiv:2605.29402v1 Announce Type: cross Abstract: Understanding long-form egocentric videos remains challenging for multimodal large language models (MLLMs) due to limited context length and insuffici

Solved in Unit Domain: JacobiNet for Differentiable Coordinate-Transformed PINNs

arXiv:2508.02537v3 Announce Type: replace Abstract: Physics-Informed Neural Networks (PINNs) offer a powerful framework for solving PDEs by embedding physical laws into the learning process. However,

STAP: A Shuffle-Tokenized App Predictor with Ultra Long Context for Vocabulary-Free Mobile App Prediction

arXiv:2605.29863v1 Announce Type: new Abstract: Predicting the next mobile application a user will launch is essential for intelligent device resource management and proactive assistance. Existing mod

Steering Language Models Before They Speak: Logit-Level Interventions

arXiv:2601.10960v2 Announce Type: replace-cross Abstract: Controllable generation requires language models to realize output characteristics such as reading level, politeness, and toxicity. Existing s

Stochastic Lifting for Generating Trajectories of Stochastic Physical Systems

arXiv:2605.29194v1 Announce Type: cross Abstract: Many stochastic physical systems evolve smoothly over time in the sense that the distribution of states changes regularly across time steps. The trans

SurfFill: Completion of LiDAR Point Clouds via Gaussian Surfel Splatting

arXiv:2512.03010v2 Announce Type: replace Abstract: LiDAR-captured point clouds are often considered the gold standard in active 3D reconstruction. While their accuracy is exceptional in flat regions,

applicationsarxiv-cs-cv

Industry

TERAFAB IS GOING TO BE INSANE. We’re targeting 100–200 billion custom AI + memory chips per year at full ramp that’s 1 terawatt (1,000 GW) o…

TERAFAB IS GOING TO BE INSANE. We’re targeting 100–200 billion custom AI + memory chips per year at full ramp that’s 1 terawatt (1,000 GW) of annual AI compute capacity. Roughly 50x current global AI

industryelon-musk--x

Industry

Thanks for the call out and glad to see that our blog inspired the broader community! We at @FireworksAI_HQ have been using delta updates to…

Thanks for the call out and glad to see that our blog inspired the broader community! We at @FireworksAI_HQ have been using delta updates to scale RL globally for about a year. Including Composer 2 &

industryclem-delangue--x

The Confidence Shortcut: A Reasoning Failure Mode of Masked Diffusion Models

arXiv:2605.29123v1 Announce Type: new Abstract: Masked diffusion language models (MDMs) uniquely support any-order generation, with confidence-based decoding currently serving as the de facto standard

safetyarxiv-cs-ai

The Hamilton-Jacobi Theory of Deep Learning

arXiv:2605.28983v1 Announce Type: cross Abstract: In this paper, training a neural network is identified, exactly, as a search through Hamilton--Jacobi initial-value problems: each gradient step selec

The New Pro Se: Generative AI and the Surge in Federal Civil Self-Representation

arXiv:2605.29493v1 Announce Type: cross Abstract: Since public access to generative AI tools became widespread, federal civil litigation has seen a marked increase in pro se (self-represented) plainti

applicationsarxiv-cs-ai

The Price Reversal Phenomenon: When Cheaper Reasoning Models Cost More

arXiv:2603.23971v2 Announce Type: replace-cross Abstract: Developers and consumers increasingly choose reasoning models (RMs) based on their listed API prices. However, how accurately do these prices

Token Inflation: How Dishonest Providers Can Overcharge for Large Language Model Usage

arXiv:2605.30040v1 Announce Type: cross Abstract: Per-token billing is now the standard pricing model for commercial large language models (LLMs), so the honesty of reported token counts directly affe

Towards Consistent Video Geometry Estimation

arXiv:2605.30060v1 Announce Type: new Abstract: This work presents ViGeo, a feed-forward foundation model for recovering spatially dense and temporally consistent geometry from video sequences. Built

researcharxiv-cs-cv

Turbulence-Robust Dynamic Object Segmentation with Multi-Signal Priors and SAM2 Refinement

arXiv:2605.29292v1 Announce Type: new Abstract: This technical report presents our solution for the CVPR 2026 UG2+ Challenge Track 3: Dynamic Object Segmentation in Turbulence (DOST). We design a trai

researcharxiv-cs-cv

Local Ai

UI-KOBE: Knowledge-Oriented Behavior Exploration for Lightweight Graph-Guided GUI Agents

arXiv:2605.29534v1 Announce Type: new Abstract: Recent advances in mobile GUI agents have shown strong potential for automating mobile tasks, but most effective systems still depend on large vision-la

local-aiarxiv-cs-ai