new

Get trending papers in your email inbox once a day!

Get trending papers in your email inbox!

Daily Papers

byAK and the research community

Oct 28

Submitted by

HU22333

ChronoReplay: One Answer Interface Is Not Enough for Streaming Spatial Intelligence

LINs-lab

Submitted by

Tingyang-Chen

Multiscale Policy Optimization for Long-Horizon LLM Agents

IQuestLab

Submitted by

IntJudge

The Trinity:Capability-Driven Recursive Self-Improvement for Multimodal Agents

NationalUniversityofSingapore

National University of Singapore

Submitted by

diaomuxi

VibeVideoBench: Evaluating Multimodal Agents on Code-Driven Video Generation

·
24 authors

Submitted by

lin5547

PDRVLA: A Vision-Language-Action Model with Physical Dynamics Representations

·
15 authors

Submitted by

weizhiwang

Hy-Nomas: Towards Free-form Interactive World Generation

Tencent-Hunyuan

Tencent Hunyuan

Submitted by

hexuan21

Memory Tax: Dissecting the Net Value of Agent Memory

UIUC-CS

University of Illinois at Urbana-Champaign

Submitted by

yuezhengrong

Can Discrete Tokens Serve as Semantically Lossless Representations of Visual Signals?

meituan-longcat

Submitted by

hollyming

HABIT-Bench: Evaluating Habit Induction from Longitudinal Weak Evidence in Agent Memory

SJTU

Shanghai Jiao Tong University

Submitted by

ryancll118

Rollback: Experience Backtracking in World Models for Policy Improvement

agibot-world

Submitted by

DogNeverSleep

CodeCap-Bench: Can Code Serve as a Good Video Caption?

·
14 authors

1

Submitted by

chyun

ThinkWorld: World Imagination Reinforces Chain-of-Dynamics Reasoning

nvidia

Submitted by

yepengliu

Can Test-Time Memory Evolution Outperform Test-Time Training for Scientific Discovery?

UCSantaBarbara

University of California, Santa Barbara

Submitted by

HigherHu

LociArt: A Unified Paradigm for Free-Style Local Video Stylization

xiaomi-research

Xiaomi Research

Submitted by

WaltonFuture

Zooming without Zooming v2: From Seeing Details to Connecting Evidence for Generalizable Multimodal Perception

inclusionAI

Submitted by

zhangchenxu

RSI Arena: Towards Recursive Self-Improvement with Human Feedback

RSIArena

Submitted by

Shuyun-Wang

MatAnyone-E: Sparse Memory Propagation for Efficient High-Resolution Video Matting

mmlab-ntu

Submitted by

skandermoalla

h1: A Taste of the Bitter Lesson in Automated Theorem Proving

meta-llama

Submitted by

gmlwns5176

Channel Selection for Mixture-of-Experts Decoding Needs No Calibration Data

kaist-ai

Submitted by

ColorfulAI

Omni-InteractScope: Benchmarking Proactive and Agentic Interaction in Continuous Omnimodal Streams

Qwen

Submitted by

KaiserWhoLearns

What Effort Level Should I Choose? On the Economics of Reasoning Effort in LLM Agents

togethercomputer

Submitted by

tomaarsen

EmbeddingGemma 2: Unified and Efficient Multimodal Representations

google

Submitted by

huangsiteng

RynnWorld-Latent: A Cross-Embodiment World Model Conditioned on Universal Latent Actions

Alibaba-DAMO-Academy

Submitted by

BennoKrojer

Cross-modal Steering Vectors Exist Across Both Image and Text Generation

MilaQuebec

Mila – Quebec Artificial Intelligence Institute

Submitted by

PenPaperKeyCode

The Lexical Imprint of Pre-training Data

naver-hyperclovax

Submitted by

mujianijan

Your Policy Is Secretly a Harness: The Decision Geometry of Agent Interfaces

Submitted by

ethanning

SearchHorizonBench: A Benchmark for Search Value in AI Agents

CarnegieMellonU

Carnegie Mellon University

Submitted by

YZCS

Can LLMs Uncover Hidden Physical Laws through Free Exploration?

·
4 authors

Submitted by

EliezerS

On the SPOT: Enhancing Speculative Decoding for Diffusion Language Models via Commit-Order Prediction

·
4 authors

Submitted by

taesiri

Timestral

mistralai

Submitted by

xxyyy123

OvisOCR3 Technical Report

ATH-MaaS

1

Submitted by

bizhen

Harness-Model Collaboration for Long-Horizon Continual Learning

·
9 authors

2

Submitted by

jacklishufan

Nemotron-Labs-Diffusion-Uni: Elastic Self-Refining Unified Diffusion LLM with Cross-Paradigm On-Policy Distillation

nvidia

Submitted by

ImAnouar1

Rare, Relevant, and Still Wrong: How Memory Fails LLM Group Decisions

·
4 authors

1

Submitted by

hyunw3

Benchmarking Obstacle-Conditioned Safety in Household Robot Navigation

SeoulNatlUniv

Seoul National University

Submitted by

rayruiyang

Learning and Reasoning with Geometry: How Generation Helps Spatial Understanding

·
6 authors

Submitted by

taesiri

MLE-World: Predicting Machine Learning Experiment Outcomes with a Learned World Model

·
5 authors

Submitted by

y0urOy

Vision-Language-Action or World-Action? Just Action Models

·
4 authors

Submitted by

ziqiaow

Beyond the Estimate: Wiener-Informed Decoding for Easier Training of Pixel-Space Flow Models

NationalUniversityofSingapore

National University of Singapore

Submitted by

shanyangmie

What Outcome RL Buys: Finishing, Reliability, Rarely Reach

·
1 authors

Submitted by

jhzyu

How Much Does the Hint Matter? Scope Conditions for Hindsight Self-Distillation in Financial Reasoning

ctgt-inc

Submitted by

hduvallh

Sub-monolayer tracking of molecular states at buried interfaces

carbon-lab

Submitted by

skhanuja

T-DIG: Text-Conditioned Diversity for Image Generation

CarnegieMellonU

Carnegie Mellon University

Submitted by

xjxyys

Truncate or Not? A Scaling Study of Looped Transformers with Truncated Backpropagation

·
8 authors

Submitted by

Leigest

PhysEvolve: Verifier-grounded Self-evolution for Physical Reasoning with Executable Worlds

·
6 authors

Submitted by

SatoSakula

MoveBench: A Rubric-Based Benchmark for Evaluating Large Language Models in Sports and Health Coaching

·
5 authors

Submitted by

Sanjidh090

Lipi-Ghor: A Large-Scale, Diarized, Multi-Domain Bengali Speech Corpus and a Scalable Pipeline for Building It