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

Sep 22

Submitted by

richardxp888

RRSI: Regularized Recursive Self-Improvement of Agent Harnesses

google

Submitted by

Drexubery

WorldCrafter: Consistent Video World Model with Implicit 3D-aware Memory

TencentARC

ARC Lab, Tencent

Submitted by

taesiri

GameHorizon Suite: Multi-Horizon Data and Evaluation in Gameplay

tencent

Submitted by

MenghaoGuo

Transferring the Intelligence of VLMs to Robotic Control

·
9 authors

Submitted by

lhpku20010120

OmniEdu: Open Foundation Models for Learning and Teaching

PekingUniversity

Peking University

Submitted by

udayallu

Document Retrieval-Aware Chunking (D-RAC): Universal Retrieval-Aware Ingestion of Enterprise Documents via PDF Normalization and Multimodal Markdown Conversion

Yellow-AI-NLP

Submitted by

taesiri

VideoGen-Agent: Reinforcing Video Generation Agents

princetonu

Princeton University

Submitted by

LichengLiu03

onPanda: Efficient Annotation of On-Policy Alignment Data for LLMs and Agents via Token-Level Correction

stepfun-ai

Submitted by

Jiafei1224

Grounded Action Model: 3D Grounding as a Foundation for Robotics

NationalUniversityofSingapore

National University of Singapore

Submitted by

Williams07

One to More, More to One: Category-Aware Iterative Expert Training for Software Engineering Agents

alibabagroup

2

Submitted by

henry-yeh

Harness-Zero: Harness Distillation via Agent-as-Harness

PekingUniversity

Peking University

Submitted by

rtmdrr

Deep Persona: A Psychologically Grounded Architecture and Evaluation Framework for Role-Playing Agents and Simulations

UoHaifa

University of Haifa

Submitted by

dj220001

Jev-Mem: System-One-Controlled Agentic Memory for Efficient AI Agents

·
3 authors

Submitted by

MrBean2024

CARE: Experience-Guided Atomic Corrective Execution for Vision-Language-Action Policies

dalian-university-of-technology

DaLian University of Technology

Submitted by

paulsmith0217

Why Do Video Diffusion Models Violate Physics? Unveiling the Flaws in Attention Mechanisms

BUPT-CIST

BUPT-Center of Intelligent Science and Technology

Submitted by

HuanxinSheng

1% of Tokens Can Be Enough: On Gradient Estimation in On-Policy Distillation

antgroup

1

Submitted by

3587jjh

HuRo: Robotizing Human Videos for Scalable VLA Pretraining

RLWRLD

Submitted by

termanteus

Think Like a World Model, Act Like a VLA: Distilling World-Model Representations into Compact Robot Policies

·
5 authors

Submitted by

korbip

Complex KDA: Understanding and Enhancing the Expressivity of Kimi Delta Attention

openeurollm

Submitted by

iainzhang

SkillSpec: Intent-Masked Specification Reasoning for Agent Skill Correctness

Buaa1

beihang university

Submitted by

mrkwanzaa

The Functionalizer: Lossless Functional Decomposition for Subword Tokenization

cavelab

Submitted by

tayalmanan

ShieldVLA: Feasibility-Aware Safety Alignment for Vision-Language-Action Models

MicrosoftResearch

Microsoft Research

1

Submitted by

Yang-Tian

Mira-Scene: Pixel-Aligned Layouts for Generative 3D Scene

TheHKU

Hong Kong University

Submitted by

Elfsong

Measuring the Checker: Mutation Analysis for GPU-Kernel Benchmark Oracles

NationalUniversityofSingapore

National University of Singapore

Submitted by

hisku

A Lie Detector Test for Language Models: Reading Knowledge a Model Won't Reveal

·
1 authors

1

Submitted by

oopatow

TAPe+ML: A Compact Structured Representation for Multi-Task Computer Vision

Comexp

Comexp Research Lab

1

Submitted by

ashutosh1919

EDGEGEN: Improving Tool-Calling Agents Beyond Happy Paths with Synthetic Edge Case Generation

SAP

1

Submitted by

kokomarina

Prediction-Powered Smoothing and Validation for Disaggregated AI Evaluation

UCSantaCruz

University of California at Santa Cruz

Submitted by

Cicici1109

Streaming Video Editing with Easy Adaptation

SJTU

Shanghai Jiao Tong University

Submitted by

AdinaY

Realtime-Venus: A full-duplex interaction system with asynchronous delegation

inclusionAI

Submitted by

YiboZhang2001

UltraTex: Unleashing 2K Multi-View Diffusion for 3D Texturing

·
7 authors