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

Jun 9

Submitted by

YerbaPage

SWE-Explore: Benchmarking How Coding Agents Explore Repositories

SJTU

Shanghai Jiao Tong University

Submitted by

5456es

On the Geometry of On-Policy Distillation

hkustitd

香港科技大学

1

Submitted by

lhmd

Latent Spatial Memory for Video World Models

MicrosoftResearch

Microsoft Research

Submitted by

SorrowTea

LatentSkill: From In-Context Textual Skills to In-Weight Latent Skills for LLM Agents

·
11 authors

Submitted by

Jessamine

CoVEBench: Can Video Editing Models Handle Complex Instructions?

NJU-LINK

Submitted by

libertywing

FlashMemory-DeepSeek-V4: Lightning Index Ultra-Long Context via Lookahead Sparse Attention

tencent

Submitted by

HongchengGao

SpatialWorld: Benchmarking Interactive Spatial Reasoning of Multimodal Agents in Real-World Tasks

Tsinghua University

Submitted by

Opusdei

Human Psychometric Questionnaires Mischaracterize LLM Behavior

SeoulNatlUniv

Seoul National University

1

Submitted by

taesiri

Echo-Memory: A Controlled Study of Memory in Action World Models

·
16 authors

Submitted by

mxlin043

OmniGameArena: A Unified UE5 Benchmark for VLM Game Agents with Improvement Dynamics

TheHKU

Hong Kong University

Submitted by

taesiri

End-to-End Context Compression at Scale

·
15 authors

Submitted by

SereneC

AHA-WAM:Asynchronous Horizon-Adaptive World-Action Modeling with Observation-Guided Context Routing

SJTU

Shanghai Jiao Tong University

1

Submitted by

chxy95

SwiftVR: Real-Time One-Step Generative Video Restoration

·
8 authors

Submitted by

wuxiaojun

Bayesian-Agent: Posterior-Guided Skill Evolution for LLM Agent Harnesses

IDEA-FinAI

Submitted by

chamber111

Skill-RM: Unifying Heterogeneous Evaluation Criteria via Agent Skill

·
13 authors

Submitted by

wang-jiahao

OmniCap-IF: Benchmarking and Improving Instruction Following Abilities for Omni-Video Captioning

NJU-LINK

Submitted by

johnnyd-gensyn

DEI: Diversity in Evolutionary Inference for Quality-Diversity Search

Gensyn

Submitted by

Egorgij21

Whisper Hallucination Detection and Mitigation via Hidden Representation Steering and Sparse AutoEncoders

huawei-noah

HUAWEI Noah's Ark Lab

1

Submitted by

prayerdan

SlimSearcher: Training Efficiency-Aware Web Agents via Adaptive Reward Gating

·
7 authors

Submitted by

hzhouml

Reasoning Arena: Trace Tournaments When Verifiable Rewards Fall Short

mistralai

Submitted by

l1ziang

Light-WAM: Efficient World Action Models with State-Fusion Action Decoding

WuhanUniversity

Wuhan Univeristy

Submitted by

TroyeML

Answer Presence Drives RAG Rewriting Gains

·
11 authors

1

Submitted by

Fengzhuo

Why Muon Outperforms Adam: A Curvature Perspective

·
5 authors

Submitted by

zly-idleness

Liberating LLM Capabilities in Full-Duplex Speech Models

·
7 authors

Submitted by

Egorgij21

A Geometric Account of Activation Steering through Angle-Norm Decomposition

huawei-noah

HUAWEI Noah's Ark Lab

1

Submitted by

lingyongyan

DuMate-DeepResearch: An Auditable Multi-Agent System with Recursive Search and Rubric-Grounded Reasoning

baidu

Submitted by

cohennoa

Text-to-Image Models Need Less from Text Encoders Than You Think

·
4 authors

Submitted by

Louieworth

Trajectory-Refined Distillation

MilaQuebec

Mila – Quebec Artificial Intelligence Institute

Submitted by

XinyangDavidHan

Agents' Last Exam

Berkeley

1

Submitted by

charlesdj

Optical Reasoning: Rethinking Images as an Expressive Reasoning Medium Beyond Text

ModalityDance

Submitted by

taesiri

Evaluation Cards: An Interpretive Layer for AI Evaluation Reporting

·
48 authors

Submitted by

oguzhanercan

Phase Marginalization for Patch-Grid Instability in Vision Transformers

BILGEM-AI

Submitted by

Fascetta

Robotic Policy Adaptation via Weight-Space Meta-Learning

ItalAI

Submitted by

MarcusGu

WorldCraft: From Camera Navigation to Object Manipulation in Interactive Video World Models

tencent

1

Submitted by

manglu3935

Experience Makes Skillful: Enabling Generalizable Medical Agent Reasoning via Self-Evolving Skill Memory

·
11 authors

Submitted by

fjzzq2002

Hardening Agent Benchmarks with Adversarial Hacker-Fixer Loops

CarnegieMellonU

Carnegie Mellon University

Submitted by

yangtian6781

PBSD: Privileged Bayesian Self-Distillation for Long-Horizon Credit Assignment

SJTU

Shanghai Jiao Tong University

1

Submitted by

taesiri

OASIS: From Simulation Data Collection to Real-World Humanoid Loco-Manipulation

·
7 authors

Submitted by

XingweiT

EmpiriGraph-Psy: A Dataset and LLM Pipeline for Extracting Empirical Relation Graphs from Psychology Abstracts

·
4 authors

Submitted by

prateeksikdar

Chiaroscuro Attention: Spending Compute in the Dark

Accenture

Submitted by

lhy-zju

Skill-3D: Evolving Scene-Aware Skills for Agentic 3D Spatial Reasoning

zju

Zhejiang University

Submitted by

ThomSes

Where Rectified Flows Leak: Characterising Membership Signals Along the Interpolation Path

Telecom-Paris

Submitted by

ernlavr

SigmaScale: LLM Compression with SVD-based Low-Rank Decomposition and Learned Scaling Matrices

Aalborg Universitet

Submitted by

xiuyuz

Cosine Misleads: Auxiliary Losses Reshape Vision Language Models, Not Their Latents

NationalUniversityofSingapore

National University of Singapore

Submitted by

xiuyuz

Self-Evaluation Is Already There: Eliciting Latent Judge Calibration in Base LLMs with Minimal Data

NationalUniversityofSingapore

National University of Singapore

Submitted by

RickyDeSkywalker

Lean4Agent: Formal Modeling and Verification for Agent Workflow and Trajectory

UIUC-ScaleML

UIUC ScaleML Lab

Submitted by

suraj-ranganath

PIPE-Cypher: Automatic Enterprise Benchmark Generation for Text-to-Cypher Systems

UCSanDiego

University of California at San Diego

Submitted by

Factral

Set-Based Transformer for Atmospheric Compensation in Standoff LWIR Hyperspectral Imaging

·
4 authors

Submitted by

yurims1892

CIPER: A Unified Framework for Cross-view Image-retrieval and Pose-estimation

SeoulNatlUniv

Seoul National University

Submitted by

jisx

Reasoning over Grammar: Can Synthetic Linguistic Reasoning Traces Enhance Low-Resource Machine Translation?

OLAResearchX

Omni Language AI Research

Submitted by

pdx97

Honest Lying: Understanding Memory Confabulation in Reflexive Agents

UMBC

THE UNIVERSITY OF MARYLAND BALTIMORE COUNTY

Submitted by

jhyuckkim

Pruning and Distilling Mixture-of-Experts into Dense Language Models

KRAFTON

Submitted by

abanerj3

EMMA: Extracting Multiple physical parameters from Multimodal Data

ASU-IMPACT-Lab

IMPACT Lab at Arizona State University