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 21

Submitted by

fyzzzzzz

Focused Omni-Modal GUI Agents

NanyangTechnologicalUniversity

Nanyang Technological University

Submitted by

Yooki

PAR: Progress-Aligned Recurrence for Length Generalization in State Tracking

ant-intl

ant-international

Submitted by

taesiri

Matrix-Game 4.0: Harnessing Video Models for Real-Time Playable Game Worlds

·
18 authors

Submitted by

tqliu

Ropedia W0: Scaling Loco-Manipulation World Models across Embodiments with Human Experience

ropedia-ai

Submitted by

lifuguan

S-Agent 2.0: Spatial Reasoning with a General Coding Agent

ropedia-ai

Submitted by

sY713

Sage-A1.5: From Long-Horizon Execution to Scientific Exploration with Capability Composition

InternScience

Submitted by

nullptr04

CoEvo-RSI: When Harnesses Evolve, Which Models Should Follow?

·
8 authors

Submitted by

zwq2018

VLA-Rolling: Fewer Denoising Steps, Better Control via Adaptive Action Reuse

OmniAI-ZJU

Submitted by

hongsunghwan

Learning the 4D World through Point Trajectory Forecasting

·
13 authors

Submitted by

Enderfga

SolarRenderer: Few-Step Generative Reference Rendering

showlab

Submitted by

Nikojamesnow

SWIFT: Self-guided Weak-to-strong dIFfusion models via Target extrapolation

alibabagroup

1

Submitted by

sunovivid

Correction Training for Diffusion Models

UCBerkeley

University of California, Berkeley

Submitted by

happy8825

MMEnterprise-Bench: Evaluating Enterprise Agents over Evolving Multimodal Records

·
7 authors

Submitted by

florinshum

LLM2R: From Large Language Models to 3D Reassemblers

NationalUniversityofSingapore

National University of Singapore

Submitted by

yuanhang123

PhysSkin++: Real-Time and Generalizable Physics-Based Animation via Self-Supervised Neural Skinning

Submitted by

beannn

Specter: Fast MoE Inference via Quantization-Enabled Multi-Step Speculative Offloading

·
8 authors

Submitted by

Markus-Pobitzer

Painting Change Map as guidance for generating painting processes

MHUGLab

Multimedia and Human Understanding Group

Submitted by

wsdwJohn1231

EvaOmni: An Evidence-Grounded Benchmark for Audio-Visual Reasoning

VisionXLab

SJTU VisionXLab

Submitted by

jaesangyu

Breaking Diffusion: Parallel Reasoning with Draft Refinement

microsoft

1

Submitted by

TobyYang7

Evolving Agent Interfaces

·
8 authors

1

Submitted by

CrazyK828

From Short Clips to Real-Time Streams: Identity-Aware Supervision and Context-Efficient Generation for Video Face Swapping

·
11 authors

Submitted by

zacharyz21

Event-STVG: Spatio-Temporal Video Grounding for Long-Form Event Reasoning

·
13 authors

Submitted by

Lie24

DecisionPivot: A Diagnostic Benchmark for Decision Models in LLM Systems

startlux-models

Submitted by

pillor

EdgeKernelBench: Unified LLM-Driven Kernel Generation for Edge Backends

encodelab

Westlake ENCODE Lab

Submitted by

HowieHwong

Epi: Towards a Persistent, Self-Improving Agent for Autonomous Research

BakeLab

Submitted by

li-fly

FlowState: Execution State as Memory for Long-Horizon LLM Agents

ant-intl

ant-international

1

Submitted by

namanvats

Good Judgment, Bad Advice: Decision-Model Controllers Raise Agent Cost Without Improving Success

·
1 authors

Submitted by

qindomitable

HarnessRL: Co-Evolving Policies with a Just-in-Time Training Harness for Open-Ended RL

Tongyi-ConvAI

Submitted by

noamrot

Joints, Camera, Action! Co-Generating 3D Motion and Video for Arbitrary Rigs

nvidia

1

Submitted by

ruhwang

More Memory, Worse Agents: Error Reproduction and Anti-Persistence in LLM Agents

Indiana University Bloomington

Submitted by

chenyilun95

λ_0: Scaling Open-World Egocentric Human Data for Humanoid Loco-Manipulation

lambda-humanoid

Lambda Humanoid

Submitted by

ruhwang

Harness Self-Evolution for Long-Horizon Terminal Agents

Tencent-Hunyuan

Tencent Hunyuan

Submitted by

david3684

Large Flow Language Models

·
7 authors

Submitted by

HanNight

HybridResearcher: Bridging the Structured-Unstructured Boundary for Deep Research Agents

Snowflake

1

Submitted by

qinjr

Evaluating Agentic Capabilities Before Post-Training: An End-to-End Approach

tencent

Submitted by

deqing

Unsupervised Tool Evolution for Coding Agents in Data-Lake Analytics

google

1

Submitted by

ZhenyangLiu

OmniCam: Omni-Camera Trajectory Generation via Geometry-Grounded Pose Token Learning

·
8 authors

Submitted by

Ryenhails

Multimodal Reranking Is a System One Decision

·
3 authors

Submitted by

robinzixuan

EMBER: Predicting TCR--pMHC Binding Without a Docked Complex

NWULIST

1

Submitted by

grenadeComing

Beyond Compilation: Evaluating Faithful Natural-Language-to-Lean Statement Formalization

ucriverside

Submitted by

facias

WinARDet: Compressed Window Auto-Regression for Long-Horizon Vision-Language Grounding

NankaiUniversity

Nankai University

Submitted by

jjunsss

HumanSkill-4D: Training-Free Language Querying of Animatable Gaussian Avatars in Space and Time

UNIST

Ulsan National Institute of Science and Technology

Submitted by

SuperSupermoon

TRACE: Tracking Multimodal Clinical Trajectories for Inpatient Risk Prediction

·
2 authors

1

Submitted by

KSZG

PVBench: Quantifying Physical Plausibility in Video Generation with Object-Level Physical Measurements

·
8 authors

Submitted by

Nova2001

SpecKCast: Spectral Kalman Forecasting for Diffusion Acceleration

SamsungResearch

Samsung Research

Submitted by

YuCrazing

PhysLDM: Latent Diffusion for High-Fidelity Deformable Simulation

mmlab-ntu

1

Submitted by

nielsr

Muse Glimmer 30B: An On-Device Agentic Model

meta

Submitted by

rohitsaxena

On the Limits of Scale: Inverse Scaling in Vision-Language Models

UniversityOfEdinburgh

University of Edinburgh