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 16

Submitted by

VCLab-PolyU

OmniCompass: A Holistic Benchmark for Omni-Modal Video Grounding and Understanding

VCLab-HKPU

Submitted by

Nothing2Say

Waypoint-OPD: Transferring RL-Induced Policy Shifts via Gated On-Policy Distillation

·
7 authors

Submitted by

demolei

OpenGCUA: Toward a Fully Open Stack for Generalist Computer Use

Tongyi-MAI

Submitted by

WaylandYang

FDE-Gym: The Last Mile of AI Delivery

DeepLethe

Submitted by

QingchuanMa

RoTA-Bench: Evaluating 3D Spatial Reasoning through Interactive Projection Alignment

MAC-AutoML

Submitted by

yshenaw

Self-Evolving Agent Harness

zgcagi

Submitted by

Fr0zencr4nE

MJev-Spatial: System One Spatial Intelligence via Native Attention Matching

mindlab-research

Submitted by

JusperLee

Learning Where to Split: Dynamic Band Splitting for Audio Source Separation

thu-ml

Tsinghua Machine Learning Group

Submitted by

Edisoneh

InfiniSplat v2: Analyzing and Improving Feed-Forward 3DGS from Any Views

Submitted by

YipengGao

3DCodeVerse: Exploring 3D Worlds as Code

UniversityofSouthernCalifornia

University of Southern California

Submitted by

kongquyu

MAI-Gym: An Open All-in-One Platform for GUI Agent Evaluation and RL Training across Mobile and Computer Environments

Tongyi-MAI

Submitted by

Jinfa

SCOPE: Semantic Bounded Credit Optimization for Prompt Engineering in Text-to-image Generation

university-of-rochester

University of Rochester

Submitted by

enrue1893

Speculative Forcing for Efficient Autoregressive Video Generation

naver-ai

Submitted by

zwq2018

VLA-Δ: Learning Physical State Transition for Vision-Language-Action Models

OmniAI-ZJU

Submitted by

sapphirex

Mosaic: Multi-Session Agentic Reinforcement Learning

LARK-Lab

LARK Lab@HKUST (GZ)

Submitted by

Zane-QIU

Intern Lumina U2: A Multi-Codebook Diffusion Large Language Model for Omni-Visual Understanding and Image Generation

Shanghai-AI-Laboratory

Shanghai AI Laboratory

Submitted by

hbx

AutoSFT: An Agentic System for Automated Data Research in Supervised Fine-Tuning

TsinghuaNLP

Tsinghua NLP Group

Submitted by

BBBBCHAN

DeltaStreamer: Towards Human-Like Agentic Streaming Video Understanding

ByteDance

Submitted by

ldkong

Orbia: Do Generated Videos Form a Consistent 3D World?

tencent

Submitted by

wenhaoyan77

WorldCast: Exploring Intelligent Visual Generation in World State Prediction

meituan-longcat

Submitted by

sabulin

InfoVis: Unified Slide Image Generation and Editing

BAAI

Beijing Academy of Artificial Intelligence

Submitted by

yxc97

MLExpert: Towards a Generalist Agentic Model for Autonomous Machine Learning Engineering

InternScience

Submitted by

Zhanli-Li

Data World: Scaling Data Environments from Scratch

meituan-longcat

Submitted by

Spico

Intern-Decision: Efficient Multimodal Decision Making with Autoregressive Masked Language Modeling

internlm

Intern Large Models

Submitted by

TberiusJunyao

Terminal-RSI: Recursive Self-Improving Terminal Agents via Recursive Self-Rewrite and Policy-Refine

tencent

Submitted by

TaobaoTmall-AlgorithmProducts

CommerceGen-Studio: An End-to-End Framework for Commercial-Grade Main-Image Generation

alibabagroup

Submitted by

florinshum

Routing Policy Optimization for Online Diffusion Reinforcement Learning

NationalUniversityofSingapore

National University of Singapore

Submitted by

DarkBluee

EdgeWise: Bringing Long-Horizon Agents to the Edge via Selective Cloud Delegation

nvidia

Submitted by

comin

OmniVerifier-C1: Decomposed Graded Verifier for Rule-Governed Generative World Modeling

·
13 authors

1

Submitted by

zhaihaotian

Each Reward to Its Relevant Tokens: Reward-Specific Token-Level Credit Assignment for Multi-Reward RL

MBZUAI

Mohamed Bin Zayed University of Artificial Intelligence

Submitted by

OpenDriveLab-org

Do Better Visual Representations Always Lead to Better End-to-End Autonomous Driving?

OpenDriveLab

Submitted by

OpenDriveLab-org

SimVTA: Simulation-Grounded Video-Trajectory Alignment for Driving World Action Models

OpenDriveLab

Submitted by

Silin-Chen

SWE-at-Scale: Scalable Issue Synthesis from Historical Bug Fixes

SJTU

Shanghai Jiao Tong University

Submitted by

florinshum

Velox: Spatial Memory Makes World Models Efficient

NationalUniversityofSingapore

National University of Singapore

Submitted by

Leozkl

JevAgentHarness: Benchmarking, Deploying and Training Decision Models That Share Control with LLMs

illinois

University of Illinois at Urbana-Champaign

Submitted by

siiian

MFVSR: A Multi-Style and Fine-Grained Benchmark for Video Subtitle Removal

·
6 authors

Submitted by

terr1ble

Trajectory-Level Speculation with Semantic Verification for Tool-Using Agents

·
5 authors

Submitted by

MasterZhou

Athanor: An Agent-Native System for Autonomous Post-Training Research

ByteDance

1

Submitted by

Vincent2311

CarryOn: A Dual-Loop Harness for Personalized and Reliable Robot Assistance

·
10 authors

Submitted by

supra3d

SUPRA-TEX: 3D Generation with Super-Aligned Texture

·
8 authors

Submitted by

isminoula

CoDA: Co-Evolution of Data Engines and Action Policies

PLAN-Lab

PLAN Lab @University of Illinois Urbana-Champaign

Submitted by

Xkev

Learning Self-Orchestration in Decentralized Agent Swarms

Harvard

Harvard University

Submitted by

zhuomingliu

Memory as Weights: Internalizing Long-Term History for Streaming Videos

uw-madison

University of Wisconsin - Madison

Submitted by

JiajiDeng

Cowork-Gym: Synthesizing Interactive Work Environments at Scale

·
21 authors

Submitted by

rescuerz

WorldMemBench: A Comprehensive Benchmark for Memory of Video World Models

ZJU-REAL

Submitted by

yangwang825

Calibrated Typed Decisions for Entity, Relation, and Event Extraction

·
5 authors

Submitted by

Guoxia

Virtual Head Attention

baidu

Submitted by

Guoxia

FlashMask-3: Efficient and Expressive Mask-Aware Distributed Attention

baidu

Submitted by

j-song

PN-Mix: Pair-Norm Guided Mixed-Precision Key Cache Quantization for Efficient Long-Output LLM Inference

snu

Seoul National University

1

Submitted by

chaofanma

SERO: Learning from Executable Oracle Continuations for Long-Horizon Agentic Reinforcement Learning

Shanghai-AI-Laboratory

Shanghai AI Laboratory

Submitted by

orronai

H-DDCM: Hadamard Codebooks for Fast Generative Image and Video Compression

Technion

Technion Israel institute of technology

1

Submitted by

Halcyon-Zhang

MMSearch-OPD: Easy Made Hard, Hard Made Learnable for Long-Horizon Multimodal Search Agents

PekingUniversity

Peking University

1

Submitted by

yxgh

PINT: A Unified Model with Shared Predictive Intent for Robot Manipulation

·
5 authors

Submitted by

CoderDoge

CLIFT: Conformal Self-Verification for Web Agent Training and Test-Time Scaling

Salesforce

Salesforce AI Research

1

Submitted by

simingfu

RepForcing: Representation-Space Initialization for Autoregressive Video Distillation

zju

Zhejiang University

Submitted by

taesiri

A team of Gemini 4 Argon agents beats Minecraft

·
3 authors

Submitted by

hematteo

Learning to Read Out: Unembedding Dynamics in Language Model Pretraining

·
6 authors

Submitted by

nthngdy

Copy-Paste Language Models

lil-lab

Cornell LIL Lab

Submitted by

sekhcopenlu

Explaining the Roles of Experts in Mixture-of-Experts LLMs for Personalization

copenlu

Submitted by

dginf

Beyond Policy Rollouts: Embedding-Space Directional Exploration with Anchored Noise for LLM RLVR

·
4 authors

Submitted by

HJOK

One Bias Score, Two Factors: How Mitigations Affect Option Bias in Multiple-Choice Evaluation

·
3 authors

Submitted by

yitongx

Gen4P: Video Generators are 4D Perceivers

·
3 authors

Submitted by

selensu

Can World and Action Modeling Benefit Each Other under Unified Causal Training?

IAAI233

Interactive and Active AI

Submitted by

EXCAI

ForcingBooster: Adaptive Patch-wise Denoising for Fast Autoregressive Video Diffusion

VAST-AI

VAST-AI-Research

Submitted by

Xiaodong

SimForcing: Distilling Simulation Motion Priors into Real-Domain Robot World Models

PekingUniversity

Peking University

Submitted by

JLingxiao

Recursive Environment Scaling for General Long-Horizon Agents

·
13 authors

1

Submitted by

jhhuangchloe

VascuSR: Iterative Anatomy-Aware Diffusion for CT Super-Resolution

·
5 authors

Submitted by

JPShi

EVA-Cap: Optimizing Audiovisual Video Captioning via Event-Centric Alignment

Fudan-University

Fudan University

1

Submitted by

hmhm1229

Skill-Decoupled On-Policy Expert Distillation for Unified Vision-Language Models

·
8 authors

Submitted by

deqing

TimeRSI: Towards Self-Improved Agentic Deployment for Time-Series Foundation Models

GeorgiaTech

Georgia Institute of Technology

Submitted by

AAyano

PAIR: Bridging Perception and Action in Vision-Language-Action Models

·
3 authors

1

Submitted by

AntonioTN

When the KV Cache Is Slower Than the GPU: Scaling Remote KV Tiers for Context-Parallel Prefill

MultiverseComputingCAI

Multiverse Computing

Submitted by

theboringai-work

Speech event detector: Timestamped detection of non-verbal vocalizations and background sounds for rich speech transcription

·
2 authors

Submitted by

charlesdj

CoWorld Reasoner: Agentic Multimodal Reasoning with the Evolving Video World Model-Driven Harness

ModalityDance

Submitted by

sharmaam85

AI-Driven, Timing-Aware Power Optimization

·
1 authors

1

Submitted by

ellenyige

ANTE: Anticipate, Then Execute -- Learning Internal World Models for Coding Agents from Verified Beliefs

amazonwebservices

Amazon Web Services (AWS)

1