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

March

Submitted by

lkdhy

AI Can Learn Scientific Taste

OpenMOSS-Team

Submitted by

taesiri

Demystifing Video Reasoning

sensenova

Submitted by

tianlezeng

CARLA-Air: Fly Drones Inside a CARLA World -- A Unified Infrastructure for Air-Ground Embodied Intelligence

·
4 authors

Submitted by

csjiaya

InCoder-32B: Code Foundation Model for Industrial Scenarios

Beihang

Beihang University

Submitted by

Jinfa

SocialOmni: Benchmarking Audio-Visual Social Interactivity in Omni Models

·
12 authors

Submitted by

Luckyyy

Bootstrapping Exploration with Group-Level Natural Language Feedback in Reinforcement Learning

rednote-hilab

Submitted by

hzxllll

Heterogeneous Agent Collaborative Reinforcement Learning

ByteDance

8

Submitted by

taesiri

Helios: Real Real-Time Long Video Generation Model

ByteDance

Submitted by

oriuta

MiroThinker-1.7 & H1: Towards Heavy-Duty Research Agents via Verification

miromind-ai

Submitted by

Gofinge

Utonia: Toward One Encoder for All Point Clouds

Pointcept

Submitted by

taesiri

Attention Residuals

moonshotai

Submitted by

dkliang

Out of Sight but Not Out of Mind: Hybrid Memory for Dynamic Video World Models

H-EmbodVis

Submitted by

yawenluo

ShotStream: Streaming Multi-Shot Video Generation for Interactive Storytelling

·
8 authors

Submitted by

dongdaxiang

Qianfan-OCR: A Unified End-to-End Model for Document Intelligence

baidu

Submitted by

jyseo

Grounding World Simulation Models in a Real-World Metropolis

naver-ai

Submitted by

ZHZisZZ

dLLM: Simple Diffusion Language Modeling

Berkeley

Submitted by

yukangcao

HSImul3R: Physics-in-the-Loop Reconstruction of Simulation-Ready Human-Scene Interactions

mmlab-ntu

Submitted by

Lingaaaaaaa

OpenClaw-RL: Train Any Agent Simply by Talking

princeton-ai

Princeton AI Lab

Submitted by

OmniLottie

OmniLottie: Generating Vector Animations via Parameterized Lottie Tokens

Fudan-University

Fudan University

Submitted by

yuwendu

OpenSeeker: Democratizing Frontier Search Agents by Fully Open-Sourcing Training Data

Submitted by

BAJUKA

EnterpriseOps-Gym: Environments and Evaluations for Stateful Agentic Planning and Tool Use in Enterprise Settings

ServiceNow-AI

Submitted by

Yulin-Li

Efficient Reasoning with Balanced Thinking

·
8 authors

Submitted by

exander

Geometry-Guided Reinforcement Learning for Multi-view Consistent 3D Scene Editing

GD-ML

Submitted by

XiaotongLi97

LongCat-Next: Lexicalizing Modalities as Discrete Tokens

meituan-longcat

Submitted by

zbeeb

TAPS: Task Aware Proposal Distributions for Speculative Sampling

IVUL-KAUST

Image and Video Understanding Lab

Submitted by

Huaxiu

MetaClaw: Just Talk -- An Agent That Meta-Learns and Evolves in the Wild

UNC-ChapelHill

University of North Carolina at Chapel Hill

Submitted by

xiaochonglinghu

From Scale to Speed: Adaptive Test-Time Scaling for Image Editing

alibaba-inc

Submitted by

taesiri

MinerU-Diffusion: Rethinking Document OCR as Inverse Rendering via Diffusion Decoding

·
6 authors

Submitted by

taesiri

Intern-S1-Pro: Scientific Multimodal Foundation Model at Trillion Scale

internlm

Intern Large Models

Submitted by

xiaochonglinghu

Omni-WorldBench: Towards a Comprehensive Interaction-Centric Evaluation for World Models

alibaba-inc

Submitted by

ethanchern

Speed by Simplicity: A Single-Stream Architecture for Fast Audio-Video Generative Foundation Model

·
45 authors

Submitted by

Qinsi1

T2S-Bench & Structure-of-Thought: Benchmarking and Prompting Comprehensive Text-to-Structure Reasoning

·
15 authors

Submitted by

taesiri

Penguin-VL: Exploring the Efficiency Limits of VLM with LLM-based Vision Encoders

tencent

Submitted by

wchengad

PixelSmile: Toward Fine-Grained Facial Expression Editing

Fudan-University

Fudan University

Submitted by

Franklinzhang

Astrolabe: Steering Forward-Process Reinforcement Learning for Distilled Autoregressive Video Models

·
9 authors

Submitted by

shenzhi-wang

HopChain: Multi-Hop Data Synthesis for Generalizable Vision-Language Reasoning

Qwen

5

Submitted by

tsbpp

Beyond Language Modeling: An Exploration of Multimodal Pretraining

6

Submitted by

taesiri

CUA-Suite: Massive Human-annotated Video Demonstrations for Computer-Use Agents

ServiceNow

5

Submitted by

taesiri

CUDA Agent: Large-Scale Agentic RL for High-Performance CUDA Kernel Generation

ByteDance-Seed

Submitted by

dkliang

Generation Models Know Space: Unleashing Implicit 3D Priors for Scene Understanding

H-EmbodVis

Submitted by

ZhuofengLi

OpenResearcher: A Fully Open Pipeline for Long-Horizon Deep Research Trajectory Synthesis

TIGER-Lab

Submitted by

JerryWzh

Thinking in Uncertainty: Mitigating Hallucinations in MLRMs with Latent Entropy-Aware Decoding

cornell

Cornell University

Submitted by

mozhu

Lost in Stories: Consistency Bugs in Long Story Generation by LLMs

·
6 authors

Submitted by

Ningyu

SkillNet: Create, Evaluate, and Connect AI Skills

Zhejiang University

Submitted by

taesiri

WildWorld: A Large-Scale Dataset for Dynamic World Modeling with Actions and Explicit State toward Generative ARPG

ShandaAI

Shanda AI Research Tokyo

Submitted by

Liuff23

Spatial-TTT: Streaming Visual-based Spatial Intelligence with Test-Time Training

Tencent-Hunyuan

Tencent Hunyuan

Submitted by

Byzzz0301

Towards a Medical AI Scientist

·
8 authors

4

Submitted by

ZonglinY

MOOSE-Star: Unlocking Tractable Training for Scientific Discovery by Breaking the Complexity Barrier

miromind-ai

Submitted by

djalexj

SWE-rebench V2: Language-Agnostic SWE Task Collection at Scale

nebius

Submitted by

ligongh

MosaicMem: Hybrid Spatial Memory for Controllable Video World Models

·
11 authors

5

Submitted by

NimrodShabtay1986

Look Where It Matters: High-Resolution Crops Retrieval for Efficient VLMs

ibm-research

Submitted by

yshenaw

UniG2U-Bench: Do Unified Models Advance Multimodal Understanding?

·
14 authors

Submitted by

lifuguan

Holi-Spatial: Evolving Video Streams into Holistic 3D Spatial Intelligence

·
17 authors

Submitted by

xihc-ucb

Flash-KMeans: Fast and Memory-Efficient Exact K-Means

Berkeley

Submitted by

LianghuiZhu

Mixture-of-Depths Attention

ByteDance-Seed

Submitted by

bibona

Lingshu-Cell: A generative cellular world model for transcriptome modeling toward virtual cells

Alibaba-DAMO-Academy

7

Submitted by

wjn1996

LongCat-Flash-Prover: Advancing Native Formal Reasoning via Agentic Tool-Integrated Reinforcement Learning

meituan-longcat

Submitted by

zorik

Thinking to Recall: How Reasoning Unlocks Parametric Knowledge in LLMs

google

4

Submitted by

Yuki131

LMEB: Long-horizon Memory Embedding Benchmark

KaLM-Embedding

Submitted by

yukangcao

Kinema4D: Kinematic 4D World Modeling for Spatiotemporal Embodied Simulation

mmlab-ntu

Submitted by

k-sobolev

Calibri: Enhancing Diffusion Transformers via Parameter-Efficient Calibration

Visual Generative AI group

Submitted by

syxbb

SAMA: Factorized Semantic Anchoring and Motion Alignment for Instruction-Guided Video Editing

baidu

Submitted by

taesiri

Nemotron-Cascade 2: Post-Training LLMs with Cascade RL and Multi-Domain On-Policy Distillation

nvidia

Submitted by

Borchmann

Strategic Navigation or Stochastic Search? How Agents and Humans Reason Over Document Collections

Snowflake

Submitted by

ksshumab

Qwen3-Coder-Next Technical Report

Qwen

Submitted by

Junyi42

LoGeR: Long-Context Geometric Reconstruction with Hybrid Memory

deepmind

Submitted by

DonJoey

RubricBench: Aligning Model-Generated Rubrics with Human Standards

Tencent-Hunyuan

Tencent Hunyuan

Submitted by

Jinfa

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning

·
6 authors

Submitted by

taesiri

WorldCam: Interactive Autoregressive 3D Gaming Worlds with Camera Pose as a Unifying Geometric Representation

adobe

Submitted by

taesiri

Voxtral TTS

mistralai

Submitted by

hbx

How Far Can Unsupervised RLVR Scale LLM Training?

Tsinghua University

Submitted by

Falcary

Enhancing Spatial Understanding in Image Generation via Reward Modeling

PKU

Peking University

Submitted by

taesiri

Gen-Searcher: Reinforcing Agentic Search for Image Generation

·
10 authors

Submitted by

yxlu0

FASTER: Rethinking Real-Time Flow VLAs

hkuhk

The University of Hong Kong

Submitted by

Zhouhc

Memento-Skills: Let Agents Design Agents

UniversityCollegeLondon

University College London

Submitted by

lanikoworld

3DreamBooth: High-Fidelity 3D Subject-Driven Video Generation Model

yonseiworld

Yonsei University

Submitted by

unilm

Online Experiential Learning for Language Models

MicrosoftResearch

Microsoft Research

3

Submitted by

DadaCloud01

TRUST-SQL: Tool-Integrated Multi-Turn Reinforcement Learning for Text-to-SQL over Unknown Schemas

meituan

2

Submitted by

wchengad

RealRestorer: Towards Generalizable Real-World Image Restoration with Large-Scale Image Editing Models

SUSTech

Southern university of science and technology

Submitted by

GuoxinChen

BeyondSWE: Can Current Code Agent Survive Beyond Single-Repo Bug Fixing?

AweAI-Team

Submitted by

TianHongZXY

CHIMERA: Compact Synthetic Data for Generalizable LLM Reasoning

apple

3

Submitted by

LeoYML

From Static Templates to Dynamic Runtime Graphs: A Survey of Workflow Optimization for LLM Agents

ibm

Submitted by

beanie00

Why Does Self-Distillation (Sometimes) Degrade the Reasoning Capability of LLMs?

MicrosoftResearch

Microsoft Research

Submitted by

kpzhang996

PackForcing: Short Video Training Suffices for Long Video Sampling and Long Context Inference

ShandaAI

Shanda AI Research Tokyo

Submitted by

bys0318

IndexCache: Accelerating Sparse Attention via Cross-Layer Index Reuse

zai-org

Submitted by

zli12321

MM-Zero: Self-Evolving Multi-Model Vision Language Models From Zero Data

nvidia

Submitted by

Stephen-SMJ

DARE: Aligning LLM Agents with the R Statistical Ecosystem via Distribution-Aware Retrieval

PolyUHK

The Hong Kong Polytechnic University

Submitted by

HowieHwong

Emergent Social Intelligence Risks in Generative Multi-Agent Systems

·
15 authors

Submitted by

JingweiNi

Trace2Skill: Distill Trajectory-Local Lessons into Transferable Agent Skills

·
9 authors

Submitted by

Min-Jaewon

DA-Flow: Degradation-Aware Optical Flow Estimation with Diffusion Models

kaist-ai

Submitted by

taesiri

Hyperagents

·
8 authors

Submitted by

sy1998

TerraScope: Pixel-Grounded Visual Reasoning for Earth Observation

·
7 authors

Submitted by

daria0gal

OpenAutoNLU: Open Source AutoML Library for NLU

MTSAIR

Submitted by

BradyFU

VideoDetective: Clue Hunting via both Extrinsic Query and Intrinsic Relevance for Long Video Understanding

nanjinguniv

Nanjing University

Submitted by

BradyFU

Omni-Diffusion: Unified Multimodal Understanding and Generation with Masked Discrete Diffusion

nanjinguniv

Nanjing University

Submitted by

taesiri

InternVL-U: Democratizing Unified Multimodal Models for Understanding, Reasoning, Generation and Editing

·
29 authors

Submitted by

Virgilllll

MSA: Memory Sparse Attention for Efficient End-to-End Memory Model Scaling to 100M Tokens

EverMindAI

Submitted by

Huacan-Wang

EpochX: Building the Infrastructure for an Emergent Agent Civilization

QuantaAlpha

Submitted by

taesiri

UI-Voyager: A Self-Evolving GUI Agent Learning via Failed Experience

tencent

Submitted by

onground

Repurposing Geometric Foundation Models for Multi-view Diffusion

kaist-ai

Submitted by

Septzzz

MMR-Life: Piecing Together Real-life Scenes for Multimodal Multi-image Reasoning

UCAS

Submitted by

rooty2020

SpatialBoost: Enhancing Visual Representation through Language-Guided Reasoning

kaist-ai

Submitted by

EilamSha

Alignment Makes Language Models Normative, Not Descriptive

Technion

Technion Israel institute of technology

Submitted by

Warrieryes

AgentVista: Evaluating Multimodal Agents in Ultra-Challenging Realistic Visual Scenarios

HKUST NLP Group

Submitted by

parishadbehnam

LLM2Vec-Gen: Generative Embeddings from Large Language Models

McGill-NLP

McGill NLP Group