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

February

Submitted by

taesiri

A Very Big Video Reasoning Suite

Video-Reason

Submitted by

YoungXuan

OPUS: Towards Efficient and Principled Data Selection in Large Language Model Pre-training in Every Iteration

Qwen

3

Submitted by

2pd

Green-VLA: Staged Vision-Language-Action Model for Generalist Robots

SberRoboticsCenter

Sber Robotics Center

Submitted by

Yikunb

Weak-Driven Learning: How Weak Agents make Strong Agents Stronger

·
11 authors

Submitted by

taesiri

ERNIE 5.0 Technical Report

·
438 authors

Submitted by

Yikunb

Does Your Reasoning Model Implicitly Know When to Stop Thinking?

ByteDance

9

Submitted by

taesiri

Kimi K2.5: Visual Agentic Intelligence

moonshotai

Submitted by

Zhongzhi1228

Less is Enough: Synthesizing Diverse Data in Feature Space of LLMs

·
5 authors

Submitted by

floyed

VESPO: Variational Sequence-Level Soft Policy Optimization for Stable Off-Policy LLM Training

rednote-hilab

Submitted by

taesiri

PaperBanana: Automating Academic Illustration for AI Scientists

google

Submitted by

March07

TermiGen: High-Fidelity Environment and Robust Trajectory Synthesis for Terminal Agents

ucsbnlp

UC Santa Barbara NLP Group

Submitted by

taesiri

Code2World: A GUI World Model via Renderable Code Generation

GD-ML

Submitted by

chengtan9907

The Trinity of Consistency as a Defining Principle for General World Models

opendatalab

Submitted by

xunyoyo

The Devil Behind Moltbook: Anthropic Safety is Always Vanishing in Self-Evolving AI Societies

·
13 authors

9

Submitted by

taesiri

Step 3.5 Flash: Open Frontier-Level Intelligence with 11B Active Parameters

stepfun-ai

Submitted by

yangzhi1

QuantaAlpha: An Evolutionary Framework for LLM-Driven Alpha Mining

QuantaAlpha

Submitted by

berlin8587

SQuTR: A Robustness Benchmark for Spoken Query to Text Retrieval under Acoustic Noise

·
7 authors

Submitted by

zhangxgu

UI-Venus-1.5 Technical Report

inclusionAI

Submitted by

tianyilt

MOVA: Towards Scalable and Synchronized Video-Audio Generation

OpenMOSS-Team

Submitted by

YuZeng260

Vision-DeepResearch: Incentivizing DeepResearch Capability in Multimodal Large Language Models

·
15 authors

Submitted by

xiaochonglinghu

FASA: Frequency-aware Sparse Attention

alibaba-inc

Submitted by

jcy

From Blind Spots to Gains: Diagnostic-Driven Iterative Training for Large Multimodal Models

·
4 authors

Submitted by

Yu2020

Modality Gap-Driven Subspace Alignment Training Paradigm For Multimodal Large Language Models

·
15 authors

Submitted by

issart12345

VidVec: Unlocking Video MLLM Embeddings for Video-Text Retrieval

·
3 authors

Submitted by

YuZeng260

Vision-DeepResearch Benchmark: Rethinking Visual and Textual Search for Multimodal Large Language Models

·
16 authors

3

Submitted by

taesiri

GLM-5: from Vibe Coding to Agentic Engineering

·
186 authors

Submitted by

Ximing

Golden Goose: A Simple Trick to Synthesize Unlimited RLVR Tasks from Unverifiable Internet Text

nvidia

5

Submitted by

xiaochonglinghu

MobilityBench: A Benchmark for Evaluating Route-Planning Agents in Real-World Mobility Scenarios

alibaba-inc

Submitted by

renjiepi

On Data Engineering for Scaling LLM Terminal Capabilities

nvidia

Submitted by

zelaix

WideSeek-R1: Exploring Width Scaling for Broad Information Seeking via Multi-Agent Reinforcement Learning

RLinf

Submitted by

YerbaPage

CodeOCR: On the Effectiveness of Vision Language Models in Code Understanding

SJTU

Shanghai Jiao Tong University

Submitted by

xx18

Composition-RL: Compose Your Verifiable Prompts for Reinforcement Learning of Large Language Models

Tencent-Hunyuan

Tencent Hunyuan

Submitted by

Aurorra1123

AOrchestra: Automating Sub-Agent Creation for Agentic Orchestration

·
11 authors

Submitted by

Luo2003

Closing the Loop: Universal Repository Representation with RPG-Encoder

·
13 authors

Submitted by

johanneskirmayr

CAR-bench: Evaluating the Consistency and Limit-Awareness of LLM Agents under Real-World Uncertainty

BMW-LLM-Research-Group

BMW LLM Research Group

Submitted by

myownskyW7

DeepGen 1.0: A Lightweight Unified Multimodal Model for Advancing Image Generation and Editing

sii-research

Shanghai Innovation Institute

Submitted by

bhavul

AIRS-Bench: a Suite of Tasks for Frontier AI Research Science Agents

Submitted by

taesiri

UniReason 1.0: A Unified Reasoning Framework for World Knowledge Aligned Image Generation and Editing

·
11 authors

Submitted by

shrango

Training Data Efficiency in Multimodal Process Reward Models

·
7 authors

Submitted by

yangzhi1

Chain of Mindset: Reasoning with Adaptive Cognitive Modes

QuantaAlpha

Submitted by

taesiri

InternAgent-1.5: A Unified Agentic Framework for Long-Horizon Autonomous Scientific Discovery

InternScience

Submitted by

Myashka

F-GRPO: Don't Let Your Policy Learn the Obvious and Forget the Rare

t-tech

3

Submitted by

lxucs

No Global Plan in Chain-of-Thought: Uncover the Latent Planning Horizon of LLMs

tencent

Submitted by

MaksimSTW

Experiential Reinforcement Learning

microsoft

Submitted by

utdawn

LLaDA2.1: Speeding Up Text Diffusion via Token Editing

inclusionAI

Submitted by

richardxp888

SkillRL: Evolving Agents via Recursive Skill-Augmented Reinforcement Learning

UNC-ChapelHill

University of North Carolina at Chapel Hill

Submitted by

Jiafei1224

Recurrent-Depth VLA: Implicit Test-Time Compute Scaling of Vision-Language-Action Models via Latent Iterative Reasoning

allenai

Submitted by

yangzhi1

Spider-Sense: Intrinsic Risk Sensing for Efficient Agent Defense with Hierarchical Adaptive Screening

AIFin-Lab

Submitted by

KaiWu123

MedXIAOHE: A Comprehensive Recipe for Building Medical MLLMs

ByteDance

Submitted by

taesiri

MARS: Modular Agent with Reflective Search for Automated AI Research

google

Submitted by

Phoebux

3D-Aware Implicit Motion Control for View-Adaptive Human Video Generation

KlingTeam

Submitted by

fairyang

Baichuan-M3: Modeling Clinical Inquiry for Reliable Medical Decision-Making

baichuan-inc

Baichuan Intelligent Technology

Submitted by

Kushnareva

AudioSAE: Towards Understanding of Audio-Processing Models with Sparse AutoEncoders

huawei-noah

HUAWEI Noah's Ark Lab

Submitted by

Keven16

Learning beyond Teacher: Generalized On-Policy Distillation with Reward Extrapolation

Tencent-Hunyuan

Tencent Hunyuan

Submitted by

Jeff-Wang

GigaBrain-0.5M*: a VLA That Learns From World Model-Based Reinforcement Learning

open-gigaai

Submitted by

xufangzhi

OdysseyArena: Benchmarking Large Language Models For Long-Horizon, Active and Inductive Interactions

·
19 authors

Submitted by

XaiverZ

MemSkill: Learning and Evolving Memory Skills for Self-Evolving Agents

NanyangTechnologicalUniversity

Nanyang Technological University

Submitted by

chenmouxiang

SWE-Universe: Scale Real-World Verifiable Environments to Millions

Qwen

3

Submitted by

BlackSamorez

Quartet II: Accurate LLM Pre-Training in NVFP4 by Improved Unbiased Gradient Estimation

ISTA-DASLab

IST Austria Distributed Algorithms and Systems Lab

Submitted by

Emperorizzis

ASTRA: Automated Synthesis of agentic Trajectories and Reinforcement Arenas

·
15 authors

Submitted by

taesiri

Unified Latents (UL): How to train your latents

google

Submitted by

ChenlongDeng

DeepImageSearch: Benchmarking Multimodal Agents for Context-Aware Image Retrieval in Visual Histories

RUC-NLPIR

NLPIR Lab @ RUC

Submitted by

Elliott

P1-VL: Bridging Visual Perception and Scientific Reasoning in Physics Olympiads

Submitted by

jt-zhang

SLA2: Sparse-Linear Attention with Learnable Routing and QAT

Berkeley

5

Submitted by

MindscapeRAG

Query-focused and Memory-aware Reranker for Long Context Processing

tencent

4

Submitted by

taesiri

SkyReels-V4: Multi-modal Video-Audio Generation, Inpainting and Editing model

Skywork

Submitted by

MilaWang

SkillOrchestra: Learning to Route Agents via Skill Transfer

uw-madison

University of Wisconsin - Madison

Submitted by

yhzheng1031

HyTRec: A Hybrid Temporal-Aware Attention Architecture for Long Behavior Sequential Recommendation

WuhanUniversity

Wuhan Univeristy

Submitted by

harryjo97

MolHIT: Advancing Molecular-Graph Generation with Hierarchical Discrete Diffusion Models

kaist-ai

3

Submitted by

therem

Sanity Checks for Sparse Autoencoders: Do SAEs Beat Random Baselines?

·
6 authors

3

Submitted by

mickyhimself

GENIUS: Generative Fluid Intelligence Evaluation Suite

·
11 authors

Submitted by

zoeyuchao

RLinf-USER: A Unified and Extensible System for Real-World Online Policy Learning in Embodied AI

RLinf

Submitted by

lyhisme

Towards Universal Video MLLMs with Attribute-Structured and Quality-Verified Instructions

AudioVisual-Caption

Submitted by

lixiaoxi45

OmniGAIA: Towards Native Omni-Modal AI Agents

·
11 authors

Submitted by

taesiri

BitDance: Scaling Autoregressive Generative Models with Binary Tokens

ByteDance

Submitted by

txiong23

PhyCritic: Multimodal Critic Models for Physical AI

nvidia

2

Submitted by

yiyexy

OneVision-Encoder: Codec-Aligned Sparsity as a Foundational Principle for Multimodal Intelligence

lmms-lab

Submitted by

mhjiang0408

daVinci-Agency: Unlocking Long-Horizon Agency Data-Efficiently

GAIR

Submitted by

Evanwu50020

AutoWebWorld: Synthesizing Infinite Verifiable Web Environments via Finite State Machines

Hong Kong University of Science and Technology(GuangZhou)

2

Submitted by

ChilleD

Agent World Model: Infinity Synthetic Environments for Agentic Reinforcement Learning

Snowflake

Submitted by

liufanfanlff

Length-Unbiased Sequence Policy Optimization: Revealing and Controlling Response Length Variation in RLVR

·
6 authors

Submitted by

IgnoraZ

FS-Researcher: Test-Time Scaling for Long-Horizon Research Tasks with File-System-Based Agents

muset-ai

Submitted by

xhyandwyy

Mobile-Agent-v3.5: Multi-platform Fundamental GUI Agents

AlibabaTongyiLab

3

Submitted by

fdugyt

MOSS-Audio-Tokenizer: Scaling Audio Tokenizers for Future Audio Foundation Models

OpenMOSS-Team

Submitted by

zbhpku

Research on World Models Is Not Merely Injecting World Knowledge into Specific Tasks

KlingTeam

4

Submitted by

Jungang

OmniSIFT: Modality-Asymmetric Token Compression for Efficient Omni-modal Large Language Models

·
15 authors

3

Submitted by

LongMountain

HySparse: A Hybrid Sparse Attention Architecture with Oracle Token Selection and KV Cache Sharing

XiaomiMiMo

3

Submitted by

Huan0825

Towards Agentic Intelligence for Materials Science

·
21 authors

Submitted by

victor

DualPath: Breaking the Storage Bandwidth Bottleneck in Agentic LLM Inference

deepseek-ai

4

Submitted by

zhaocheng

NarraScore: Bridging Visual Narrative and Musical Dynamics via Hierarchical Affective Control

ByteDance

2

Submitted by

Canyu

Alleviating Sparse Rewards by Modeling Step-Wise and Long-Term Sampling Effects in Flow-Based GRPO

zju

Zhejiang University

Submitted by

zehongma

PixelGen: Pixel Diffusion Beats Latent Diffusion with Perceptual Loss

PekingUniversity

Peking University

Submitted by

AngLv

SPARKLING: Balancing Signal Preservation and Symmetry Breaking for Width-Progressive Learning

ByteDance-Seed

Submitted by

TianheWu

Diversity-Preserved Distribution Matching Distillation for Fast Visual Synthesis

CityU-HongKong

City University of Hong Kong

Submitted by

jt-zhang

SpargeAttention2: Trainable Sparse Attention via Hybrid Top-k+Top-p Masking and Distillation Fine-Tuning

Tsinghua University

6

Submitted by

jianchen0311

DFlash: Block Diffusion for Flash Speculative Decoding

z-lab

Submitted by

Sicong

RynnBrain: Open Embodied Foundation Models

Alibaba-DAMO-Academy

Submitted by

Michael4933

Imagination Helps Visual Reasoning, But Not Yet in Latent Space

Tsinghua University

Submitted by

Zijie-chen

Improving Data and Reward Design for Scientific Reasoning in Large Language Models

microsoft

2

Submitted by

Ricardoly

Rethinking Generative Recommender Tokenizer: Recsys-Native Encoding and Semantic Quantization Beyond LLMs

Personalized&Inclusive Intelligence Research

Submitted by

shizhuo2

Good SFT Optimizes for SFT, Better SFT Prepares for Reinforcement Learning

UIUC-CS

University of Illinois at Urbana-Champaign

3

Submitted by

Ayanami0730

WildGraphBench: Benchmarking GraphRAG with Wild-Source Corpora

muset-ai

Submitted by

jshin49

Generative Visual Code Mobile World Models

trillionlabs

Submitted by

mickyhimself

GEBench: Benchmarking Image Generation Models as GUI Environments

stepfun-ai

Submitted by

tellarin

EgoActor: Grounding Task Planning into Spatial-aware Egocentric Actions for Humanoid Robots via Visual-Language Models

BAAI

Beijing Academy of Artificial Intelligence

3