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

Mar 4

Submitted by

Gofinge

Utonia: Toward One Encoder for All Point Clouds

Pointcept

Submitted by

yshenaw

UniG2U-Bench: Do Unified Models Advance Multimodal Understanding?

·
14 authors

Submitted by

tsbpp

Beyond Language Modeling: An Exploration of Multimodal Pretraining

2

Submitted by

GuoxinChen

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

AweAI-Team

Submitted by

DonJoey

Beyond Length Scaling: Synergizing Breadth and Depth for Generative Reward Models

Tencent-Hunyuan

Tencent Hunyuan

Submitted by

taesiri

Kling-MotionControl Technical Report

KlingTeam

Submitted by

Ningyu

How Controllable Are Large Language Models? A Unified Evaluation across Behavioral Granularities

alibaba-inc

Submitted by

taesiri

PRISM: Pushing the Frontier of Deep Think via Process Reward Model-Guided Inference

·
4 authors

Submitted by

ksshumab

Qwen3-Coder-Next Technical Report

Qwen

Submitted by

linyq

Kiwi-Edit: Versatile Video Editing via Instruction and Reference Guidance

showlab

Submitted by

GeorgeBredis

Next Embedding Prediction Makes World Models Stronger

t-tech

Submitted by

grvkamath

Humans and LLMs Diverge on Probabilistic Inferences

McGill-NLP

McGill NLP Group

Submitted by

linius

Surgical Post-Training: Cutting Errors, Keeping Knowledge

hkuhk

The University of Hong Kong

Submitted by

Fancylalala

InfoPO: Information-Driven Policy Optimization for User-Centric Agents

·
6 authors

Submitted by

rmokady

BBQ-to-Image: Numeric Bounding Box and Qolor Control in Large-Scale Text-to-Image Models

briaai

Submitted by

adrianrob

Spilled Energy in Large Language Models

·
3 authors

Submitted by

ldiex

NOVA: Sparse Control, Dense Synthesis for Pair-Free Video Editing

·
10 authors

Submitted by

hanyang-21

CFG-Ctrl: Control-Based Classifier-Free Diffusion Guidance

Tsinghua-IVG

Submitted by

akshaynambi

Learning When to Act or Refuse: Guarding Agentic Reasoning Models for Safe Multi-Step Tool Use

MicrosoftResearch

Microsoft Research

Submitted by

Piang

Track4World: Feedforward World-centric Dense 3D Tracking of All Pixels

TencentARC

ARC Lab, Tencent PCG

Submitted by

bruiiii

Chain of World: World Model Thinking in Latent Motion

·
9 authors

Submitted by

taesiri

DREAM: Where Visual Understanding Meets Text-to-Image Generation

·
11 authors

Submitted by

leonardklin

SciDER: Scientific Data-centric End-to-end Researcher

AI4Research

Submitted by

joshseth

Transformers converge to invariant algorithmic cores

·
1 authors

1

Submitted by

taesiri

APRES: An Agentic Paper Revision and Evaluation System

metaresearch

Submitted by

mchraba

Conditioned Activation Transport for T2I Safety Steering

·
6 authors

Submitted by

UVSKKR

HateMirage: An Explainable Multi-Dimensional Dataset for Decoding Faux Hate and Subtle Online Abuse

·
4 authors

Submitted by

Goekdeniz-Guelmez

DynaMoE: Dynamic Token-Level Expert Activation with Layer-Wise Adaptive Capacity for Mixture-of-Experts Neural Networks

·
1 authors

1

Submitted by

TyroneDragon

Token Reduction via Local and Global Contexts Optimization for Efficient Video Large Language Models

·
4 authors

1

Submitted by

EliotShen

GroupGPT: A Token-efficient and Privacy-preserving Agentic Framework for Multi-User Chat Assistant

·
5 authors

Submitted by

mandipgoswami

Whisper-RIR-Mega: A Paired Clean-Reverberant Speech Benchmark for ASR Robustness to Room Acoustics

·
1 authors

Submitted by

solstice0621

SGDC: Structurally-Guided Dynamic Convolution for Medical Image Segmentation

·
3 authors

Submitted by

nsschw

Towards Simulating Social Media Users with LLMs: Evaluating the Operational Validity of Conditioned Comment Prediction

cl-trier

Computational Linguistics | Trier University

Submitted by

herbwood27

Easy to Learn, Yet Hard to Forget: Towards Robust Unlearning Under Bias

·
6 authors

Submitted by

qqggez

QEDBENCH: Quantifying the Alignment Gap in Automated Evaluation of University-Level Mathematical Proofs

YaleUniversity

Yale University

Submitted by

Ysjtu

AgentConductor: Topology Evolution for Multi-Agent Competition-Level Code Generation

SJTU

Shanghai Jiao Tong University

Submitted by

taesiri

Code2Math: Can Your Code Agent Effectively Evolve Math Problems Through Exploration?

·
10 authors

Submitted by

qqggez

ParEVO: Synthesizing Code for Irregular Data: High-Performance Parallelism through Agentic Evolution

YaleUniversity

Yale University

Submitted by

dronperminov

Fast Matrix Multiplication in Small Formats: Discovering New Schemes with an Open-Source Flip Graph Framework

·
1 authors

Submitted by

jeertmans

Transform-Invariant Generative Ray Path Sampling for Efficient Radio Propagation Modeling

UCLouvain

Université Catholique de Louvain

Submitted by

ZhenhHuang

Multi-Domain Riemannian Graph Gluing for Building Graph Foundation Models

·
7 authors