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 6

Submitted by

vvasilev

Kandinsky 6.0 Video: Foundation Models for Synchronized Video and Audio Generation

kandinskylab

Submitted by

Liancheng2

ALoDLM: Adaptively Looped Diffusion Language Models

amazon

Submitted by

Qing145

Memadapter: Counterfactual Adaptation Against Memory-induced Sycophancy

·
7 authors

Submitted by

youngyoon911

In-Distribution Forcing for Long Video Generation at Test Time

SeoulNatlUniv

Seoul National University

Submitted by

harryoh99

Foundations of Proactive Agents: Principles, Technical Layers, and Proactivity-Gym

minnesotanlp

1

Submitted by

luochengzz

Self-Generated Feedback Destabilizes Test-Time Training: A Causal Decomposition of Long-Horizon Adaptation

KAUST

King Abdullah University of Science and Technology

Submitted by

ofantomas

Optimizing the Optimizer: Language Models Discover Faster Molecular Relaxation

·
7 authors

Submitted by

hyojeongyun

CANOPY: Adaptive-Granularity Evidence Compression for Multimodal RAG

POSTECH

Pohang University of Science and Technology

1

Submitted by

LeonMeon

Data Unlearning via Inverse Distillation

·
6 authors

Submitted by

jeff024

ASCENT: Online Test-Time Training of Long-Horizon Agents via Self-Distillation of Verified Experience

·
2 authors

1

Submitted by

yifAI

OSWorld-Pro: Process-based Evaluation for Computer Use Agents

nvidia

1

Submitted by

free001style

Representation-Space MMD for Diffusion Language Models

yresearch

Yandex Research

Submitted by

HuskyDoge

Towards Looped Models Done Right, Part II: Rethinking at Fixed Points

IFM

Institute of Foundation Models

Submitted by

mjuicem

OmniReasoning: Pushing the Limits of Audio-Visual Joint Reasoning

Qwen

Submitted by

youganglyu

SearchJev: A Fast and Calibrated System-1 Model for Search Agents

·
9 authors

Submitted by

Leejh123e

RobotUse: Allocating Computation, Context, and Decisions

kaist-ai

Submitted by

Sarim-Hash

Certification of Real Images through Calibrated Content Authentication

MBZUAI

Mohamed Bin Zayed University of Artificial Intelligence

Submitted by

Sarim-Hash

Noise Out, Bias In: Targeted Bias Injection in Diffusion Language Models via Closed-Loop Activation Steering

MBZUAI

Mohamed Bin Zayed University of Artificial Intelligence

Submitted by

SeonghoonYu

When to Switch: Reliable Action-Chunk Extension for Vision-Language-Action Models

POSTECH

Pohang University of Science and Technology

Submitted by

TanmaySah

UndoBench: Separating Task Competence from Recovery Capability in Tool-Using AI Agents

·
4 authors

Submitted by

mack-williams

RealtimeWAM: One-Step Asynchronous World Action Models

NanyangTechnologicalUniversity

Nanyang Technological University

Submitted by

sophicle

Base Models Can Reason By Taking a Cue From Training Data

MIT

Massachusetts Institute of Technology

Submitted by

Ironieser

Rethinking Long-Video Efficiency: A Joint Allocation Perspective on Frames, Pixels, and Front-End Latency

University of Central Florida

Submitted by

MingyuLiu

PerturBot: Breaking Shortcut Priors in Vision-Language-Action Models with Perturbative Training

·
7 authors

Submitted by

alexeychernysh

QuantCode Model: Specializing Language Models for Executable Algorithmic Trading Code

·
3 authors

1

Submitted by

taesiri

Dynamic Harness Search: Building Multi-Agent Systems Per-Query via Prediction

google

Submitted by

gpolitis

What Gradients Add to Text Leakage in Split Language Models, Counted per Token and per Document

Setloop

1

Submitted by

pedrocurvo

LiFT: Loop Flow Transformers

VISLab-Amsterdam

Video & Image Sense Lab

Submitted by

youali

What Matters for Latent Reasoning with Flow Matching

SamsungResearch

Samsung Research

Submitted by

Fulucheng

From Knowledge Access to Source Learning: Developing Source-Specific Competence

GeorgiaTech

Georgia Institute of Technology

Submitted by

PeterLauLukCh

Training Numerical Intelligence via Auto-Diagnosis and Skill Discovery

·
2 authors

1

Submitted by

FrancisRing

Prism: Dynamic Sparse Attention for Native 2K Joint Video-Audio Generation Model Training

Tencent-Hunyuan

Tencent Hunyuan

Submitted by

Lumos-Jiateng

LMBuild: Evaluating LLM Agents for Generating Buildable and Functional Structures

UIUC-CS

University of Illinois at Urbana-Champaign

Submitted by

Debargha

How to Loop MoE: Flatten the Experts, Untie the Attention

·
9 authors

Submitted by

SteveZeyuZhang

Code2Games: Enabling Coding Agents for Gaming World Generation

PekingUniversity

Peking University

Submitted by

shuoxing

The Missing Primitive: Diagnosing and Repairing Mathematical Reasoning in Large Language Models

TexasAMUniversity

Texas A&M University

Submitted by

yuxuanli

PaLoRA: Paced Low-Rank Adaptation for Continual Learning

·
3 authors

Submitted by

yrshi

PluginRSI: Recursive Improvement of Agent Harnesses with Reusable Plugins

·
8 authors

Submitted by

JeonghyeokDo

GeoSET: Generalist Foundation Model for SAR-to-EO Image Translation

KAIST-VIClab

KAIST VIClab (Prof. Munchurl Kim)

Submitted by

chenfeng1271

HLA-WM: Hybrid Linear Attention for Long-Horizon Video World Models

Monash University

Submitted by

lennartcb

Learning to Learn a Language

·
2 authors

Submitted by

chuzy

OpenRUA: Robot-Use Agents Are Zero-Shot Visuomotor Policies

·
7 authors

Submitted by

mohammad2012191

Periscope: Extending Frozen Language Models Beyond Their Context Window

KAUST

King Abdullah University of Science and Technology

Submitted by

Fulucheng

TextReg: Mitigating Prompt Distributional Overfitting via Regularized Text-Space Optimization

GeorgiaTech

Georgia Institute of Technology

Submitted by

Eric1231

OmniConfess: Eliciting Token Confessions to Mitigate Omni-Modal Hallucination

·
7 authors

Submitted by

Saidgurbuz

DeskForge: Dense Supervision from Desktop Environments for Computer-Use Agents

docling-project

1

Submitted by

Kivy

MM-ABC: Towards Generalist Mobile Manipulation via Seeing, Coordinating and Imagining

·
9 authors

Submitted by

JeonghyeokDo

GeoCR: Learning a Generalist Cloud Removal Prior from Heterogeneous Observations

KAIST-VIClab

KAIST VIClab (Prof. Munchurl Kim)

Submitted by

OniReimu

SoK: Semantic Decision Engines in Network Control Loops

·
6 authors

Submitted by

OniReimu

Intent Interpretation at RIC Timescales: Jev Decision Models versus Large Language Models in 6G Open RAN

·
5 authors

Submitted by

gurayozgur

Learning Steadily: Accumulating Relative Point Margin Scores for Face Image Quality Assessment

FraunhoferIGD

1

Submitted by

myshi

CurveCodec 2: Skeleton-agnostic animation compression with a learned entropy model

HKU-CGVU

Submitted by

MrBean2024

Arm-wise Compositional Generalization in Dual-Arm Vision-Language-Action Models

·
13 authors

Submitted by

Maximilian2001

Adapting prior-data fitted networks for tabular anomaly detection

·
2 authors

Submitted by

mapooon

Empirical Variational Autoencoder

·
1 authors

Submitted by

Agcs12

MEA: A Reward-Driven Multi-Agent System for Faithful Model Explanations

UVABiology

University Of Virginia

Submitted by

asiimo

DEPICT: Scoring Text-to-Image Alignment by Answer Agreement

swordhealth

Submitted by

xusirui

InterMimicGen: Scaling Humanoid Loco-Manipulation through Self-Evolving Motion Imitation

·
11 authors