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

Apr 28

Submitted by

fredreick

From Skills to Talent: Organising Heterogeneous Agents as a Real-World Company

·
8 authors

Submitted by

lhmd

World-R1: Reinforcing 3D Constraints for Text-to-Video Generation

MicrosoftResearch

Microsoft Research

Submitted by

eamonn-zh

ReVSI: Rebuilding Visual Spatial Intelligence Evaluation for Accurate Assessment of VLM 3D Reasoning

SFU

Simon Fraser University

Submitted by

LIQIIIII

Vision-Language-Action Safety: Threats, Challenges, Evaluations, and Mechanisms

·
9 authors

Submitted by

FanqingM

ClawMark: A Living-World Benchmark for Multi-Turn, Multi-Day, Multimodal Coworker Agents

·
47 authors

Submitted by

taesiri

Tuna-2: Pixel Embeddings Beat Vision Encoders for Multimodal Understanding and Generation

·
15 authors

Submitted by

Ningyu

Rewarding the Scientific Process: Process-Level Reward Modeling for Agentic Data Analysis

antgroup

Submitted by

taesiri

SketchVLM: Vision language models can annotate images to explain thoughts and guide users

·
6 authors

Submitted by

itaynakash

Efficient Agent Evaluation via Diversity-Guided User Simulation

ibm-research

Submitted by

HanNight

Stabilizing Efficient Reasoning with Step-Level Advantage Selection

amd

Submitted by

HikariDawn

OmniShotCut: Holistic Relational Shot Boundary Detection with Shot-Query Transformer

·
5 authors

Submitted by

dwenlong

For-Value: Efficient Forward-Only Data Valuation for finetuning LLMs and VLMs

·
8 authors

Submitted by

BradNLP

Taming Actor-Observer Asymmetry in Agents via Dialectical Alignment

NationalUniversityofSingapore

National University of Singapore

Submitted by

EliverQ

Improving Vision-language Models with Perception-centric Process Reward Models

RUC-AIBOX

Submitted by

taesiri

ProEval: Proactive Failure Discovery and Efficient Performance Estimation for Generative AI Evaluation

deepmind

Submitted by

vicgalle

Discovering Agentic Safety Specifications from 1-Bit Danger Signals

·
1 authors

Submitted by

gurayozgur

ATTN-FIQA: Interpretable Attention-based Face Image Quality Assessment with Vision Transformers

FraunhoferIGD

Submitted by

gurayozgur

EX-FIQA: Leveraging Intermediate Early eXit Representations from Vision Transformers for Face Image Quality Assessment

FraunhoferIGD

Submitted by

taesiri

Zero-to-CAD: Agentic Synthesis of Interpretable CAD Programs at Million-Scale Without Real Data

·
4 authors

Submitted by

taesiri

Stochastic KV Routing: Enabling Adaptive Depth-Wise Cache Sharing

apple