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 27

Submitted by

HU22333

Trajectory2Gym: Decision-Grounded Compositional Interaction Synthesis for Tool-Use Agents

ant-intl

ant-international

Submitted by

LCZZZZ

StreamVibe: Proactive Video Agents for Evolving User Intent

google

Submitted by

HengjieZhu

Trial by Trace: Benchmarking AI Agent Harnesses for Reliability and Failure Attribution

TraceLite-AI

Submitted by

LIQIIIII

DynaFlow: Learning Free-Form Multi-Agent Orchestration

NationalUniversityofSingapore

National University of Singapore

Submitted by

Qinsi1

InteractiveWorld: A Verifiable Human-Centric Environment for Building Better Interactive MLLM Assistants

meta

Submitted by

peng25

WorldCast: World-Consistent Character Video Generation with Large-Angle Camera Control

·
10 authors

Submitted by

WJM-George

Spatial Audio Generation and Editing with Executable Plans and Self-Distilling Transfusion

NoizAI

Submitted by

shenjunhao

Work Agents: Systems, Capabilities, and Optimization

·
16 authors

Submitted by

hyun1905

Eat Your Precision: Precision Objective Mitigates Reward Hacking in Rubric-based RL

CarnegieMellonU

Carnegie Mellon University

Submitted by

DongkiKim

Principia: Test-Time Hypotheses to Principle Construction for Scientific Discovery

kaist-ai

1

Submitted by

Harold328

World-in-Context: Interactive Video World Modeling with In-Context Conditioning

AlibabaTongyiLab

Submitted by

pujiayue

Trustworthy Recursive Self-Improvement in the Era of Large Models: Foundation, Risks, Evaluation and Open Challenges

·
12 authors

Submitted by

GoooDte

Δτ-Bench: Pushing Agents Toward a Moving Capability

UniversityCollegeLondon

University College London

Submitted by

JustinZekai

Qwen-Image-PE: Learning Visual Design for Text-to-Image Prompt Enhancement

Qwen

Submitted by

HaoyuZhao

Continual Learning Exam: Benchmarking Long-Horizon RSI and Experiential Learning in LLM Agents

UniversityCollegeLondon

University College London

Submitted by

penny123

S3AC: Stabilized, Selective, and Scalable Actor-Critic for LLM Reasoning

Shanghai-AI-Laboratory

Shanghai AI Laboratory

Submitted by

HongchengGao

EmbodiedLab: Benchmarking Embodied Agents That Act to Know

HexaFuture

Submitted by

huangsiteng

RynnVLA-Latent: Scaling Robotic Manipulation Foundation Models with 3D-Grounded Latent Actions

Alibaba-DAMO-Academy

Submitted by

Daniel21Ding

ProAS: A Progress-Aware Advantage Shaping Framework for Multi-Turn Agentic Reinforcement Learning

LINs-lab

Submitted by

Rizling

Build, Don't Play: Self-Improving LLM Research Agents Develop a Hearthstone Battlegrounds AI That Rivals Top Players

·
2 authors

Submitted by

Skywalker0410

Rethinking Video Storage for AI Training: Breaking Cross-Frame Dependencies for Efficient Random Access

PKU

Peking University

Submitted by

kkakkkka

HiP3R: Hierarchical Pose Graphs for Long-Horizon 3D Reconstruction

nanyang-technological-university-singapore

Nanyang Technological University Singapore

Submitted by

QiYao-Wang

Novel or Not? PatEx-Gym: Agentic Prior-Art Search for Novelty Assessment in Patent Examination

IPIntelligence

IP Intelligence

Submitted by

zcahjl3

Extreme Compression of Per-Layer Embeddings via Cross-Layer Latent Vector Quantization

·
8 authors

Submitted by

ethanning

AgentProofBench: Benchmarking Reusable Verifiers for Web Agents

CarnegieMellonU

Carnegie Mellon University

Submitted by

EthanLI24

miniTE: Diagnosing and Bridging the Execution Gap in Native NVFP4 Reinforcement Learning

InfiX-ai

1

Submitted by

julia-bel

Mycelium: A Meta-Study of Hallucination Detection Across Provenance, Typology, and Granularities

·
2 authors

Submitted by

taowangcheng

Skill^{infty}: Recursive Self-Improvement for Agent Skill Optimization

NationalUniversityofSingapore

National University of Singapore

1

Submitted by

Jayfeather1024

Steering Black-Box Large Language Model Agents with Pitfall Anticipation World Model

·
7 authors

Submitted by

BinceQu

RoboHarness: A Simple Harness Outperforms VLA and World Action Models

zju

Zhejiang University

1

Submitted by

tianzhechu

Nested Sparsity for Scalable Pretraining and Continual Learning

hkuhk

The University of Hong Kong

Submitted by

gtc-hf

Humans in Space: Can MLLMs Construct Human-Centric Spatial Understanding via Active Exocentric Exploration?

·
5 authors

Submitted by

Suchenl

More Memory, Better Videos? A Controlled Study of Visual Memory in Long-Video Generation

·
9 authors

Submitted by

bdqnghi

Thinking-in-Bets at Test Time Reduces Sycophantic Revisions in LLM Agents

google

Submitted by

lonelyinnovator

Assessing the CAD Capabilities of General-Purpose AI Models: A Case Study of GPT-6 Astra

Zhejiang University

1

Submitted by

jainofficials

Beyond Benchmarks: Can Small Open-Weight Models Get Real Coding Work Done?

·
1 authors

1