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

Feb 6

Submitted by

yangzhi1

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

AIFin-Lab

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

liufanfanlff

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

·
6 authors

Submitted by

ShuoChen20

Context Forcing: Consistent Autoregressive Video Generation with Long Context

TIGER-Lab

Submitted by

melisa

Accurate Failure Prediction in Agents Does Not Imply Effective Failure Prevention

Writer

2

Submitted by

XaiverZ

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

NanyangTechnologicalUniversity

Nanyang Technological University

Submitted by

yangxue

RISE-Video: Can Video Generators Decode Implicit World Rules?

·
12 authors

Submitted by

taesiri

ProAct: Agentic Lookahead in Interactive Environments

Tencent-Hunyuan

Tencent Hunyuan

Submitted by

PeterV09

Dr. Kernel: Reinforcement Learning Done Right for Triton Kernel Generations

HKUST NLP Group

Submitted by

Vilin97

Semantic Search over 9 Million Mathematical Theorems

uw-math-ai

University of Washington Math AI Lab

Submitted by

zhouenyu04

Steering LLMs via Scalable Interactive Oversight

nex-agi

Submitted by

ppEmiliano

Privileged Information Distillation for Language Models

ServiceNow-AI

Submitted by

Steven-Shaobo

Grounding and Enhancing Informativeness and Utility in Dataset Distillation

SJTU

Shanghai Jiao Tong University

3

Submitted by

keyangx3

SocialVeil: Probing Social Intelligence of Language Agents under Communication Barriers

·
6 authors

Submitted by

ShawnYing

Retrieval-Infused Reasoning Sandbox: A Benchmark for Decoupling Retrieval and Reasoning Capabilities

ByteDance-Seed

Submitted by

chengzu

Thinking in Frames: How Visual Context and Test-Time Scaling Empower Video Reasoning

CambUni

University of Cambridge

Submitted by

xusirui

InterPrior: Scaling Generative Control for Physics-Based Human-Object Interactions

·
7 authors

2

Submitted by

Baolin

Reinforcement World Model Learning for LLM-based Agents

MicrosoftResearch

Microsoft Research

Submitted by

vincentleebang

Reinforced Attention Learning

google

Submitted by

shilinyan

SwimBird: Eliciting Switchable Reasoning Mode in Hybrid Autoregressive MLLMs

Accio-Lab

Submitted by

HughieHu

SAGE: Benchmarking and Improving Retrieval for Deep Research Agents

·
5 authors

2

Submitted by

Xiaoye08

LatentMem: Customizing Latent Memory for Multi-Agent Systems

·
9 authors

Submitted by

menik1126

V-Retrver: Evidence-Driven Agentic Reasoning for Universal Multimodal Retrieval

·
9 authors

Submitted by

changdae

Towards Reducible Uncertainty Modeling for Reliable Large Language Model Agents

uw-madison

University of Wisconsin - Madison

2

Submitted by

hba123

Multi-Task GRPO: Reliable LLM Reasoning Across Tasks

·
8 authors

Submitted by

jianchen0311

DFlash: Block Diffusion for Flash Speculative Decoding

z-lab

Submitted by

taesiri

BABE: Biology Arena BEnchmark

ByteDance-Seed

Submitted by

haoranli-ml

CoPE: Clipped RoPE as A Scalable Free Lunch for Long Context LLMs

CarnegieMellonU

Carnegie Mellon University

Submitted by

Jimlkh

Breaking the Static Graph: Context-Aware Traversal for Robust Retrieval-Augmented Generation

·
7 authors

2

Submitted by

clyu

Approximation of Log-Partition Function in Policy Mirror Descent Induces Implicit Regularization for LLM Post-Training

amazon

Submitted by

taesiri

Late-to-Early Training: LET LLMs Learn Earlier, So Faster and Better

ByteDance-Seed

Submitted by

wuruiqi0722

Infinite-World: Scaling Interactive World Models to 1000-Frame Horizons via Pose-Free Hierarchical Memory

Submitted by

taesiri

Pathwise Test-Time Correction for Autoregressive Long Video Generation

·
10 authors

Submitted by

qingyuanwu

A Unified Framework for Rethinking Policy Divergence Measures in GRPO

·
9 authors

Submitted by

RayY

Do Vision-Language Models Respect Contextual Integrity in Location Disclosure?

GeorgiaTech

Georgia Institute of Technology

Submitted by

Harahan

Light Forcing: Accelerating Autoregressive Video Diffusion via Sparse Attention

·
6 authors

2

Submitted by

AghaTizi

Failing to Explore: Language Models on Interactive Tasks

·
6 authors

Submitted by

taesiri

FastVMT: Eliminating Redundancy in Video Motion Transfer

·
12 authors

Submitted by

wlfeng

Fast-SAM3D: 3Dfy Anything in Images but Faster

·
13 authors

Submitted by

Speeeed

Learning Rate Matters: Vanilla LoRA May Suffice for LLM Fine-tuning

·
4 authors

Submitted by

Dongchao

UniAudio 2.0: A Unified Audio Language Model with Text-Aligned Factorized Audio Tokenization

·
6 authors

Submitted by

tytyt

Adaptive 1D Video Diffusion Autoencoder

·
6 authors

Submitted by

lucadellalib

Beyond Fixed Frames: Dynamic Character-Aligned Speech Tokenization

MilaQuebec

Mila – Quebec Artificial Intelligence Institute

Submitted by

Kavanavnlp

PhysicsAgentABM: Physics-Guided Generative Agent-Based Modeling

Virginia Polytechnic Institute and State University

Submitted by

Harahan

Focus-dLLM: Accelerating Long-Context Diffusion LLM Inference via Confidence-Guided Context Focusing

·
7 authors

Submitted by

abhishek9909

Assessing Domain-Level Susceptibility to Emergent Misalignment from Narrow Finetuning

·
6 authors

3