Papers7

All Beginner Intermediate Advanced

All Sources arXiv

#POMDP

TIDE: Trajectory-based Diagnostic Evaluation of Test-Time Improvement in LLM Agents

Intermediate

Hang Yan, Xinyu Che et al.Feb 2arXiv

This paper studies how AI agents get better while they are working, not just whether they finish the job.

#Test-Time Improvement#LLM agents#trajectory analysis

Spark: Strategic Policy-Aware Exploration via Dynamic Branching for Long-Horizon Agentic Learning

Intermediate

Jinyang Wu, Shuo Yang et al.Jan 28arXiv

SPARK is a new way to train AI agents that saves compute by exploring more only at the most important moments.

#SPARK#dynamic branching#strategic exploration

EvoCUA: Evolving Computer Use Agents via Learning from Scalable Synthetic Experience

Intermediate

Taofeng Xue, Chong Peng et al.Jan 22arXiv

Before this work, computer-using AIs mostly copied old examples and struggled with long step-by-step tasks on real computers.

#computer use agent#verifiable synthesis#validator

Agentic Uncertainty Quantification

Intermediate

Jiaxin Zhang, Prafulla Kumar Choubey et al.Jan 22arXiv

Long AI tasks can go wrong early and keep getting worse, like a snowball of mistakes called the Spiral of Hallucination.

#Agentic Uncertainty Quantification#Spiral of Hallucination#Dual-Process Architecture

KAGE-Bench: Fast Known-Axis Visual Generalization Evaluation for Reinforcement Learning

Intermediate

Egor Cherepanov, Daniil Zelezetsky et al.Jan 20arXiv

KAGE-Bench is a fast, carefully controlled benchmark that tests how well reinforcement learning (RL) agents trained on pixels handle specific visual changes, like new backgrounds or lighting, without changing the actual game rules.

#reinforcement learning#visual generalization#KAGE-Env

Imagine-then-Plan: Agent Learning from Adaptive Lookahead with World Models

Intermediate

Youwei Liu, Jian Wang et al.Jan 13arXiv

Agents often act like tourists without a map: they react to what they see now and miss long-term consequences.

#Imagine-then-Plan#world models#adaptive lookahead

Active Intelligence in Video Avatars via Closed-loop World Modeling

Intermediate

Xuanhua He, Tianyu Yang et al.Dec 23arXiv

The paper turns video avatars from passive puppets into active doers that can plan, act, check their own work, and fix mistakes over many steps.

#ORCA#L-IVA#Internal World Model