Project Lifelong Agent

Lifelong Agents: Learning, Aligning, Evolving

"Project Lifelong Agent" is an ongoing research effort driven by the dream of superintelligent agents that continuously learn, adapt, and evolve. This page collects latest papers from researchers diligently working to advance that mission.

Learning Aligning Evolving
Tool-R0: Self-Evolving LLM Agents for Tool-Learning from Zero Data
πŸ“„ PDF    🌐 Website    𝕏 Post
Under Review @ COLM
Evolving Feb 2026
Current Agents Fail to Leverage World Model as Tool for Foresight
πŸ“„ PDF    🌐 Website
ACL 2026
Aligning Jan 2026
TT-SI: Self-Improving LLM Agents with Test-Time Training
πŸ“„ PDF    🌐 Website    𝕏 Post
ACL 2026
Evolving Oct 2025
ToolRL: Reward is All Tool Learning Needs
πŸ“„ PDF    🌐 Website
NeurIPS 2025
Learning Apr 2025
SMART: Self-Aware Agent for Tool Overuse Mitigation
πŸ“„ PDF    🌐 Website
ACL 2025
Aligning Feb 2025
Can a Single Model Master Both Multi-turn Conversations and Tool Use? CoALM: A Unified Conversational Agentic Language Model
πŸ“„ PDF    🌐 Website    𝕏 Post
ACL 2025
Learning Feb 2025
UserRL: Training Interactive User-Centric Agent via Reinforcement Learning
πŸ“„ PDF    🌐 Website
arXiv 2025
Aligning Jan 2026