Research

@inproceedings{li2026combee,
  title = {Combee: Scaling Prompt Learning for Self-Improving Language Model Agents},
  author = {Li, Hanchen and He, Runyuan and Zhang, Qizheng and Ji, Changxiu and Mang, Qiuyang and Chen, Xiaokun and Agrawal, Lakshya A. and Liao, Wei-Liang and Yang, Eric and Cheung, Alvin and Zou, James and Olukotun, Kunle and Stoica, Ion and Gonzalez, Joseph E.},
  booktitle = {COLM 2026 (Conference on Language Modeling)},
  year = {2026},
  doi = {10.48550/arXiv.2604.04247},
}

@inproceedings{ji2026sentry,
  title = {Sentry: A Self-evolving Failure Manager for Agentic LLM Systems},
  author = {Ji, Changxiu and Lu, Amy},
  booktitle = {COLM 2026 Workshop on Agent Behavior (WAB); COLM 2026 The 2nd Workshop on Lifelong Agents: Learning, Aligning, and Evolving (LLA)},
  year = {2026},
  note = {* Equal contribution},
}

@inproceedings{zhang2026learningwhattolearn,
  title = {Learning What to Learn: Curriculum Curation for Test-Time Agent Learning},
  author = {Zhang, Qizheng and Ruan, Sherry and Upasani, Shubhangi and Hong, Fenglu and Ji, Changxiu and Hu, Changran and Li, Bo and Li, Hanchen and Olukotun, Kunle},
  booktitle = {ICLR 2026 Workshop on AI with Recursive Self-Improvement},
  year = {2026},
  note = {Poster},
}

@article{lu2026art,
  title = {Association Restoration Test: Revealing Restorable Shortcuts after Unlearning},
  author = {Lu, Amy and Ji, Changxiu},
  journal = {arXiv preprint arXiv:2607.05726},
  year = {2026},
  note = {* Equal contribution},
  doi = {10.48550/arXiv.2607.05726},
}