Yukun Huang

@article{Huang2025EvalDeepResearch,
  title={When Human Gold Breaks: Co-Evolving Agents and Benchmarks for Evaluating Factuality in Deep Research Reports},
  author={Huang, Yukun and Ribeiro, Leonardo and Hardalov, Momchil and Dhingra, Bhuwan and Saligrama, Venkatesh and Dreyer, Markus},
  journal={Under Review},
  year={2025},
  url={https://kkkevinkkkkk.github.io},
  note={Summary: We propose Evolve Benchmarking, a new paradigm for Deep Research Report (DRR) factuality evaluation where benchmarks and agents co-evolve through an Audit-then-Score protocol, replacing unreliable static human "gold". It yields two artifacts: DeepFact-Bench, an auditable evolving benchmark, and DeepFact-Eval, an expert-level agentic verifier for DRR claims.}
}

@misc{huang2025citepretrain,
  title={Cite Pretrain: Retrieval-Free Knowledge Attribution for Large Language Models},
  author={Huang, Yukun and Chen, Sanxing and Pei, Jian and Zaheer, Manzil and Dhingra, Bhuwan},
  year={2025},
  eprint={2506.17585},
  archivePrefix={arXiv},
  primaryClass={cs.AI},
  doi={10.48550/arXiv.2506.17585},
  url={https://arxiv.org/abs/2506.17585}
}

@article{Chen2025WhenGW,
  title={When Greedy Wins: Emergent Exploitation Bias in Meta-Bandit LLM Training},
  author={Chen, Sanxing and Chen, Xiaoyin and Huang, Yukun and Xie, Roy and Dhingra, Bhuwan},
  journal={Under Review},
  year={2025},
  volume={abs/2509.24923},
  url={https://arxiv.org/abs/2509.24923},
  note={Summary: We compares Reinforcement Learning (RL) and Supervised Fine-Tuning (SFT) for teaching LLMs to explore in multi-armed bandit tasks. We introduce strategic and algorithmic reward designs, showing that RL generalizes better across environments but tends to learn more exploitative, short-sighted strategies than SFT.}
}

@inproceedings{huang2025situatedfaithfulness,
  title={To Trust or Not to Trust? Enhancing Large Language Models' Situated Faithfulness to External Contexts},
  author={Huang, Yukun and Chen, Sanxing and Cai, Hongyi and Dhingra, Bhuwan},
  booktitle={International Conference on Learning Representations (ICLR)},
  year={2025},
  url={https://openreview.net/forum?id=K2jOacHUlO},
  note={Spotlight}
}

@inproceedings{chen-etal-2025-real,
  title={Real-time Factuality Assessment from Adversarial Feedback},
  author={Chen, Sanxing and Huang, Yukun and Dhingra, Bhuwan},
  booktitle={Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year={2025},
  pages={1610–1630},
  url={https://aclanthology.org/2025.acl-long.81/},
  doi={10.18653/v1/2025.acl-long.81}
}

@inproceedings{holsman-etal-2025-fuzzy,
  title={Fuzzy Speculative Decoding for a Tunable Accuracy-Runtime Tradeoff},
  author={Holsman, Maximilian and Huang, Yukun and Dhingra, Bhuwan},
  booktitle={Findings of the Association for Computational Linguistics: ACL 2025},
  year={2025},
  pages={26257–26273},
  url={https://aclanthology.org/2025.findings-acl.1346.pdf},
  doi={10.18653/v1/2025.findings-acl.1346}
}

@inproceedings{huang-etal-2024-calibrating,
  title={Calibrating Long-form Generations From Large Language Models},
  author={Huang, Yukun and Liu, Yixin and Thirukovalluru, Raghuveer and Cohan, Arman and Dhingra, Bhuwan},
  booktitle={Findings of the Association for Computational Linguistics: EMNLP 2024},
  year={2024},
  pages={13441–13460},
  url={https://aclanthology.org/2024.findings-emnlp.785/},
  doi={10.18653/v1/2024.findings-emnlp.785}
}

@inproceedings{thirukovalluru-etal-2024-atomic,
  title={Atomic Self-Consistency for Better Long Form Generations},
  author={Thirukovalluru, Raghuveer and Huang, Yukun and Dhingra, Bhuwan},
  booktitle={Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing},
  year={2024},
  pages={12681–12694},
  url={https://aclanthology.org/2024.emnlp-main.706.pdf},
  doi={10.18653/v1/2024.emnlp-main.706}
}

@inproceedings{toles-etal-2025-learning,
  title={Learning and Evaluating Factual Clarification Question Generation Without Examples},
  author={Toles, Matthew and Huang, Yukun and Yu, Zhou},
  booktitle={Proceedings of the Fourth Workshop on Generation, Evaluation and Metrics (GEM²)},
  year={2025},
  pages={200–211},
  url={https://aclanthology.org/2025.gem-1.15.pdf}
}

@inproceedings{huang-etal-2023-learning-better,
  title={Learning a Better Initialization for Soft Prompts via Meta-Learning},
  author={Huang, Yukun and Qian, Kun and Yu, Zhou},
  booktitle={Proceedings of the 13th International Joint Conference on Natural Language Processing (Short Papers)},
  year={2023},
  pages={67–75},
  url={https://aclanthology.org/2023.ijcnlp-short.8.pdf},
  doi={10.18653/v1/2023.ijcnlp-short.8}
}

@misc{huang2022incontext,
  title={In-context Learning Distillation: Transferring Few-shot Learning Ability of Pre-trained Language Models},
  author={Huang, Yukun and Chen, Yanda and Yu, Zhou and McKeown, Kathleen},
  year={2022},
  eprint={2212.10670},
  archivePrefix={arXiv},
  primaryClass={cs.CL},
  doi={10.48550/arXiv.2212.10670},
  url={https://arxiv.org/abs/2212.10670}
}