Elamur Research⌕
☰
home / papers / 2026 / When Tasks Share Structure: A Comparative Study of Training Strategies for Generative Event Extraction

When Tasks Share Structure: A Comparative Study of Training Strategies for Generative Event Extraction

Paper · 2026 · Rishi Ravikumar and Riza Batista-Navarro

Summary

Event extraction usually breaks into two subtasks: event detection (ED) locates trigger words and assigns each an event type, while event argument extraction (EAE) identifies participants and the roles they play. It is unclear how these two linked subtasks should be trained together when LLMs generate the extraction as text. The authors fine-tune three instruction-tuned LLMs (3B–12B parameters) with LoRA adapters. Each model receives a bare sentence with no instructions and outputs JSON. They compare eight training strategies of equal compute, grouped into three families: separate adapters per task (disjoint), one adapter set for both tasks (fully shared), and shared lower layers with task-specific upper layers (hybrid). Evaluation uses ACE2005 and RichERE under the TextEE framework. The choice of strategy has a consistent effect on accuracy, and the effect is larger for smaller models. The best approach first trains EAE adapters and then uses them to initialise ED adapters. This setup beats both standard baselines, independent pipelines and single-pass joint modelling. The results give practical guidance for designing generative IE systems built from dependent subtasks.

Key points

  • The eight strategies are:
    • Disjoint: Independent (S1), Forward Transfer ED→EAE (S2), Backward Transfer EAE→ED (S3).
    • Fully shared: Joint Modelling in one pass (S4), Mixed Training on interleaved batches (S5).
    • Hybrid: Partial Sharing of the lower 25%, 50% or 75% of layers (S6.1–S6.3).
  • Backward Transfer (S3) ranks first on trigger classification (TC) and argument classification (AC). Forward Transfer (S2) ranks first on AC|TC on both benchmarks.
  • Joint Modelling (S4) is the weakest configuration on TC, but it partly recovers on AC|TC.
  • ED gains from cross-task signal, whether through initialisation or sharing. EAE works best with dedicated adapter capacity, and its performance drops as parameter sharing increases. Hybrid strategies behave inconsistently across benchmarks.
  • Setup details: TextEE Split 1, a fixed 2-epoch schedule, and LoRA only (full fine-tuning was not tested). The code is released on GitHub (rishi-ravikumar/GenerativeEventExtractionTrainingStrategies).

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓University of ManchesterManchester · United Kingdom87 works · 44,938 citations

 Fields, topics and concepts

📚 Shelves 5The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 worksComputer Vision and Pattern Recognition · 2,869 works

🧭 Topics 1Research subjects and languages the work is about.Information extraction · 167 works

Cite

Cite as Rishi Ravikumar and Riza Batista-Navarro. 2026. When Tasks Share Structure: A Comparative Study of Training Strategies for Generative Event Extraction. Proceedings of the 9th Workshop on Event Extraction and Understanding: Challenges and Applications (EEUCA 2026).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.eeuca-1.5

@inproceedings{ravikumar2026when,
  title = {{When Tasks Share Structure: A Comparative Study of Training Strategies for Generative Event Extraction}},
  author = {Rishi Ravikumar and Riza Batista-Navarro},
  booktitle = {Proceedings of the 9th Workshop on Event Extraction and Understanding: Challenges and Applications (EEUCA 2026)},
  year = {2026},
  doi = {10.18653/v1/2026.eeuca-1.5},
  url = {https://doi.org/10.18653/v1/2026.eeuca-1.5}
}

References

24 references, 15 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Rich Caruana. 1997. Multitask learning. Machine Learning, 28(1):41–75. find
  2. LDC. 2005. Ace (automatic content extraction) english annotation guidelines for events. Accessed: 2025-02-12. find
  3. Tim Dettmers, Artidoro Pagnoni, Ari Holtzman, and Luke Zettlemoyer. 2023. Qlora: efficient finetuning of quantized llms. In Proceedings of the 37th International Conference on Neural Information Processing Systems, NIPS ’23, Red Hook, NY, USA. Curran Associates Inc.
  4. Ying Lin, Heng Ji, Fei Huang, and Lingfei Wu. 2020. A joint neural model for information extraction with global features. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 7999–8009, Online. Association for Computational Linguistics.
  5. George Doddington, Alexis Mitchell, Mark Przybocki, Lance Ramshaw, Stephanie Strassel, and Ralph Weischedel. 2004. The automatic content extraction (ACE) program – tasks, data, and evaluation. In Proceedings of the Fourth International Conference on Language Resources and Evaluation (LREC’04), Lisbo…
  6. Yaojie Lu, Hongyu Lin, Jin Xu, Xianpei Han, Jialong Tang, Annan Li, Le Sun, Meng Liao, and Shaoyi Chen. 2021. Text2Event: Controllable sequence-to-structure generation for end-to-end event extraction. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the…
  7. Xinya Du and Claire Cardie. 2020. Event extraction by answering (almost) natural questions. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 671–683, Online. Association for Computational Linguistics.
  8. Meta and 1 others. 2024. The llama 3 herd of models. Preprint, arXiv:2407.21783.
  9. Jun Gao, Huan Zhao, Changlong Yu, and Ruifeng Xu. 2023. Exploring the feasibility of chatgpt for event extraction. Preprint, arXiv:2303.03836.
  10. Mistral AI. 2024. Mistral nemo. Accessed: 2026-03-28. find
  11. I-Hung Hsu, Kuan-Hao Huang, Elizabeth Boschee, Scott Miller, Prem Natarajan, Kai-Wei Chang, and Nanyun Peng. 2022. DEGREE: A data-efficient generation-based event extraction model. In Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: H…
  12. Giovanni Paolini, Ben Athiwaratkun, Jason Krone, Jie Ma, Alessandro Achille, Rishita Anubhai, Cicero Nogueira dos Santos, Bing Xiang, and Stefano Soatto. 2021. Structured prediction as translation between augmented natural languages. Preprint, arXiv:2101.05779. find
  13. Yada Pruksachatkun, Jason Phang, Haokun Liu, Phu Mon Htut, Xiaoyi Zhang, Richard Yuanzhe Pang, Clara Vania, Katharina Kann, and Samuel R. Bowman. 2020. Intermediate-task transfer learning with pretrained language models: When and why does it work? In Proceedings of the 58th Annual Meeting of the Ass…
  14. I-Hung Hsu, Kuan-Hao Huang, Shuning Zhang, Wenxin Cheng, Prem Natarajan, Kai-Wei Chang, and Nanyun Peng. 2023. TAGPRIME: A unified framework for relational structure extraction. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages… find
  15. Qwen, An Yang, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chengyuan Li, Dayiheng Liu, Fei Huang, Haoran Wei, Huan Lin, Jian Yang, Jianhong Tu, Jianwei Zhang, Jianxin Yang, Jiaxi Yang, Jingren Zhou, Junyang Lin, and 24 others. 2025. Qwen2.5 technical report. Preprint, arXiv:2412.15… find
  16. Edward J. Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. 2021. Lora: Low-rank adaptation of large language models. Preprint, arXiv:2106.09685.
  17. Kuan-Hao Huang, I-Hung Hsu, Tanmay Parekh, Zhiyu Xie, Zixuan Zhang, Prem Natarajan, Kai-Wei Chang, Rishi Ravikumar, Nuhu Ibrahim, and Riza Batista-Navarro. 2026. Lost in formatting: How output formats skew LLM performance on information extraction. In Proceedings of the 19th Conference of the Europe…
  18. Saurabh Srivastava, Sweta Pati, and Ziyu Yao. 2025. Instruction-tuning LLMs for event extraction with annotation guidelines. In Findings of the Association for Computational Linguistics: ACL 2025, pages 13055–13071, Vienna, Austria. Association for Computational Linguistics. find
  19. The Comprehensive R Archive Network. interpret_kendalls_w: Interpret Kendall’s Coefficient of Concordance W. CRAN. Accessed: 2026-05-18. find
  20. Anna Rogers, Olga Kovaleva, and Anna Rumshisky. 2021. A primer in bertology: What we know about how bert works. Transactions of the Association for Computational Linguistics, 8:842–866.
  21. David Wadden, Ulme Wennberg, Yi Luan, and Hannaneh Hajishirzi. 2019. Entity, relation, and event extraction with contextualized span representations. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Langua…
  22. Sebastian Ruder. 2017. An overview of multi-task learning in deep neural networks. Preprint, arXiv:1706.05098.
  23. Zhiyi Song, Ann Bies, Stephanie Strassel, Tom Riese, Justin Mott, Joe Ellis, Jonathan Wright, Seth Kulick, Neville Ryant, and Xiaoyi Ma. 2015. From light to rich ERE: Annotation of entities, relations, and events. In Proceedings of the 3rd Workshop on EVENTS: Definition, Detection, Coreference, and…
  24. Wei Xiang and Bang Wang. 2019. A survey of event extraction from text. IEEE Access, 7:173111–173137. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/ravikumar-2026-when-tasks-share-structure-comparative/