@inproceedings{tortoreto-mousavi-2024-dolomites,
    title = "Dolomites@{\#}{SMM}4{H} 2024: Helping {LLM}s {\textquotedblleft}Know The Drill{\textquotedblright} in Low-Resource Settings - A Study on Social Media Posts",
    author = "Tortoreto, Giuliano  and
      Mousavi, Seyed Mahed",
    editor = "Xu, Dongfang  and
      Gonzalez-Hernandez, Graciela",
    booktitle = "Proceedings of the 9th Social Media Mining for Health Research and Applications (SMM4H 2024) Workshop and Shared Tasks",
    month = aug,
    year = "2024",
    address = "Bangkok, Thailand",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/jlcl-multiple-ingestion/2024.smm4h-1.5/",
    pages = "17--22",
    abstract = "The amount of data to fine-tune LLMs plays a crucial role in the performance of these models in downstream tasks. Consequently, it is not straightforward to deploy these models in low-resource settings. In this work, we investigate two new multi-task learning data augmentation approaches for fine-tuning LLMs when little data is available: {\textquotedblleft}In-domain Augmentation{\textquotedblright} of the training data and extracting {\textquotedblleft}Drills{\textquotedblright} as smaller tasks from the target dataset. We evaluate the proposed approaches in three natural language processing settings in the context of SMM4H 2024 competition tasks: multi-class classification, entity recognition, and information extraction. The results show that both techniques improve the performance of the models in all three settings, suggesting a positive impact from the knowledge learned in multi-task training to perform the target task."
}