@inproceedings{smilga-alabiad-2024-tuduo,
title = {{T}{\"u}{D}uo at {S}em{E}val-2024 Task 2: Flan-T5 and Data Augmentation for Biomedical {NLI}},
author = "Smilga, Veronika and
Alabiad, Hazem",
editor = {Ojha, Atul Kr. and
Do{\u{g}}ru{\"o}z, A. Seza and
Tayyar Madabushi, Harish and
Da San Martino, Giovanni and
Rosenthal, Sara and
Ros{\'a}, Aiala},
booktitle = "Proceedings of the 18th International Workshop on Semantic Evaluation (SemEval-2024)",
month = jun,
year = "2024",
address = "Mexico City, Mexico",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.semeval-1.106",
doi = "10.18653/v1/2024.semeval-1.106",
pages = "737--744",
abstract = "This paper explores using data augmentation with smaller language models under 3 billion parameters for the SemEval-2024 Task 2 on Biomedical Natural Language Inference for Clinical Trials. We fine-tune models from the Flan-T5 family with and without using augmented data automatically generated by GPT-3.5-Turbo and find that data augmentation through techniques like synonym replacement, syntactic changes, adding random facts, and meaning reversion improves model faithfulness (ability to change predictions for semantically different inputs) and consistency (ability to give same predictions for semantic preserving changes). However, data augmentation tends to decrease performance on the original dataset distribution, as measured by F1 score. Our best system is the Flan-T5 XL model fine-tuned on the original training data combined with over 6,000 augmented examples. The system ranks in the top 10 for all three metrics.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="smilga-alabiad-2024-tuduo">
<titleInfo>
<title>TüDuo at SemEval-2024 Task 2: Flan-T5 and Data Augmentation for Biomedical NLI</title>
</titleInfo>
<name type="personal">
<namePart type="given">Veronika</namePart>
<namePart type="family">Smilga</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hazem</namePart>
<namePart type="family">Alabiad</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2024-06</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 18th International Workshop on Semantic Evaluation (SemEval-2024)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Atul</namePart>
<namePart type="given">Kr.</namePart>
<namePart type="family">Ojha</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">A</namePart>
<namePart type="given">Seza</namePart>
<namePart type="family">Doğruöz</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Harish</namePart>
<namePart type="family">Tayyar Madabushi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Giovanni</namePart>
<namePart type="family">Da San Martino</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sara</namePart>
<namePart type="family">Rosenthal</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Aiala</namePart>
<namePart type="family">Rosá</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Mexico City, Mexico</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>This paper explores using data augmentation with smaller language models under 3 billion parameters for the SemEval-2024 Task 2 on Biomedical Natural Language Inference for Clinical Trials. We fine-tune models from the Flan-T5 family with and without using augmented data automatically generated by GPT-3.5-Turbo and find that data augmentation through techniques like synonym replacement, syntactic changes, adding random facts, and meaning reversion improves model faithfulness (ability to change predictions for semantically different inputs) and consistency (ability to give same predictions for semantic preserving changes). However, data augmentation tends to decrease performance on the original dataset distribution, as measured by F1 score. Our best system is the Flan-T5 XL model fine-tuned on the original training data combined with over 6,000 augmented examples. The system ranks in the top 10 for all three metrics.</abstract>
<identifier type="citekey">smilga-alabiad-2024-tuduo</identifier>
<identifier type="doi">10.18653/v1/2024.semeval-1.106</identifier>
<location>
<url>https://aclanthology.org/2024.semeval-1.106</url>
</location>
<part>
<date>2024-06</date>
<extent unit="page">
<start>737</start>
<end>744</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T TüDuo at SemEval-2024 Task 2: Flan-T5 and Data Augmentation for Biomedical NLI
%A Smilga, Veronika
%A Alabiad, Hazem
%Y Ojha, Atul Kr.
%Y Doğruöz, A. Seza
%Y Tayyar Madabushi, Harish
%Y Da San Martino, Giovanni
%Y Rosenthal, Sara
%Y Rosá, Aiala
%S Proceedings of the 18th International Workshop on Semantic Evaluation (SemEval-2024)
%D 2024
%8 June
%I Association for Computational Linguistics
%C Mexico City, Mexico
%F smilga-alabiad-2024-tuduo
%X This paper explores using data augmentation with smaller language models under 3 billion parameters for the SemEval-2024 Task 2 on Biomedical Natural Language Inference for Clinical Trials. We fine-tune models from the Flan-T5 family with and without using augmented data automatically generated by GPT-3.5-Turbo and find that data augmentation through techniques like synonym replacement, syntactic changes, adding random facts, and meaning reversion improves model faithfulness (ability to change predictions for semantically different inputs) and consistency (ability to give same predictions for semantic preserving changes). However, data augmentation tends to decrease performance on the original dataset distribution, as measured by F1 score. Our best system is the Flan-T5 XL model fine-tuned on the original training data combined with over 6,000 augmented examples. The system ranks in the top 10 for all three metrics.
%R 10.18653/v1/2024.semeval-1.106
%U https://aclanthology.org/2024.semeval-1.106
%U https://doi.org/10.18653/v1/2024.semeval-1.106
%P 737-744
Markdown (Informal)
[TüDuo at SemEval-2024 Task 2: Flan-T5 and Data Augmentation for Biomedical NLI](https://aclanthology.org/2024.semeval-1.106) (Smilga & Alabiad, SemEval 2024)
ACL