Selective Token Generation for Few-shot Natural Language Generation
DC Field | Value | Language |
---|---|---|
dc.contributor.author | Jo, Daejin | - |
dc.contributor.author | Kwon, Taehwan | - |
dc.contributor.author | Kim, Eun-Sol | - |
dc.contributor.author | Kim, Sungwoong | - |
dc.date.accessioned | 2023-08-16T07:35:28Z | - |
dc.date.available | 2023-08-16T07:35:28Z | - |
dc.date.created | 2023-08-07 | - |
dc.date.issued | 2022-10 | - |
dc.identifier.issn | 2951-2093 | - |
dc.identifier.uri | https://scholarworks.bwise.kr/hanyang/handle/2021.sw.hanyang/189021 | - |
dc.description.abstract | Natural language modeling with limited training data is a challenging problem, and many algorithms make use of large-scale pretrained language models (PLMs) for this due to its great generalization ability. Among them, additive learning that incorporates a task-specific adapter on top of the fixed large-scale PLM has been popularly used in the few-shot setting. However, this added adapter is still easy to disregard the knowledge of the PLM especially for few-shot natural language generation (NLG) since an entire sequence is usually generated by only the newly trained adapter. Therefore, in this work, we develop a novel additive learning algorithm based on reinforcement learning (RL) that selectively outputs language tokens between the task-general PLM and the task-specific adapter during both training and inference. This output token selection over the two generators allows the adapter to take into account solely the task-relevant parts in sequence generation, and therefore makes it more robust to overfitting as well as more stable in RL training. In addition, to obtain the complementary adapter from the PLM for each few-shot task, we exploit a separate selecting module that is also simultaneously trained using RL. Experimental results on various few-shot NLG tasks including question answering, data-to-text generation and text summarization demonstrate that the proposed selective token generation significantly outperforms the previous additive learning algorithms based on the PLMs. | - |
dc.language | 영어 | - |
dc.language.iso | en | - |
dc.publisher | Association for Computational Linguistics (ACL) | - |
dc.title | Selective Token Generation for Few-shot Natural Language Generation | - |
dc.type | Article | - |
dc.contributor.affiliatedAuthor | Kim, Eun-Sol | - |
dc.identifier.doi | 10.48550/arXiv.2209.08206 | - |
dc.identifier.scopusid | 2-s2.0-85165747575 | - |
dc.identifier.bibliographicCitation | Proceedings - International Conference on Computational Linguistics, COLING, v.29, no.1, pp.5837 - 5856 | - |
dc.relation.isPartOf | Proceedings - International Conference on Computational Linguistics, COLING | - |
dc.citation.title | Proceedings - International Conference on Computational Linguistics, COLING | - |
dc.citation.volume | 29 | - |
dc.citation.number | 1 | - |
dc.citation.startPage | 5837 | - |
dc.citation.endPage | 5856 | - |
dc.type.rims | ART | - |
dc.type.docType | Conference paper | - |
dc.description.journalClass | 1 | - |
dc.description.isOpenAccess | N | - |
dc.description.journalRegisteredClass | scopus | - |
dc.subject.keywordPlus | Additives | - |
dc.subject.keywordPlus | Computational linguistics | - |
dc.subject.keywordPlus | Inference engines | - |
dc.subject.keywordPlus | Modeling languages | - |
dc.subject.keywordPlus | Natural language processing systems | - |
dc.subject.keywordPlus | Reinforcement learning | - |
dc.subject.keywordPlus | Learning algorithms | - |
dc.subject.keywordPlus | Generalization ability | - |
dc.subject.keywordPlus | Language model | - |
dc.subject.keywordPlus | Large-scales | - |
dc.subject.keywordPlus | Limited training data | - |
dc.subject.keywordPlus | Natural language generation | - |
dc.subject.keywordPlus | Natural language model | - |
dc.subject.keywordPlus | Overfitting | - |
dc.subject.keywordPlus | Reinforcement learnings | - |
dc.subject.keywordPlus | Sequence generation | - |
dc.subject.keywordPlus | Task relevant | - |
dc.identifier.url | https://arxiv.org/abs/2209.08206 | - |
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.
222, Wangsimni-ro, Seongdong-gu, Seoul, 04763, Korea+82-2-2220-1365
COPYRIGHT © 2021 HANYANG UNIVERSITY.
Certain data included herein are derived from the © Web of Science of Clarivate Analytics. All rights reserved.
You may not copy or re-distribute this material in whole or in part without the prior written consent of Clarivate Analytics.