Abstract
We propose a pre-training objective based on question answering (QA) for learning general-purpose contextual representations, motivated by the intuition that the representation of a phrase in a passage should encode all questions that the phrase can answer in context. To this end, we train a bi-encoder QA model, which independently encodes passages and questions, to match the predictions of a more accurate cross-encoder model on 80 million synthesized QA pairs. By encoding QA-relevant information, the bi-encoder’s token-level representations are useful for non-QA downstream tasks without extensive (or in some cases, any) fine-tuning. We show large improvements over both RoBERTa-large and previous state-of-the-art results on zero-shot and few-shot paraphrase detection on four datasets, few-shot named entity recognition on two datasets, and zero-shot sentiment analysis on three datasets.- Anthology ID:
- 2022.findings-acl.59
- Volume:
- Findings of the Association for Computational Linguistics: ACL 2022
- Month:
- May
- Year:
- 2022
- Address:
- Dublin, Ireland
- Editors:
- Smaranda Muresan, Preslav Nakov, Aline Villavicencio
- Venue:
- Findings
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 711–728
- Language:
- URL:
- https://aclanthology.org/2022.findings-acl.59
- DOI:
- 10.18653/v1/2022.findings-acl.59
- Cite (ACL):
- Robin Jia, Mike Lewis, and Luke Zettlemoyer. 2022. Question Answering Infused Pre-training of General-Purpose Contextualized Representations. In Findings of the Association for Computational Linguistics: ACL 2022, pages 711–728, Dublin, Ireland. Association for Computational Linguistics.
- Cite (Informal):
- Question Answering Infused Pre-training of General-Purpose Contextualized Representations (Jia et al., Findings 2022)
- PDF:
- https://preview.aclanthology.org/naacl24-info/2022.findings-acl.59.pdf
- Code
- facebookresearch/quip
- Data
- HotpotQA, MRPC, MRQA, Natural Questions, NewsQA, PAWS, SQuAD, SST, SST-2, SearchQA