How much pretraining data do language models need to learn syntax?
Open accessThe DOI matches its Crossref record; 19 of the 19 references it cites were found in a registry or an index.
Abstract
Transformers-based pretrained language models achieve outstanding results in many wellknown NLU benchmarks.However, while pretraining methods are very convenient, they are expensive in terms of time and resources.This calls for a study of the impact of pretraining data size on the knowledge of the models.We explore this impact on the syntactic capabilities of RoBERTa, using models trained on incremental sizes of raw text data.First, we use syntactic structural probes to determine whether models pretrained on more data encode a higher amount of syntactic information.Second, we perform a targeted syntactic evaluation to analyze the impact of pretraining data size on the syntactic generalization performance of the models.Third, we compare the performance of the different models on three downstream applications: part-of-speech tagging, dependency parsing and paraphrase identification.We complement our study with an analysis of the cost-benefit trade-off of training such models.Our experiments show that while models pretrained on more data encode more syntactic knowledge and perform better on downstream applications, they do not always offer a better performance across the different syntactic phenomena and come at a higher financial and environmental cost.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2019) · Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) · doi:10.18653/v1/n19-1423 · Cited by 54,325
Deep Contextualized Word Representations
Peters, M., Neumann, M., Iyyer, M., Gardner, M., Clark, C., Lee, K., & Zettlemoyer, L. (2018) · Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers) · doi:10.18653/v1/n18-1202 · Cited by 11,983
Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer
Raffel, C., Shazeer, N., Roberts, A., Lee, K., Narang, S., Matena, M., Zhou, Y., Li, W., & Liu, P. J. (2019) · arXiv · doi:10.48550/arxiv.1910.10683 · Cited by 10,929
Energy and Policy Considerations for Deep Learning in NLP
Strubell, E., Ganesh, A., & McCallum, A. (2019) · Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics · doi:10.18653/v1/p19-1355 · Cited by 3,684 · cited 9×
“As the resources used to train the MiniBERTas are not publicly available, we rely on the data provided in (Strubell et al., 2019)”
BackgroundMethodInfluentialKnow What You Don't Know: Unanswerable Questions for SQuAD
Rajpurkar, P., Jia, R., & Liang, P. (2018) · arXiv · doi:10.48550/arxiv.1806.03822 · Cited by 2,242
Know What You Don’t Know: Unanswerable Questions for SQuAD
Rajpurkar, P., Jia, R., & Liang, P. (2018) · Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers) · doi:10.18653/v1/p18-2124 · Cited by 2,240
A primer in BERTology: What we know about how BERT works
Rogers, A., Kovaleva, O., & Rumshisky, A. (2020) · Transactions of the Association for Computational Linguistics · doi:10.1162/tacl_a_00349 · Cited by 1,552
(2019) · doi:10.18653/v1/n19-1419 · Cited by 601
Reimers, N., & Gurevych, I. (2017) · Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/d17-1035 · Cited by 463 · cited 1×
“the variance in performance induced by weight initialization and training data order (Dodge et al., 2020; Reimers and Gurevych, 2017), we repeat this process 5 times per task with different random seeds and average results.”
MethodDesigning and Interpreting Probes with Control Tasks
Hewitt, J., & Liang, P. (2019) · Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP) · doi:10.18653/v1/d19-1275 · Cited by 393 · cited 1×
“(Hewitt and Liang, 2019), and can fail to determine whether the detected features are actually used (Voita and Titov, 2020; Pimentel et al.”
Background
Cited by
What Artificial Neural Networks Can Tell Us about Human Language Acquisition
Warstadt, A., & Bowman, S. R. (2022) · Algebraic Structures in Natural Language · doi:10.1201/9781003205388-2 · Cited by 111
Language Model Behavior: A Comprehensive Survey
Chang, T. A., & Bergen, B. K. (2024) · Computational Linguistics · doi:10.1162/coli_a_00492 · Cited by 105
Hosseini, E. A., Schrimpf, M., Zhang, Y., Bowman, S., Zaslavsky, N., & Fedorenko, E. (2024) · Neurobiology of Language · doi:10.1162/nol_a_00137 · Cited by 63
“Pérez-Mayos et al. (2021) find that a RoBERTa model (Liu et al., 2019) trained on 100 million words performs similarly to a model trained on 1 billion words on several syntactic benchmarks.”
BackgroundMethodDistilled GPT for source code summarization
Su, C.-Y., & McMillan, C. (2024) · Automated Software Engineering · doi:10.1007/s10515-024-00421-4 · Cited by 49
The Life Cycle of Knowledge in Big Language Models: A Survey
Cao, B., Lin, H., Han, X., & Sun, L. (2023) · arXiv · doi:10.48550/arxiv.2303.07616 · Cited by 18
Metadata
- Authors
- Pérez-Mayos, L., Ballesteros, M., & Wanner, L.
- Venue
- Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing
- Published
- 2021
- Pages
- 1571-1582
- OpenAlex
- W3198757395
- Cited by
- 26
- References
- 19
- FWCI
- 2.84
- Type
- conference paper
- Language
- en
- Open access
- hybrid
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: How much pretraining data do language models need to learn syntax? https://app.cochat.ai/papers/10.18653/v1/2021.emnlp-main.118
- The paper
- Pérez-Mayos, L., Ballesteros, M., & Wanner, L. (2021). How much pretraining data do language models need to learn syntax? Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing. https://doi.org/10.18653/v1/2021.emnlp-main.118
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.