A Warm Start and a Clean Crawled Corpus - A Recipe for Good Language Models
Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review
We train several language models for Icelandic, including IceBERT, that achieve state-of-the-art performance in a variety of downstream tasks, including part-of-speech tagging, named entity recognition, grammatical error detection and constituency parsing. To train the models we introduce a new corpus of Icelandic text, the Icelandic Common Crawl Corpus (IC3), a collection of high quality texts found online by targeting the Icelandic top-level-domain.is. Several other public data sources are also collected for a total of 16GB of Icelandic text. To enhance the evaluation of model performance and to raise the bar in baselines for Icelandic, we manually translate and adapt the WinoGrande commonsense reasoning dataset. Through these efforts we demonstrate that a properly cleaned crawled corpus is sufficient to achieve state-of-the-art results in NLP applications for low to medium resource languages, by comparison with models trained on a curated corpus. We further show that initializing models using existing multilingual models can lead to state-of-the-art results for some downstream tasks.
Original language | English |
---|---|
Title of host publication | 2022 Language Resources and Evaluation Conference, LREC 2022 |
Editors | Nicoletta Calzolari, Frederic Bechet, Philippe Blache, Khalid Choukri, Christopher Cieri, Thierry Declerck, Sara Goggi, Hitoshi Isahara, Bente Maegaard, Joseph Mariani, Helene Mazo, Jan Odijk, Stelios Piperidis |
Number of pages | 11 |
Publisher | European Language Resources Association (ELRA) |
Publication date | 2022 |
Pages | 4356-4366 |
ISBN (Electronic) | 9791095546726 |
Publication status | Published - 2022 |
Externally published | Yes |
Event | 13th International Conference on Language Resources and Evaluation Conference, LREC 2022 - Marseille, France Duration: 20 Jun 2022 → 25 Jun 2022 |
Conference
Conference | 13th International Conference on Language Resources and Evaluation Conference, LREC 2022 |
---|---|
Land | France |
By | Marseille |
Periode | 20/06/2022 → 25/06/2022 |
Sponsor | 3M, Emvista, et al., Google, SADILAR, Vocapia |
Bibliographical note
Funding Information:
We thank Prof. Dr.-Ing. Morris Riedel and his team for providing access to the DEEP super-computer at Forschungszentrum Jülich. We also thank the Icelandic Language Technology Program (Nikulásdóttir et al., 2020). It has enabled the authors to focus on work in Icelandic NLP. Finally, we thank the anonymous reviewers for their helpful feedback.
Publisher Copyright:
© European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0.
- co-reference resolution, corpus, IceBERT, Icelandic, language model, named entity recognition, natural language understanding, parsing, part of speech
Research areas
ID: 371184890