Pre-trained Flair and BERT weights (Spanish biomedical literature, SciELO) and corpus

Published: 19 August 2020| Version 2 | DOI: 10.17632/vf6jmvz83b.2
Contributor:
Liliya Akhtyamova

Description

This zipped folder includes pre-trained weights of BERT and Flair models. Both models were trained from scratch on Spanish biomedical literature texts obtained from SciELO website. The corpus on which the models were trained is also included. It consists of over 86B tokens. More details on the corpus and training process could be found in our paper "Testing Contextualized Word Embeddings to Improve NER in Spanish Clinical Case Narratives", doi: 10.21203/rs.2.22697/v2

Files

Categories

Natural Language Processing, Language Modeling, Representation Language, Word Embedding

Licence