bnosac/golgotha: Contextualised Embeddings and Language Modelling using BERT

Wraps the transformers Python module described in "HuggingFace's Transformers: State-of-the-art Natural Language Processing" by Wolf T. et al. (2020) <arXiv:1910.03771> in order to easily obtain contextualised embeddings of sentences. This work was done in order to ease the work of building predictive models using BERT-like embeddings. BERT stands for "Bidirectional Encoder Representations from Transformers" as described in Devlin J. et al. (2018) <arXiv:1810.04805>. It is a deep learning model which provides for pieces of words in a text a set of numbers called embeddings. These embedding capture the meaning of the words and depend on the context in which each word appears. The package provides an interface to BERT-like "Transformer" models such that R users can just download these pretrained models and retrieve the embeddings of text with a straightforward call to predict.

install.packages("remotes")
remotes::install_github("bnosac/golgotha")