BERT: Load a BERT-like Transformer model

In bnosac/golgotha: Contextualised Embeddings and Language Modelling using BERT

Description

Load a BERT-like Transformer model stored on disk

Usage

BERT(model_name, path = system.file(package = "golgotha", "models"))

Arguments

model_name	character string with the name of the model. E.g. 'bert-base-uncased', 'bert-base-multilingual-uncased', 'bert-base-multilingual-cased', 'bert-base-dutch-cased'. Defaults to 'bert-base-multilingual-uncased'.
path	path to a directory on disk where the model is stored

Value

the directory where the model is saved to

Examples

bert_download_model("bert-base-multilingual-uncased")
model <- BERT("bert-base-multilingual-uncased")

x <- data.frame(doc_id = c("doc_1", "doc_2"),
                text = c("provide some words to embed", "another sentence of text"),
                stringsAsFactors = FALSE)
predict(model, x, type = "tokenise")
embedding <- predict(model, x, type = "embed-sentence")
dim(embedding)
embedding <- predict(model, x, type = "embed-token")
str(embedding)



model_dir <- file.path(getwd(), "inst", "models")
bert_download_model("bert-base-multilingual-uncased", path = model_dir)
path  <- file.path(getwd(), "inst", "models", "bert-base-multilingual-uncased")
model <- BERT(model_name = "bert-base-multilingual-uncased", path = path)


unlink(file.path(system.file(package = "golgotha", "models"),
       "bert-base-multilingual-uncased"), recursive = TRUE)

bnosac/golgotha documentation built on May 28, 2020, 4:06 a.m.