(Down)Load BERT models (useless for GPU).

Load BERT models from local cache folder "%USERPROFILE%/.cache/huggingface". Models that have not been downloaded can also be automatically downloaded (but silently). For GPU Acceleration, please directly use FMAT_run instead.

Usage

FMAT_load(models)

Arguments

models

Model names at HuggingFace.

Value

A named list of fill-mask pipelines obtained from the models. The returned object cannot be saved as any RData. You will need to rerun this function if you restart the R session.

See also

BERT_download

FMAT_query

FMAT_query_bind

FMAT_run

Examples

if (FALSE) {
models = c("bert-base-uncased", "bert-base-cased")
models = FMAT_load(models)  # load models from cache
}