flambe.nlp.transformers.roberta

Intergation of the pytorch_transformers roberta module.

Note that these objects are only to be used to load pretrained models. The pytorch-transformers library wasn’t designed to train these models from scratch.

Module Contents

class flambe.nlp.transformers.roberta.RobertaTextField[source]

Bases: flambe.nlp.transformers.utils.TransformerTextField

Integrate the pytorch_transformers RobertaTokenizer.

Currently available aliases:
. roberta-base . roberta-large . roberta-large-mnli
_cls[source]
class flambe.nlp.transformers.roberta.RobertaEmbedder[source]

Bases: flambe.nlp.transformers.utils.TransformerEmbedder

Integrate the pytorch_transformers RobertaModel.

Currently available aliases:
. roberta-base . roberta-large . roberta-large-mnli
_cls[source]