"""
Intergation of the pytorch_transformers xlnet 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.
"""
import pytorch_transformers as pt
from flambe.nlp.transformers.utils import TransformerTextField, TransformerEmbedder
[docs]class XLNetTextField(TransformerTextField):
"""Integrate the pytorch_transformers XLNetTokenizer.
Currently available aliases:
. `xlnet-base-cased`
. `xlnet-large-cased`
"""
[docs] _cls = pt.XLNetTokenizer
[docs]class XLNetEmbedder(TransformerEmbedder):
"""Integrate the pytorch_transformers XLNetModel.
Currently available aliases:
. `xlnet-base-cased`
. `xlnet-large-cased`
"""