Skip to content

Latest commit

 

History

History
 
 

models

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
README.md
README.md
 
 
__init__.py
__init__.py
 
 
bert_classifier.py
bert_classifier.py
 
 
bert_classifier_test.py
bert_classifier_test.py
 
 
bert_pretrainer.py
bert_pretrainer.py
 
 
bert_pretrainer_test.py
bert_pretrainer_test.py
 
 
bert_span_labeler.py
bert_span_labeler.py
 
 
bert_span_labeler_test.py
bert_span_labeler_test.py
 
 
bert_token_classifier.py
bert_token_classifier.py
 
 
bert_token_classifier_test.py
bert_token_classifier_test.py
 
 
dual_encoder.py
dual_encoder.py
 
 
dual_encoder_test.py
dual_encoder_test.py
 
 
electra_pretrainer.py
electra_pretrainer.py
 
 
electra_pretrainer_test.py
electra_pretrainer_test.py
 
 
seq2seq_transformer.py
seq2seq_transformer.py
 
 
seq2seq_transformer_test.py
seq2seq_transformer_test.py
 
 
xlnet.py
xlnet.py
 
 
xlnet_test.py
xlnet_test.py
 
 

README.md

Models

Models are combinations of tf.keras layers and models that can be trained.

Several pre-built canned models are provided to train encoder networks. These models are intended as both convenience functions and canonical examples.

  • BertClassifier implements a simple classification model containing a single classification head using the Classification network. It can be used as a regression model as well.

  • BertTokenClassifier implements a simple token classification model containing a single classification head over the sequence output embeddings.

  • BertSpanLabeler implementats a simple single-span start-end predictor (that is, a model that predicts two values: a start token index and an end token index), suitable for SQuAD-style tasks.

  • BertPretrainer implements a masked LM and a classification head using the Masked LM and Classification networks, respectively.

  • DualEncoder implements a dual encoder model, suitbale for retrieval tasks.