def __call__(self,
input_word_ids,
input_mask=None,
input_type_ids=None,
**kwargs):
inputs = tf_utils.pack_inputs([input_word_ids, input_mask, input_type_ids])
return super(AlbertModel, self).__call__(inputs, **kwargs)
def __call__(self,
pooled_output,
sequence_output=None,
masked_lm_positions=None):
inputs = tf_utils.pack_inputs(
[pooled_output, sequence_output, masked_lm_positions])
return super(BertPretrainLayer, self).__call__(inputs)
def __call__(self,
sequence_output,
p_mask,
cls_index,
start_positions=None,
**kwargs):
inputs = tf_utils.pack_inputs(
[sequence_output, p_mask, cls_index, start_positions])
return super(ALBertQALayer, self).__call__(inputs, **kwargs)
def __call__(self,
lm_output,
sentence_output=None,
lm_label_ids=None,
lm_label_weights=None,
sentence_labels=None):
inputs = tf_utils.pack_inputs([
lm_output, sentence_output, lm_label_ids, lm_label_weights,
sentence_labels
])
return super(BertPretrainLossAndMetricLayer, self).__call__(inputs)
def __call__(self, input_tensor, attention_mask=None, **kwargs):
inputs = tf_utils.pack_inputs([input_tensor, attention_mask])
return super(Transformer, self).__call__(inputs=inputs, **kwargs)
def __call__(self, from_tensor, to_tensor, attention_mask=None, **kwargs):
inputs = tf_utils.pack_inputs([from_tensor, to_tensor, attention_mask])
return super(Attention, self).__call__(inputs, **kwargs)
def __call__(self, word_embeddings, token_type_ids=None, **kwargs):
inputs = tf_utils.pack_inputs([word_embeddings, token_type_ids])
return super(EmbeddingPostprocessor, self).__call__(inputs, **kwargs)
