def bert_trans():
"""bert_trans"""
net = BertTransformer(batch_size=1,
hidden_size=768,
seq_length=128,
num_hidden_layers=1,
num_attention_heads=12,
intermediate_size=768,
attention_probs_dropout_prob=0.1,
use_one_hot_embeddings=False,
initializer_range=0.02,
use_relative_positions=False,
hidden_act="gelu",
compute_type=mstype.float32,
return_all_encoders=True)
net.set_train()
return net
initializer_range=0.02,
max_position_embeddings=512,
dropout_prob=0.0)
}, {
'id':
'BertTransformer',
'group':
'BertTransformer',
'block':
BertTransformer(batch_size=1,
hidden_size=1024,
seq_length=128,
num_hidden_layers=2,
num_attention_heads=16,
intermediate_size=4096,
attention_probs_dropout_prob=0.0,
use_one_hot_embeddings=False,
initializer_range=0.02,
use_relative_positions=True,
hidden_act="gelu",
compute_type=mstype.float32,
return_all_encoders=True)
}, {
'id':
'BertEncoderCell',
'group':
'BertEncoderCell',
'block':
BertEncoderCell(batch_size=1,
hidden_size=1024,
seq_length=128,