File size: 2,307 Bytes
2d06dcc | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 | class Param():
def __init__(self, args):
self.hyper_param = self.get_hyper_parameters(args)
def get_hyper_parameters(self, args):
"""
Args:
bert_model (directory): The path for the pre-trained bert model.
max_seq_length (autofill): The maximum total input sequence length after tokenization. Sequences longer than this will be truncated, sequences shorter will be padded.
num_train_epochs (int): The number of training epochs.
num_pretrain_epochs (int): The number of pre-training epochs.
num_labels (autofill): The output dimension.
freeze_bert_parameters (binary): Whether to freeze all parameters but the last layer.
feat_dim (int): The feature dimension.
warmup_proportion (float): The warmup ratio for learning rate.
lr_pre (float): The learning rate for pre-training the backbone.
lr (float): The learning rate for training the backbone.
loss_fct (str): The loss function for training.
activation (str): The activation function of the hidden layer (support 'relu' and 'tanh').
train_batch_size (int): The batch size for training.
eval_batch_size (int): The batch size for evaluation.
test_batch_size (int): The batch size for testing.
wait_patient (int): Patient steps for Early Stop.
"""
hyper_parameters = {
'pretrained_bert_model': '/home/sharing/disk1/pretrained_embedding/bert/uncased_L-12_H-768_A-12/',
'max_seq_length': None,
'num_pretrain_epochs': 100,
'num_train_epochs': 100,
'update_per_epoch' : 5,
'num_labels': None,
'freeze_backbone_parameters': False,
'feat_dim': 768,
'warmup_proportion': 0.1,
'lr_pre': 5e-5,
'lr': 1e-5,
'activation': 'tanh',
'train_batch_size': 128,
'pretrain_batch_size': 64,
'eval_batch_size': 64,
'test_batch_size': 64,
'wait_patient': 10,
'grad_clip': 1,
'nuu': 0.1,
'k': 'rbf',
'use_cls': False
}
return hyper_parameters
|