-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtrain.py
170 lines (141 loc) · 6.3 KB
/
train.py
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
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch import optim
import logging
import random
import math
import os
from tqdm import tqdm
from model import EncoderRNN,DecoderRNN
from config import SOS_token,EOS_token,PAD_token,MAXLEN,teacher_forcing_ratio
USE_CUDA = torch.cuda.is_available()
device = torch.device("cuda" if USE_CUDA else "cpu")
# one iteration
def train_iteration(input_batch_tensor,input_len,target_batch_tensor,max_target_len,mask,keywords_batch_tensor,encoder,decoder,
encoder_optimizer,decoder_optimizer,batch_size,use_ATTN,max_length=MAXLEN):
# Gradient reset
encoder_optimizer.zero_grad()
decoder_optimizer.zero_grad()
input_batch_tensor = input_batch_tensor.to(device)
target_batch_tensor = target_batch_tensor.to(device)
keywords_batch_tensor = keywords_batch_tensor.to(device)
mask = mask.to(device)
loss = 0
print_losses = []
n_totals = 0
encoder_outputs,(encoder_h,encoder_c) = encoder(input_batch_tensor,input_len,None)
# convert input
decoder_input = torch.LongTensor([[SOS_token] for _ in range(batch_size)])
decoder_input = decoder_input.to(device)
decoder_hidden = (encoder_h[:decoder.n_layers],encoder_c[:decoder.n_layers]) # because of the bidirection
#use_teach_forcing = True if random.random() < teacher_forcing_ratio else False
use_teach_forcing = False
# use teach forcing
if use_teach_forcing:
for t in range(max_target_len):
if use_ATTN:
decoder_output,decoder_hidden,_ = decoder(encoder_outputs,decoder_input,keywords_batch_tensor,decoder_hidden)
else:
decoder_output,decoder_hidden = decoder(encoder_outputs,decoder_input,keywords_batch_tensor,decoder_hidden)
decoder_input = target_batch_tensor[:,t].view(1,-1)
loss += F.cross_entropy(decoder_output,target_batch_tensor[:,t], ignore_index=EOS_token)
# without forcing
else:
for t in range(max_target_len):
if use_ATTN:
decoder_output,decoder_hidden,_ = decoder(encoder_outputs,decoder_input,keywords_batch_tensor,decoder_hidden)
else:
decoder_output,decoder_hidden = decoder(encoder_outputs,decoder_input,keywords_batch_tensor,decoder_hidden)
_,topi = decoder_output.topk(1)
decoder_input = torch.LongTensor([[topi[i][0]] for i in range(batch_size)])
decoder_input = decoder_input.to(device)
loss += F.cross_entropy(decoder_output, target_batch_tensor[:,t], ignore_index=EOS_token)
loss.backward()
# gradient clip
gradient_clip = 50.0
_ = torch.nn.utils.clip_grad_norm_(encoder.parameters(),gradient_clip)
_ = torch.nn.utils.clip_grad_norm_(decoder.parameters(),gradient_clip)
encoder_optimizer.step()
decoder_optimizer.step()
return loss.item()/max_target_len
def train(load_pretrain,src_voc,tag_voc,pair_batches,n_iteration,learning_rate,batch_size,n_layers,input_size,hidden_size,print_every,
save_every,dropout,src_embeddings,tag_embeddings,rnn_type='LSTM',bidirectional=True,use_ATTN=True,decoder_learning_ratio=1.0):
"""
src_voc:source vocabulary
tag_voc:target vocabulary
"""
checkpoint = None
# build the model
source_embedding = nn.Embedding(src_voc.n_words,input_size)
source_embedding.weight = nn.Parameter(src_embeddings)
target_embedding = nn.Embedding(tag_voc.n_words,input_size)
target_embedding.weight = nn.Parameter(tag_embeddings)
#whether fine-tune embedding
for p in source_embedding.parameters():
p.requires_grad = False
for p in target_embedding.parameters():
p.requires_grad = False
print("INFO:Building Encoder and Decoder ... ")
encoder = EncoderRNN(source_embedding,input_size,hidden_size,n_layers,bidirectional=bidirectional,dropout=dropout,rnn_type=rnn_type)
decoder = DecoderRNN(target_embedding,input_size,hidden_size,tag_voc.n_words,n_layers,rnn_type=rnn_type,use_ATTN=use_ATTN,dropout=dropout)
#load pre-training
if load_pretrain!=None:
checkpoint = torch.load(load_pretrain)
encoder.load_state_dict(checkpoint['en'])
decoder.load_state_dict(checkpoint['de'])
# train on GPU
encoder = encoder.to(device)
decoder = decoder.to(device)
print("INFO:Building optimizers ...")
encoder_optimizer = optim.Adam(encoder.parameters(),lr=learning_rate)
decoder_optimizer = optim.Adam(decoder.parameters(),lr=learning_rate*decoder_learning_ratio)
# load pre-training
if load_pretrain!=None:
encoder_optimizer.load_state_dict(checkpoint['en_opt'])
decoder_optimizer.load_state_dict(checkpoint['de_opt'])
print("INFO:Initializing...")
start_iteration = 1
perplexity = []
print_loss = 0
# logger
logger = logging.getLogger()
logger.setLevel(level = logging.INFO)
handler = logging.FileHandler("log.txt")
handler.setLevel(logging.INFO)
formatter = logging.Formatter('%(asctime)s - %(name)s - %(levelname)s - %(message)s')
handler.setFormatter(formatter)
logger.addHandler(handler)
if load_pretrain!=None:
start_iteration = checkpoint['iteration'] + 1
perplexity = checkpoint['plt']
# iteration
for iteration in tqdm(range(start_iteration,n_iteration+1)):
#
training_batch = pair_batches[iteration-1]
input_batch_tensor,input_len,output_batch_tensor,max_target_len,mask,keywords_batch_tensor = training_batch
loss = train_iteration(input_batch_tensor,input_len,output_batch_tensor,max_target_len,mask,keywords_batch_tensor,encoder,
decoder,encoder_optimizer,decoder_optimizer,batch_size,use_ATTN=use_ATTN)
print_loss += loss
perplexity.append(loss)
if iteration % print_every == 0:
print_loss_average = math.exp(print_loss/print_every)
print('INFO:loss %d %d%% %.4f' % (iteration, iteration / n_iteration * 100, print_loss_average))
logger.info('INFO:loss %d %d%% %.4f' % (iteration, iteration / n_iteration * 100, print_loss_average))
print_loss = 0.0
# save model
if (iteration % save_every == 0):
directory = os.path.join('model_param','{}_{}_{}'.format(n_layers,n_layers,hidden_size))
if not os.path.exists(directory):
os.makedirs(directory)
torch.save(
{
'iteration':iteration,
'encoder':encoder.state_dict(),
'decoder':decoder.state_dict(),
'encoder_optim':encoder_optimizer.state_dict(),
'decoder_optim':decoder_optimizer.state_dict(),
'loss':loss,
'plt':perplexity
},os.path.join(directory,'{}_{}.tar'.format(n_iteration,'seq2seq_bidir_model'))
)