-
Notifications
You must be signed in to change notification settings - Fork 1
/
trainer.py
189 lines (156 loc) · 7.27 KB
/
trainer.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
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
import os
import numpy as np
import torch.nn.utils
import torch.optim as optim
from time import time
from tqdm import tqdm
from torch.utils.tensorboard import SummaryWriter
from logging import getLogger
from utils import early_stopping, ensure_dir
from optim import ScheduledOptimizer
from evaluator import Evaluator
class Trainer:
def __init__(self, config, model):
self.config = config
self.model = model
self.logger = getLogger()
# training settings
self.DDP = config['DDP']
self.epochs = config['epochs']
self.learner = config['learner'].lower()
self.learning_rate = config['learning_rate']
self.eval_step = min(config['eval_step'], self.epochs)
self.stopping_step = config['stopping_step']
self.grad_clip = config['grad_clip']
self.plot = config['plot']
if self.plot:
self.writer = SummaryWriter(log_dir='./runs/{}'.format(config['filename']))
self.start_epoch = 0
self.cur_step = 0
self.best_valid_score = 1e9
self.best_valid_result = None
self.optimizer = self._build_optimizer()
self.checkpoint_dir = config['checkpoint_dir']
ensure_dir(self.checkpoint_dir)
saved_model_file = config['filename'] + '.pth'
self.saved_model_file = os.path.join(self.checkpoint_dir, saved_model_file)
self.generated_text_dir = config['generated_text_dir']
ensure_dir(self.generated_text_dir)
saved_text_file = self.config['filename'] + '.txt'
self.saved_text_file = os.path.join(self.generated_text_dir, saved_text_file)
self.evaluator = Evaluator()
self.is_logger = not self.DDP
def _build_optimizer(self):
if self.learner == 'adam':
return optim.Adam(self.model.parameters(), lr=self.learning_rate)
elif self.learner == 'schedule':
return ScheduledOptimizer(optim.Adam(self.model.parameters(), lr=self.learning_rate), self.config)
def _train_epoch(self, train_data):
self.model.train()
total_loss = 0.
pbar = train_data
if self.is_logger:
pbar = tqdm(pbar)
for data in pbar:
self.optimizer.zero_grad()
loss = self.model(data)
total_loss += loss.item()
loss.backward()
torch.nn.utils.clip_grad_norm_(self.model.parameters(), self.grad_clip)
self.optimizer.step()
train_loss = total_loss / len(train_data)
return train_loss
def _valid_epoch(self, valid_data):
with torch.no_grad():
self.model.eval()
total_loss = 0.
for data in valid_data:
loss = self.model(data)
total_loss += loss.item()
valid_loss = total_loss / len(valid_data)
ppl = np.exp(valid_loss)
return valid_loss, ppl
def _save_checkpoint(self, epoch):
state = {
'epoch': epoch,
'cur_step': self.cur_step,
'best_valid_score': self.best_valid_score,
'state_dict': self.model.state_dict(),
'optimizer': self.optimizer.state_dict()
}
torch.save(state, self.saved_model_file)
def resume_checkpoint(self, resume_file):
resume_file = str(resume_file)
checkpoint = torch.load(resume_file)
self.start_epoch = checkpoint['epoch'] + 1
self.cur_step = checkpoint['cur_step']
self.best_valid_score = checkpoint['best_valid_score']
self.model.load_state_dict(checkpoint['state_dict'])
self.optimizer.load_state_dict(checkpoint['optimizer'])
if self.is_logger:
self.logger.info('Checkpoint loaded. Resume training from epoch {}'.format(self.start_epoch))
def _save_generated_text(self, generated_corpus):
with open(self.saved_text_file, 'w') as fin:
for tokens in generated_corpus:
fin.write(' '.join(tokens) + '\n')
def fit(self, train_data, valid_data, saved=True):
if self.start_epoch >= self.epochs or self.epochs <= 0:
self._save_checkpoint(-1)
for epoch_idx in range(self.start_epoch, self.epochs):
training_start_time = time()
train_loss = self._train_epoch(train_data)
training_end_time = time()
train_loss_output = "epoch %d training [time: %.2fs, train_loss: %.4f]" % \
(epoch_idx, training_end_time - training_start_time, train_loss)
if self.is_logger:
self.logger.info(train_loss_output)
if self.plot:
self.writer.add_scalar('loss', train_loss, epoch_idx)
if (epoch_idx + 1) % self.eval_step == 0:
valid_start_time = time()
valid_score, valid_result = self._valid_epoch(valid_data)
self.best_valid_score, self.cur_step, stop_flag, update_flag = early_stopping(
valid_score, self.best_valid_score, self.cur_step, max_step=self.stopping_step
)
valid_end_time = time()
valid_score_output = "epoch %d evaluating [time: %.2fs, valid_loss: %f]" % \
(epoch_idx, valid_end_time - valid_start_time, valid_score)
valid_result_output = 'valid ppl: {}'.format(valid_result)
if self.is_logger:
self.logger.info(valid_score_output)
self.logger.info(valid_result_output)
if update_flag:
if saved:
self._save_checkpoint(epoch_idx)
update_output = 'Saving current best: %s' % self.saved_model_file
if self.is_logger:
self.logger.info(update_output)
self.best_valid_result = valid_result
if stop_flag:
stop_output = ('Finished training, best eval result in epoch %d' %
(epoch_idx - self.cur_step * self.eval_step))
if self.is_logger:
self.logger.info(stop_output)
break
return self.best_valid_score, self.best_valid_result
@torch.no_grad()
def evaluate(self, eval_data, model_file=None):
if model_file:
checkpoint_file = model_file
else:
checkpoint_file = self.saved_model_file
checkpoint = torch.load(checkpoint_file)
self.model.load_state_dict(checkpoint['state_dict'])
message_output = 'Loading model structure and parameters from {}'.format(checkpoint_file)
if self.is_logger:
self.logger.info(message_output)
self.model.eval()
generated_corpus = []
with torch.no_grad():
for data in tqdm(eval_data):
generated = self.model.generate(data)
generated_corpus.extend(generated)
self._save_generated_text(generated_corpus)
reference_corpus = eval_data.get_reference()
result = self.evaluator.evaluate(generated_corpus, reference_corpus)
return result