/
main.py
191 lines (161 loc) · 8.47 KB
/
main.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
190
191
import argparse
import math
import numpy as np
import torch
import torch.nn as nn
import time
import data
import model as lmModel
import os
import hashlib
from utils import batchify, model_save, model_load
from train_utils import evaluate, train
'''
This code is adapted from https://github.com/salesforce/awd-lstm-lm/
'''
###############################################################################
# Training script
###############################################################################
def trainEvalLM(args):
fn = 'corpus.{}.data'.format(hashlib.md5(args.data.encode()).hexdigest())
if os.path.exists(fn):
print('Loading cached dataset...')
corpus = torch.load(fn)
else:
print('Producing dataset...')
corpus = data.Corpus(args.data)
torch.save(corpus, fn)
if torch.cuda.is_available():
args.cuda = True
ntokens = len(corpus.dictionary)
eval_batch_size = 10
train_data = batchify(corpus.train, args.batch_size, args)
val_data = batchify(corpus.valid, eval_batch_size, args)
# Build the model and loss function
model = lmModel.RNNModel(args.model, ntokens, args.emsize, args.nhid, args.nlayers,
args.dropout, args.tied, g=args.g, k=args.k)
criterion = nn.CrossEntropyLoss()
if torch.cuda.is_available():
model = model.cuda()
criterion = criterion.cuda()
#compute network parameters
params = list(model.parameters())
total_params = np.sum([np.prod(p.size()) for p in params])
print('\033[1;32;40mTotal parameters (in million):\033[0m\033[1;31;40m {:0.2f} \033[0m\n'.format(total_params / 1e6, 2))
optimizer = torch.optim.SGD(params, lr=args.lr, weight_decay=args.wdecay)
start_epoch = 1
if args.resume:
print('Resuming model ...')
model, criterion, optimizer, start_epoch = model_load(args.resume)
optimizer.param_groups[0]['lr'] = args.lr
model.dropout = args.dropout
# At any point you can hit Ctrl + C to break out of training early.
try:
#Create folder for saving model and log files
args.saveDir += '_' + args.model
# =====================
if not os.path.isdir(args.saveDir):
os.mkdir(args.saveDir)
save_str = 'nl_' + str(args.nlayers) + '_nh_' + str(args.nhid) + '_g_' + str(args.g) + '_k_' + str(args.k)
args.save = args.saveDir + '/model_' + save_str + '.pt'
logFileLoc = args.saveDir + '/logs_' + save_str + '.txt'
logger = open(logFileLoc, 'w')
logger.write(str(args))
logger.write('\n Total parameters (in million): {:0.2f}'.format(
total_params / 1e6, 2))
logger.write('\n\n')
logger.write("\n%s\t%s\t%s\t%s\t%s" % ('Epoch', 'Loss(Tr)', 'Loss(val)', 'ppl (tr)', 'ppl (val)'))
logger.flush()
best_val_loss = []
stored_loss = 100000000
# Loop over epochs.
for epoch in range(start_epoch, args.epochs + 1):
epoch_start_time = time.time()
train_loss = train(args, model, criterion, optimizer, epoch, train_data, ntokens)
### TRAIN WITH ASGD
if 't0' in optimizer.param_groups[0]:
tmp = {}
for prm in model.parameters():
tmp[prm] = prm.data.clone()
prm.data = optimizer.state[prm]['ax'].clone()
val_loss = evaluate(args, model, criterion, val_data, ntokens, eval_batch_size)
print('-' * 89)
print('| end of epoch {:3d} | time: {:5.2f}s | valid loss {:5.2f} | '
'valid ppl {:8.2f}'.format(
epoch, (time.time() - epoch_start_time), val_loss, math.exp(val_loss)))
print('-' * 89)
logger.write("\n%d\t\t%.4f\t\t%.4f\t\t%.4f\t\t%.4f" % (
epoch, train_loss, val_loss,
math.exp(train_loss), math.exp(val_loss)))
logger.flush()
if val_loss < stored_loss:
model_save(args.save, model, criterion, optimizer, epoch)
print('Saving Averaged (new best validation)')
stored_loss = val_loss
for prm in model.parameters():
prm.data = tmp[prm].clone()
else:
val_loss = evaluate(args, model, criterion, val_data, ntokens, eval_batch_size)
print('-' * 89)
print('| end of epoch {:3d} | time: {:5.2f}s | valid loss {:5.2f} | '
'valid ppl {:8.2f}'.format(
epoch, (time.time() - epoch_start_time), val_loss, math.exp(val_loss)))
print('-' * 89)
logger.write("\n%d\t\t%.4f\t\t%.4f\t\t%.4f\t\t%.4f" % (
epoch, train_loss, val_loss,
math.exp(train_loss), math.exp(val_loss)))
logger.flush()
if val_loss < stored_loss:
model_save(args.save, model, criterion, optimizer, epoch)
print('Saving model (new best validation)')
stored_loss = val_loss
if 't0' not in optimizer.param_groups[0] and (
len(best_val_loss) > args.nonmono and val_loss > min(best_val_loss[:-args.nonmono])):
print('Switching to ASGD')
optimizer = torch.optim.ASGD(model.parameters(), lr=args.lr, t0=0, lambd=0.,
weight_decay=args.wdecay)
best_val_loss.append(val_loss)
except KeyboardInterrupt:
print('-' * 89)
print('Exiting from training early')
if __name__ == '__main__':
parser = argparse.ArgumentParser(description='Langauge model based on PRU')
#=============General hyper-parameters================================================
parser.add_argument('--data', type=str, default='data/penn/', help='location of the data corpus')
parser.add_argument('--seed', type=int, default=1111, help='random seed')
parser.add_argument('--log-interval', type=int, default=200, help='report interval')
parser.add_argument('--saveDir', type=str, default='./results', help='Directory to save the models')
parser.add_argument('--resume', type=str, default='', help='path of model to resume')
#=============Model related hyper-parameters===========================================
parser.add_argument('--model', type=str, default='PRU', help='type of recurrent networks (LSTM or PRU)')
parser.add_argument('--emsize', type=int, default=400, help='size of word embeddings')
parser.add_argument('--nhid', type=int, default=1400, help='number of hidden units per layer')
parser.add_argument('--nlayers', type=int, default=3, help='number of RNN layers')
parser.add_argument('--g', type=int, default=4, help='No. of groups in Grouped Linear Transformation')
parser.add_argument('--k', type=int, default=2, help='No. of pyramidal levels in Pyramidal Transformation')
parser.add_argument('--dropout', type=float, default=0.4, help='dropout applied to layers (0 = no dropout)')
#======================Training related hyper-parameters==============================
parser.add_argument('--lr', type=float, default=30, help='initial learning rate')
parser.add_argument('--clip', type=float, default=0.25, help='gradient clipping')
parser.add_argument('--batch_size', type=int, default=20, help='batch size')
parser.add_argument('--bptt', type=int, default=70, help='sequence length')
parser.add_argument('--epochs', type=int, default=200, help='Max. number of epochs')
parser.add_argument('--nonmono', type=int, default=5, help='non monotonous interval')
parser.add_argument('--alpha', type=float, default=2,
help='alpha L2 regularization on RNN activation (alpha = 0 means no regularization)')
parser.add_argument('--beta', type=float, default=1,
help='beta slowness regularization applied on RNN activiation (beta = 0 means no regularization)')
parser.add_argument('--wdecay', type=float, default=1.2e-6, help='weight decay applied to all weights')
args = parser.parse_args()
args.tied = True
# Set the random seed manually for reproducibility.
np.random.seed(args.seed)
torch.manual_seed(args.seed)
if torch.cuda.is_available():
torch.cuda.manual_seed(args.seed)
if args.model == 'LSTM':
args.g = 1
args.k = 1
print(args)
# train and evaluate on the validation set
trainEvalLM(args)