im2code/train.py at master · Destinia/im2code · GitHub

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
from options.train_options import TrainOptions
from dataloader import UI2codeDataloader

import numpy as np
import math
import torch
import torch.nn as nn
import pickle
import time
import os
from tqdm import tqdm
from torch.autograd import Variable
from torch.nn.utils import clip_grad_norm
from torch.nn.utils.rnn import pack_padded_sequence
from models.model import EncoderCNN, EncoderRNN, AttentionDecoder
import util.util as utils
from eval_utils import wordErrorRate

try:
    import tensorflow as tf
except ImportError:
    print("Tensorflow not installed; No tensorboard logging.")
    tf = None

def add_summary_value(writer, key, value, iteration):
    summary = tf.Summary(value=[tf.Summary.Value(tag=key, simple_value=value)])
    writer.add_summary(summary, iteration)

# def validate(model, loader):


def train(opt):
    data_loader = UI2codeDataloader(opt)
    opt.vocab = data_loader.get_vocab()
    opt.rev_vocab = [t[0] for t in list(sorted(opt.vocab.items(), key=lambda x: x[1]))]
    opt.target_vocab_size = len(opt.vocab)
    val_data_loader = UI2codeDataloader(opt, phase='val')
    dataset = data_loader.load_data()
    dataset_size = len(data_loader)
    print('#training images = %d' % dataset_size)
    tf_summary_writer = tf and tf.summary.FileWriter(opt.checkpoint_path)

    # model = create_model(opt)
    # visualizer = Visualizer(opt)
    encoderCNN = EncoderCNN(opt)
    encoderRNN = EncoderRNN(opt)
    decoder = AttentionDecoder(opt)
    if torch.cuda.is_available():
        encoderCNN.cuda()
        encoderRNN.cuda()
        decoder.cuda()

    criterion = utils.LanguageModelCriterion()
    params = list(decoder.parameters()) + \
        list(encoderRNN.parameters()) + list(encoderCNN.parameters())
    opt.current_lr = opt.learning_rate
    optimizer = torch.optim.SGD(params, lr=opt.current_lr)
    total_steps = 0
    infos = {}
    histories = {}
    best_val_loss = math.inf

    for epoch in range(opt.num_epochs):
        epoch_start_time = time.time()
        iter_data_time = time.time()
        epoch_iter = 0
        print('Epoch [%d/%d]' % (epoch + 1, opt.num_epochs))
        pbar = tqdm(data_loader)
        for (images, captions, masks) in pbar:
            iter_start_time = time.time()
            total_steps += opt.batch_size
            images = Variable(images, requires_grad=False).cuda()
            captions = Variable(captions, requires_grad=False).cuda()
            masks = Variable(masks, requires_grad=False).cuda()
            # targets = pack_padded_sequence(captions, lengths, batch_first=True)[0]
            features = encoderCNN(images)
            encoded_features = encoderRNN(features)
            outputs = decoder(encoded_features, captions)
            loss = criterion(outputs[:, :-1], captions[:,1:], masks[:,1:])
            optimizer.zero_grad()
            loss.backward()
            clip_grad_norm(params, opt.norm_grad_clip)
            optimizer.step()
            train_loss = loss.data[0]
            pbar.set_description('Loss: %.4f'
                                 % (loss.data[0]))
            pbar.refresh()
            if total_steps % opt.print_freq == 0:
                # print('Epoch [%d/%d], Step [%d/%d], Loss: %.4f'
                #       % (epoch+1, opt.num_epochs, total_steps, dataset_size,
                #          loss.data[0]))
                if tf is not None:
                    add_summary_value(tf_summary_writer, 'train_loss', train_loss, total_steps)
                    add_summary_value(tf_summary_writer, 'learning_rate', opt.current_lr, total_steps)
                    # add_summary_value(tf_summary_writer, 'scheduled_sampling_prob', model.ss_prob, total_steps)
                    tf_summary_writer.flush()
            ## validation
            # if total_steps % opt.save_latest_freq == 0:
            #     print('saving the latest model (epoch %d, total_steps %d)' %
            #         (epoch, total_steps))
            #     torch.save(decoder.state_dict(), os.path.join(opt.checkpoint_path, 'decoder-%d-%d.pkl' % (epoch + 1, total_steps)))
            #     torch.save(encoderRNN.state_dict(), os.path.join(opt.checkpoint_path, 'encoder-rnn-%d-%d.pkl' % (epoch + 1, total_steps)))
            #     torch.save(encoderCNN.state_dict(), os.path.join(opt.checkpoint_path, 'encoder-cnn-%d-%d.pkl' % (epoch + 1, total_steps)))

        print('the end of epoch %d, iters %d' % (epoch, total_steps))
        val_accuracy = []
        for (images, captions, masks) in val_data_loader:
            images = Variable(images, requires_grad=False).cuda()
            captions = Variable(captions, requires_grad=False).cuda()
            masks = Variable(masks, requires_grad=False).cuda()
            # targets = pack_padded_sequence(captions, lengths, batch_first=True)[0]
            features = encoderCNN(images)
            encoded_features = encoderRNN(features)
            outputs = decoder(encoded_features, captions)
            beam_output, beam_score = decoder.decode_beam(encoded_features)
            # print(beam_output, beam_score)
            # loss = criterion(beam_output[:, :-1], captions[:,1:], masks[:,1:])
            accuracy = wordErrorRate(beam_output[:, :-1], captions.data[:, 1:], opt.rev_vocab)
            # val_losses.append(loss.data[0])
            val_accuracy.append(accuracy)
        cur_val_loss = sum(val_accuracy) / len(val_accuracy)
        print('validation loss: %.4f\nBest validation loss: %.4f'% (cur_val_loss, best_val_loss))
        add_summary_value(tf_summary_writer, 'val_accuracy', cur_val_loss, epoch)
        if cur_val_loss < best_val_loss:
            print('save best model at Epoch: %d'%(epoch))
            torch.save(decoder.state_dict(), os.path.join(opt.checkpoint_path, 'decoder-%d-%d.pkl' % (epoch + 1, total_steps)))
            torch.save(encoderRNN.state_dict(), os.path.join(opt.checkpoint_path, 'encoder-rnn-%d-%d.pkl' % (epoch + 1, total_steps)))
            torch.save(encoderCNN.state_dict(), os.path.join(opt.checkpoint_path, 'encoder-cnn-%d-%d.pkl' % (epoch + 1, total_steps)))

            ## update learning rate
            opt.current_lr = opt.current_lr * opt.lr_decay
            print('update learning rate: %.4f' % (opt.current_lr))
            utils.set_lr(optimizer, opt.current_lr)


        torch.save(decoder.state_dict(),
                   os.path.join(opt.checkpoint_path,
                                'decoder-%d.pkl' % (epoch + 1)))
        torch.save(encoderRNN.state_dict(),
                   os.path.join(opt.checkpoint_path,
                                'encoder-rnn-%d.pkl' % (epoch + 1)))
        torch.save(encoderCNN.state_dict(),
                   os.path.join(opt.checkpoint_path,
                                'encoder-cnn-%d.pkl' % (epoch + 1)))

        print('End of epoch %d \t Time Taken: %d sec' %
            (epoch, time.time() - epoch_start_time))

if __name__ == '__main__':
    opt = TrainOptions().parse()
    train(opt)