CANet/process.py at main · icantnamemyself/CANet · 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
154
155
156
157
158
159
160
161
162
163
164
165
166
167
import time
import torch
import torch.nn as nn
from tqdm import tqdm
from loss import CE
from torch.optim.lr_scheduler import LambdaLR


class Trainer():
    def __init__(self, args, model, train_loader, val_loader, test_loader):
        self.args = args
        self.device = args.device
        print(self.device)
        self.model = model.to(torch.device(self.device))

        self.train_loader = train_loader
        self.val_loader = val_loader
        self.test_loader = test_loader
        self.lr_decay = args.lr_decay_rate
        self.lr_decay_steps = args.lr_decay_steps

        self.cr = CE(self.model)
        self.test_loss = nn.CrossEntropyLoss(ignore_index=0)

        self.num_epoch = args.num_epoch
        self.epoch = 0
        self.metric_ks = args.metric_ks
        self.eval_per_steps = args.eval_per_steps
        self.save_path = args.save_path
        if self.num_epoch:
            self.result_file = open(self.save_path + '/result.txt', 'w')
            self.result_file.close()

        self.step = 0
        self.metric = args.best_metric
        self.best_metric = -1e9

        self.labels = torch.zeros(512, self.args.num_item + 1).to(self.device)

    def train(self):
        self.optimizer = torch.optim.Adam(self.model.parameters(), lr=self.args.lr, weight_decay=self.args.weight_decay)
        self.scheduler = LambdaLR(self.optimizer, lr_lambda=lambda step: self.lr_decay ** step, verbose=True)
        for epoch in range(self.num_epoch):
            loss_epoch, time_cost= self._train_one_epoch()
            self.result_file = open(self.save_path + '/result.txt', 'a+')
            print('Model train epoch:{0},loss:{1},training_time:{2}'.format(epoch + 1, loss_epoch,
                                                                                             time_cost))
            print('Model train epoch:{0},loss:{1},training_time:{2}'.format(epoch + 1, loss_epoch,
                                                                                             time_cost),
                  file=self.result_file)
            self.result_file.close()

    def _train_one_epoch(self):
        self.epoch += 1
        optim = self.optimizer
        t0 = time.perf_counter()
        self.model.train()
        tqdm_dataloader = tqdm(self.train_loader)

        loss_sum = 0
        for idx, batch in enumerate(tqdm_dataloader):
            batch = [x.to(self.device) for x in batch]

            optim.zero_grad()
            loss = self.cr.compute(batch)
            loss_sum += loss.cpu().item()
            loss.backward()
            optim.step()

            self.step += 1
            if self.step % self.lr_decay_steps == 0:
                self.scheduler.step()
                self.cr.step()
                metric = {}
                for mode in ['test']:
                    depth, metric[mode] = self.eval_model(mode)
                print(metric, depth)
                self.result_file = open(self.save_path + '/result.txt', 'a+')
                print('step{0}'.format(self.step), file=self.result_file)
                print(metric, file=self.result_file)
                print(depth, file=self.result_file)
                self.result_file.close()
                if metric['test'][self.metric] > self.best_metric:
                    torch.save(self.model.state_dict(), self.save_path + '/model.pkl')
                    self.result_file = open(self.save_path + '/result.txt', 'a+')
                    print('saving model of step{0}'.format(self.step), file=self.result_file)
                    self.result_file.close()
                    self.best_metric = metric['test'][self.metric]
                self.model.train()

        return loss_sum / idx, time.perf_counter() - t0

    def eval_model(self, mode):
        self.model.eval()
        tqdm_data_loader = tqdm(self.val_loader) if mode == 'val' else tqdm(self.test_loader)
        metrics = {}
        depth_dict = {i: 0 for i in range(self.model.config_num)}

        with torch.no_grad():
            for idx, batch in enumerate(tqdm_data_loader):
                batch = [x.to(self.device) for x in batch]
                metrics_batch = self.compute_metrics(batch)
                if len(metrics_batch) == 2:
                    metrics_batch, index = metrics_batch
                    for i in index:
                        depth_dict[i.item()] += 1

                for k, v in metrics_batch.items():
                    if not metrics.__contains__(k):
                        metrics[k] = v
                    else:
                        metrics[k] += v

        for k, v in metrics.items():
            metrics[k] = v / (idx + 1)
        return depth_dict, metrics

    def compute_metrics(self, batch):
        seqs, answers = batch
        ret = self.model(seqs, 'test')

        scores, choices = ret
        index = choices.cpu().int()
        scores = scores[:, -1, :]
        test_loss = self.test_loss(scores.view(-1, scores.shape[-1]), answers.view(-1))
        row = []
        col = []
        seqs = seqs.tolist()
        answers = answers.tolist()
        for i in range(len(answers)):
            seq = list(set(seqs[i] + answers[i]))
            seq.remove(answers[i][0])
            if self.args.num_item + 1 in seq:
                seq.remove(self.args.num_item + 1)
            row += [i] * len(seq)
            col += seq
            self.labels[i][answers[i]] = 1
        scores[row, col] = -1e9
        metrics = recalls_and_ndcgs_for_ks(scores, self.labels[:len(seqs)], self.metric_ks)
        metrics['test_loss'] = test_loss.item()
        self.labels[self.labels == 1] = 0
        return metrics, index


def recalls_and_ndcgs_for_ks(scores, labels, ks):
    metrics = {}

    answer_count = labels.sum(1)

    labels_float = labels.float()
    rank = (-scores).argsort(dim=1)
    cut = rank
    for k in sorted(ks, reverse=True):
        cut = cut[:, :k]
        hits = labels_float.gather(1, cut)
        metrics['Recall@%d' % k] = \
            (hits.sum(1) / torch.min(torch.Tensor([k]).to(labels.device),
                                     labels.sum(1).float())).mean().cpu().item()

        position = torch.arange(2, 2 + k)
        weights = 1 / torch.log2(position.float())
        dcg = (hits * weights.to(hits.device)).sum(1)
        idcg = torch.Tensor([weights[:min(int(n), k)].sum() for n in answer_count]).to(dcg.device)
        ndcg = (dcg / idcg).mean()
        metrics['NDCG@%d' % k] = ndcg.cpu().item()

    return metrics