File size: 10,519 Bytes
7900c16
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
"""
This script provides an example to wrap TencentPretrain for SimCSE.
"""
import sys
import os
import random
import argparse
import math
import scipy.stats
import torch
import torch.nn as nn
import numpy as np

tencentpretrain_dir = os.path.abspath(os.path.join(os.path.dirname(__file__), ".."))
sys.path.append(tencentpretrain_dir)

from tencentpretrain.embeddings import *
from tencentpretrain.encoders import *
from tencentpretrain.targets import *
from tencentpretrain.utils.vocab import Vocab
from tencentpretrain.utils.constants import *
from tencentpretrain.utils import *
from tencentpretrain.utils.optimizers import *
from tencentpretrain.utils.config import load_hyperparam
from tencentpretrain.utils.seed import set_seed
from tencentpretrain.utils.logging import init_logger
from tencentpretrain.model_saver import save_model
from tencentpretrain.opts import finetune_opts, tokenizer_opts
from finetune.run_classifier import count_labels_num, build_optimizer, load_or_initialize_parameters
from finetune.run_classifier_siamese import batch_loader


class SimCSE(nn.Module):
    def __init__(self, args):
        super(SimCSE, self).__init__()
        self.embedding = Embedding(args)
        for embedding_name in args.embedding:
            tmp_emb = str2embedding[embedding_name](args, len(args.tokenizer.vocab))
            self.embedding.update(tmp_emb, embedding_name)
        self.encoder = str2encoder[args.encoder](args)

        self.pooling_type = args.pooling

    def forward(self, src, seg):
        """
        Args:
            src: [batch_size x seq_length]
            tgt: [batch_size]
            seg: [batch_size x seq_length]
        """
        # Embedding.
        emb_0 = self.embedding(src[0], seg[0])
        emb_1 = self.embedding(src[1], seg[1])
        # Encoder.
        output_0 = self.encoder(emb_0, seg[0])
        output_1 = self.encoder(emb_1, seg[1])
        # Target.
        features_0 = self.pooling(output_0, seg[0], self.pooling_type)
        features_1 = self.pooling(output_1, seg[1], self.pooling_type)

        return features_0, features_1

    def pooling(self, memory_bank, seg, pooling_type):
        seg = torch.unsqueeze(seg, dim=-1).type(torch.float)
        memory_bank = memory_bank * seg
        if pooling_type == "mean":
            features = torch.sum(memory_bank, dim=1)
            features = torch.div(features, torch.sum(seg, dim=1))
        elif pooling_type == "last":
            features = memory_bank[torch.arange(memory_bank.shape[0]), torch.squeeze(torch.sum(seg, dim=1).type(torch.int64) - 1), :]
        elif pooling_type == "max":
            features = torch.max(memory_bank + (seg - 1) * sys.maxsize, dim=1)[0]
        else:
            features = memory_bank[:, 0, :]
        return features


def read_dataset(args, path):
    dataset, columns = [], {}
    with open(path, mode="r", encoding="utf-8") as f:
        for line_id, line in enumerate(f):
            if line_id == 0:
                for i, column_name in enumerate(line.rstrip("\r\n").split("\t")):
                    columns[column_name] = i
                continue
            line = line.rstrip("\r\n").split("\t")

            if "text_b" in columns:
                text_a, text_b = line[columns["text_a"]], line[columns["text_b"]]
            else:
                text_a = line[columns["text_a"]]
                text_b = text_a
            src_a = args.tokenizer.convert_tokens_to_ids([CLS_TOKEN] + args.tokenizer.tokenize(text_a) + [SEP_TOKEN])
            src_b = args.tokenizer.convert_tokens_to_ids([CLS_TOKEN] + args.tokenizer.tokenize(text_b) + [SEP_TOKEN])
            seg_a = [1] * len(src_a)
            seg_b = [1] * len(src_b)
            PAD_ID = args.tokenizer.convert_tokens_to_ids([PAD_TOKEN])[0]

            if len(src_a) >= args.seq_length:
                src_a = src_a[:args.seq_length]
                seg_a = seg_a[:args.seq_length]
            while len(src_a) < args.seq_length:
                src_a.append(PAD_ID)
                seg_a.append(0)

            if len(src_b) >= args.seq_length:
                src_b = src_b[:args.seq_length]
                seg_b = seg_b[:args.seq_length]
            while len(src_b) < args.seq_length:
                src_b.append(PAD_ID)
                seg_b.append(0)

            if "label" in columns:
                tgt = float(line[columns["label"]])
                dataset.append(((src_a, src_b), tgt, (seg_a, seg_b)))
            else:
                dataset.append(((src_a, src_a), -1, (seg_a, seg_a)))
    return dataset


def evaluate(args, dataset):
    src_a = torch.LongTensor([example[0][0] for example in dataset])
    src_b = torch.LongTensor([example[0][1] for example in dataset])
    tgt = torch.FloatTensor([example[1] for example in dataset])
    seg_a = torch.LongTensor([example[2][0] for example in dataset])
    seg_b = torch.LongTensor([example[2][1] for example in dataset])

    all_similarities = []
    batch_size = args.batch_size
    args.model.eval()

    for i, (src_batch, tgt_batch, seg_batch) in enumerate(batch_loader(batch_size, (src_a, src_b), tgt, (seg_a, seg_b))):

        src_a_batch, src_b_batch = src_batch
        seg_a_batch, seg_b_batch = seg_batch

        src_a_batch = src_a_batch.to(args.device)
        src_b_batch = src_b_batch.to(args.device)

        seg_a_batch = seg_a_batch.to(args.device)
        seg_b_batch = seg_b_batch.to(args.device)

        with torch.no_grad():
            features_0, features_1 = args.model((src_a_batch, src_b_batch), (seg_a_batch, seg_b_batch))
        similarity_matrix = similarity(features_0, features_1, 1)

        for j in range(similarity_matrix.size(0)):
            all_similarities.append(similarity_matrix[j][j].item())

    corrcoef = scipy.stats.spearmanr(tgt, all_similarities).correlation
    args.logger.info("Spearman's correlation: {:.4f}".format(corrcoef))
    return corrcoef


def similarity(x, y, temperature):
    x = x / x.norm(dim=-1, keepdim=True)
    y = y / y.norm(dim=-1, keepdim=True)
    return torch.matmul(x, y.transpose(-2, -1)) / temperature


def main():
    parser = argparse.ArgumentParser(formatter_class=argparse.ArgumentDefaultsHelpFormatter)

    finetune_opts(parser)

    tokenizer_opts(parser)

    parser.add_argument("--temperature", type=float, default=0.05)
    parser.add_argument("--eval_steps", type=int, default=200, help="Evaluate frequency.")

    args = parser.parse_args()

    # Load the hyperparameters from the config file.
    args = load_hyperparam(args)

    set_seed(args.seed)

    # Build tokenizer.
    args.tokenizer = str2tokenizer[args.tokenizer](args)

    # Build classification model.
    model = SimCSE(args)

    # Load or initialize parameters.
    load_or_initialize_parameters(args, model)

    # Get logger.
    args.logger = init_logger(args)

    args.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
    model = model.to(args.device)

    # Training phase.
    trainset = read_dataset(args, args.train_path)
    instances_num = len(trainset)
    batch_size = args.batch_size

    args.train_steps = int(instances_num * args.epochs_num / batch_size) + 1

    args.logger.info("Batch size: {}".format(batch_size))
    args.logger.info("The number of training instances: {}".format(instances_num))

    optimizer, scheduler = build_optimizer(args, model)

    if args.fp16:
        try:
            from apex import amp
        except ImportError:
            raise ImportError("Please install apex from https://www.github.com/nvidia/apex to use fp16 training.")
        model, optimizer = amp.initialize(model, optimizer, opt_level=args.fp16_opt_level)
        args.amp = amp

    if torch.cuda.device_count() > 1:
        args.logger.info("{} GPUs are available. Let's use them.".format(torch.cuda.device_count()))
        model = torch.nn.DataParallel(model)
    args.model = model

    total_loss, result, best_result = 0.0, 0.0, 0.0

    args.logger.info("Start training.")
    for epoch in range(1, args.epochs_num + 1):
        random.shuffle(trainset)
        src_a = torch.LongTensor([example[0][0] for example in trainset])
        src_b = torch.LongTensor([example[0][1] for example in trainset])
        tgt = torch.FloatTensor([example[1] for example in trainset])
        seg_a = torch.LongTensor([example[2][0] for example in trainset])
        seg_b = torch.LongTensor([example[2][1] for example in trainset])

        model.train()
        for i, (src_batch, tgt_batch, seg_batch) in enumerate(batch_loader(batch_size, (src_a, src_b), tgt, (seg_a, seg_b))):
            model.zero_grad()

            src_a_batch, src_b_batch = src_batch
            seg_a_batch, seg_b_batch = seg_batch

            src_a_batch = src_a_batch.to(args.device)
            src_b_batch = src_b_batch.to(args.device)

            seg_a_batch = seg_a_batch.to(args.device)
            seg_b_batch = seg_b_batch.to(args.device)

            features_0, features_1 = model((src_a_batch, src_b_batch), (seg_a_batch, seg_b_batch))

            similarity_matrix = similarity(features_0, features_1, args.temperature)
            tgt_batch = torch.arange(similarity_matrix.size(0), device=similarity_matrix.device, dtype=torch.long)
            loss = nn.CrossEntropyLoss()(similarity_matrix, tgt_batch)

            if args.fp16:
                with args.amp.scale_loss(loss, optimizer) as scaled_loss:
                    scaled_loss.backward()
            else:
                loss.backward()

            optimizer.step()
            scheduler.step()

            total_loss += loss.item()
            if (i + 1) % args.report_steps == 0:
                args.logger.info("Epoch id: {}, Training steps: {}, Avg loss: {:.3f}"
                                 .format(epoch, i + 1, total_loss / args.report_steps))
                total_loss = 0.0

            if (i + 1) % args.eval_steps == 0 or (i + 1) == math.ceil(instances_num / batch_size):
                result = evaluate(args, read_dataset(args, args.dev_path))
                args.logger.info("Epoch id: {}, Training steps: {}, Evaluate result: {}, Best result: {}"
                                 .format(epoch, i + 1, result, best_result))
                if result > best_result:
                    best_result = result
                    save_model(model, args.output_model_path)
                    args.logger.info("It is the best model until now. Save it to {}".format(args.output_model_path))


if __name__ == "__main__":
    main()