Context2Vector/word2vec.py at main · binbinjiang/Context2Vector · 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
from input_data import InputData
import argparse
from model import SkipGramModel
from torch.autograd import Variable
import torch
import torch.optim as optim
from torch.nn.utils import clip_grad_norm_
from tqdm import tqdm
import random
import numpy as np
import os

class Word2Vec:
    def __init__(self, input_file_name, output_file_name , output_model_dir, output_dir, emb_dimension=300, batch_size=50,
                 window_size=5, iteration=5, initial_lr=0.025, neg_num=5, min_count=5):

        self.data = InputData(input_file_name, min_count)
        self.output_file_name = output_file_name
        self.output_model_dir = output_model_dir
        self.output_dir = output_dir

        self.emb_size = len(self.data.word2id)
        self.emb_dimension = emb_dimension
        self.batch_size = batch_size
        self.window_size = window_size
        self.iteration = iteration
        self.initial_lr = initial_lr
        self.neg_num = neg_num
        self.skip_gram_model = SkipGramModel(self.emb_size, self.emb_dimension)
        self.skip_gram_model.cuda()
        self.optimizer = optim.SGD(self.skip_gram_model.parameters(), lr=self.initial_lr)

    def train(self):

        START = 0
        if not os.path.exists(self.output_dir+"/log_final.txt"):
            if os.path.exists(self.output_dir+"/log_temp.txt"):
                with open(self.output_dir+"/log_temp.txt","r") as f:
                    log_dic0 = eval(f.readlines()[0])
                    START = log_dic0['laststep']
                    restore_checkpoint = log_dic0["lastcheckpoint"]
                    self.skip_gram_model.load_state_dict(torch.load(restore_checkpoint))
                    print(f"Restore from the checkpoint:{restore_checkpoint}! Last Step:{START}")

        else:
            print("!!!! Please remove the log_final.txt !!!")
            exit(0)

        pair_count = self.data.evaluate_pair_count(self.window_size)
        batch_count = self.iteration * pair_count / self.batch_size
        process_bar = tqdm(range(int(batch_count-START)))

        count = int(batch_count) // 30
        for i0 in process_bar:
            i= i0 + START
            pos_pairs = self.data.get_batch_pairs(self.batch_size, self.window_size)
            neg_v = self.data.get_neg_v_neg_sampling(pos_pairs, self.neg_num)

            pos_u = [pair[0] for pair in pos_pairs]
            pos_v = [pair[1] for pair in pos_pairs]
            u_bert = [pair[2] for pair in pos_pairs]

            pos_u = Variable(torch.LongTensor(pos_u)).cuda()
            pos_v = Variable(torch.LongTensor(pos_v)).cuda()
            neg_v = Variable(torch.LongTensor(neg_v)).cuda()

            self.optimizer.zero_grad()
            loss = self.skip_gram_model.forward(pos_u, pos_v, neg_v, u_bert,)
            loss.backward()
            clip_grad_norm_(self.skip_gram_model.parameters(), max_norm=5)
            self.optimizer.step()

            process_bar.set_description("Loss: %0.8f, lr: %0.6f" %
                                        (loss.item(), self.optimizer.param_groups[0]['lr']))

            if i * self.batch_size % 100000 == 0:
                lr = self.initial_lr * (1.0 - 1.0 * i / batch_count)
                for param_group in self.optimizer.param_groups:
                    param_group['lr'] = lr

            if i != 0 and i % count == 0:
                log_dic = dict()
                self.skip_gram_model.save_embedding(self.data.id2word, self.output_file_name + "data.emb300." +str(i) +'_'+ str(self.initial_lr))
                torch.save(self.skip_gram_model.state_dict(), self.output_model_dir+"checkpoint_" + str(i) +".pt")
                log_dic["laststep"] = i
                log_dic["lastcheckpoint"] = self.output_model_dir+"checkpoint_" + str(i) +".pt"
                log_dic["lastembedding"] = self.output_file_name + "data.emb300." +str(i) +'_'+ str(self.initial_lr)

                with open(self.output_dir+"/log_temp.txt","w") as f_log:
                    f_log.writelines(str(log_dic))
                    # print(str(log_dic))

        # Save the final embedding
        self.skip_gram_model.save_embedding(self.data.id2word, self.output_dir + "data.emb300." +'_final')
        torch.save(self.skip_gram_model.state_dict(), self.output_dir+"checkpoint_final.pt")

        log_dic_final = dict()
        log_dic_final["finalstep"] = i
        log_dic_final["finalcheckpoint"] = self.output_dir+"checkpoint_final.pt"
        log_dic_final["finalembedding"] = self.output_dir + "data.emb300." +'_final'

        with open(self.output_dir+"/log_final.txt","w") as f_log:
            f_log.writelines(log_dic_final)


if __name__ == '__main__':
    parser = argparse.ArgumentParser()
    parser.add_argument("--input_file", default="../data_common/data_new.txt", type=str)
    parser.add_argument("--output_dir_root", default="../output/bert2vec_mean_contexts_2/", type=str)
    parser.add_argument("--emb_dim", default=300, type=int)
    parser.add_argument("--batch_size", default=1000, type=int)
    parser.add_argument("--window_size", default=5, type=int)
    parser.add_argument("--iteration", default=1, type=int)
    parser.add_argument("--initial_lr", default=0.08, type=float)
    parser.add_argument("--neg_num", default=5, type=int)
    parser.add_argument("--min_count", default=5, type=int)
    parser.add_argument('--seed', type=int, default=12345)
    args = parser.parse_args()

    torch.manual_seed(args.seed)
    random.seed(args.seed)
    torch.cuda.manual_seed(args.seed)
    np.random.seed(args.seed)

    output_dir = args.output_dir_root + "batch_"+str(args.batch_size)+"/"

    output_file = output_dir +"embeddings/"
    output_model_dir = output_dir +"checkpoints/"
    if not os.path.exists(output_file):
        os.makedirs(output_file)
    if not os.path.exists(output_model_dir):
        os.makedirs(output_model_dir)

    w2v = Word2Vec(args.input_file, output_file, output_model_dir, output_dir, args.emb_dim, args.batch_size, args.window_size, args.iteration,
                   args.initial_lr, args.min_count)
    w2v.train()