Style_Transfer/utils.py at main · C-Ritam98/Style_Transfer · 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
from torchtext.vocab import FastText
import os
# embedding_glove = GloVe(name='6B', dim=100)

import torch
from torchtext.legacy.data import LabelField, Field, Dataset, TabularDataset, BucketIterator, Pipeline
from torchtext.vocab import Vocab

import pandas as pd
import dill

import random
import math

device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')

PATH = os.getcwd()

def embedding_loader(type = "fasttext"):

    cache = PATH + "/vector_cache"

    if not os.path.exists(cache):
        os.mkdir(cache)

    if type == "fasttext":
        vectors = FastText(language='en', cache=cache)

    if type == "glove":
        vectors = GloVe(name='6B', cache=cache)


    if type == "word2vec":
        vectors = Word2vec(language='en', cache=cache)

    return vectors


def build_iterators(dataset="yelp",BATCH_SIZE = 32, return_review_object = True):

	if dataset.lower() == "yelp":
		dataset = "Yelp"
	if dataset.lower() == "imdb":
		dataset = "IMDB"
	if dataset.lower() == "amazon":
		dataset = "Amazon"

	REVIEW = Field(sequential=True,
	      tokenize = lambda x: x.split(),
	      use_vocab = True,
	      init_token = '<sos>',
	      eos_token = '<eos>',
	      pad_token = '<pad>',
	      # fix_length = 22,
	      batch_first = True,
	      lower = True)

	LABEL = Field(sequential=False, dtype=torch.int64, batch_first=True, use_vocab=False,\
		   preprocessing=Pipeline(lambda x: int(x)))

	fields = [('review', REVIEW),('label',LABEL)]

	train_data, val_data, test_data = TabularDataset.splits(
		                        path = PATH,
		                        train = './data/'+dataset+'/train.csv',
		                        validation = './data/'+dataset+'/val.csv',
		                        test = './data/'+dataset+'/test.csv',
		                        format = 'csv',
		                        fields = fields,
		                        skip_header = True)

	vectors = embedding_loader("fasttext")

	REVIEW.build_vocab(train_data, val_data, test_data, vectors = vectors, min_freq = 2)
	LABEL.build_vocab(train_data)

	print("Meta Data :")
	print(f"Vocab size: {len(REVIEW.vocab)}")
	print(f"<eos> index: {REVIEW.vocab.stoi['<eos>']}")
	print(f"<sos> index: {REVIEW.vocab.stoi['<sos>']}")
	print(f"<pad> index: {REVIEW.vocab.stoi['<pad>']}")
	print(f"<unk> index: {REVIEW.vocab.stoi['<unk>']}")

	# print("Training data instance:")
	# print(train_data[0].__dict__.keys())
	# print(train_data[21].__dict__.values())

	train_iterator, val_iterator, test_iterator = BucketIterator.splits(
		                                        (train_data, val_data, test_data),
		                                        batch_size = BATCH_SIZE,
		                                        sort_key = lambda x : len(x.review),
		                                        shuffle = False,
		                                        device = device
		                                        )

	if return_review_object:
	  return REVIEW, LABEL, train_iterator, val_iterator, test_iterator
	else:
	  return train_iterator, val_iterator, test_iterator


def remove_redundant_tokens(x):
  return ' '.join([tok for tok in x.split() if tok not in ["<unk>","<sos>","<eos>","<pad>"]])


def Create_data(dataset = "Yelp"):

  print(f"Current directory: {os.getcwd()}")

  PATH = os.getcwd() #"/media/rahul/DATA-2/my_CPTG"

  if dataset.lower() == "yelp":
    dataset = "Yelp"
  if dataset.lower() == "imdb":
    dataset = "IMDB"
  if dataset.lower() == "amazon":
    dataset = "Amazon"

  if os.path.exists(PATH+'/data/'+dataset+'/train.csv') and os.path.exists(PATH+'/data/'+dataset+'/val.csv') and os.path.exists(PATH+'/data/'+dataset+'/test.csv'):
    # use pre-built datasets
    return

  # PATH = "/home/cds-2/Desktop/ri/CPTG/data/"

  with open(PATH+"/data/sentiment.train.1",'r') as f:
    pos_data = f.readlines()[:10000]
    pos_data = [x.split('\n')[0] for x in pos_data]


  with open(PATH+"/data/sentiment.train.0",'r') as f:
    neg_data = f.readlines()[:10000]
    neg_data = [x.split('\n')[0] for x in neg_data]

  print(f"Pos size = {len(pos_data)}, Neg size = {len(neg_data)}")

  df_train = pd.DataFrame({"review":pos_data[:8000]+neg_data[:8000], "label":[1]*8000 + [0]*8000})
  df_val = pd.DataFrame({"review":pos_data[8000:9000]+neg_data[8000:9000], "label":[1]*1000 + [0]*1000})
  df_test = pd.DataFrame({"review":pos_data[9000:]+neg_data[9000:], "label":[1]*1000 + [0]*1000})

  df_train = df_train.sample(frac=1)
  # df_val = df_val.sample(frac=1)
  # df_test = df_test.sample(frac=1)

  PATH = os.getcwd()

  df_train.to_csv("train.csv",index=False)
  df_val.to_csv("val.csv",index=False)
  df_test.to_csv("test.csv",index=False)