-
Notifications
You must be signed in to change notification settings - Fork 0
/
data.py
248 lines (191 loc) · 7.16 KB
/
data.py
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
# -*- coding: utf-8 -*-
"""
Created on Sat Dec 16 14:27:15 2023
@author: VuralBayraklii
"""
from io import open
import unicodedata
import string
import re
import random
import os
import torch
from torch import nn
import torch.nn.functional as F
from torch.nn.utils.rnn import pad_sequence
from torch.utils.data import DataLoader
from torchtext.data.utils import get_tokenizer
from torchtext.vocab import Vocab, build_vocab_from_iterator
from collections import Counter
from tqdm.notebook import tqdm
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from matplotlib import ticker
from tr_tokenizer import tokenize, analyze
from save_pickle_format import SavePickle
MAX_SENTENCE_LENGTH = 20
FILTER_TO_BASIC_PREFIXES = False
SAVE_DIR = os.path.join(".", "models")
ENCODER_EMBEDDING_DIM = 256
ENCODER_HIDDEN_SIZE = 256
DECODER_EMBEDDING_DIM = 256
DECODER_HIDDEN_SIZE = 256
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
with open('tur-eng/tur.txt', encoding='utf-8') as f:
lines = f.read().strip().split('\n')
print(f"{len(lines):,} English-Turkish phrase pairs.\n")
print("~~~~~ Examples: ~~~~~")
for example in random.choices(lines, k=5):
pair = example.split('\t')
print(f"English: {pair[0]}")
print(f"Turkish: {pair[1]}")
print()
def unicodeToAscii(s, language):
s = ''.join(
c for c in unicodedata.normalize('NFD', s)
if unicodedata.category(c) != 'Mn'
)
if language.lower() == 'tr':
turkish_mapping = {'ı': 'i', 'İ': 'I', 'ğ': 'g', 'Ğ': 'G', 'ü': 'u', 'Ü': 'U', 'ş': 's', 'Ş': 'S', 'ö': 'o', 'Ö': 'O', 'ç': 'c', 'Ç': 'C'}
s = ''.join(turkish_mapping.get(c, c) for c in s)
return s
def normalizeString(s):
#s = unicodeToAscii(s.lower().strip(), language)
s = s.lower()
s = re.sub(r"[^a-zA-ZıİÇçğÜüÖöŞş.!?]+", " ", s)
return s
def filterPair(p, max_length, prefixes):
good_length = (len(p[0].split(' ')) < max_length) and (len(p[1].split(' ')) < max_length)
if len(prefixes) == 0:
return good_length
else:
return good_length and p[0].startswith(prefixes)
def filterPairs(pairs, max_length, prefixes=()):
return [pair for pair in pairs if filterPair(pair, max_length, prefixes)]
def prepareData(lines, filter=False, reverse=False, max_length=10, prefixes=()):
pairs = [[normalizeString(s) for s in l.split('\t')] for l in lines]
print(f"Given {len(pairs):,} sentence pairs.")
if filter:
pairs = filterPairs(pairs, max_length=max_length, prefixes=prefixes)
print(f"After filtering, {len(pairs):,} remain.")
return pairs
basic_prefixes = (
"i am ", "i m ",
"he is", "he s ",
"she is", "she s ",
"you are", "you re ",
"we are", "we re ",
"they are", "they re ",
'are you', 'am i ',
'were you', 'was i ',
'where are', 'where is',
'what is', 'what are'
)
pairs = prepareData(lines,
filter=True,
max_length=MAX_SENTENCE_LENGTH,
prefixes=basic_prefixes if FILTER_TO_BASIC_PREFIXES else ())
en_tokenizer = get_tokenizer('spacy', language='en_core_web_sm')
def tr_tokenizerr(text):
words = []
tokens = tokenize(text)
for word in tokens:
token = word.token
words.append(token)
return words
SPECIALS = ['<unk>', '<pad>', '<bos>', '<eos>']
en_list = []
tr_list = []
en_counter = Counter()
tr_counter = Counter()
en_lengths = []
tr_lengths = []
for idx, (en, tr, info) in enumerate(pairs):
en_toks = en_tokenizer(en)
print(en_toks)
tr_toks = tr_tokenizerr(tr)
en_list += [en_toks]
tr_list += [tr_toks]
en_counter.update(en_toks)
tr_counter.update(tr_toks)
en_lengths.append(len(en_toks))
tr_lengths.append(len(tr_toks))
if idx % 1000 == 0:
print(idx)
en_vocab = build_vocab_from_iterator(en_list, specials=SPECIALS)
tr_vocab = build_vocab_from_iterator(tr_list, specials=SPECIALS)
VALID_PCT = 0.1
TEST_PCT = 0.1
train_data = []
valid_data = []
test_data = []
random.seed(6547)
for (en, tr, info) in pairs:
en_tensor_ = torch.tensor([en_vocab[token] for token in en_tokenizer(en)])
tr_tensor_ = torch.tensor([tr_vocab[token] for token in tr_tokenizerr(tr)])
random_draw = random.random()
if random_draw <= VALID_PCT:
valid_data.append((en_tensor_, tr_tensor_))
elif random_draw <= VALID_PCT + TEST_PCT:
test_data.append((en_tensor_, tr_tensor_))
else:
train_data.append((en_tensor_, tr_tensor_))
print(f"""
Training pairs: {len(train_data):,}
Validation pairs: {len(valid_data):,}
Test pairs: {len(test_data):,}""")
PAD_IDX = en_vocab['<pad>']
BOS_IDX = en_vocab['<bos>']
EOS_IDX = en_vocab['<eos>']
for en_id, tr_id in zip(en_vocab.lookup_indices(SPECIALS), tr_vocab.lookup_indices(SPECIALS)):
assert en_id == tr_id
def generate_batch(data_batch):
'''
Prepare English and French examples for batch-friendly modeling by appending
BOS/EOS tokens to each, stacking the tensors, and filling trailing spaces of
shorter sentences with the <pad> token. To be used as the collate_fn in the
English-to-French DataLoader.
Input:
- data_batch, an iterable of (English, French) tuples from the datasets
created above
Outputs
- en_batch: a (max length X batch size) tensor of English token IDs
- fr_batch: a (max length X batch size) tensor of French token IDs
'''
en_batch, tr_batch = [], []
for (en_item, tr_item) in data_batch:
en_batch.append(torch.cat([torch.tensor([BOS_IDX]), en_item, torch.tensor([EOS_IDX])], dim=0))
tr_batch.append(torch.cat([torch.tensor([BOS_IDX]), tr_item, torch.tensor([EOS_IDX])], dim=0))
en_batch = pad_sequence(en_batch, padding_value=PAD_IDX, batch_first=False)
tr_batch = pad_sequence(tr_batch, padding_value=PAD_IDX, batch_first=False)
return en_batch, tr_batch
BATCH_SIZE = 16
train_iter = DataLoader(train_data, batch_size=BATCH_SIZE, shuffle=True, collate_fn=generate_batch)
valid_iter = DataLoader(valid_data, batch_size=BATCH_SIZE, shuffle=False, collate_fn=generate_batch)
test_iter = DataLoader(test_data, batch_size=BATCH_SIZE, shuffle=False, collate_fn=generate_batch)
for i, (en_id, tr_id) in enumerate(train_iter):
print('English:', ' '.join([en_vocab.lookup_token(idx) for idx in en_id[:, 0]]))
print('Turkish:', ' '.join([tr_vocab.lookup_token(idx) for idx in tr_id[:, 0]]))
if i == 4:
break
else:
print()
save_pairs = SavePickle("pairs", pairs)
save_pairs.save_process()
save_en_vocab = SavePickle("en_vocab", en_vocab)
save_en_vocab.save_process()
save_tr_vocab =SavePickle("tr_vocab", tr_vocab)
save_tr_vocab.save_process()
save_train_data = SavePickle("train_data", train_data)
save_train_data.save_process()
save_valid_data = SavePickle("valid_data", valid_data)
save_valid_data.save_process()
save_test_data = SavePickle("test_data", test_data)
save_test_data.save_process()
save_train_iter = SavePickle("train_iter", train_iter)
save_train_iter.save_process()
save_test_iter = SavePickle("test_iter", test_iter)
save_test_iter.save_process()
save_valid_iter = SavePickle("valid_iter", valid_iter)
save_valid_iter.save_process()