/
embed.py
389 lines (326 loc) · 14.3 KB
/
embed.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
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
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
#!/usr/bin/python3
# Copyright (c) Facebook, Inc. and its affiliates.
# All rights reserved.
#
# This source code is licensed under the BSD-style license found in the
# LICENSE file in the root directory of this source tree.
#
# LASER Language-Agnostic SEntence Representations
# is a toolkit to calculate multilingual sentence embeddings
# and to use them for document classification, bitext filtering
# and mining
#
# --------------------------------------------------------
#
# Tool to calculate to embed a text file
# The functions can be also imported into another Python code
import re
import os
import tempfile
import sys
import time
import argparse
import numpy as np
from collections import namedtuple
import torch
import torch.nn as nn
# get environment
assert os.environ.get('LASER'), 'Please set the enviornment variable LASER'
LASER = os.environ['LASER']
sys.path.append(LASER + '/source/lib')
from text_processing import Token, BPEfastApply
SPACE_NORMALIZER = re.compile("\s+")
Batch = namedtuple('Batch', 'srcs tokens lengths')
def buffered_read(fp, buffer_size):
buffer = []
for src_str in fp:
buffer.append(src_str.strip())
if len(buffer) >= buffer_size:
yield buffer
buffer = []
if len(buffer) > 0:
yield buffer
def buffered_arange(max):
if not hasattr(buffered_arange, 'buf'):
buffered_arange.buf = torch.LongTensor()
if max > buffered_arange.buf.numel():
torch.arange(max, out=buffered_arange.buf)
return buffered_arange.buf[:max]
# TODO Do proper padding from the beginning
def convert_padding_direction(src_tokens, padding_idx, right_to_left=False, left_to_right=False):
assert right_to_left ^ left_to_right
pad_mask = src_tokens.eq(padding_idx)
if not pad_mask.any():
# no padding, return early
return src_tokens
if left_to_right and not pad_mask[:, 0].any():
# already right padded
return src_tokens
if right_to_left and not pad_mask[:, -1].any():
# already left padded
return src_tokens
max_len = src_tokens.size(1)
range = buffered_arange(max_len).type_as(src_tokens).expand_as(src_tokens)
num_pads = pad_mask.long().sum(dim=1, keepdim=True)
if right_to_left:
index = torch.remainder(range - num_pads, max_len)
else:
index = torch.remainder(range + num_pads, max_len)
return src_tokens.gather(1, index)
class SentenceEncoder:
def __init__(self, model_path, max_sentences=None, max_tokens=None, cpu=False, fp16=False, verbose=False,
sort_kind='quicksort'):
self.use_cuda = torch.cuda.is_available() and not cpu
self.max_sentences = max_sentences
self.max_tokens = max_tokens
if self.max_tokens is None and self.max_sentences is None:
self.max_sentences = 1
state_dict = torch.load(model_path)
self.encoder = Encoder(**state_dict['params'])
self.encoder.load_state_dict(state_dict['model'])
self.dictionary = state_dict['dictionary']
self.pad_index = self.dictionary['<pad>']
self.eos_index = self.dictionary['</s>']
self.unk_index = self.dictionary['<unk>']
if fp16:
self.encoder.half()
if self.use_cuda:
if verbose:
print(' - transfer encoder to GPU')
self.encoder.cuda()
self.sort_kind = sort_kind
def _process_batch(self, batch):
tokens = batch.tokens
lengths = batch.lengths
if self.use_cuda:
tokens = tokens.cuda()
lengths = lengths.cuda()
self.encoder.eval()
embeddings = self.encoder(tokens, lengths)['sentemb']
return embeddings.detach().cpu().numpy()
def _tokenize(self, line):
tokens = SPACE_NORMALIZER.sub(" ", line).strip().split()
ntokens = len(tokens)
ids = torch.LongTensor(ntokens + 1)
for i, token in enumerate(tokens):
ids[i] = self.dictionary.get(token, self.unk_index)
ids[ntokens] = self.eos_index
return ids
def _make_batches(self, lines):
tokens = [self._tokenize(line) for line in lines]
lengths = np.array([t.numel() for t in tokens])
indices = np.argsort(-lengths, kind=self.sort_kind)
def batch(tokens, lengths, indices):
toks = tokens[0].new_full((len(tokens), tokens[0].shape[0]), self.pad_index)
for i in range(len(tokens)):
toks[i, -tokens[i].shape[0]:] = tokens[i]
return Batch(
srcs=None,
tokens=toks,
lengths=torch.LongTensor(lengths)
), indices
batch_tokens, batch_lengths, batch_indices = [], [], []
ntokens = nsentences = 0
for i in indices:
if nsentences > 0 and ((self.max_tokens is not None and ntokens + lengths[i] > self.max_tokens) or
(self.max_sentences is not None and nsentences == self.max_sentences)):
yield batch(batch_tokens, batch_lengths, batch_indices)
ntokens = nsentences = 0
batch_tokens, batch_lengths, batch_indices = [], [], []
batch_tokens.append(tokens[i])
batch_lengths.append(lengths[i])
batch_indices.append(i)
ntokens += tokens[i].shape[0]
nsentences += 1
if nsentences > 0:
yield batch(batch_tokens, batch_lengths, batch_indices)
def encode_sentences(self, sentences):
indices = []
results = []
for batch, batch_indices in self._make_batches(sentences):
indices.extend(batch_indices)
results.append(self._process_batch(batch))
return np.vstack(results)[np.argsort(indices, kind=self.sort_kind)]
class Encoder(nn.Module):
def __init__(
self, num_embeddings, padding_idx, embed_dim=320, hidden_size=512, num_layers=1, bidirectional=False,
left_pad=True, padding_value=0.
):
super().__init__()
self.num_layers = num_layers
self.bidirectional = bidirectional
self.hidden_size = hidden_size
self.padding_idx = padding_idx
self.embed_tokens = nn.Embedding(num_embeddings, embed_dim, padding_idx=self.padding_idx)
self.lstm = nn.LSTM(
input_size=embed_dim,
hidden_size=hidden_size,
num_layers=num_layers,
bidirectional=bidirectional,
)
self.left_pad = left_pad
self.padding_value = padding_value
self.output_units = hidden_size
if bidirectional:
self.output_units *= 2
def forward(self, src_tokens, src_lengths):
if self.left_pad:
# convert left-padding to right-padding
src_tokens = convert_padding_direction(
src_tokens,
self.padding_idx,
left_to_right=True,
)
bsz, seqlen = src_tokens.size()
# embed tokens
x = self.embed_tokens(src_tokens)
# B x T x C -> T x B x C
x = x.transpose(0, 1)
# pack embedded source tokens into a PackedSequence
packed_x = nn.utils.rnn.pack_padded_sequence(x, src_lengths.data.tolist())
# apply LSTM
if self.bidirectional:
state_size = 2 * self.num_layers, bsz, self.hidden_size
else:
state_size = self.num_layers, bsz, self.hidden_size
h0 = x.data.new(*state_size).zero_()
c0 = x.data.new(*state_size).zero_()
packed_outs, (final_hiddens, final_cells) = self.lstm(packed_x, (h0, c0))
# unpack outputs and apply dropout
x, _ = nn.utils.rnn.pad_packed_sequence(packed_outs, padding_value=self.padding_value)
assert list(x.size()) == [seqlen, bsz, self.output_units]
if self.bidirectional:
def combine_bidir(outs):
return torch.cat([
torch.cat([outs[2 * i], outs[2 * i + 1]], dim=0).view(1, bsz, self.output_units)
for i in range(self.num_layers)
], dim=0)
final_hiddens = combine_bidir(final_hiddens)
final_cells = combine_bidir(final_cells)
encoder_padding_mask = src_tokens.eq(self.padding_idx).t()
# Set padded outputs to -inf so they are not selected by max-pooling
padding_mask = src_tokens.eq(self.padding_idx).t().unsqueeze(-1)
if padding_mask.any():
x = x.float().masked_fill_(padding_mask, float('-inf')).type_as(x)
# Build the sentence embedding by max-pooling over the encoder outputs
sentemb = x.max(dim=0)[0]
return {
'sentemb': sentemb,
'encoder_out': (x, final_hiddens, final_cells),
'encoder_padding_mask': encoder_padding_mask if encoder_padding_mask.any() else None
}
def EncodeLoad(args):
args.buffer_size = max(args.buffer_size, 1)
assert not args.max_sentences or args.max_sentences <= args.buffer_size, \
'--max-sentences/--batch-size cannot be larger than --buffer-size'
print(' - loading encoder', args.encoder)
return SentenceEncoder(args.encoder,
max_sentences=args.max_sentences,
max_tokens=args.max_tokens,
cpu=args.cpu,
verbose=args.verbose)
def EncodeTime(t):
t = int(time.time() - t)
if t < 1000:
print(' in {:d}s'.format(t))
else:
print(' in {:d}m{:d}s'.format(t // 60, t % 60))
# Encode sentences (existing file pointers)
def EncodeFilep(encoder, inp_file, out_file, buffer_size=10000, verbose=False):
n = 0
t = time.time()
for sentences in buffered_read(inp_file, buffer_size):
encoder.encode_sentences(sentences).tofile(out_file)
n += len(sentences)
if verbose and n % 10000 == 0:
print('\r - Encoder: {:d} sentences'.format(n), end='')
if verbose:
print('\r - Encoder: {:d} sentences'.format(n), end='')
EncodeTime(t)
# Encode sentences (file names)
def EncodeFile(encoder, inp_fname, out_fname,
buffer_size=10000, verbose=False, over_write=False,
inp_encoding='utf-8'):
# TODO :handle over write
if not os.path.isfile(out_fname):
if verbose:
print(' - Encoder: {} to {}'.
format(os.path.basename(inp_fname) if len(inp_fname) > 0 else 'stdin',
os.path.basename(out_fname)))
fin = open(inp_fname, 'r', encoding=inp_encoding, errors='surrogateescape') if len(inp_fname) > 0 else sys.stdin
fout = open(out_fname, mode='wb')
EncodeFilep(encoder, fin, fout, buffer_size=buffer_size, verbose=verbose)
fin.close()
fout.close()
elif not over_write and verbose:
print(' - Encoder: {} exists already'.format(os.path.basename(out_fname)))
# Load existing embeddings
def EmbedLoad(fname, dim=1024, verbose=False):
x = np.fromfile(fname, dtype=np.float32, count=-1)
x.resize(x.shape[0] // dim, dim)
if verbose:
print(' - Embeddings: {:s}, {:d}x{:d}'.format(fname, x.shape[0], dim))
return x
# Get memory mapped embeddings
def EmbedMmap(fname, dim=1024, dtype=np.float32, verbose=False):
nbex = int(os.path.getsize(fname) / dim / np.dtype(dtype).itemsize)
E = np.memmap(fname, mode='r', dtype=dtype, shape=(nbex, dim))
if verbose:
print(' - embeddings on disk: {:s} {:d} x {:d}'.format(fname, nbex, dim))
return E
if __name__ == '__main__':
parser = argparse.ArgumentParser(description='LASER: Embed sentences')
parser.add_argument('--encoder', type=str, required=True,
help='encoder to be used')
parser.add_argument('--token-lang', type=str, default='--',
help="Perform tokenization with given language ('--' for no tokenization)")
parser.add_argument('--bpe-codes', type=str, default=None,
help='Apply BPE using specified codes')
parser.add_argument('-v', '--verbose', action='store_true',
help='Detailed output')
parser.add_argument('-o', '--output', required=True,
help='Output sentence embeddings')
parser.add_argument('--buffer-size', type=int, default=10000,
help='Buffer size (sentences)')
parser.add_argument('--max-tokens', type=int, default=12000,
help='Maximum number of tokens to process in a batch')
parser.add_argument('--max-sentences', type=int, default=None,
help='Maximum number of sentences to process in a batch')
parser.add_argument('--cpu', action='store_true',
help='Use CPU instead of GPU')
parser.add_argument('--stable', action='store_true',
help='Use stable merge sort instead of quick sort')
args = parser.parse_args()
args.buffer_size = max(args.buffer_size, 1)
assert not args.max_sentences or args.max_sentences <= args.buffer_size, \
'--max-sentences/--batch-size cannot be larger than --buffer-size'
if args.verbose:
print(' - Encoder: loading {}'.format(args.encoder))
encoder = SentenceEncoder(args.encoder,
max_sentences=args.max_sentences,
max_tokens=args.max_tokens,
sort_kind='mergesort' if args.stable else 'quicksort',
cpu=args.cpu)
with tempfile.TemporaryDirectory() as tmpdir:
ifname = '' # stdin will be used
if args.token_lang != '--':
tok_fname = os.path.join(tmpdir, 'tok')
Token(ifname,
tok_fname,
lang=args.token_lang,
romanize=True if args.token_lang == 'el' else False,
lower_case=True, gzip=False,
verbose=args.verbose, over_write=False)
ifname = tok_fname
if args.bpe_codes:
bpe_fname = os.path.join(tmpdir, 'bpe')
BPEfastApply(ifname,
bpe_fname,
args.bpe_codes,
verbose=args.verbose, over_write=False)
ifname = bpe_fname
EncodeFile(encoder,
ifname,
args.output,
verbose=args.verbose, over_write=False,
buffer_size=args.buffer_size)