-
Notifications
You must be signed in to change notification settings - Fork 0
/
decode_line_multi_best.py
169 lines (141 loc) · 5.89 KB
/
decode_line_multi_best.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
# Copyright 2016 Stanford University
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# ==============================================================================
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import os
import time
import numpy as np
from six.moves import xrange
import tensorflow as tf
from multiprocessing import Pool
from os.path import join as pjoin
import nlc_model_multiple as nlc_model
import nlc_data
from levenshtein import align, align_one2many
from util import initialize_vocabulary, get_tokenizer
from flag import FLAGS
reverse_vocab, vocab = None, None
def create_model(session, vocab_size, forward_only):
model = nlc_model.NLCModel(
vocab_size, FLAGS.size, FLAGS.num_layers, FLAGS.max_gradient_norm, FLAGS.batch_size,
FLAGS.learning_rate, FLAGS.learning_rate_decay_factor, FLAGS.dropout,
forward_only=forward_only)
ckpt = tf.train.get_checkpoint_state(FLAGS.train_dir)
if ckpt and tf.gfile.Exists(ckpt.model_checkpoint_path):
print("Reading model parameters from %s" % ckpt.model_checkpoint_path)
model.saver.restore(session, ckpt.model_checkpoint_path)
else:
print("Created model with fresh parameters.")
session.run(tf.initialize_all_variables())
return model
def padded(tokens, depth):
maxlen = max(map(lambda x: len(x), tokens))
align = pow(2, depth - 1)
padlen = maxlen + (align - maxlen) % align
return map(lambda token_list: token_list + [0] * (padlen - len(token_list)), tokens)
def tokenize(sents, vocab, depth=FLAGS.num_layers):
token_ids = []
for sent in sents:
token_ids.append(nlc_data.sentence_to_token_ids(sent, vocab, get_tokenizer(FLAGS.tokenizer)))
token_ids = padded(token_ids, depth)
source = np.array(token_ids).T
source_mask = (source != 0).astype(np.int32)
return source, source_mask
def detokenize(sents, reverse_vocab):
# TODO: char vs word
def detok_sent(sent):
outsent = ''
for t in sent:
if t >= len(nlc_data._START_VOCAB):
outsent += reverse_vocab[t]
return outsent
return [detok_sent(s) for s in sents]
def decode_beam(model, sess, encoder_output, max_beam_size, len_input):
toks, probs, prob_trans = model.decode_beam(sess, encoder_output, max_beam_size, len_input)
return toks.tolist(), probs.tolist(), prob_trans
def decode_lattice(beamstrs, probs, prob_trans):
num_str = len(beamstrs)
dict_str = {}
for i in range(num_str):
cur_str = beamstrs[i]
for j in range(0, len(cur_str)):
prefix = cur_str[:j]
if prefix not in dict_str:
dict_str[prefix] = {}
dict_str[prefix][cur_str[j]] = prob_trans[i][j + 1]
else:
dict_str[prefix][cur_str[j]] = prob_trans[i][j + 1]
if cur_str not in dict_str:
dict_str[cur_str] = {}
dict_str[cur_str]['<eos>'] = prob_trans[i][len(cur_str) + 1]
dict_str[cur_str + '<eos>'] = probs[i]
return dict_str
def fix_sent(model, sess, sent):
# Tokenize
input_toks, mask = tokenize(sent, vocab)
# Encode
encoder_output = model.encode(sess, input_toks, mask)
s1, s2, s3 = encoder_output.shape
encoder_output = np.transpose(encoder_output, (1, 0, 2))
encoder_output = np.reshape(encoder_output, [s2, s1, 1, s3])
len_input = mask.shape[0]
# Decode
beam_toks, probs, prob_trans = decode_beam(model, sess, encoder_output, FLAGS.beam_size, len_input)
# De-tokenize
beam_strs = detokenize(beam_toks, reverse_vocab)
return beam_strs, probs
def decode():
# Prepare NLC data.
global reverse_vocab, vocab
folder_out = FLAGS.out_dir
if not os.path.exists(folder_out):
os.makedirs(folder_out)
print("Preparing NLC data in %s" % FLAGS.data_dir)
vocab_path = pjoin(FLAGS.data_dir, "vocab.dat")
vocab, reverse_vocab = initialize_vocabulary(vocab_path)
vocab_size = len(vocab)
print("Vocabulary size: %d" % vocab_size)
if FLAGS.gpu_frac == 1:
sess = tf.Session()
else:
gpu_options = tf.GPUOptions(per_process_gpu_memory_fraction=FLAGS.gpu_frac)
sess = tf.Session(config=tf.ConfigProto(gpu_options=gpu_options, allow_soft_placement=True))
print("Creating %d layers of %d units." % (FLAGS.num_layers, FLAGS.size))
model = create_model(sess, vocab_size, True)
tic = time.time()
with open(pjoin(FLAGS.data_dir, FLAGS.dev + '.x.txt'), 'r') as f_:
lines = [ele.strip() for ele in f_.readlines()]
f_o = open(pjoin(folder_out, FLAGS.dev + '.avg.best.' + '.o.txt.' + str(FLAGS.start) + '_' + str(FLAGS.end)), 'w')
for line_id in range(FLAGS.start, FLAGS.end):
line = lines[line_id]
sents = [ele for ele in line.strip('\n').split('\t')][:20]
sents = [ele for ele in sents if len(ele.strip()) > 0]
#sents = [ele.replace('-', '_') for ele in sents if len(ele.strip()) > 0]
if len(sents) > 0:
output_sents, output_probs = fix_sent(model, sess, sents)
#output_sents = [ele.replace('_', '-') for ele in output_sents]
f_o.write('\t'.join(output_sents[:20]) + '\n')
else:
f_o.write('\n')
if line_id % 100 == 0:
toc = time.time()
print(toc - tic)
tic = time.time()
f_o.close()
def main(_):
decode()
if __name__ == "__main__":
tf.app.run()