forked from HKUST-KnowComp/R-Net
-
Notifications
You must be signed in to change notification settings - Fork 0
/
config.py
137 lines (117 loc) · 5.7 KB
/
config.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
import os
import tensorflow as tf
from prepro import prepro
from main import train, test
flags = tf.flags
os.environ["TF_CPP_MIN_LOG_LEVEL"] = "3"
home = os.path.expanduser("~")
train_file = os.path.join(home, "data", "squad", "train-v2.0.json")
dev_file = os.path.join(home, "data", "squad", "dev-v2.0.json")
test_file = os.path.join(home, "data", "squad", "dev-v2.0.json")
glove_word_file = os.path.join(home, "data", "glove", "glove.840B.300d.txt")
target_dir = "data"
log_dir = "log/event"
save_dir = "log/model"
answer_dir = "log/answer"
train_record_file = os.path.join(target_dir, "train.tfrecords")
dev_record_file = os.path.join(target_dir, "dev.tfrecords")
test_record_file = os.path.join(target_dir, "test.tfrecords")
word_emb_file = os.path.join(target_dir, "word_emb.json")
char_emb_file = os.path.join(target_dir, "char_emb.json")
train_eval = os.path.join(target_dir, "train_eval.json")
dev_eval = os.path.join(target_dir, "dev_eval.json")
test_eval = os.path.join(target_dir, "test_eval.json")
dev_meta = os.path.join(target_dir, "dev_meta.json")
test_meta = os.path.join(target_dir, "test_meta.json")
word2idx_file = os.path.join(target_dir, "word2idx.json")
char2idx_file = os.path.join(target_dir, "char2idx.json")
answer_file = os.path.join(answer_dir, "answer.json")
if not os.path.exists(target_dir):
os.makedirs(target_dir)
if not os.path.exists(log_dir):
os.makedirs(log_dir)
if not os.path.exists(save_dir):
os.makedirs(save_dir)
if not os.path.exists(answer_dir):
os.makedirs(answer_dir)
flags.DEFINE_string("mode", "train", "train/debug/test")
flags.DEFINE_string("target_dir", target_dir, "")
flags.DEFINE_string("log_dir", log_dir, "")
flags.DEFINE_string("save_dir", save_dir, "")
flags.DEFINE_string("train_file", train_file, "")
flags.DEFINE_string("dev_file", dev_file, "")
flags.DEFINE_string("test_file", test_file, "")
flags.DEFINE_string("glove_word_file", glove_word_file, "")
flags.DEFINE_string("train_record_file", train_record_file, "")
flags.DEFINE_string("dev_record_file", dev_record_file, "")
flags.DEFINE_string("test_record_file", test_record_file, "")
flags.DEFINE_string("word_emb_file", word_emb_file, "")
flags.DEFINE_string("char_emb_file", char_emb_file, "")
flags.DEFINE_string("train_eval_file", train_eval, "")
flags.DEFINE_string("dev_eval_file", dev_eval, "")
flags.DEFINE_string("test_eval_file", test_eval, "")
flags.DEFINE_string("dev_meta", dev_meta, "")
flags.DEFINE_string("test_meta", test_meta, "")
flags.DEFINE_string("word2idx_file", word2idx_file, "")
flags.DEFINE_string("char2idx_file", char2idx_file, "")
flags.DEFINE_string("answer_file", answer_file, "")
flags.DEFINE_integer("glove_char_size", 94, "Corpus size for Glove")
flags.DEFINE_integer("glove_word_size", int(2.2e6), "Corpus size for Glove")
flags.DEFINE_integer("glove_dim", 300, "Embedding dimension for Glove")
flags.DEFINE_integer("char_dim", 8, "Embedding dimension for char")
flags.DEFINE_integer("para_limit", 400, "Limit length for paragraph")
flags.DEFINE_integer("ques_limit", 50, "Limit length for question")
flags.DEFINE_integer("ans_limit", 30, "Limit length for answer")
flags.DEFINE_integer("test_para_limit", 1000,
"Max length for paragraph in test")
flags.DEFINE_integer("test_ques_limit", 100, "Max length of questions in test")
flags.DEFINE_integer("char_limit", 16, "Limit length for character")
flags.DEFINE_integer("word_count_limit", -1, "Min count for word")
flags.DEFINE_integer("char_count_limit", -1, "Min count for char")
flags.DEFINE_integer("capacity", 15000, "Batch size of dataset shuffle")
flags.DEFINE_integer("num_threads", 4, "Number of threads in input pipeline")
flags.DEFINE_boolean("use_cudnn", True, "Whether to use cudnn (only for GPU)")
flags.DEFINE_boolean("is_bucket", False, "Whether to use bucketing")
flags.DEFINE_list("bucket_range", [40, 361, 40], "range of bucket")
flags.DEFINE_integer("batch_size", 64, "Batch size")
flags.DEFINE_integer("num_steps", 40000, "Number of steps")
flags.DEFINE_integer("checkpoint", 1000, "checkpoint for evaluation")
flags.DEFINE_integer("period", 100, "period to save batch loss")
flags.DEFINE_integer("val_num_batches", 150, "Num of batches for evaluation")
flags.DEFINE_float("init_lr", 0.5, "Initial lr for Adadelta")
flags.DEFINE_float("keep_prob", 0.7, "Keep prob in rnn")
flags.DEFINE_float("ptr_keep_prob", 0.7, "Keep prob for pointer network")
flags.DEFINE_float("grad_clip", 5.0, "Global Norm gradient clipping rate")
flags.DEFINE_integer("hidden", 75, "Hidden size")
flags.DEFINE_integer("char_hidden", 100, "GRU dim for char")
flags.DEFINE_integer("patience", 3, "Patience for lr decay")
flags.DEFINE_boolean("use_squad_v2", True, "Whether to use SQuAD 2.0")
# Extensions (Uncomment corresponding line in download.sh to download the required data)
glove_char_file = os.path.join(
home, "data", "glove", "glove.840B.300d-char.txt")
flags.DEFINE_string("glove_char_file", glove_char_file,
"Glove character embedding")
flags.DEFINE_boolean("pretrained_char", False,
"Whether to use pretrained char embedding")
fasttext_file = os.path.join(home, "data", "fasttext", "wiki-news-300d-1M.vec")
flags.DEFINE_string("fasttext_file", fasttext_file, "Fasttext word embedding")
flags.DEFINE_boolean("fasttext", False, "Whether to use fasttext")
def main(_):
config = flags.FLAGS
if config.mode == "train":
train(config)
elif config.mode == "prepro":
prepro(config)
elif config.mode == "debug":
config.num_steps = 2
config.val_num_batches = 1
config.checkpoint = 1
config.period = 1
train(config)
elif config.mode == "test":
test(config)
else:
print("Unknown mode")
exit(0)
if __name__ == "__main__":
tf.app.run()