-
Notifications
You must be signed in to change notification settings - Fork 7
/
Copy pathconfig.py
152 lines (118 loc) · 5.25 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
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
# -*- coding: utf-8 -*-
"""
@author: alexyang
@contact: [email protected]
@file: config.py
@time: 2019/2/2 14:05
@desc:
"""
from os import path
from keras.optimizers import Adam
RAW_DATA_DIR = './raw_data'
PROCESSED_DATA_DIR = './data'
LOG_DIR = './log'
MODEL_SAVED_DIR = './ckpt'
FEATURE_DIR = './feature'
IMG_DIR = './img'
SNLI_DIR = path.join(RAW_DATA_DIR, 'snli_1.0/')
SNLI_TRAIN_FILENAME = path.join(SNLI_DIR, 'snli_1.0_train.jsonl')
SNLI_DEV_FILENAME = path.join(SNLI_DIR, 'snli_1.0_dev.jsonl')
SNLI_TEST_FILENAME = path.join(SNLI_DIR, 'snli_1.0_test.jsonl')
MULTINLI_DIR = path.join(RAW_DATA_DIR, 'multinli_1.0/')
MULTINLI_TRAIN_FILENAME = path.join(MULTINLI_DIR, 'multinli_1.0_train.jsonl')
MULTINLI_DEV_FILENAME = path.join(MULTINLI_DIR, 'multinli_1.0_dev_matched.jsonl')
MLI_DIR = path.join(RAW_DATA_DIR, 'mednli_1.0/')
MLI_TRAIN_FILENAME = path.join(MLI_DIR, 'mli_train_v1.jsonl')
MLI_DEV_FILENAME = path.join(MLI_DIR, 'mli_dev_v1.jsonl')
MLI_TEST_FILENAME = path.join(MLI_DIR, 'mli_test_v1.jsonl')
TRAIN_DATA_TEMPLATE = 'genre_{}_train.pkl'
DEV_DATA_TEMPLATE = 'genre_{}_dev.pkl'
TEST_DATA_TEMPLATE = 'genre_{}_test.pkl'
TRAIN_IDS_MATRIX_TEMPLATE = 'genre_{}_level_{}_ids_train.pkl'
DEV_IDS_MATRIX_TEMPLATE = 'genre_{}_level_{}_ids_dev.pkl'
TEST_IDS_MATRIX_TEMPLATE = 'genre_{}_level_{}_ids_test.pkl'
TRAIN_FEATURES_TEMPLATE = 'genre_{}_feature_{}_train.pkl'
DEV_FEATURES_TEMPLATE = 'genre_{}_feature_{}_dev.pkl'
TEST_FEATURES_TEMPLATE = 'genre_{}_feature_{}_test.pkl'
EMBEDDING_MATRIX_TEMPLATE = 'genre_{}_type_{}_embeddings.npy'
TOKENIZER_TEMPLATE = 'genre_{}_level_{}_tokenizer.pkl'
VOCABULARY_TEMPLATE = 'genre_{}_level_{}_vocab.pkl'
ANALYSIS_LOG_TEMPLATE = 'genre_{}_analysis.log'
PERFORMANCE_LOG = 'genre_{}_performance.log'
EXTERNAL_WORD_VECTORS_DIR = path.join(RAW_DATA_DIR, 'word_embeddings/')
EXTERNAL_WORD_VECTORS_FILENAME = {
'glove_cc': path.join(EXTERNAL_WORD_VECTORS_DIR, 'glove.840B.300d.txt'),
'fasttext_cc': path.join(EXTERNAL_WORD_VECTORS_DIR, 'fasttext-wiki-news-300d-1M-subword.vec'),
'fasttext_wiki': path.join(EXTERNAL_WORD_VECTORS_DIR, 'fasttext-crawl-300d-2M-subword.vec'),
'tfhub_elmo_2': path.join(EXTERNAL_WORD_VECTORS_DIR, 'tfhub_elmo_2'),
'tfhub_bert': path.join(EXTERNAL_WORD_VECTORS_DIR, 'bert_uncased_L_12_H_768_A_12'),
'original_elmo_5.5B': {'weights': path.join(EXTERNAL_WORD_VECTORS_DIR,
'elmo_2x4096_512_2048cnn_2xhighway_5.5B_weights.hdf5'),
'options': path.join(EXTERNAL_WORD_VECTORS_DIR,
'elmo_2x4096_512_2048cnn_2xhighway_5.5B_options.json')
}
}
CACHE_DIR = path.join(PROCESSED_DATA_DIR, 'cache')
LABELS = {'contradiction': 0, 'neutral': 1, 'entailment': 2}
GENRES = ['fiction', 'government', 'slate', 'telephone', 'travel', 'snli', 'multinli', 'mednli']
class ProcessConfig(object):
def __init__(self):
self.clean = False
self.stem = False
self.lowercase = True
self.word_max_len = None
self.char_max_len = None
self.padding = 'post'
self.truncating = 'post'
self.n_class = 3
self.word_cut_func = lambda x: x.split()
self.char_cut_func = lambda x: list(x)
class ModelConfig(object):
def __init__(self):
# input configuration
self.genre = 'snli'
self.input_level = 'word'
self.word_max_len = {'snli': 82, 'mednli': 202}
self.char_max_len = {'snli': 406, 'mednli': 1132}
self.max_len = 0
self.word_embed_type = 'glove'
self.word_embed_dim = 300
self.word_embed_trainable = False
self.word_embeddings = None
self.add_features = False # whether to add additional statistical features
self.feature_len = 79 # dimension of statistical features
# elmo embedding configuration
self.elmo_model_url = EXTERNAL_WORD_VECTORS_FILENAME['tfhub_elmo_2']
self.elmo_options_file = EXTERNAL_WORD_VECTORS_FILENAME['original_elmo_5.5B']['options']
self.elmo_weight_file = EXTERNAL_WORD_VECTORS_FILENAME['original_elmo_5.5B']['weights']
self.cache_dir = CACHE_DIR
self.idx2token = None # used for get ELMo embedding
# model structure configuration
self.exp_name = None
self.model_name = None
self.rnn_units = 300
self.dense_units = 128
# model training configuration
self.batch_size = 512
self.n_epoch = 16
self.learning_rate = 0.001
self.optimizer = Adam(self.learning_rate)
self.dropout = 0.5
self.l2_reg = 0.001
self.min_lr = 0.0005
self.max_lr = 0.001
# output configuration
self.n_class = 3
# checkpoint configuration
self.checkpoint_dir = MODEL_SAVED_DIR
self.checkpoint_monitor = 'val_acc'
self.checkpoint_save_best_only = True
self.checkpoint_save_weights_only = True
self.checkpoint_save_weights_mode = 'max'
self.checkpoint_verbose = 1
# early_stoping configuration
self.early_stopping_monitor = 'val_acc'
self.early_stopping_mode = 'max'
self.early_stopping_patience = 5
self.early_stopping_verbose = 1
self.callbacks_to_add = ['modelcheckpoint']