-
Notifications
You must be signed in to change notification settings - Fork 4
/
args_kmeans.py
249 lines (248 loc) · 12 KB
/
args_kmeans.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
import argparse
def get_args(description='Youtube-Text-Video'):
parser = argparse.ArgumentParser(description=description)
parser.add_argument(
'--train_csv',
type=str,
default='data/HowTo100M_1166_videopaths.txt',
#default='/home/brian27/nobackup/data/howto100m/HowTo100M_1166_videopaths.txt',
help='train csv')
parser.add_argument(
'--features_path',
type=str,
default='parsed_videos/',
help='path for visual features (2D, 3D) visual features')
parser.add_argument(
'--features_path_audio',
type=str,
default='',
help='path for audio files (defaults to --features_path)')
parser.add_argument(
'--caption_path',
type=str,
default='data/caption.pickle',
help='HowTo100M caption pickle file path')
parser.add_argument(
'--word2vec_path',
type=str,
default='data/GoogleNews-vectors-negative300.bin',
help='word embedding path')
parser.add_argument(
'--pretrain_path',
type=str,
default='',
help='pre train model path')
parser.add_argument(
'--checkpoint_dir',
type=str,
default='',
help='checkpoint model folder')
parser.add_argument('--eval_lang_retrieval', type=int, default=0,
help='if 1, eval language retrieval instead of video retrieval')
parser.add_argument('--tri_modal', type=int, default=0,
help='use vision, speech, and text')
parser.add_argument('--tri_modal_fuse', type=int, default=0,
help='use speech and text features (tri_modal must be 1)')
parser.add_argument('--fuse_videoaudio_additive', type=int, default=0,
help='eval T->A+V with tri-modal modal \
set tri_modal=1, tri_modal_fuse=0')
parser.add_argument('--loss', type=int, default=0,
help='0 for Masked Margin Softmax (MMS) loss')
parser.add_argument('--apex_level', type=int, default=0,
help='Apex (mixed precision) level: chose 0 for none, 1 for O1.')
parser.add_argument('--random_audio_windows', type=int, default=1,
help='1 to use random audio windows, 0 to use HowTo100M ASR clips')
parser.add_argument('--howto_audio_frames', type=int, default=1024,
help='number of frames to use for loading howto100m audio')
parser.add_argument('--youcook_num_frames_multiplier', type=int, default=5,
help='use 1024 * x audio frames for youcook2')
parser.add_argument('--msrvtt_num_frames_multiplier', type=int, default=3,
help='use 1024 * x audio frames for msrvtt')
parser.add_argument('--lsmdc_num_frames_multiplier', type=int, default=3,
help='use 1024 * x audio frames for lsmdc')
parser.add_argument('--num_thread_reader', type=int, default=1,
help='')
parser.add_argument('--embd_dim', type=int, default=2048,
help='embedding dim')
parser.add_argument('--lr', type=float, default=0.0001,
help='initial learning rate')
parser.add_argument('--epochs', type=int, default=20,
help='upper epoch limit')
parser.add_argument('--batch_size', type=int, default=256,
help='batch size')
parser.add_argument('--batch_size_val', type=int, default=3500,
help='batch size eval')
parser.add_argument('--lr_decay', type=float, default=0.9,
help='Learning rate exp epoch decay')
parser.add_argument('--n_display', type=int, default=200,
help='Information display frequence')
parser.add_argument('--feature_dim', type=int, default=4096,
help='video feature dimension')
parser.add_argument('--we_dim', type=int, default=300,
help='word embedding dimension')
parser.add_argument('--seed', type=int, default=1,
help='random seed')
parser.add_argument('--verbose', type=int, default=1,
help='')
parser.add_argument('--max_words', type=int, default=20,
help='')
parser.add_argument('--min_words', type=int, default=0,
help='')
parser.add_argument('--feature_framerate', type=int, default=1,
help='')
parser.add_argument('--min_time', type=float, default=5.0,
help='Gather small clips')
parser.add_argument('--n_pair', type=int, default=1,
help='Number of video clips to use per video')
parser.add_argument('--lsmdc', type=int, default=0,
help='Train on LSDMC data')
parser.add_argument('--youcook', type=int, default=0,
help='Train on YouCook2 data')
parser.add_argument('--msrvtt', type=int, default=0,
help='Train on MSRVTT data')
parser.add_argument('--eval_lsmdc', type=int, default=0,
help='Evaluate on LSMDC data')
parser.add_argument('--eval_msrvtt', type=int, default=0,
help='Evaluate on MSRVTT data')
parser.add_argument('--eval_youcook', type=int, default=0,
help='Evaluate on YouCook2 data')
parser.add_argument('--eval_ucf', type=int, default=0,
help='Evaluate on UCF-101 data')
parser.add_argument('--eval_hmdb', type=int, default=0,
help='Evaluate on HMDB data')
parser.add_argument('--eval_cross', type=int, default=0,
help='Evaluate on CrossTask data')
parser.add_argument('--eval_how', type=int, default=0,
help='Evaluate on how2 data')
parser.add_argument('--sentence_dim', type=int, default=-1,
help='sentence dimension')
parser.add_argument('--cluster', type=int, default=0,
help='cluster loss')
parser.add_argument('--queue_size', type=int, default=3,
help='queue size')
parser.add_argument('--start_queue', type=int, default=0,
help='start_queue')
parser.add_argument('--start_cluster', type=int, default=0,
help='start_cluster')
parser.add_argument('--num_candidates', type=int, default=1,
help='num candidates for MILNCE loss')
parser.add_argument('--use_queue', type=int, default=0,
help='use_queue')
parser.add_argument('--cluster_size', type=int, default=256,
help='cluster_size')
parser.add_argument('--layer', type=int, default=0,
help='classification layer')
parser.add_argument('--soft_label', type=int, default=0,
help='soft_label')
parser.add_argument('--multi_cluster', type=int, default=0,
help='multi_cluster')
parser.add_argument('--pure_cluster', type=int, default=0,
help='pure_cluster')
parser.add_argument('--project', type=int, default=0,
help='project')
parser.add_argument('--proto_nce', type=int, default=0,
help='proto_nce')
parser.add_argument('--switch_loss_h', type=int, default=0,
help='switch_loss_h')
parser.add_argument('--switch_loss_s', type=int, default=0,
help='switch_loss_s')
parser.add_argument('--self_prediction', type=int, default=0,
help='self_prediction')
parser.add_argument('--soft_contrast', type=int, default=0,
help='soft_contrast')
parser.add_argument('--soft_contrast_only', type=int, default=0,
help='soft_contrast_only')
parser.add_argument('--nce', type=int, default=0,
help='nce')
parser.add_argument('--nce_only', type=int, default=0,
help='nce_only')
parser.add_argument('--pseudo_contrast', type=int, default=0,
help='pseudo_contrast')
parser.add_argument('--cooperative', type=int, default=0,
help='cooperative')
parser.add_argument('--project_dim', type=int, default=6000,
help='project_dim')
parser.add_argument('--no_audio', type=int, default=0,
help='no_audio')
parser.add_argument('--no_va', type=int, default=0,
help='no_va')
parser.add_argument('--rand', type=int, default=0,
help='random drop')
parser.add_argument('--joint', type=int, default=0,
help='joint cluster')
parser.add_argument('--kmeans', type=int, default=0,
help='kmeans cluster')
parser.add_argument('--fastC', type=int, default=0,
help='fast cluster')
parser.add_argument('--withMLP', type=int, default=0,
help='withMLP cluster')
parser.add_argument('--recon', type=int, default=0,
help='recon ')
parser.add_argument('--mms', type=int, default=0,
help='mms ')
parser.add_argument('--mean', type=int, default=0,
help='mean ')
parser.add_argument('--lamb', type=float, default=0.5,
help='lambda ')
parser.add_argument('--tri_loss', type=int, default=0,
help='tri_loss ')
parser.add_argument('--recon_size', type=int, default=768,
help='recon_size ')
parser.add_argument('--clu_lamb', type=int, default=1,
help='clu_lamb ')
parser.add_argument('--noC', type=int, default=0,
help='noC ')
parser.add_argument('--cos', type=int, default=1,
help='cos ')
parser.add_argument("--base_lr", default=4.8, type=float, help="base learning rate")
parser.add_argument("--final_lr", type=float, default=0, help="final learning rate")
parser.add_argument("--freeze_prototypes_niters", default=313, type=int,
help="freeze the prototypes during this many iterations from the start")
parser.add_argument("--wd", default=1e-6, type=float, help="weight decay")
parser.add_argument("--warmup_epochs", default=10, type=int, help="number of warmup epochs")
parser.add_argument("--start_warmup", default=0, type=float,
help="initial warmup learning rate")
parser.add_argument('--warmup_steps', type=int, default=5000,
help='')
parser.add_argument(
'--youcook_train_path',
type=str,
default='data/youcook_train_audio.pkl',
help='')
parser.add_argument(
'--youcook_val_path',
type=str,
default='data/youcook_val_audio.pkl',
help='')
parser.add_argument(
'--msrvtt_test_path',
type=str,
default='data/msrvtt_jsfusion_test.pkl',
help='')
parser.add_argument(
'--msrvtt_train_path',
type=str,
default='data/msrvtt_train.pkl',
help='')
parser.add_argument(
'--lsmdc_test_path',
type=str,
default='data/lsmdc_test.pkl',
help='')
parser.add_argument(
'--lsmdc_train_path',
type=str,
default='data/lsmdc_train.pkl',
help='')
parser.add_argument(
'--ucf_test_path',
type=str,
default='data/UCF101_data.pkl',
help='')
parser.add_argument(
'--hmdb_test_path',
type=str,
default='data/HMDB_data.pkl',
help='')
args = parser.parse_args()
return args