-
Notifications
You must be signed in to change notification settings - Fork 4
/
hyperparams.yml
277 lines (235 loc) · 4.27 KB
/
hyperparams.yml
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
roberta_base:
model_name: deepset/roberta-base-squad2
# model_name: roberta-base
# model_name: xlm-roberta-base
gpus: 1
batch_size: 16
max_epochs: 6
lr: 0.00005
weight_decay: 1.0
val_check_interval: 5
kl_loss: True
swa: True
bigbird:
model_name: google/bigbird-roberta-large
gpus: 1
batch_size: 16
max_epochs: 6
lr: 0.000025
weight_decay: 1.0
val_check_interval: 5
kl_loss: True
swa: True
roberta_large:
model_name: roberta-large
# model_name: xlm-roberta-large
gpus: 1
batch_size: 16
max_epochs: 6
lr: 0.000025
weight_decay: 1.0
val_check_interval: 5
kl_loss: True
swa: True
roberta_large_squad:
model_name: deepset/roberta-large-squad2
gpus: 1
batch_size: 16
max_epochs: 6
lr: 0.000025
weight_decay: 1.0
val_check_interval: 5
kl_loss: True
swa: True
distilroberta:
model_name: distilroberta-base
gpus: 1
batch_size: 16
max_epochs: 6
lr: 0.00005
weight_decay: 1.0
val_check_interval: 5
kl_loss: False
swa: True
albert:
model_name: albert-large-v2
gpus: 1
batch_size: 16
max_epochs: 6
lr: 0.00003
weight_decay: 1.0
val_check_interval: 5
eps: 0.000001
betas: !!python/tuple [0.9, 0.999]
kl_loss: True
swa: True
warmup: 100
grad_clip_val: 1.0
funnel:
model_name: funnel-transformer/large-base
gpus: 1
batch_size: 16
max_epochs: 6
lr: 0.00005
weight_decay: 1.0
val_check_interval: 5
eps: 0.0001
betas: !!python/tuple [0.9, 0.999]
kl_loss: True
swa: True
warmup: 100
bert_base:
model_name: bert-base-uncased
gpus: 1
batch_size: 16
max_epochs: 6
lr: 0.00005
weight_decay: 1.0
val_check_interval: 5
kl_loss: False
swa: True
bert_large:
model_name: bert-large-cased
gpus: 1
batch_size: 16
max_epochs: 6
lr: 0.00005
weight_decay: 1.0
val_check_interval: 5
kl_loss: False
swa: True
bert_large_wm:
model_name: bert-large-cased-whole-word-masking
gpus: 1
batch_size: 16
max_epochs: 6
lr: 0.00005
weight_decay: 1.0
val_check_interval: 5
kl_loss: False
swa: True
deberta_base:
model_name: microsoft/deberta-base
gpus: 1
batch_size: 16
max_epochs: 6
lr: 0.000025
weight_decay: 1.0
val_check_interval: 5
kl_loss: True
swa: True
deberta_large:
model_name: microsoft/deberta-large
gpus: 1
batch_size: 12
max_epochs: 6
lr: 0.000025
weight_decay: 1.0
val_check_interval: 5
kl_loss: True
swa: True
deberta_xlarge:
model_name: microsoft/deberta-v2-xlarge
gpus: 1
plugins: ddp_sharded
accelerator: ddp
precision: 16
batch_size: 10
max_epochs: 6
lr: 0.000025
weight_decay: 1.0
val_check_interval: 5
kl_loss: False
swa: True
bart_base:
model_name: facebook/bart-base
gpus: 1
batch_size: 16
max_epochs: 6
lr: 0.000025
weight_decay: 1.0
val_check_interval: 5
kl_loss: True
swa: True
bart_large:
model_name: facebook/bart-large
gpus: 1
batch_size: 12
max_epochs: 6
lr: 0.000025
weight_decay: 1.0
val_check_interval: 5
kl_loss: False
swa: True
st_labse:
model_name: sentence-transformers/LaBSE
gpus: 1
batch_size: 16
max_epochs: 6
lr: 0.000025
weight_decay: 1.0
val_check_interval: 5
kl_loss: True
swa: True
# https://www.sbert.net/docs/pretrained_models.html
st_paraphrase:
model_name: sentence-transformers/paraphrase-mpnet-base-v2
gpus: 1
batch_size: 32
max_epochs: 6
lr: 0.00005
weight_decay: 1.0
val_check_interval: 5
kl_loss: True
swa: True
sentence_bert:
model_name: deepset/sentence_bert
gpus: 1
batch_size: 16
max_epochs: 6
lr: 0.000025
weight_decay: 1.0
val_check_interval: 5
kl_loss: False
swa: True
gpt2:
model_name: gpt2
gpus: 1
batch_size: 16
max_epochs: 6
lr: 0.00005
weight_decay: 1.0
val_check_interval: 5
kl_loss: False
swa: True
electra:
model_name: google/electra-large-discriminator
gpus: 1
batch_size: 16
max_epochs: 6
lr: 0.00001
warmup: 100
weight_decay: 1.0
val_check_interval: 5
kl_loss: True
swa: True
convbert:
model_name: YituTech/conv-bert-base
gpus: 1
batch_size: 16
max_epochs: 6
lr: 0.00005
weight_decay: 1.0
val_check_interval: 5
kl_loss: False
swa: True
ctrl:
model_name: ctrl # 6.5GB!!
gpus: 1
batch_size: 16
max_epochs: 6
lr: 0.00005
weight_decay: 1.0
val_check_interval: 5
kl_loss: False
swa: True