/
config.ini
293 lines (271 loc) · 6.14 KB
/
config.ini
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
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
[conll04]
datasets = conll04
model_name_or_path = t5-base
num_train_epochs = 100
max_seq_length = 256
max_seq_length_eval = 512
per_device_train_batch_size = 4
per_device_eval_batch_size = 4
do_train = True
do_eval = True
do_predict = True
[ade]
datasets = ade
model_name_or_path = t5-base
num_train_epochs = 100
max_seq_length = 256
max_seq_length_eval = 512
per_device_train_batch_size = 4
per_device_eval_batch_size = 4
do_train = True
do_eval = False
do_predict = True
episodes = 1-10
[nyt]
datasets = nyt
model_name_or_path = t5-base
num_train_epochs = 10
max_seq_length = 256
max_seq_length_eval = 512
per_device_train_batch_size = 4
per_device_eval_batch_size = 4
do_train = True
do_eval = True
do_predict = True
[ace2005_joint_er]
datasets = ace2005_joint_er
model_name_or_path = t5-base
num_train_epochs = 10
max_seq_length = 256
max_seq_length_eval = 512
per_device_train_batch_size = 4
per_device_eval_batch_size = 4
do_train = True
do_eval = True
do_predict = True
[multi_dataset_joint_er]
datasets = conll04,ade,nyt,ace2005_joint_er
model_name_or_path = t5-base
num_train_epochs = 10
max_seq_length = 256
max_seq_length_eval = 512
per_device_train_batch_size = 4
per_device_eval_batch_size = 4
do_train = True
do_eval = False
do_predict = True
multitask = True
[ace2005_ner]
datasets = ace2005_ner
model_name_or_path = t5-base
num_train_epochs = 50
max_seq_length = 256
max_seq_length_eval = 512
per_device_train_batch_size = 4
per_device_eval_batch_size = 4
do_train = True
do_eval = True
do_predict = True
[conll03]
datasets = conll03
model_name_or_path = t5-base
num_train_epochs = 10
max_seq_length = 256
max_seq_length_eval = 512
per_device_train_batch_size = 4
per_device_eval_batch_size = 4
do_train = True
do_eval = True
do_predict = True
[ontonotes]
datasets = ontonotes
model_name_or_path = t5-base
num_train_epochs = 10
max_seq_length = 256
max_seq_length_eval = 256
per_device_train_batch_size = 4
per_device_eval_batch_size = 8
do_train = True
do_eval = True
do_predict = True
[genia]
datasets = genia
model_name_or_path = t5-base
num_train_epochs = 10
max_seq_length = 256
max_seq_length_eval = 512
per_device_train_batch_size = 4
per_device_eval_batch_size = 4
do_train = True
do_eval = True
do_predict = True
[multi_dataset_ner]
datasets = conll03,ontonotes,genia,ace2005_ner
model_name_or_path = t5-base
num_train_epochs = 10
max_seq_length = 256
max_seq_length_eval = 512
per_device_train_batch_size = 4
per_device_eval_batch_size = 4
do_train = True
do_eval = True
do_predict = True
multitask = True
[fewrel_meta]
datasets = FewRelFull
model_name_or_path = t5-base
num_train_epochs = 1
max_seq_length = 256
per_device_train_batch_size = 4
per_device_eval_batch_size = 4
do_train = True
do_eval = True
do_predict = False
episodes = 1
[fewrel_1shot_5way]
datasets = FewRelEpisodic
model_name_or_path = experiments/fewrel_meta-t5-base-ep1-len256-b4-train/episode1/
tokenizer_name = t5-base
num_train_epochs = 500
max_seq_length = 256
per_device_train_batch_size = 4
do_train = True
do_eval = False
do_predict = True
episodes = 1-10
num_ways = 5
num_shots = 1
num_query = 5
[tacred]
datasets = tacred
multitask = True
model_name_or_path = t5-base
num_train_epochs = 10
max_seq_length = 256
train_split = train
per_device_train_batch_size = 4
do_train = True
do_eval = True
do_predict = True
[conll05_srl]
datasets = CoNLL2005-SRL
model_name_or_path = t5-base
num_train_epochs = 1
max_seq_length = 256
per_device_train_batch_size = 4
do_train = True
do_eval = True
do_predict = True
[conll05_srl_brown]
datasets = CoNLL2005-SRL
eval_datasets = CoNLL2005-SRL-Brown
model_name_or_path = t5-base
num_train_epochs = 1
max_seq_length = 256
per_device_train_batch_size = 4
do_train = True
do_eval = False
do_predict = True
[conll12_srl]
datasets = CoNLL2012-SRL
model_name_or_path = t5-base
num_train_epochs = 1
max_seq_length = 256
per_device_train_batch_size = 4
do_train = True
do_eval = True
do_predict = True
[ace2005_event]
multitask = True
datasets = ace2005event_argument,ace2005event_trigger
eval_datasets = ace2005event
model_name_or_path = t5-base
num_train_epochs = 10
max_seq_length = 256
max_seq_length_eval = 512
per_device_train_batch_size = 4
per_device_eval_batch_size = 2
do_train = True
do_eval = True
do_predict = True
[conll12_coref]
datasets = conll12_coref
model_name_or_path = t5-base
num_train_epochs = 1
max_seq_length = 256
max_seq_length_eval = 256
per_device_train_batch_size = 4
per_device_eval_batch_size = 4
do_train = True
do_eval = True
do_predict = True
chunk_size = 128
chunk_overlap = 64
chunk_size_eval = 128
chunk_overlap_eval = 64
[multi_woz]
datasets = multi_woz
model_name_or_path = t5-base
do_train = True
do_predict = True
num_train_epochs = 10
max_seq_length = 512
per_device_train_batch_size = 3
per_device_eval_batch_size = 2
overwrite_cache = True
[conll04_final]
datasets = conll04
model_name_or_path = t5-base
num_train_epochs = 200
max_seq_length = 256
max_seq_length_eval = 512
train_split = train,dev
per_device_train_batch_size = 8
per_device_eval_batch_size = 16
do_train = True
do_eval = False
do_predict = True
episodes = 1-10
num_beams = 8
[snips]
datasets = snips
model_name_or_path = t5-base
do_train = True
do_predict = True
do_eval = True
num_train_epochs = 20
max_seq_length = 256
max_seq_length_eval = 512
per_device_train_batch_size = 4
per_device_eval_batch_size = 4
[atis]
datasets = atis
model_name_or_path = t5-base
do_train = True
do_predict = True
do_eval = True
num_train_epochs = 20
max_seq_length = 256
max_seq_length_eval = 512
per_device_train_batch_size = 4
per_device_eval_batch_size = 4
[multitask]
datasets = conll04,ade,nyt,conll03,ontonotes,genia,ace2005event_argument,ace2005event_trigger,conll12_coref,tacred,multi_woz,CoNLL2005-SRL,CoNLL2012-SRL
eval_datasets = conll04,ade,nyt,conll03,ontonotes,genia,ace2005event,conll12_coref,tacred,multi_woz,CoNLL2005-SRL,CoNLL2012-SRL,CoNLL2005-SRL-Brown
model_name_or_path = t5-base
num_train_epochs = 50
max_seq_length = 512
max_seq_length_eval = 512
train_split = train,dev
per_device_train_batch_size = 8
per_device_eval_batch_size = 4
do_train = True
do_eval = False
do_predict = True
chunk_size = 128
chunk_overlap = 96
chunk_size_eval = 192
chunk_overlap_eval = 96
multitask = True
episodes = 1
num_beams = 8