system HF staff commited on
Commit
47abc14
1 Parent(s): 1359b44

Update config.json

Browse files
Files changed (1) hide show
  1. config.json +335 -0
config.json ADDED
@@ -0,0 +1,335 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dump_path": "/checkpoint/aconneau/dumped/xlm_17_100_big.3/16656237",
3
+ "exp_name": "xlm_17_100_big.3",
4
+ "save_periodic": 0,
5
+ "exp_id": "16656237",
6
+ "fp16": true,
7
+ "amp": 2,
8
+ "encoder_only": true,
9
+ "emb_dim": 1280,
10
+ "n_layers": 16,
11
+ "n_heads": 16,
12
+ "dropout": 0.1,
13
+ "attention_dropout": 0.1,
14
+ "gelu_activation": true,
15
+ "share_inout_emb": true,
16
+ "sinusoidal_embeddings": false,
17
+ "use_lang_emb": false,
18
+ "use_memory": false,
19
+ "asm": false,
20
+ "context_size": 0,
21
+ "word_pred": 0.15,
22
+ "sample_alpha": 0.5,
23
+ "word_mask_keep_rand": "0.8,0.1,0.1",
24
+ "word_shuffle": 0.0,
25
+ "word_dropout": 0.0,
26
+ "word_blank": 0.0,
27
+ "data_path": "/private/home/aconneau/projects/XLM/data/wiki/17/175k",
28
+ "lgs": "en-fr-es-de-it-pt-nl-sv-pl-ru-ar-tr-zh-ja-ko-hi-vi",
29
+ "max_vocab": 200000,
30
+ "min_count": 0,
31
+ "lg_sampling_factor": 0.7,
32
+ "bptt": 256,
33
+ "max_len": 200,
34
+ "group_by_size": true,
35
+ "batch_size": 16,
36
+ "max_batch_size": 0,
37
+ "tokens_per_batch": -1,
38
+ "split_data": true,
39
+ "optimizer": "adam_inverse_sqrt,lr=0.00005,warmup_updates=30000,beta1=0.9,beta2=0.999,weight_decay=0.01,eps=0.000001",
40
+ "clip_grad_norm": 1.0,
41
+ "epoch_size": 200000,
42
+ "max_epoch": 100000,
43
+ "stopping_criterion": "_valid_zh_mlm_ppl,25",
44
+ "validation_metrics": "_valid_en_mlm_ppl,_valid_mlm_ppl,_valid_zh_mlm_ppl",
45
+ "accumulate_gradients": 4,
46
+ "lambda_mlm": 1.0,
47
+ "lambda_clm": 1.0,
48
+ "lambda_pc": 1.0,
49
+ "lambda_ae": 1.0,
50
+ "lambda_mt": 1.0,
51
+ "lambda_bt": 1.0,
52
+ "clm_steps": [],
53
+ "mlm_steps": [
54
+ [
55
+ "en",
56
+ null
57
+ ],
58
+ [
59
+ "fr",
60
+ null
61
+ ],
62
+ [
63
+ "es",
64
+ null
65
+ ],
66
+ [
67
+ "de",
68
+ null
69
+ ],
70
+ [
71
+ "it",
72
+ null
73
+ ],
74
+ [
75
+ "pt",
76
+ null
77
+ ],
78
+ [
79
+ "nl",
80
+ null
81
+ ],
82
+ [
83
+ "sv",
84
+ null
85
+ ],
86
+ [
87
+ "pl",
88
+ null
89
+ ],
90
+ [
91
+ "ru",
92
+ null
93
+ ],
94
+ [
95
+ "ar",
96
+ null
97
+ ],
98
+ [
99
+ "tr",
100
+ null
101
+ ],
102
+ [
103
+ "zh",
104
+ null
105
+ ],
106
+ [
107
+ "ja",
108
+ null
109
+ ],
110
+ [
111
+ "ko",
112
+ null
113
+ ],
114
+ [
115
+ "hi",
116
+ null
117
+ ],
118
+ [
119
+ "vi",
120
+ null
121
+ ]
122
+ ],
123
+ "mt_steps": [],
124
+ "ae_steps": [],
125
+ "bt_steps": [],
126
+ "pc_steps": [],
127
+ "reload_emb": "",
128
+ "reload_model": "/checkpoint/aconneau/dumped/xlm_17_100_240_big_model_upper.2/14884510/best-valid_zh_mlm_ppl.pth",
129
+ "reload_checkpoint": "",
130
+ "beam_size": 1,
131
+ "length_penalty": 1,
132
+ "early_stopping": false,
133
+ "eval_bleu": false,
134
+ "eval_only": false,
135
+ "debug_train": false,
136
+ "debug_slurm": false,
137
+ "debug": false,
138
+ "local_rank": 0,
139
+ "master_port": 14148,
140
+ "langs": [
141
+ "en",
142
+ "fr",
143
+ "es",
144
+ "de",
145
+ "it",
146
+ "pt",
147
+ "nl",
148
+ "sv",
149
+ "pl",
150
+ "ru",
151
+ "ar",
152
+ "tr",
153
+ "zh",
154
+ "ja",
155
+ "ko",
156
+ "hi",
157
+ "vi"
158
+ ],
159
+ "id2lang": {
160
+ "0": "ar",
161
+ "1": "de",
162
+ "2": "en",
163
+ "3": "es",
164
+ "4": "fr",
165
+ "5": "hi",
166
+ "6": "it",
167
+ "7": "ja",
168
+ "8": "ko",
169
+ "9": "nl",
170
+ "10": "pl",
171
+ "11": "pt",
172
+ "12": "ru",
173
+ "13": "sv",
174
+ "14": "tr",
175
+ "15": "vi",
176
+ "16": "zh"
177
+ },
178
+ "lang2id": {
179
+ "ar": 0,
180
+ "de": 1,
181
+ "en": 2,
182
+ "es": 3,
183
+ "fr": 4,
184
+ "hi": 5,
185
+ "it": 6,
186
+ "ja": 7,
187
+ "ko": 8,
188
+ "nl": 9,
189
+ "pl": 10,
190
+ "pt": 11,
191
+ "ru": 12,
192
+ "sv": 13,
193
+ "tr": 14,
194
+ "vi": 15,
195
+ "zh": 16
196
+ },
197
+ "n_langs": 17,
198
+ "bt_src_langs": [],
199
+ "mono_dataset": {
200
+ "en": {
201
+ "train": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/train.en.pth",
202
+ "valid": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/valid.en.pth",
203
+ "test": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/test.en.pth"
204
+ },
205
+ "fr": {
206
+ "train": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/train.fr.pth",
207
+ "valid": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/valid.fr.pth",
208
+ "test": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/test.fr.pth"
209
+ },
210
+ "es": {
211
+ "train": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/train.es.pth",
212
+ "valid": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/valid.es.pth",
213
+ "test": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/test.es.pth"
214
+ },
215
+ "de": {
216
+ "train": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/train.de.pth",
217
+ "valid": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/valid.de.pth",
218
+ "test": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/test.de.pth"
219
+ },
220
+ "it": {
221
+ "train": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/train.it.pth",
222
+ "valid": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/valid.it.pth",
223
+ "test": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/test.it.pth"
224
+ },
225
+ "pt": {
226
+ "train": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/train.pt.pth",
227
+ "valid": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/valid.pt.pth",
228
+ "test": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/test.pt.pth"
229
+ },
230
+ "nl": {
231
+ "train": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/train.nl.pth",
232
+ "valid": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/valid.nl.pth",
233
+ "test": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/test.nl.pth"
234
+ },
235
+ "sv": {
236
+ "train": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/train.sv.pth",
237
+ "valid": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/valid.sv.pth",
238
+ "test": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/test.sv.pth"
239
+ },
240
+ "pl": {
241
+ "train": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/train.pl.pth",
242
+ "valid": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/valid.pl.pth",
243
+ "test": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/test.pl.pth"
244
+ },
245
+ "ru": {
246
+ "train": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/train.ru.pth",
247
+ "valid": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/valid.ru.pth",
248
+ "test": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/test.ru.pth"
249
+ },
250
+ "ar": {
251
+ "train": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/train.ar.pth",
252
+ "valid": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/valid.ar.pth",
253
+ "test": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/test.ar.pth"
254
+ },
255
+ "tr": {
256
+ "train": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/train.tr.pth",
257
+ "valid": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/valid.tr.pth",
258
+ "test": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/test.tr.pth"
259
+ },
260
+ "zh": {
261
+ "train": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/train.zh.pth",
262
+ "valid": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/valid.zh.pth",
263
+ "test": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/test.zh.pth"
264
+ },
265
+ "ja": {
266
+ "train": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/train.ja.pth",
267
+ "valid": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/valid.ja.pth",
268
+ "test": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/test.ja.pth"
269
+ },
270
+ "ko": {
271
+ "train": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/train.ko.pth",
272
+ "valid": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/valid.ko.pth",
273
+ "test": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/test.ko.pth"
274
+ },
275
+ "hi": {
276
+ "train": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/train.hi.pth",
277
+ "valid": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/valid.hi.pth",
278
+ "test": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/test.hi.pth"
279
+ },
280
+ "vi": {
281
+ "train": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/train.vi.pth",
282
+ "valid": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/valid.vi.pth",
283
+ "test": "/private/home/aconneau/projects/XLM/data/wiki/17/175k/test.vi.pth"
284
+ }
285
+ },
286
+ "para_dataset": {},
287
+ "word_mask": 0.8,
288
+ "word_keep": 0.1,
289
+ "word_rand": 0.1,
290
+ "is_slurm_job": true,
291
+ "n_nodes": 4,
292
+ "node_id": 0,
293
+ "global_rank": 0,
294
+ "world_size": 32,
295
+ "n_gpu_per_node": 8,
296
+ "master_addr": "learnfair1605",
297
+ "is_master": true,
298
+ "multi_node": true,
299
+ "multi_gpu": true,
300
+ "command": "python /private/home/aconneau/workdir/xlm_17_100_big.3/2019_08_10_19_23_42/train.py --n_heads 16 --bt_steps '' --max_vocab 200000 --word_mask_keep_rand '0.8,0.1,0.1' --use_lang_emb false --data_path '/private/home/aconneau/projects/XLM/data/wiki/17/175k' --save_periodic 0 --max_len 200 --bptt 256 --ae_steps '' --fp16 true --share_inout_emb true --sinusoidal_embeddings false --word_shuffle 0 --tokens_per_batch '-1' --accumulate_gradients 4 --validation_metrics '_valid_en_mlm_ppl,_valid_mlm_ppl,_valid_zh_mlm_ppl' --attention_dropout '0.1' --split_data true --max_epoch 100000 --stopping_criterion '_valid_zh_mlm_ppl,25' --dump_path '/checkpoint/aconneau/dumped' --epoch_size 200000 --word_blank 0 --gelu_activation true --n_layers 16 --optimizer 'adam_inverse_sqrt,lr=0.00005,warmup_updates=30000,beta1=0.9,beta2=0.999,weight_decay=0.01,eps=0.000001' --mlm_steps 'en,fr,es,de,it,pt,nl,sv,pl,ru,ar,tr,zh,ja,ko,hi,vi' --eval_bleu false --dropout '0.1' --mt_steps '' --batch_size 16 --word_dropout 0 --reload_model '/checkpoint/aconneau/dumped/xlm_17_100_240_big_model_upper.2/14884510/best-valid_zh_mlm_ppl.pth' --min_count 0 --amp 2 --group_by_size true --asm false --sample_alpha '0.5' --word_pred '0.15' --clip_grad_norm 1 --emb_dim 1280 --encoder_only true --lgs 'en-fr-es-de-it-pt-nl-sv-pl-ru-ar-tr-zh-ja-ko-hi-vi' --clm_steps '' --exp_name 'xlm_17_100_big.3' --lg_sampling_factor '0.7' --eval_only false --exp_id 16656237 --master_port 14148 --exp_id \"16656237\"",
301
+ "n_words": 200000,
302
+ "bos_index": 0,
303
+ "eos_index": 1,
304
+ "pad_index": 2,
305
+ "unk_index": 3,
306
+ "mask_index": 5,
307
+ "lambda_clm_config": null,
308
+ "lambda_mlm_config": null,
309
+ "lambda_pc_config": null,
310
+ "lambda_ae_config": null,
311
+ "lambda_mt_config": null,
312
+ "lambda_bt_config": null,
313
+ "hyp_path": "/checkpoint/aconneau/dumped/xlm_17_100_big.3/16656237/hypotheses",
314
+ "ref_paths": {},
315
+ "mono_list": [
316
+ "en",
317
+ "fr",
318
+ "es",
319
+ "de",
320
+ "it",
321
+ "pt",
322
+ "nl",
323
+ "sv",
324
+ "pl",
325
+ "ru",
326
+ "ar",
327
+ "tr",
328
+ "zh",
329
+ "ja",
330
+ "ko",
331
+ "hi",
332
+ "vi"
333
+ ],
334
+ "para_list": []
335
+ }