ethanpai commited on
Commit
4d5bdaa
1 Parent(s): 3fe82a3

Upload 6 files

Browse files
236a1657485f406c94cab962265e3f6c_tokenizer.vocab ADDED
@@ -0,0 +1,1024 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ <unk> 0
2
+ s -3.01191
3
+ ▁the -3.3064
4
+ ▁a -3.88763
5
+ t -3.9109
6
+ ▁to -3.9418
7
+ ▁and -4.00695
8
+ ▁that -4.1472
9
+ ' -4.14887
10
+ ▁of -4.21192
11
+ ▁i -4.28338
12
+ ▁in -4.3299
13
+ d -4.45986
14
+ ed -4.50274
15
+ ing -4.58105
16
+ ▁it -4.61438
17
+ ▁you -4.66118
18
+ n -4.71658
19
+ e -4.81146
20
+ ▁is -4.82865
21
+ y -4.84298
22
+ er -4.95427
23
+ ▁for -5.02561
24
+ m -5.07592
25
+ ▁was -5.09809
26
+ ▁we -5.12267
27
+ re -5.13215
28
+ ly -5.17261
29
+ r -5.19104
30
+ ▁he -5.20264
31
+ ▁s -5.29823
32
+ ▁be -5.30402
33
+ c -5.31557
34
+ ▁this -5.33569
35
+ ▁so -5.34881
36
+ ▁on -5.40047
37
+ ▁have -5.40956
38
+ g -5.41208
39
+ a -5.41635
40
+ ▁not -5.46278
41
+ or -5.46927
42
+ p -5.50585
43
+ ▁they -5.5213
44
+ ▁with -5.54336
45
+ o -5.55076
46
+ al -5.55786
47
+ ▁but -5.57088
48
+ ▁c -5.60206
49
+ ▁are -5.66857
50
+ ▁re -5.708
51
+ ▁there -5.7197
52
+ ar -5.72039
53
+ b -5.7282
54
+ l -5.73367
55
+ le -5.74736
56
+ ▁know -5.75832
57
+ ▁as -5.76391
58
+ in -5.77182
59
+ u -5.78957
60
+ ▁what -5.82451
61
+ ▁do -5.84196
62
+ ▁or -5.873
63
+ ▁ -5.87542
64
+ ▁at -5.90446
65
+ ation -5.9276
66
+ w -5.95113
67
+ ll -5.97891
68
+ f -5.98177
69
+ ▁one -5.98197
70
+ ▁e -5.994
71
+ ▁f -6.00195
72
+ ri -6.01002
73
+ ▁if -6.0142
74
+ ▁b -6.01851
75
+ ent -6.01951
76
+ ▁all -6.05459
77
+ ▁de -6.06291
78
+ ▁can -6.07891
79
+ it -6.09132
80
+ ▁would -6.09439
81
+ ▁me -6.09718
82
+ il -6.11202
83
+ ▁like -6.12096
84
+ ▁an -6.12651
85
+ ce -6.13747
86
+ ▁no -6.14346
87
+ ▁t -6.14376
88
+ ment -6.15481
89
+ i -6.19644
90
+ ve -6.20529
91
+ k -6.2056
92
+ ▁from -6.20712
93
+ ▁just -6.20928
94
+ ▁by -6.22157
95
+ ▁his -6.22687
96
+ ▁had -6.22896
97
+ ion -6.23268
98
+ ▁think -6.23785
99
+ es -6.26358
100
+ ch -6.28135
101
+ on -6.30523
102
+ ▁about -6.30789
103
+ th -6.32557
104
+ ▁my -6.3408
105
+ ck -6.35024
106
+ en -6.35335
107
+ ▁were -6.35478
108
+ ▁your -6.3676
109
+ ur -6.36794
110
+ ▁will -6.37536
111
+ ic -6.37739
112
+ ▁go -6.38958
113
+ ▁con -6.39725
114
+ ▁she -6.4097
115
+ li -6.41085
116
+ an -6.41939
117
+ ▁out -6.42092
118
+ ter -6.43866
119
+ ▁because -6.44229
120
+ se -6.4599
121
+ la -6.46081
122
+ ▁when -6.4621
123
+ ▁st -6.46785
124
+ ▁has -6.48439
125
+ ▁don -6.4864
126
+ ▁which -6.48954
127
+ ir -6.48961
128
+ ra -6.50111
129
+ ▁p -6.50674
130
+ ▁two -6.53343
131
+ ▁right -6.53353
132
+ ▁court -6.5447
133
+ ▁some -6.55176
134
+ ▁pa -6.55739
135
+ ▁pro -6.56366
136
+ lo -6.5659
137
+ ate -6.5719
138
+ ▁time -6.58352
139
+ ▁uh -6.594
140
+ te -6.60626
141
+ ▁up -6.60743
142
+ ▁our -6.60816
143
+ ▁g -6.6195
144
+ ▁case -6.62197
145
+ ▁w -6.62264
146
+ ▁her -6.63933
147
+ ▁well -6.64425
148
+ ro -6.64443
149
+ ▁their -6.66105
150
+ ▁who -6.66799
151
+ ▁been -6.66992
152
+ ▁yeah -6.67251
153
+ ▁said -6.68264
154
+ ers -6.69412
155
+ ▁mo -6.69492
156
+ ▁other -6.69575
157
+ ▁see -6.69594
158
+ ive -6.69949
159
+ tion -6.70273
160
+ el -6.7078
161
+ ▁more -6.73953
162
+ ▁se -6.74843
163
+ vi -6.76224
164
+ ▁any -6.76504
165
+ ▁get -6.76799
166
+ ▁also -6.77858
167
+ ▁fa -6.78673
168
+ ▁here -6.78945
169
+ ▁un -6.80325
170
+ ▁now -6.80498
171
+ ▁co -6.81085
172
+ ▁people -6.81307
173
+ age -6.81378
174
+ ist -6.8281
175
+ ul -6.84209
176
+ ▁how -6.84637
177
+ ▁going -6.85305
178
+ h -6.85447
179
+ ▁very -6.85772
180
+ ▁then -6.85982
181
+ ▁could -6.86291
182
+ de -6.87571
183
+ ▁ma -6.8759
184
+ ver -6.88014
185
+ ▁mean -6.89035
186
+ ▁su -6.89079
187
+ id -6.89223
188
+ ng -6.8924
189
+ ▁lo -6.89385
190
+ ▁sa -6.89406
191
+ ▁la -6.89707
192
+ ▁sp -6.8989
193
+ ▁us -6.90189
194
+ ol -6.90262
195
+ ▁pre -6.90805
196
+ ance -6.9084
197
+ ▁want -6.9086
198
+ ▁work -6.91911
199
+ ld -6.91946
200
+ ▁did -6.92095
201
+ ge -6.9262
202
+ ▁state -6.92731
203
+ ▁po -6.92829
204
+ ▁dis -6.92862
205
+ ity -6.93644
206
+ able -6.93702
207
+ ▁say -6.93874
208
+ ▁where -6.94155
209
+ et -6.95194
210
+ ▁ra -6.95885
211
+ us -6.96114
212
+ ▁li -6.96142
213
+ ry -6.96232
214
+ ction -6.9729
215
+ ▁um -6.97651
216
+ nd -6.97673
217
+ ne -6.98047
218
+ ▁even -6.98963
219
+ ▁him -6.9918
220
+ ▁ba -7.00031
221
+ ad -7.00067
222
+ ▁these -7.00076
223
+ ▁part -7.00877
224
+ ▁really -7.01556
225
+ lu -7.02494
226
+ ▁them -7.02693
227
+ ▁those -7.03211
228
+ ▁ho -7.03778
229
+ v -7.03819
230
+ ▁bo -7.03958
231
+ ▁ex -7.04583
232
+ ect -7.04619
233
+ ▁over -7.04859
234
+ ▁make -7.06531
235
+ ru -7.07049
236
+ ▁ro -7.07859
237
+ is -7.08085
238
+ ▁fi -7.09063
239
+ ▁m -7.09156
240
+ ut -7.093
241
+ ke -7.10501
242
+ ence -7.10553
243
+ ▁bu -7.11489
244
+ ▁ch -7.12158
245
+ ▁comp -7.12338
246
+ me -7.12813
247
+ ted -7.12898
248
+ ▁en -7.13022
249
+ ▁fact -7.13166
250
+ z -7.13457
251
+ ▁into -7.13497
252
+ ▁need -7.13935
253
+ ure -7.14199
254
+ ▁way -7.14219
255
+ ies -7.14845
256
+ end -7.15016
257
+ ▁mi -7.15039
258
+ ▁take -7.15331
259
+ un -7.15407
260
+ ▁only -7.15644
261
+ ▁should -7.15675
262
+ as -7.15799
263
+ ▁look -7.16519
264
+ ▁three -7.16796
265
+ one -7.1741
266
+ ▁v -7.17909
267
+ co -7.18057
268
+ ▁com -7.18358
269
+ ▁man -7.18977
270
+ ta -7.19395
271
+ ▁good -7.19512
272
+ ated -7.20268
273
+ ▁back -7.21128
274
+ ▁first -7.22203
275
+ ▁mr -7.22322
276
+ ▁point -7.23126
277
+ x -7.2324
278
+ ▁new -7.23993
279
+ ant -7.24022
280
+ ▁vi -7.24441
281
+ ▁dr -7.24764
282
+ ff -7.25476
283
+ ▁k -7.25479
284
+ ci -7.25611
285
+ ▁app -7.25994
286
+ um -7.26048
287
+ ▁n -7.2612
288
+ ▁exp -7.2663
289
+ ▁under -7.26684
290
+ ▁after -7.27957
291
+ po -7.28204
292
+ ight -7.28593
293
+ ▁d -7.2918
294
+ est -7.29294
295
+ mp -7.29424
296
+ ▁okay -7.29546
297
+ ine -7.30407
298
+ ti -7.31565
299
+ mo -7.3327
300
+ qu -7.33473
301
+ ▁da -7.33717
302
+ ▁come -7.33754
303
+ ▁day -7.33773
304
+ ▁oh -7.34295
305
+ ▁issue -7.3462
306
+ at -7.34798
307
+ ▁question -7.34902
308
+ ▁than -7.35427
309
+ ▁pri -7.35552
310
+ ho -7.35832
311
+ ten -7.37007
312
+ ▁ca -7.37394
313
+ ▁off -7.37399
314
+ ain -7.37402
315
+ ▁di -7.37622
316
+ ▁o -7.37697
317
+ ▁much -7.40521
318
+ ▁five -7.40649
319
+ ow -7.409
320
+ ▁ha -7.41643
321
+ ▁thousand -7.41742
322
+ ▁six -7.41909
323
+ ▁give -7.41948
324
+ ▁let -7.41955
325
+ ▁le -7.42609
326
+ ary -7.43195
327
+ ous -7.43602
328
+ sh -7.43735
329
+ ▁yes -7.43739
330
+ ▁four -7.43883
331
+ ▁before -7.44371
332
+ ���call -7.44795
333
+ ▁ga -7.45028
334
+ ▁ne -7.45328
335
+ ▁law -7.45555
336
+ ma -7.45726
337
+ ▁tra -7.45852
338
+ ish -7.46122
339
+ ical -7.4669
340
+ ▁something -7.47402
341
+ iv -7.47625
342
+ ▁why -7.48319
343
+ ▁may -7.49437
344
+ ▁talk -7.49684
345
+ ▁actually -7.50003
346
+ ig -7.50192
347
+ ▁act -7.50548
348
+ ▁got -7.50646
349
+ ▁car -7.50913
350
+ ▁through -7.51522
351
+ ful -7.51968
352
+ ▁made -7.52059
353
+ ▁hundred -7.5212
354
+ om -7.52654
355
+ man -7.52947
356
+ ia -7.53144
357
+ ▁didn -7.53557
358
+ ual -7.53782
359
+ ▁kind -7.54275
360
+ ▁sta -7.54474
361
+ ting -7.54877
362
+ ▁things -7.55296
363
+ ell -7.55834
364
+ ▁down -7.5587
365
+ im -7.5635
366
+ ▁years -7.56811
367
+ ▁lot -7.56935
368
+ ▁ri -7.57354
369
+ pe -7.57371
370
+ ard -7.57492
371
+ he -7.59531
372
+ ▁gra -7.59967
373
+ ▁year -7.60166
374
+ ▁every -7.60566
375
+ ▁eight -7.61078
376
+ ▁reason -7.6164
377
+ ▁seven -7.62338
378
+ ca -7.62471
379
+ ise -7.63057
380
+ ▁twenty -7.63185
381
+ ▁many -7.63236
382
+ ▁per -7.6346
383
+ op -7.63775
384
+ ▁little -7.64017
385
+ ▁mar -7.64481
386
+ ▁same -7.64687
387
+ ▁find -7.64884
388
+ ▁different -7.65231
389
+ ▁wa -7.65629
390
+ pp -7.65688
391
+ ▁nine -7.65728
392
+ if -7.65919
393
+ ▁does -7.65963
394
+ ie -7.66457
395
+ ach -7.66468
396
+ cu -7.66524
397
+ ▁never -7.66947
398
+ ▁thing -7.66974
399
+ ▁whether -7.67551
400
+ ward -7.67865
401
+ ▁again -7.6788
402
+ ▁put -7.68119
403
+ ▁claim -7.68143
404
+ ▁op -7.68506
405
+ ugh -7.68672
406
+ ▁show -7.68989
407
+ tic -7.69278
408
+ ial -7.69396
409
+ ▁still -7.69406
410
+ ▁government -7.69561
411
+ ok -7.69643
412
+ ▁th -7.69739
413
+ ▁long -7.70185
414
+ ▁gu -7.70497
415
+ ▁being -7.70527
416
+ ▁plan -7.7093
417
+ ▁own -7.71567
418
+ am -7.71603
419
+ ian -7.72075
420
+ ▁evidence -7.72253
421
+ ot -7.72507
422
+ mb -7.72698
423
+ ▁most -7.73078
424
+ ▁ru -7.73615
425
+ mit -7.73651
426
+ ▁thank -7.73695
427
+ ious -7.73763
428
+ ▁person -7.73973
429
+ em -7.74697
430
+ ▁use -7.74951
431
+ na -7.75065
432
+ ▁sha -7.75634
433
+ ▁r -7.75793
434
+ ▁believe -7.76261
435
+ mi -7.76681
436
+ ▁school -7.76682
437
+ ▁record -7.77672
438
+ ition -7.78101
439
+ ▁imp -7.79068
440
+ ound -7.79235
441
+ ▁place -7.79729
442
+ ities -7.79917
443
+ ▁happen -7.80132
444
+ ▁ph -7.80617
445
+ ▁great -7.80849
446
+ ative -7.8109
447
+ ha -7.81685
448
+ no -7.8192
449
+ ▁cha -7.82092
450
+ ite -7.82093
451
+ ▁clear -7.82261
452
+ ▁help -7.82338
453
+ dic -7.82599
454
+ ea -7.83059
455
+ ▁high -7.83497
456
+ ▁judge -7.83862
457
+ ▁al -7.84074
458
+ ▁last -7.84579
459
+ di -7.84803
460
+ qui -7.84818
461
+ ▁sure -7.85092
462
+ tra -7.85174
463
+ go -7.85242
464
+ ▁sh -7.85274
465
+ ▁both -7.86145
466
+ ▁read -7.86794
467
+ ving -7.87457
468
+ ▁district -7.87967
469
+ ▁ref -7.88067
470
+ ▁sub -7.88088
471
+ ▁hi -7.88215
472
+ ▁such -7.88573
473
+ ▁ten -7.88929
474
+ ▁start -7.89603
475
+ ▁doesn -7.91334
476
+ ph -7.91563
477
+ ▁change -7.91568
478
+ ay -7.91582
479
+ ▁count -7.91591
480
+ ▁ob -7.92581
481
+ ▁found -7.92942
482
+ ▁du -7.93296
483
+ ▁ki -7.93473
484
+ ▁pay -7.94173
485
+ ▁order -7.9429
486
+ ▁another -7.94315
487
+ for -7.94509
488
+ ▁hand -7.94629
489
+ lic -7.94858
490
+ ▁number -7.94898
491
+ ▁too -7.95303
492
+ ▁ta -7.9548
493
+ ni -7.95975
494
+ ens -7.96071
495
+ ▁tell -7.96191
496
+ ven -7.96383
497
+ port -7.96504
498
+ ▁fun -7.96575
499
+ ▁h -7.96811
500
+ ▁term -7.96982
501
+ ▁end -7.98241
502
+ rr -7.98451
503
+ ▁must -7.9904
504
+ ▁home -7.99293
505
+ ▁hu -7.99697
506
+ ▁wi -7.99775
507
+ side -8.01026
508
+ ▁certain -8.01058
509
+ ▁name -8.01068
510
+ ▁understand -8.01129
511
+ ▁money -8.01343
512
+ ize -8.01367
513
+ ▁pu -8.01477
514
+ ▁interest -8.01561
515
+ ip -8.01876
516
+ ▁came -8.01967
517
+ ▁house -8.02009
518
+ land -8.02103
519
+ ▁men -8.02226
520
+ ▁require -8.02343
521
+ ▁says -8.0254
522
+ min -8.02578
523
+ ▁member -8.02731
524
+ our -8.02825
525
+ ating -8.02935
526
+ ber -8.02962
527
+ ▁jo -8.03073
528
+ ▁might -8.03257
529
+ vo -8.0329
530
+ ▁support -8.03469
531
+ ries -8.03892
532
+ son -8.03904
533
+ ▁problem -8.04279
534
+ ling -8.0446
535
+ ▁old -8.04515
536
+ ▁pe -8.04547
537
+ ▁second -8.04767
538
+ ▁def -8.0477
539
+ ness -8.04891
540
+ ▁decision -8.05608
541
+ row -8.05741
542
+ ▁dollars -8.05786
543
+ ▁mu -8.05867
544
+ ▁conf -8.06149
545
+ use -8.06401
546
+ ▁ah -8.06759
547
+ ▁va -8.069
548
+ ▁life -8.07021
549
+ ▁inter -8.075
550
+ ▁play -8.07627
551
+ va -8.07788
552
+ ▁saying -8.08265
553
+ ▁pi -8.08295
554
+ ▁far -8.08369
555
+ ▁went -8.08624
556
+ ▁public -8.09081
557
+ ▁war -8.09371
558
+ ▁trans -8.09644
559
+ ▁live -8.09769
560
+ ▁ask -8.09779
561
+ ▁real -8.10281
562
+ rate -8.10664
563
+ ▁important -8.10777
564
+ ▁thought -8.11341
565
+ ▁done -8.11365
566
+ ▁between -8.11583
567
+ ▁argument -8.11685
568
+ ▁rep -8.12465
569
+ ▁consider -8.12509
570
+ ▁anything -8.12752
571
+ ▁direct -8.13753
572
+ less -8.1379
573
+ ▁ja -8.14377
574
+ ▁doing -8.14735
575
+ ▁course -8.14967
576
+ line -8.1498
577
+ ▁month -8.15089
578
+ ▁word -8.15292
579
+ ▁care -8.15391
580
+ ▁provide -8.15495
581
+ ap -8.15614
582
+ ▁honor -8.15843
583
+ ba -8.16065
584
+ ator -8.16143
585
+ j -8.16579
586
+ ▁report -8.16587
587
+ ▁exc -8.17717
588
+ ▁service -8.17727
589
+ ▁tri -8.17995
590
+ nk -8.18562
591
+ ▁world -8.18778
592
+ ▁without -8.19845
593
+ ▁ju -8.19868
594
+ ▁against -8.19954
595
+ ible -8.2001
596
+ ▁used -8.20453
597
+ ▁always -8.20522
598
+ ▁able -8.20625
599
+ ship -8.20658
600
+ ▁rule -8.21223
601
+ ▁matter -8.21235
602
+ ▁while -8.21387
603
+ par -8.21484
604
+ ▁maybe -8.21651
605
+ ▁miss -8.21888
606
+ ▁around -8.22213
607
+ ▁u -8.22807
608
+ ▁today -8.23514
609
+ ▁friend -8.23574
610
+ ▁ti -8.23677
611
+ ▁big -8.23908
612
+ bu -8.25207
613
+ ▁next -8.25218
614
+ ju -8.25437
615
+ ▁thirty -8.2615
616
+ ▁pass -8.26664
617
+ ▁y -8.26709
618
+ ▁business -8.2706
619
+ ▁test -8.27192
620
+ ▁fo -8.27206
621
+ ▁nothing -8.27358
622
+ ▁turn -8.27658
623
+ ▁process -8.277
624
+ com -8.27878
625
+ ▁el -8.28
626
+ ▁better -8.28086
627
+ ▁commission -8.28106
628
+ ▁week -8.28551
629
+ ▁move -8.28817
630
+ ▁trial -8.2933
631
+ ▁open -8.29365
632
+ ▁each -8.29395
633
+ ▁specific -8.29571
634
+ ▁feel -8.29613
635
+ ▁cap -8.29661
636
+ ency -8.29835
637
+ ▁particular -8.29895
638
+ ability -8.30361
639
+ ▁system -8.30875
640
+ gra -8.3112
641
+ ▁present -8.31276
642
+ ▁looking -8.3154
643
+ duc -8.32012
644
+ ▁fra -8.32345
645
+ ▁correct -8.32378
646
+ ▁allow -8.32432
647
+ ▁side -8.33521
648
+ ign -8.33656
649
+ serv -8.34918
650
+ ▁city -8.35114
651
+ ▁agree -8.35225
652
+ ▁having -8.35541
653
+ ▁defendant -8.35977
654
+ ▁keep -8.36067
655
+ ▁follow -8.36086
656
+ ▁invest -8.36103
657
+ ▁add -8.36234
658
+ ▁information -8.36336
659
+ ▁whole -8.36452
660
+ ▁general -8.36773
661
+ hi -8.37286
662
+ ▁away -8.37759
663
+ ▁upon -8.37843
664
+ ▁win -8.37876
665
+ ▁based -8.38012
666
+ ▁program -8.38366
667
+ ▁develop -8.38375
668
+ ▁country -8.38516
669
+ ified -8.38819
670
+ ▁sort -8.39013
671
+ ▁board -8.3916
672
+ ▁qua -8.39656
673
+ ex -8.39857
674
+ ▁family -8.39988
675
+ ▁close -8.40114
676
+ ▁dec -8.4027
677
+ ▁though -8.40303
678
+ ▁hard -8.40316
679
+ ▁answer -8.40517
680
+ ▁guess -8.40829
681
+ ▁job -8.41187
682
+ ▁trying -8.4167
683
+ ▁je -8.42679
684
+ ▁asked -8.42774
685
+ ▁during -8.42911
686
+ ▁fee -8.42947
687
+ ▁love -8.43046
688
+ ▁already -8.4339
689
+ ▁address -8.43928
690
+ ▁ever -8.44234
691
+ ▁bill -8.44324
692
+ gue -8.45149
693
+ ▁cause -8.45364
694
+ ▁gen -8.45576
695
+ ▁free -8.45745
696
+ ▁continue -8.45946
697
+ ▁fifty -8.46775
698
+ ▁contract -8.46969
699
+ ▁european -8.47013
700
+ ▁rest -8.47015
701
+ ▁few -8.47047
702
+ ▁water -8.47107
703
+ ▁line -8.47546
704
+ ▁position -8.47646
705
+ ▁percent -8.47781
706
+ ▁enough -8.48453
707
+ ▁sea -8.48764
708
+ ▁har -8.48974
709
+ ▁america -8.49179
710
+ ▁full -8.50284
711
+ ▁since -8.50427
712
+ ▁everything -8.50663
713
+ ▁bring -8.50735
714
+ ▁lead -8.50767
715
+ ▁probably -8.5078
716
+ ▁children -8.50966
717
+ ▁tax -8.51183
718
+ ▁face -8.51201
719
+ ▁best -8.51422
720
+ ▁concern -8.51433
721
+ ▁sign -8.51531
722
+ ▁mhm -8.5172
723
+ ▁minute -8.5245
724
+ ▁bit -8.52588
725
+ ▁stand -8.53396
726
+ ▁situation -8.53488
727
+ ▁speak -8.53944
728
+ ▁hope -8.53986
729
+ ▁yet -8.54098
730
+ ▁discuss -8.54652
731
+ ▁standard -8.54739
732
+ ▁brief -8.54929
733
+ ▁statute -8.55189
734
+ ▁officer -8.55328
735
+ ▁ok -8.55347
736
+ ▁respect -8.55373
737
+ ▁together -8.55387
738
+ ▁agreement -8.55436
739
+ ▁room -8.55507
740
+ ▁least -8.55609
741
+ ▁market -8.56207
742
+ ▁however -8.56304
743
+ ▁getting -8.5648
744
+ ▁else -8.56715
745
+ ▁true -8.5735
746
+ ▁counsel -8.57393
747
+ ttle -8.57534
748
+ ▁err -8.57653
749
+ ▁effect -8.58066
750
+ ▁example -8.58199
751
+ ▁less -8.58583
752
+ ▁cost -8.58671
753
+ ▁left -8.587
754
+ ▁group -8.58704
755
+ ▁level -8.58828
756
+ ▁policy -8.59079
757
+ ▁town -8.59082
758
+ ▁small -8.59091
759
+ ▁child -8.5911
760
+ ▁president -8.59252
761
+ ▁large -8.5941
762
+ ▁later -8.5968
763
+ ▁protect -8.59839
764
+ ▁decide -8.59865
765
+ way -8.60153
766
+ ▁making -8.60336
767
+ ▁regard -8.60362
768
+ ▁book -8.60408
769
+ ▁pretty -8.60445
770
+ ▁author -8.60544
771
+ ▁hour -8.61483
772
+ ▁idea -8.61874
773
+ ▁include -8.61964
774
+ ▁exactly -8.61994
775
+ ▁power -8.62075
776
+ ▁limit -8.62432
777
+ ▁watch -8.62485
778
+ ▁night -8.62807
779
+ ▁ya -8.63635
780
+ ▁jury -8.63694
781
+ ▁leave -8.63812
782
+ ▁quite -8.63956
783
+ ▁cover -8.64034
784
+ ▁grow -8.64087
785
+ ▁result -8.64918
786
+ ▁singapore -8.6529
787
+ ▁either -8.65431
788
+ ▁company -8.65657
789
+ ▁stop -8.66169
790
+ ▁community -8.66982
791
+ ▁million -8.67059
792
+ ▁current -8.67357
793
+ ▁federal -8.67921
794
+ ▁united -8.681
795
+ ▁mind -8.68365
796
+ ▁step -8.68439
797
+ ▁class -8.69014
798
+ ▁police -8.69578
799
+ ▁remain -8.70047
800
+ ▁health -8.70131
801
+ ▁final -8.7037
802
+ ▁represent -8.70732
803
+ ▁light -8.7173
804
+ ▁treat -8.72451
805
+ ▁review -8.72717
806
+ ▁testimony -8.72966
807
+ ▁bank -8.73168
808
+ ▁product -8.7353
809
+ ▁appeal -8.73612
810
+ ▁half -8.73638
811
+ ▁cannot -8.74236
812
+ ▁involve -8.7433
813
+ ▁charge -8.7451
814
+ ▁knew -8.74528
815
+ ▁learn -8.74844
816
+ ▁eye -8.7509
817
+ ▁language -8.75207
818
+ ▁individual -8.75576
819
+ ▁receive -8.7588
820
+ ▁further -8.76324
821
+ ▁mister -8.76849
822
+ ▁view -8.77119
823
+ ▁council -8.77187
824
+ ▁until -8.77872
825
+ ▁human -8.78283
826
+ ▁stuff -8.78933
827
+ ▁student -8.79001
828
+ ▁become -8.79074
829
+ ▁walk -8.79421
830
+ ▁control -8.79606
831
+ ▁check -8.7976
832
+ ▁seems -8.80605
833
+ ▁strong -8.80668
834
+ ▁period -8.80835
835
+ ▁please -8.81052
836
+ ▁food -8.81059
837
+ ▁type -8.81071
838
+ ▁complete -8.81714
839
+ ▁purpose -8.8189
840
+ ▁account -8.82531
841
+ ▁project -8.82686
842
+ ▁sense -8.82864
843
+ ▁national -8.82924
844
+ ▁appear -8.8307
845
+ ▁simply -8.8355
846
+ ▁return -8.83742
847
+ ▁benefit -8.8388
848
+ ▁employ -8.84023
849
+ ▁create -8.84074
850
+ ▁wrong -8.84311
851
+ ▁whatever -8.84539
852
+ ▁morning -8.84604
853
+ ▁suggest -8.8498
854
+ ▁condition -8.85096
855
+ ▁amount -8.85197
856
+ ▁object -8.85388
857
+ ▁third -8.85609
858
+ ▁remember -8.85706
859
+ ▁himself -8.8648
860
+ ▁property -8.86667
861
+ ▁itself -8.87131
862
+ ▁somebody -8.87283
863
+ ▁suppose -8.87729
864
+ ▁future -8.8805
865
+ ▁short -8.88536
866
+ ▁opinion -8.88537
867
+ ▁several -8.8888
868
+ ▁common -8.89514
869
+ ▁taking -8.8994
870
+ ▁notice -8.90291
871
+ ▁office -8.90755
872
+ ▁almost -8.91074
873
+ ▁father -8.91522
874
+ ▁document -8.93135
875
+ ▁judgment -8.93517
876
+ ▁experience -8.93708
877
+ ▁uhhuh -8.94282
878
+ ▁committee -8.94464
879
+ ▁parent -8.94658
880
+ ▁legal -8.95078
881
+ ▁client -8.95282
882
+ ▁major -8.95286
883
+ ▁congress -8.95347
884
+ ▁rather -8.95844
885
+ ▁proceed -8.95903
886
+ ▁possible -8.96184
887
+ ▁plea -8.96986
888
+ ▁subject -8.97233
889
+ ▁along -8.97325
890
+ ▁mother -8.97346
891
+ ▁increase -8.97621
892
+ ▁meeting -8.97677
893
+ ▁girl -8.97813
894
+ ▁couple -8.98003
895
+ ▁north -8.98205
896
+ ▁local -8.98769
897
+ ▁sometimes -8.98901
898
+ ▁brought -8.99432
899
+ ▁fifteen -8.99524
900
+ ▁organ -8.9989
901
+ ▁difference -8.99953
902
+ ▁entire -9.01016
903
+ ▁security -9.01059
904
+ ▁department -9.01178
905
+ ▁south -9.01416
906
+ ▁vote -9.01545
907
+ ▁basis -9.01546
908
+ ▁twelve -9.02257
909
+ ▁conduct -9.02285
910
+ ▁street -9.02304
911
+ ▁social -9.02308
912
+ ▁wonder -9.02355
913
+ ▁supreme -9.02736
914
+ ▁difficult -9.02982
915
+ ▁travel -9.03377
916
+ ▁opportunity -9.03544
917
+ ▁defense -9.03614
918
+ ▁plaintiff -9.03962
919
+ ▁white -9.04234
920
+ ▁focus -9.04245
921
+ ▁therefore -9.04306
922
+ ▁challenge -9.04331
923
+ ▁value -9.04772
924
+ ▁except -9.05048
925
+ ▁among -9.06512
926
+ ▁ground -9.06849
927
+ ▁serve -9.07044
928
+ ▁attorney -9.07851
929
+ ▁education -9.08087
930
+ ▁credit -9.08163
931
+ ▁request -9.08265
932
+ ▁women -9.088
933
+ ▁basically -9.09031
934
+ ▁political -9.09232
935
+ ▁determine -9.09492
936
+ ▁access -9.10648
937
+ ▁across -9.10743
938
+ ▁sentence -9.10825
939
+ ▁perform -9.10892
940
+ ▁everybody -9.10925
941
+ ▁black -9.10926
942
+ ▁drug -9.10967
943
+ ▁accept -9.1116
944
+ ▁eleven -9.11364
945
+ ▁establish -9.11478
946
+ ▁design -9.11658
947
+ ▁success -9.11797
948
+ ▁including -9.12041
949
+ ▁john -9.12115
950
+ ▁employee -9.13045
951
+ ▁nice -9.13525
952
+ ▁crime -9.13568
953
+ ibility -9.13792
954
+ ▁circumstance -9.14069
955
+ ▁occur -9.14127
956
+ ▁citizen -9.14442
957
+ ▁woman -9.14602
958
+ ▁witness -9.14622
959
+ ▁explain -9.14644
960
+ ▁additional -9.14665
961
+ ▁history -9.14927
962
+ ▁front -9.15337
963
+ ▁medical -9.15616
964
+ ▁death -9.15855
965
+ ification -9.16062
966
+ ▁financial -9.16478
967
+ ▁break -9.16866
968
+ ▁connect -9.17213
969
+ ▁perhaps -9.1739
970
+ ▁sorry -9.17422
971
+ ▁especially -9.1758
972
+ ▁trust -9.17594
973
+ ▁insurance -9.18285
974
+ ▁economic -9.18607
975
+ ▁effort -9.19186
976
+ ▁circuit -9.19196
977
+ ▁cross -9.19591
978
+ ▁instead -9.19723
979
+ ▁affect -9.19903
980
+ ▁zero -9.20141
981
+ ▁available -9.20769
982
+ ▁team -9.21706
983
+ ▁application -9.2189
984
+ ▁separate -9.22192
985
+ ▁absolutely -9.22227
986
+ ▁christ -9.22667
987
+ ization -9.23734
988
+ ▁europe -9.23791
989
+ ▁budget -9.23896
990
+ ▁damage -9.23976
991
+ ▁myself -9.24506
992
+ ▁private -9.24722
993
+ ▁figure -9.25259
994
+ ▁provision -9.25344
995
+ ▁collect -9.25536
996
+ ▁similar -9.25629
997
+ ▁regulation -9.26179
998
+ ▁approach -9.26478
999
+ ▁themselves -9.27531
1000
+ ▁construct -9.27705
1001
+ ▁petition -9.27737
1002
+ ▁material -9.27916
1003
+ ▁improve -9.27992
1004
+ ▁york -9.28463
1005
+ ▁parliament -9.29099
1006
+ ▁broad -9.29759
1007
+ ▁companies -9.29798
1008
+ ▁nature -9.29833
1009
+ ▁piece -9.30056
1010
+ ▁threat -9.30893
1011
+ ▁energy -9.31421
1012
+ ▁behind -9.32084
1013
+ ▁significant -9.32477
1014
+ ▁response -9.33036
1015
+ ▁definitely -9.33051
1016
+ ▁guide -9.33241
1017
+ ▁beyond -9.33485
1018
+ ▁amendment -9.33552
1019
+ ▁measure -9.33664
1020
+ ▁criminal -9.33766
1021
+ ▁original -9.3398
1022
+ ▁brother -9.34819
1023
+ ▁impact -9.35544
1024
+ q -9.42163
2503ec8f45bf4a04b852c0b302a973e8_tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6fdbb245a2b0020c7466e1f2b2913fe5bc282a6981c78f2366b6662ed0905f6d
3
+ size 260690
a4398a6af7324038a05d5930e49f4cf2_vocab.txt ADDED
@@ -0,0 +1,1023 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ##s
2
+ the
3
+ a
4
+ ##t
5
+ to
6
+ and
7
+ that
8
+ ##'
9
+ of
10
+ i
11
+ in
12
+ ##d
13
+ ##ed
14
+ ##ing
15
+ it
16
+ you
17
+ ##n
18
+ ##e
19
+ is
20
+ ##y
21
+ ##er
22
+ for
23
+ ##m
24
+ was
25
+ we
26
+ ##re
27
+ ##ly
28
+ ##r
29
+ he
30
+ s
31
+ be
32
+ ##c
33
+ this
34
+ so
35
+ on
36
+ have
37
+ ##g
38
+ ##a
39
+ not
40
+ ##or
41
+ ##p
42
+ they
43
+ with
44
+ ##o
45
+ ##al
46
+ but
47
+ c
48
+ are
49
+ re
50
+ there
51
+ ##ar
52
+ ##b
53
+ ##l
54
+ ##le
55
+ know
56
+ as
57
+ ##in
58
+ ##u
59
+ what
60
+ do
61
+ or
62
+
63
+ at
64
+ ##ation
65
+ ##w
66
+ ##ll
67
+ ##f
68
+ one
69
+ e
70
+ f
71
+ ##ri
72
+ if
73
+ b
74
+ ##ent
75
+ all
76
+ de
77
+ can
78
+ ##it
79
+ would
80
+ me
81
+ ##il
82
+ like
83
+ an
84
+ ##ce
85
+ no
86
+ t
87
+ ##ment
88
+ ##i
89
+ ##ve
90
+ ##k
91
+ from
92
+ just
93
+ by
94
+ his
95
+ had
96
+ ##ion
97
+ think
98
+ ##es
99
+ ##ch
100
+ ##on
101
+ about
102
+ ##th
103
+ my
104
+ ##ck
105
+ ##en
106
+ were
107
+ your
108
+ ##ur
109
+ will
110
+ ##ic
111
+ go
112
+ con
113
+ she
114
+ ##li
115
+ ##an
116
+ out
117
+ ##ter
118
+ because
119
+ ##se
120
+ ##la
121
+ when
122
+ st
123
+ has
124
+ don
125
+ which
126
+ ##ir
127
+ ##ra
128
+ p
129
+ two
130
+ right
131
+ court
132
+ some
133
+ pa
134
+ pro
135
+ ##lo
136
+ ##ate
137
+ time
138
+ uh
139
+ ##te
140
+ up
141
+ our
142
+ g
143
+ case
144
+ w
145
+ her
146
+ well
147
+ ##ro
148
+ their
149
+ who
150
+ been
151
+ yeah
152
+ said
153
+ ##ers
154
+ mo
155
+ other
156
+ see
157
+ ##ive
158
+ ##tion
159
+ ##el
160
+ more
161
+ se
162
+ ##vi
163
+ any
164
+ get
165
+ also
166
+ fa
167
+ here
168
+ un
169
+ now
170
+ co
171
+ people
172
+ ##age
173
+ ##ist
174
+ ##ul
175
+ how
176
+ going
177
+ ##h
178
+ very
179
+ then
180
+ could
181
+ ##de
182
+ ma
183
+ ##ver
184
+ mean
185
+ su
186
+ ##id
187
+ ##ng
188
+ lo
189
+ sa
190
+ la
191
+ sp
192
+ us
193
+ ##ol
194
+ pre
195
+ ##ance
196
+ want
197
+ work
198
+ ##ld
199
+ did
200
+ ##ge
201
+ state
202
+ po
203
+ dis
204
+ ##ity
205
+ ##able
206
+ say
207
+ where
208
+ ##et
209
+ ra
210
+ ##us
211
+ li
212
+ ##ry
213
+ ##ction
214
+ um
215
+ ##nd
216
+ ##ne
217
+ even
218
+ him
219
+ ba
220
+ ##ad
221
+ these
222
+ part
223
+ really
224
+ ##lu
225
+ them
226
+ those
227
+ ho
228
+ ##v
229
+ bo
230
+ ex
231
+ ##ect
232
+ over
233
+ make
234
+ ##ru
235
+ ro
236
+ ##is
237
+ fi
238
+ m
239
+ ##ut
240
+ ##ke
241
+ ##ence
242
+ bu
243
+ ch
244
+ comp
245
+ ##me
246
+ ##ted
247
+ en
248
+ fact
249
+ ##z
250
+ into
251
+ need
252
+ ##ure
253
+ way
254
+ ##ies
255
+ ##end
256
+ mi
257
+ take
258
+ ##un
259
+ only
260
+ should
261
+ ##as
262
+ look
263
+ three
264
+ ##one
265
+ v
266
+ ##co
267
+ com
268
+ man
269
+ ##ta
270
+ good
271
+ ##ated
272
+ back
273
+ first
274
+ mr
275
+ point
276
+ ##x
277
+ new
278
+ ##ant
279
+ vi
280
+ dr
281
+ ##ff
282
+ k
283
+ ##ci
284
+ app
285
+ ##um
286
+ n
287
+ exp
288
+ under
289
+ after
290
+ ##po
291
+ ##ight
292
+ d
293
+ ##est
294
+ ##mp
295
+ okay
296
+ ##ine
297
+ ##ti
298
+ ##mo
299
+ ##qu
300
+ da
301
+ come
302
+ day
303
+ oh
304
+ issue
305
+ ##at
306
+ question
307
+ than
308
+ pri
309
+ ##ho
310
+ ##ten
311
+ ca
312
+ off
313
+ ##ain
314
+ di
315
+ o
316
+ much
317
+ five
318
+ ##ow
319
+ ha
320
+ thousand
321
+ six
322
+ give
323
+ let
324
+ le
325
+ ##ary
326
+ ##ous
327
+ ##sh
328
+ yes
329
+ four
330
+ before
331
+ call
332
+ ga
333
+ ne
334
+ law
335
+ ##ma
336
+ tra
337
+ ##ish
338
+ ##ical
339
+ something
340
+ ##iv
341
+ why
342
+ may
343
+ talk
344
+ actually
345
+ ##ig
346
+ act
347
+ got
348
+ car
349
+ through
350
+ ##ful
351
+ made
352
+ hundred
353
+ ##om
354
+ ##man
355
+ ##ia
356
+ didn
357
+ ##ual
358
+ kind
359
+ sta
360
+ ##ting
361
+ things
362
+ ##ell
363
+ down
364
+ ##im
365
+ years
366
+ lot
367
+ ri
368
+ ##pe
369
+ ##ard
370
+ ##he
371
+ gra
372
+ year
373
+ every
374
+ eight
375
+ reason
376
+ seven
377
+ ##ca
378
+ ##ise
379
+ twenty
380
+ many
381
+ per
382
+ ##op
383
+ little
384
+ mar
385
+ same
386
+ find
387
+ different
388
+ wa
389
+ ##pp
390
+ nine
391
+ ##if
392
+ does
393
+ ##ie
394
+ ##ach
395
+ ##cu
396
+ never
397
+ thing
398
+ whether
399
+ ##ward
400
+ again
401
+ put
402
+ claim
403
+ op
404
+ ##ugh
405
+ show
406
+ ##tic
407
+ ##ial
408
+ still
409
+ government
410
+ ##ok
411
+ th
412
+ long
413
+ gu
414
+ being
415
+ plan
416
+ own
417
+ ##am
418
+ ##ian
419
+ evidence
420
+ ##ot
421
+ ##mb
422
+ most
423
+ ru
424
+ ##mit
425
+ thank
426
+ ##ious
427
+ person
428
+ ##em
429
+ use
430
+ ##na
431
+ sha
432
+ r
433
+ believe
434
+ ##mi
435
+ school
436
+ record
437
+ ##ition
438
+ imp
439
+ ##ound
440
+ place
441
+ ##ities
442
+ happen
443
+ ph
444
+ great
445
+ ##ative
446
+ ##ha
447
+ ##no
448
+ cha
449
+ ##ite
450
+ clear
451
+ help
452
+ ##dic
453
+ ##ea
454
+ high
455
+ judge
456
+ al
457
+ last
458
+ ##di
459
+ ##qui
460
+ sure
461
+ ##tra
462
+ ##go
463
+ sh
464
+ both
465
+ read
466
+ ##ving
467
+ district
468
+ ref
469
+ sub
470
+ hi
471
+ such
472
+ ten
473
+ start
474
+ doesn
475
+ ##ph
476
+ change
477
+ ##ay
478
+ count
479
+ ob
480
+ found
481
+ du
482
+ ki
483
+ pay
484
+ order
485
+ another
486
+ ##for
487
+ hand
488
+ ##lic
489
+ number
490
+ too
491
+ ta
492
+ ##ni
493
+ ##ens
494
+ tell
495
+ ##ven
496
+ ##port
497
+ fun
498
+ h
499
+ term
500
+ end
501
+ ##rr
502
+ must
503
+ home
504
+ hu
505
+ wi
506
+ ##side
507
+ certain
508
+ name
509
+ understand
510
+ money
511
+ ##ize
512
+ pu
513
+ interest
514
+ ##ip
515
+ came
516
+ house
517
+ ##land
518
+ men
519
+ require
520
+ says
521
+ ##min
522
+ member
523
+ ##our
524
+ ##ating
525
+ ##ber
526
+ jo
527
+ might
528
+ ##vo
529
+ support
530
+ ##ries
531
+ ##son
532
+ problem
533
+ ##ling
534
+ old
535
+ pe
536
+ second
537
+ def
538
+ ##ness
539
+ decision
540
+ ##row
541
+ dollars
542
+ mu
543
+ conf
544
+ ##use
545
+ ah
546
+ va
547
+ life
548
+ inter
549
+ play
550
+ ##va
551
+ saying
552
+ pi
553
+ far
554
+ went
555
+ public
556
+ war
557
+ trans
558
+ live
559
+ ask
560
+ real
561
+ ##rate
562
+ important
563
+ thought
564
+ done
565
+ between
566
+ argument
567
+ rep
568
+ consider
569
+ anything
570
+ direct
571
+ ##less
572
+ ja
573
+ doing
574
+ course
575
+ ##line
576
+ month
577
+ word
578
+ care
579
+ provide
580
+ ##ap
581
+ honor
582
+ ##ba
583
+ ##ator
584
+ ##j
585
+ report
586
+ exc
587
+ service
588
+ tri
589
+ ##nk
590
+ world
591
+ without
592
+ ju
593
+ against
594
+ ##ible
595
+ used
596
+ always
597
+ able
598
+ ##ship
599
+ rule
600
+ matter
601
+ while
602
+ ##par
603
+ maybe
604
+ miss
605
+ around
606
+ u
607
+ today
608
+ friend
609
+ ti
610
+ big
611
+ ##bu
612
+ next
613
+ ##ju
614
+ thirty
615
+ pass
616
+ y
617
+ business
618
+ test
619
+ fo
620
+ nothing
621
+ turn
622
+ process
623
+ ##com
624
+ el
625
+ better
626
+ commission
627
+ week
628
+ move
629
+ trial
630
+ open
631
+ each
632
+ specific
633
+ feel
634
+ cap
635
+ ##ency
636
+ particular
637
+ ##ability
638
+ system
639
+ ##gra
640
+ present
641
+ looking
642
+ ##duc
643
+ fra
644
+ correct
645
+ allow
646
+ side
647
+ ##ign
648
+ ##serv
649
+ city
650
+ agree
651
+ having
652
+ defendant
653
+ keep
654
+ follow
655
+ invest
656
+ add
657
+ information
658
+ whole
659
+ general
660
+ ##hi
661
+ away
662
+ upon
663
+ win
664
+ based
665
+ program
666
+ develop
667
+ country
668
+ ##ified
669
+ sort
670
+ board
671
+ qua
672
+ ##ex
673
+ family
674
+ close
675
+ dec
676
+ though
677
+ hard
678
+ answer
679
+ guess
680
+ job
681
+ trying
682
+ je
683
+ asked
684
+ during
685
+ fee
686
+ love
687
+ already
688
+ address
689
+ ever
690
+ bill
691
+ ##gue
692
+ cause
693
+ gen
694
+ free
695
+ continue
696
+ fifty
697
+ contract
698
+ european
699
+ rest
700
+ few
701
+ water
702
+ line
703
+ position
704
+ percent
705
+ enough
706
+ sea
707
+ har
708
+ america
709
+ full
710
+ since
711
+ everything
712
+ bring
713
+ lead
714
+ probably
715
+ children
716
+ tax
717
+ face
718
+ best
719
+ concern
720
+ sign
721
+ mhm
722
+ minute
723
+ bit
724
+ stand
725
+ situation
726
+ speak
727
+ hope
728
+ yet
729
+ discuss
730
+ standard
731
+ brief
732
+ statute
733
+ officer
734
+ ok
735
+ respect
736
+ together
737
+ agreement
738
+ room
739
+ least
740
+ market
741
+ however
742
+ getting
743
+ else
744
+ true
745
+ counsel
746
+ ##ttle
747
+ err
748
+ effect
749
+ example
750
+ less
751
+ cost
752
+ left
753
+ group
754
+ level
755
+ policy
756
+ town
757
+ small
758
+ child
759
+ president
760
+ large
761
+ later
762
+ protect
763
+ decide
764
+ ##way
765
+ making
766
+ regard
767
+ book
768
+ pretty
769
+ author
770
+ hour
771
+ idea
772
+ include
773
+ exactly
774
+ power
775
+ limit
776
+ watch
777
+ night
778
+ ya
779
+ jury
780
+ leave
781
+ quite
782
+ cover
783
+ grow
784
+ result
785
+ singapore
786
+ either
787
+ company
788
+ stop
789
+ community
790
+ million
791
+ current
792
+ federal
793
+ united
794
+ mind
795
+ step
796
+ class
797
+ police
798
+ remain
799
+ health
800
+ final
801
+ represent
802
+ light
803
+ treat
804
+ review
805
+ testimony
806
+ bank
807
+ product
808
+ appeal
809
+ half
810
+ cannot
811
+ involve
812
+ charge
813
+ knew
814
+ learn
815
+ eye
816
+ language
817
+ individual
818
+ receive
819
+ further
820
+ mister
821
+ view
822
+ council
823
+ until
824
+ human
825
+ stuff
826
+ student
827
+ become
828
+ walk
829
+ control
830
+ check
831
+ seems
832
+ strong
833
+ period
834
+ please
835
+ food
836
+ type
837
+ complete
838
+ purpose
839
+ account
840
+ project
841
+ sense
842
+ national
843
+ appear
844
+ simply
845
+ return
846
+ benefit
847
+ employ
848
+ create
849
+ wrong
850
+ whatever
851
+ morning
852
+ suggest
853
+ condition
854
+ amount
855
+ object
856
+ third
857
+ remember
858
+ himself
859
+ property
860
+ itself
861
+ somebody
862
+ suppose
863
+ future
864
+ short
865
+ opinion
866
+ several
867
+ common
868
+ taking
869
+ notice
870
+ office
871
+ almost
872
+ father
873
+ document
874
+ judgment
875
+ experience
876
+ uhhuh
877
+ committee
878
+ parent
879
+ legal
880
+ client
881
+ major
882
+ congress
883
+ rather
884
+ proceed
885
+ possible
886
+ plea
887
+ subject
888
+ along
889
+ mother
890
+ increase
891
+ meeting
892
+ girl
893
+ couple
894
+ north
895
+ local
896
+ sometimes
897
+ brought
898
+ fifteen
899
+ organ
900
+ difference
901
+ entire
902
+ security
903
+ department
904
+ south
905
+ vote
906
+ basis
907
+ twelve
908
+ conduct
909
+ street
910
+ social
911
+ wonder
912
+ supreme
913
+ difficult
914
+ travel
915
+ opportunity
916
+ defense
917
+ plaintiff
918
+ white
919
+ focus
920
+ therefore
921
+ challenge
922
+ value
923
+ except
924
+ among
925
+ ground
926
+ serve
927
+ attorney
928
+ education
929
+ credit
930
+ request
931
+ women
932
+ basically
933
+ political
934
+ determine
935
+ access
936
+ across
937
+ sentence
938
+ perform
939
+ everybody
940
+ black
941
+ drug
942
+ accept
943
+ eleven
944
+ establish
945
+ design
946
+ success
947
+ including
948
+ john
949
+ employee
950
+ nice
951
+ crime
952
+ ##ibility
953
+ circumstance
954
+ occur
955
+ citizen
956
+ woman
957
+ witness
958
+ explain
959
+ additional
960
+ history
961
+ front
962
+ medical
963
+ death
964
+ ##ification
965
+ financial
966
+ break
967
+ connect
968
+ perhaps
969
+ sorry
970
+ especially
971
+ trust
972
+ insurance
973
+ economic
974
+ effort
975
+ circuit
976
+ cross
977
+ instead
978
+ affect
979
+ zero
980
+ available
981
+ team
982
+ application
983
+ separate
984
+ absolutely
985
+ christ
986
+ ##ization
987
+ europe
988
+ budget
989
+ damage
990
+ myself
991
+ private
992
+ figure
993
+ provision
994
+ collect
995
+ similar
996
+ regulation
997
+ approach
998
+ themselves
999
+ construct
1000
+ petition
1001
+ material
1002
+ improve
1003
+ york
1004
+ parliament
1005
+ broad
1006
+ companies
1007
+ nature
1008
+ piece
1009
+ threat
1010
+ energy
1011
+ behind
1012
+ significant
1013
+ response
1014
+ definitely
1015
+ guide
1016
+ beyond
1017
+ amendment
1018
+ measure
1019
+ criminal
1020
+ original
1021
+ brother
1022
+ impact
1023
+ ##q
model_config.yaml ADDED
@@ -0,0 +1,3276 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ sample_rate: 16000
2
+ compute_eval_loss: false
3
+ log_prediction: true
4
+ model_defaults:
5
+ enc_hidden: 512
6
+ pred_hidden: 640
7
+ joint_hidden: 640
8
+ train_ds:
9
+ manifest_filepath:
10
+ - - /raid/local//bucket1/tarred_audio_manifest.json
11
+ - - /raid/local//bucket2/tarred_audio_manifest.json
12
+ - - /raid/local//bucket3/tarred_audio_manifest.json
13
+ - - /raid/local//bucket4/tarred_audio_manifest.json
14
+ - - /raid/local//bucket5/tarred_audio_manifest.json
15
+ - - /raid/local//bucket6/tarred_audio_manifest.json
16
+ - - /raid/local//bucket7/tarred_audio_manifest.json
17
+ - - /raid/local//bucket8/tarred_audio_manifest.json
18
+ sample_rate: 16000
19
+ batch_size: 1
20
+ shuffle: true
21
+ num_workers: 4
22
+ pin_memory: true
23
+ use_start_end_token: false
24
+ trim_silence: false
25
+ max_duration: 25
26
+ min_duration: 0.1
27
+ is_tarred: true
28
+ tarred_audio_filepaths:
29
+ - - /data2/nemo_asr/nemo_asr_set_3.0//bucket1/audio__OP_0..8191_CL_.tar
30
+ - - /data2/nemo_asr/nemo_asr_set_3.0//bucket2/audio__OP_0..8191_CL_.tar
31
+ - - /data2/nemo_asr/nemo_asr_set_3.0//bucket3/audio__OP_0..8191_CL_.tar
32
+ - - /data2/nemo_asr/nemo_asr_set_3.0//bucket4/audio__OP_0..8191_CL_.tar
33
+ - - /data2/nemo_asr/nemo_asr_set_3.0//bucket5/audio__OP_0..8191_CL_.tar
34
+ - - /data2/nemo_asr/nemo_asr_set_3.0//bucket6/audio__OP_0..8191_CL_.tar
35
+ - - /data2/nemo_asr/nemo_asr_set_3.0//bucket7/audio__OP_0..8191_CL_.tar
36
+ - - /data2/nemo_asr/nemo_asr_set_3.0//bucket8/audio__OP_0..8191_CL_.tar
37
+ shuffle_n: 2048
38
+ bucketing_strategy: fully_randomized
39
+ bucketing_batch_size:
40
+ - 72
41
+ - 64
42
+ - 56
43
+ - 48
44
+ - 40
45
+ - 32
46
+ - 24
47
+ - 16
48
+ validation_ds:
49
+ manifest_filepath:
50
+ - /manifests/librispeech/librivox-dev-other.json
51
+ - /manifests/librispeech/librivox-dev-clean.json
52
+ - /manifests/librispeech/librivox-test-other.json
53
+ - /manifests/librispeech/librivox-test-clean.json
54
+ sample_rate: 16000
55
+ batch_size: 16
56
+ shuffle: false
57
+ num_workers: 8
58
+ pin_memory: true
59
+ use_start_end_token: false
60
+ test_ds:
61
+ manifest_filepath:
62
+ - /manifests/librispeech/librivox-dev-other.json
63
+ - /manifests/librispeech/librivox-dev-clean.json
64
+ - /manifests/librispeech/librivox-test-other.json
65
+ - /manifests/librispeech/librivox-test-clean.json
66
+ sample_rate: 16000
67
+ batch_size: 16
68
+ shuffle: false
69
+ num_workers: 8
70
+ pin_memory: true
71
+ use_start_end_token: false
72
+ tokenizer:
73
+ dir: /tokenizers/NeMo_ASR_SET/English/asr_set_3.0/tokenizer_spe_unigram_v1024/
74
+ type: bpe
75
+ model_path: nemo:2503ec8f45bf4a04b852c0b302a973e8_tokenizer.model
76
+ vocab_path: nemo:a4398a6af7324038a05d5930e49f4cf2_vocab.txt
77
+ spe_tokenizer_vocab: nemo:236a1657485f406c94cab962265e3f6c_tokenizer.vocab
78
+ preprocessor:
79
+ _target_: nemo.collections.asr.modules.AudioToMelSpectrogramPreprocessor
80
+ sample_rate: 16000
81
+ normalize: NA
82
+ window_size: 0.025
83
+ window_stride: 0.01
84
+ window: hann
85
+ features: 80
86
+ n_fft: 512
87
+ frame_splicing: 1
88
+ dither: 1.0e-05
89
+ pad_to: 0
90
+ spec_augment:
91
+ _target_: nemo.collections.asr.modules.SpectrogramAugmentation
92
+ freq_masks: 2
93
+ time_masks: 10
94
+ freq_width: 27
95
+ time_width: 0.05
96
+ encoder:
97
+ _target_: nemo.collections.asr.modules.ConformerEncoder
98
+ feat_in: 80
99
+ feat_out: -1
100
+ n_layers: 17
101
+ d_model: 512
102
+ subsampling: dw_striding
103
+ subsampling_factor: 8
104
+ subsampling_conv_channels: 256
105
+ causal_downsampling: true
106
+ reduction: null
107
+ reduction_position: null
108
+ reduction_factor: 1
109
+ ff_expansion_factor: 4
110
+ self_attention_model: rel_pos
111
+ n_heads: 8
112
+ att_context_size:
113
+ - - 70
114
+ - 13
115
+ - - 70
116
+ - 6
117
+ - - 70
118
+ - 1
119
+ - - 70
120
+ - 0
121
+ att_context_style: chunked_limited
122
+ xscaling: true
123
+ untie_biases: true
124
+ pos_emb_max_len: 5000
125
+ conv_kernel_size: 9
126
+ conv_norm_type: layer_norm
127
+ conv_context_size: causal
128
+ dropout: 0.1
129
+ dropout_pre_encoder: 0.1
130
+ dropout_emb: 0.0
131
+ dropout_att: 0.1
132
+ decoder:
133
+ _target_: nemo.collections.asr.modules.RNNTDecoder
134
+ normalization_mode: null
135
+ random_state_sampling: false
136
+ blank_as_pad: true
137
+ prednet:
138
+ pred_hidden: 640
139
+ pred_rnn_layers: 1
140
+ t_max: null
141
+ dropout: 0.2
142
+ vocab_size: 1024
143
+ joint:
144
+ _target_: nemo.collections.asr.modules.RNNTJoint
145
+ log_softmax: null
146
+ preserve_memory: false
147
+ fuse_loss_wer: true
148
+ fused_batch_size: 2
149
+ jointnet:
150
+ joint_hidden: 640
151
+ activation: relu
152
+ dropout: 0.2
153
+ encoder_hidden: 512
154
+ pred_hidden: 640
155
+ num_classes: 1024
156
+ vocabulary:
157
+ - <unk>
158
+ - s
159
+ - ▁the
160
+ - ▁a
161
+ - t
162
+ - ▁to
163
+ - ▁and
164
+ - ▁that
165
+ - ''''
166
+ - ▁of
167
+ - ▁i
168
+ - ▁in
169
+ - d
170
+ - ed
171
+ - ing
172
+ - ▁it
173
+ - ▁you
174
+ - 'n'
175
+ - e
176
+ - ▁is
177
+ - 'y'
178
+ - er
179
+ - ▁for
180
+ - m
181
+ - ▁was
182
+ - ▁we
183
+ - re
184
+ - ly
185
+ - r
186
+ - ▁he
187
+ - ▁s
188
+ - ▁be
189
+ - c
190
+ - ▁this
191
+ - ▁so
192
+ - ▁on
193
+ - ▁have
194
+ - g
195
+ - a
196
+ - ▁not
197
+ - or
198
+ - p
199
+ - ▁they
200
+ - ▁with
201
+ - o
202
+ - al
203
+ - ▁but
204
+ - ▁c
205
+ - ▁are
206
+ - ▁re
207
+ - ▁there
208
+ - ar
209
+ - b
210
+ - l
211
+ - le
212
+ - ▁know
213
+ - ▁as
214
+ - in
215
+ - u
216
+ - ▁what
217
+ - ▁do
218
+ - ▁or
219
+ - ▁
220
+ - ▁at
221
+ - ation
222
+ - w
223
+ - ll
224
+ - f
225
+ - ▁one
226
+ - ��e
227
+ - ▁f
228
+ - ri
229
+ - ▁if
230
+ - ▁b
231
+ - ent
232
+ - ▁all
233
+ - ▁de
234
+ - ▁can
235
+ - it
236
+ - ▁would
237
+ - ▁me
238
+ - il
239
+ - ▁like
240
+ - ▁an
241
+ - ce
242
+ - ▁no
243
+ - ▁t
244
+ - ment
245
+ - i
246
+ - ve
247
+ - k
248
+ - ▁from
249
+ - ▁just
250
+ - ▁by
251
+ - ▁his
252
+ - ▁had
253
+ - ion
254
+ - ▁think
255
+ - es
256
+ - ch
257
+ - 'on'
258
+ - ▁about
259
+ - th
260
+ - ▁my
261
+ - ck
262
+ - en
263
+ - ▁were
264
+ - ▁your
265
+ - ur
266
+ - ▁will
267
+ - ic
268
+ - ▁go
269
+ - ▁con
270
+ - ▁she
271
+ - li
272
+ - an
273
+ - ▁out
274
+ - ter
275
+ - ▁because
276
+ - se
277
+ - la
278
+ - ▁when
279
+ - ▁st
280
+ - ▁has
281
+ - ▁don
282
+ - ▁which
283
+ - ir
284
+ - ra
285
+ - ▁p
286
+ - ▁two
287
+ - ▁right
288
+ - ▁court
289
+ - ▁some
290
+ - ▁pa
291
+ - ▁pro
292
+ - lo
293
+ - ate
294
+ - ▁time
295
+ - ▁uh
296
+ - te
297
+ - ▁up
298
+ - ▁our
299
+ - ▁g
300
+ - ▁case
301
+ - ▁w
302
+ - ▁her
303
+ - ▁well
304
+ - ro
305
+ - ▁their
306
+ - ▁who
307
+ - ▁been
308
+ - ▁yeah
309
+ - ▁said
310
+ - ers
311
+ - ▁mo
312
+ - ▁other
313
+ - ▁see
314
+ - ive
315
+ - tion
316
+ - el
317
+ - ▁more
318
+ - ▁se
319
+ - vi
320
+ - ▁any
321
+ - ▁get
322
+ - ▁also
323
+ - ▁fa
324
+ - ▁here
325
+ - ▁un
326
+ - ▁now
327
+ - ▁co
328
+ - ▁people
329
+ - age
330
+ - ist
331
+ - ul
332
+ - ▁how
333
+ - ▁going
334
+ - h
335
+ - ▁very
336
+ - ▁then
337
+ - ▁could
338
+ - de
339
+ - ▁ma
340
+ - ver
341
+ - ▁mean
342
+ - ▁su
343
+ - id
344
+ - ng
345
+ - ▁lo
346
+ - ▁sa
347
+ - ▁la
348
+ - ▁sp
349
+ - ▁us
350
+ - ol
351
+ - ▁pre
352
+ - ance
353
+ - ▁want
354
+ - ▁work
355
+ - ld
356
+ - ▁did
357
+ - ge
358
+ - ▁state
359
+ - ▁po
360
+ - ▁dis
361
+ - ity
362
+ - able
363
+ - ▁say
364
+ - ▁where
365
+ - et
366
+ - ▁ra
367
+ - us
368
+ - ▁li
369
+ - ry
370
+ - ction
371
+ - ▁um
372
+ - nd
373
+ - ne
374
+ - ▁even
375
+ - ▁him
376
+ - ▁ba
377
+ - ad
378
+ - ▁these
379
+ - ▁part
380
+ - ▁really
381
+ - lu
382
+ - ▁them
383
+ - ▁those
384
+ - ▁ho
385
+ - v
386
+ - ▁bo
387
+ - ▁ex
388
+ - ect
389
+ - ▁over
390
+ - ▁make
391
+ - ru
392
+ - ▁ro
393
+ - is
394
+ - ▁fi
395
+ - ▁m
396
+ - ut
397
+ - ke
398
+ - ence
399
+ - ▁bu
400
+ - ▁ch
401
+ - ▁comp
402
+ - me
403
+ - ted
404
+ - ▁en
405
+ - ▁fact
406
+ - z
407
+ - ▁into
408
+ - ▁need
409
+ - ure
410
+ - ▁way
411
+ - ies
412
+ - end
413
+ - ▁mi
414
+ - ▁take
415
+ - un
416
+ - ▁only
417
+ - ▁should
418
+ - as
419
+ - ▁look
420
+ - ▁three
421
+ - one
422
+ - ▁v
423
+ - co
424
+ - ▁com
425
+ - ▁man
426
+ - ta
427
+ - ▁good
428
+ - ated
429
+ - ▁back
430
+ - ▁first
431
+ - ▁mr
432
+ - ▁point
433
+ - x
434
+ - ▁new
435
+ - ant
436
+ - ▁vi
437
+ - ▁dr
438
+ - ff
439
+ - ▁k
440
+ - ci
441
+ - ▁app
442
+ - um
443
+ - ▁n
444
+ - ▁exp
445
+ - ▁under
446
+ - ▁after
447
+ - po
448
+ - ight
449
+ - ▁d
450
+ - est
451
+ - mp
452
+ - ▁okay
453
+ - ine
454
+ - ti
455
+ - mo
456
+ - qu
457
+ - ▁da
458
+ - ▁come
459
+ - ▁day
460
+ - ▁oh
461
+ - ▁issue
462
+ - at
463
+ - ▁question
464
+ - ▁than
465
+ - ▁pri
466
+ - ho
467
+ - ten
468
+ - ▁ca
469
+ - ▁off
470
+ - ain
471
+ - ▁di
472
+ - ▁o
473
+ - ▁much
474
+ - ▁five
475
+ - ow
476
+ - ▁ha
477
+ - ▁thousand
478
+ - ▁six
479
+ - ▁give
480
+ - ▁let
481
+ - ▁le
482
+ - ary
483
+ - ous
484
+ - sh
485
+ - ▁yes
486
+ - ▁four
487
+ - ▁before
488
+ - ▁call
489
+ - ▁ga
490
+ - ▁ne
491
+ - ▁law
492
+ - ma
493
+ - ▁tra
494
+ - ish
495
+ - ical
496
+ - ▁something
497
+ - iv
498
+ - ▁why
499
+ - ▁may
500
+ - ▁talk
501
+ - ▁actually
502
+ - ig
503
+ - ▁act
504
+ - ▁got
505
+ - ▁car
506
+ - ▁through
507
+ - ful
508
+ - ▁made
509
+ - ▁hundred
510
+ - om
511
+ - man
512
+ - ia
513
+ - ▁didn
514
+ - ual
515
+ - ▁kind
516
+ - ▁sta
517
+ - ting
518
+ - ▁things
519
+ - ell
520
+ - ▁down
521
+ - im
522
+ - ▁years
523
+ - ▁lot
524
+ - ▁ri
525
+ - pe
526
+ - ard
527
+ - he
528
+ - ▁gra
529
+ - ▁year
530
+ - ▁every
531
+ - ▁eight
532
+ - ▁reason
533
+ - ▁seven
534
+ - ca
535
+ - ise
536
+ - ▁twenty
537
+ - ▁many
538
+ - ▁per
539
+ - op
540
+ - ▁little
541
+ - ▁mar
542
+ - ▁same
543
+ - ▁find
544
+ - ▁different
545
+ - ▁wa
546
+ - pp
547
+ - ▁nine
548
+ - if
549
+ - ▁does
550
+ - ie
551
+ - ach
552
+ - cu
553
+ - ▁never
554
+ - ▁thing
555
+ - ▁whether
556
+ - ward
557
+ - ▁again
558
+ - ▁put
559
+ - ▁claim
560
+ - ▁op
561
+ - ugh
562
+ - ▁show
563
+ - tic
564
+ - ial
565
+ - ▁still
566
+ - ▁government
567
+ - ok
568
+ - ▁th
569
+ - ▁long
570
+ - ▁gu
571
+ - ▁being
572
+ - ▁plan
573
+ - ▁own
574
+ - am
575
+ - ian
576
+ - ▁evidence
577
+ - ot
578
+ - mb
579
+ - ▁most
580
+ - ▁ru
581
+ - mit
582
+ - ▁thank
583
+ - ious
584
+ - ▁person
585
+ - em
586
+ - ▁use
587
+ - na
588
+ - ▁sha
589
+ - ▁r
590
+ - ▁believe
591
+ - mi
592
+ - ▁school
593
+ - ▁record
594
+ - ition
595
+ - ▁imp
596
+ - ound
597
+ - ▁place
598
+ - ities
599
+ - ▁happen
600
+ - ▁ph
601
+ - ▁great
602
+ - ative
603
+ - ha
604
+ - 'no'
605
+ - ▁cha
606
+ - ite
607
+ - ▁clear
608
+ - ▁help
609
+ - dic
610
+ - ea
611
+ - ▁high
612
+ - ▁judge
613
+ - ▁al
614
+ - ▁last
615
+ - di
616
+ - qui
617
+ - ▁sure
618
+ - tra
619
+ - go
620
+ - ▁sh
621
+ - ▁both
622
+ - ▁read
623
+ - ving
624
+ - ▁district
625
+ - ▁ref
626
+ - ▁sub
627
+ - ▁hi
628
+ - ▁such
629
+ - ▁ten
630
+ - ▁start
631
+ - ▁doesn
632
+ - ph
633
+ - ▁change
634
+ - ay
635
+ - ▁count
636
+ - ▁ob
637
+ - ▁found
638
+ - ▁du
639
+ - ▁ki
640
+ - ▁pay
641
+ - ▁order
642
+ - ▁another
643
+ - for
644
+ - ▁hand
645
+ - lic
646
+ - ▁number
647
+ - ▁too
648
+ - ▁ta
649
+ - ni
650
+ - ens
651
+ - ▁tell
652
+ - ven
653
+ - port
654
+ - ▁fun
655
+ - ▁h
656
+ - ▁term
657
+ - ▁end
658
+ - rr
659
+ - ▁must
660
+ - ▁home
661
+ - ▁hu
662
+ - ▁wi
663
+ - side
664
+ - ▁certain
665
+ - ▁name
666
+ - ▁understand
667
+ - ▁money
668
+ - ize
669
+ - ▁pu
670
+ - ▁interest
671
+ - ip
672
+ - ▁came
673
+ - ▁house
674
+ - land
675
+ - ▁men
676
+ - ▁require
677
+ - ▁says
678
+ - min
679
+ - ▁member
680
+ - our
681
+ - ating
682
+ - ber
683
+ - ▁jo
684
+ - ▁might
685
+ - vo
686
+ - ▁support
687
+ - ries
688
+ - son
689
+ - ▁problem
690
+ - ling
691
+ - ▁old
692
+ - ▁pe
693
+ - ▁second
694
+ - ▁def
695
+ - ness
696
+ - ▁decision
697
+ - row
698
+ - ▁dollars
699
+ - ▁mu
700
+ - ▁conf
701
+ - use
702
+ - ▁ah
703
+ - ▁va
704
+ - ▁life
705
+ - ▁inter
706
+ - ▁play
707
+ - va
708
+ - ▁saying
709
+ - ▁pi
710
+ - ▁far
711
+ - ▁went
712
+ - ▁public
713
+ - ▁war
714
+ - ▁trans
715
+ - ▁live
716
+ - ▁ask
717
+ - ▁real
718
+ - rate
719
+ - ▁important
720
+ - ▁thought
721
+ - ▁done
722
+ - ▁between
723
+ - ▁argument
724
+ - ▁rep
725
+ - ▁consider
726
+ - ▁anything
727
+ - ▁direct
728
+ - less
729
+ - ▁ja
730
+ - ▁doing
731
+ - ▁course
732
+ - line
733
+ - ▁month
734
+ - ▁word
735
+ - ▁care
736
+ - ▁provide
737
+ - ap
738
+ - ▁honor
739
+ - ba
740
+ - ator
741
+ - j
742
+ - ▁report
743
+ - ▁exc
744
+ - ▁service
745
+ - ▁tri
746
+ - nk
747
+ - ▁world
748
+ - ▁without
749
+ - ▁ju
750
+ - ▁against
751
+ - ible
752
+ - ▁used
753
+ - ▁always
754
+ - ▁able
755
+ - ship
756
+ - ▁rule
757
+ - ▁matter
758
+ - ▁while
759
+ - par
760
+ - ▁maybe
761
+ - ▁miss
762
+ - ▁around
763
+ - ▁u
764
+ - ▁today
765
+ - ▁friend
766
+ - ▁ti
767
+ - ▁big
768
+ - bu
769
+ - ▁next
770
+ - ju
771
+ - ▁thirty
772
+ - ▁pass
773
+ - ▁y
774
+ - ▁business
775
+ - ▁test
776
+ - ▁fo
777
+ - ▁nothing
778
+ - ▁turn
779
+ - ▁process
780
+ - com
781
+ - ▁el
782
+ - ▁better
783
+ - ▁commission
784
+ - ▁week
785
+ - ▁move
786
+ - ▁trial
787
+ - ▁open
788
+ - ▁each
789
+ - ▁specific
790
+ - ▁feel
791
+ - ▁cap
792
+ - ency
793
+ - ▁particular
794
+ - ability
795
+ - ▁system
796
+ - gra
797
+ - ▁present
798
+ - ▁looking
799
+ - duc
800
+ - ▁fra
801
+ - ▁correct
802
+ - ▁allow
803
+ - ▁side
804
+ - ign
805
+ - serv
806
+ - ▁city
807
+ - ▁agree
808
+ - ▁having
809
+ - ▁defendant
810
+ - ▁keep
811
+ - ▁follow
812
+ - ▁invest
813
+ - ▁add
814
+ - ▁information
815
+ - ▁whole
816
+ - ▁general
817
+ - hi
818
+ - ▁away
819
+ - ▁upon
820
+ - ▁win
821
+ - ▁based
822
+ - ▁program
823
+ - ▁develop
824
+ - ▁country
825
+ - ified
826
+ - ▁sort
827
+ - ▁board
828
+ - ▁qua
829
+ - ex
830
+ - ▁family
831
+ - ▁close
832
+ - ▁dec
833
+ - ▁though
834
+ - ▁hard
835
+ - ▁answer
836
+ - ▁guess
837
+ - ▁job
838
+ - ▁trying
839
+ - ▁je
840
+ - ▁asked
841
+ - ▁during
842
+ - ▁fee
843
+ - ▁love
844
+ - ▁already
845
+ - ▁address
846
+ - ▁ever
847
+ - ▁bill
848
+ - gue
849
+ - ▁cause
850
+ - ▁gen
851
+ - ▁free
852
+ - ▁continue
853
+ - ▁fifty
854
+ - ▁contract
855
+ - ▁european
856
+ - ▁rest
857
+ - ▁few
858
+ - ▁water
859
+ - ▁line
860
+ - ▁position
861
+ - ▁percent
862
+ - ▁enough
863
+ - ▁sea
864
+ - ▁har
865
+ - ▁america
866
+ - ▁full
867
+ - ▁since
868
+ - ▁everything
869
+ - ▁bring
870
+ - ▁lead
871
+ - ▁probably
872
+ - ▁children
873
+ - ▁tax
874
+ - ▁face
875
+ - ▁best
876
+ - ▁concern
877
+ - ▁sign
878
+ - ▁mhm
879
+ - ▁minute
880
+ - ▁bit
881
+ - ▁stand
882
+ - ▁situation
883
+ - ▁speak
884
+ - ▁hope
885
+ - ▁yet
886
+ - ▁discuss
887
+ - ▁standard
888
+ - ▁brief
889
+ - ▁statute
890
+ - ▁officer
891
+ - ▁ok
892
+ - ▁respect
893
+ - ▁together
894
+ - ▁agreement
895
+ - ▁room
896
+ - ▁least
897
+ - ▁market
898
+ - ▁however
899
+ - ▁getting
900
+ - ▁else
901
+ - ▁true
902
+ - ▁counsel
903
+ - ttle
904
+ - ▁err
905
+ - ▁effect
906
+ - ▁example
907
+ - ▁less
908
+ - ▁cost
909
+ - ▁left
910
+ - ▁group
911
+ - ▁level
912
+ - ▁policy
913
+ - ▁town
914
+ - ▁small
915
+ - ▁child
916
+ - ▁president
917
+ - ▁large
918
+ - ▁later
919
+ - ▁protect
920
+ - ▁decide
921
+ - way
922
+ - ▁making
923
+ - ▁regard
924
+ - ▁book
925
+ - ▁pretty
926
+ - ▁author
927
+ - ▁hour
928
+ - ▁idea
929
+ - ▁include
930
+ - ▁exactly
931
+ - ▁power
932
+ - ▁limit
933
+ - ▁watch
934
+ - ▁night
935
+ - ▁ya
936
+ - ▁jury
937
+ - ▁leave
938
+ - ▁quite
939
+ - ▁cover
940
+ - ▁grow
941
+ - ▁result
942
+ - ▁singapore
943
+ - ▁either
944
+ - ▁company
945
+ - ▁stop
946
+ - ▁community
947
+ - ▁million
948
+ - ▁current
949
+ - ▁federal
950
+ - ▁united
951
+ - ▁mind
952
+ - ▁step
953
+ - ▁class
954
+ - ▁police
955
+ - ▁remain
956
+ - ▁health
957
+ - ▁final
958
+ - ▁represent
959
+ - ▁light
960
+ - ▁treat
961
+ - ▁review
962
+ - ▁testimony
963
+ - ▁bank
964
+ - ▁product
965
+ - ▁appeal
966
+ - ▁half
967
+ - ▁cannot
968
+ - ▁involve
969
+ - ▁charge
970
+ - ▁knew
971
+ - ▁learn
972
+ - ▁eye
973
+ - ▁language
974
+ - ▁individual
975
+ - ▁receive
976
+ - ▁further
977
+ - ▁mister
978
+ - ▁view
979
+ - ▁council
980
+ - ▁until
981
+ - ▁human
982
+ - ▁stuff
983
+ - ▁student
984
+ - ▁become
985
+ - ▁walk
986
+ - ▁control
987
+ - ▁check
988
+ - ▁seems
989
+ - ▁strong
990
+ - ▁period
991
+ - ▁please
992
+ - ▁food
993
+ - ▁type
994
+ - ▁complete
995
+ - ▁purpose
996
+ - ▁account
997
+ - ▁project
998
+ - ▁sense
999
+ - ▁national
1000
+ - ▁appear
1001
+ - ▁simply
1002
+ - ▁return
1003
+ - ▁benefit
1004
+ - ▁employ
1005
+ - ▁create
1006
+ - ▁wrong
1007
+ - ▁whatever
1008
+ - ▁morning
1009
+ - ▁suggest
1010
+ - ▁condition
1011
+ - ▁amount
1012
+ - ▁object
1013
+ - ▁third
1014
+ - ▁remember
1015
+ - ▁himself
1016
+ - ▁property
1017
+ - ▁itself
1018
+ - ▁somebody
1019
+ - ▁suppose
1020
+ - ▁future
1021
+ - ▁short
1022
+ - ▁opinion
1023
+ - ▁several
1024
+ - ▁common
1025
+ - ▁taking
1026
+ - ▁notice
1027
+ - ▁office
1028
+ - ▁almost
1029
+ - ▁father
1030
+ - ▁document
1031
+ - ▁judgment
1032
+ - ▁experience
1033
+ - ▁uhhuh
1034
+ - ▁committee
1035
+ - ▁parent
1036
+ - ▁legal
1037
+ - ▁client
1038
+ - ▁major
1039
+ - ▁congress
1040
+ - ▁rather
1041
+ - ▁proceed
1042
+ - ▁possible
1043
+ - ▁plea
1044
+ - ▁subject
1045
+ - ▁along
1046
+ - ▁mother
1047
+ - ▁increase
1048
+ - ▁meeting
1049
+ - ▁girl
1050
+ - ▁couple
1051
+ - ▁north
1052
+ - ▁local
1053
+ - ▁sometimes
1054
+ - ▁brought
1055
+ - ▁fifteen
1056
+ - ▁organ
1057
+ - ▁difference
1058
+ - ▁entire
1059
+ - ▁security
1060
+ - ▁department
1061
+ - ▁south
1062
+ - ▁vote
1063
+ - ▁basis
1064
+ - ▁twelve
1065
+ - ▁conduct
1066
+ - ▁street
1067
+ - ▁social
1068
+ - ▁wonder
1069
+ - ▁supreme
1070
+ - ▁difficult
1071
+ - ▁travel
1072
+ - ▁opportunity
1073
+ - ▁defense
1074
+ - ▁plaintiff
1075
+ - ▁white
1076
+ - ▁focus
1077
+ - ▁therefore
1078
+ - ▁challenge
1079
+ - ▁value
1080
+ - ▁except
1081
+ - ▁among
1082
+ - ▁ground
1083
+ - ▁serve
1084
+ - ▁attorney
1085
+ - ▁education
1086
+ - ▁credit
1087
+ - ▁request
1088
+ - ▁women
1089
+ - ▁basically
1090
+ - ▁political
1091
+ - ▁determine
1092
+ - ▁access
1093
+ - ▁across
1094
+ - ▁sentence
1095
+ - ▁perform
1096
+ - ▁everybody
1097
+ - ▁black
1098
+ - ▁drug
1099
+ - ▁accept
1100
+ - ▁eleven
1101
+ - ▁establish
1102
+ - ▁design
1103
+ - ▁success
1104
+ - ▁including
1105
+ - ▁john
1106
+ - ▁employee
1107
+ - ▁nice
1108
+ - ▁crime
1109
+ - ibility
1110
+ - ▁circumstance
1111
+ - ▁occur
1112
+ - ▁citizen
1113
+ - ▁woman
1114
+ - ▁witness
1115
+ - ▁explain
1116
+ - ▁additional
1117
+ - ▁history
1118
+ - ▁front
1119
+ - ▁medical
1120
+ - ▁death
1121
+ - ification
1122
+ - ▁financial
1123
+ - ▁break
1124
+ - ▁connect
1125
+ - ▁perhaps
1126
+ - ▁sorry
1127
+ - ▁especially
1128
+ - ▁trust
1129
+ - ▁insurance
1130
+ - ▁economic
1131
+ - ▁effort
1132
+ - ▁circuit
1133
+ - ▁cross
1134
+ - ▁instead
1135
+ - ▁affect
1136
+ - ▁zero
1137
+ - ▁available
1138
+ - ▁team
1139
+ - ▁application
1140
+ - ▁separate
1141
+ - ▁absolutely
1142
+ - ▁christ
1143
+ - ization
1144
+ - ▁europe
1145
+ - ▁budget
1146
+ - ▁damage
1147
+ - ▁myself
1148
+ - ▁private
1149
+ - ▁figure
1150
+ - ▁provision
1151
+ - ▁collect
1152
+ - ▁similar
1153
+ - ▁regulation
1154
+ - ▁approach
1155
+ - ▁themselves
1156
+ - ▁construct
1157
+ - ▁petition
1158
+ - ▁material
1159
+ - ▁improve
1160
+ - ▁york
1161
+ - ▁parliament
1162
+ - ▁broad
1163
+ - ▁companies
1164
+ - ▁nature
1165
+ - ▁piece
1166
+ - ▁threat
1167
+ - ▁energy
1168
+ - ▁behind
1169
+ - ▁significant
1170
+ - ▁response
1171
+ - ▁definitely
1172
+ - ▁guide
1173
+ - ▁beyond
1174
+ - ▁amendment
1175
+ - ▁measure
1176
+ - ▁criminal
1177
+ - ▁original
1178
+ - ▁brother
1179
+ - ▁impact
1180
+ - q
1181
+ decoding:
1182
+ strategy: greedy_batch
1183
+ greedy:
1184
+ max_symbols: 10
1185
+ beam:
1186
+ beam_size: 5
1187
+ return_best_hypothesis: false
1188
+ score_norm: true
1189
+ tsd_max_sym_exp: 50
1190
+ alsd_max_target_len: 2.0
1191
+ aux_ctc:
1192
+ ctc_loss_weight: 0.3
1193
+ use_cer: false
1194
+ ctc_reduction: mean_batch
1195
+ decoder:
1196
+ _target_: nemo.collections.asr.modules.ConvASRDecoder
1197
+ feat_in: 512
1198
+ num_classes: 1024
1199
+ vocabulary:
1200
+ - <unk>
1201
+ - s
1202
+ - ▁the
1203
+ - ▁a
1204
+ - t
1205
+ - ▁to
1206
+ - ▁and
1207
+ - ▁that
1208
+ - ''''
1209
+ - ▁of
1210
+ - ▁i
1211
+ - ▁in
1212
+ - d
1213
+ - ed
1214
+ - ing
1215
+ - ▁it
1216
+ - ▁you
1217
+ - 'n'
1218
+ - e
1219
+ - ▁is
1220
+ - 'y'
1221
+ - er
1222
+ - ▁for
1223
+ - m
1224
+ - ▁was
1225
+ - ▁we
1226
+ - re
1227
+ - ly
1228
+ - r
1229
+ - ▁he
1230
+ - ▁s
1231
+ - ▁be
1232
+ - c
1233
+ - ▁this
1234
+ - ▁so
1235
+ - ▁on
1236
+ - ▁have
1237
+ - g
1238
+ - a
1239
+ - ▁not
1240
+ - or
1241
+ - p
1242
+ - ▁they
1243
+ - ▁with
1244
+ - o
1245
+ - al
1246
+ - ▁but
1247
+ - ▁c
1248
+ - ▁are
1249
+ - ▁re
1250
+ - ▁there
1251
+ - ar
1252
+ - b
1253
+ - l
1254
+ - le
1255
+ - ▁know
1256
+ - ▁as
1257
+ - in
1258
+ - u
1259
+ - ▁what
1260
+ - ▁do
1261
+ - ▁or
1262
+ - ▁
1263
+ - ▁at
1264
+ - ation
1265
+ - w
1266
+ - ll
1267
+ - f
1268
+ - ▁one
1269
+ - ▁e
1270
+ - ▁f
1271
+ - ri
1272
+ - ▁if
1273
+ - ▁b
1274
+ - ent
1275
+ - ▁all
1276
+ - ▁de
1277
+ - ▁can
1278
+ - it
1279
+ - ▁would
1280
+ - ▁me
1281
+ - il
1282
+ - ▁like
1283
+ - ▁an
1284
+ - ce
1285
+ - ▁no
1286
+ - ▁t
1287
+ - ment
1288
+ - i
1289
+ - ve
1290
+ - k
1291
+ - ▁from
1292
+ - ▁just
1293
+ - ▁by
1294
+ - ▁his
1295
+ - ▁had
1296
+ - ion
1297
+ - ▁think
1298
+ - es
1299
+ - ch
1300
+ - 'on'
1301
+ - ▁about
1302
+ - th
1303
+ - ▁my
1304
+ - ck
1305
+ - en
1306
+ - ▁were
1307
+ - ▁your
1308
+ - ur
1309
+ - ▁will
1310
+ - ic
1311
+ - ▁go
1312
+ - ▁con
1313
+ - ▁she
1314
+ - li
1315
+ - an
1316
+ - ▁out
1317
+ - ter
1318
+ - ▁because
1319
+ - se
1320
+ - la
1321
+ - ▁when
1322
+ - ▁st
1323
+ - ▁has
1324
+ - ▁don
1325
+ - ▁which
1326
+ - ir
1327
+ - ra
1328
+ - ▁p
1329
+ - ▁two
1330
+ - ▁right
1331
+ - ▁court
1332
+ - ▁some
1333
+ - ▁pa
1334
+ - ▁pro
1335
+ - lo
1336
+ - ate
1337
+ - ▁time
1338
+ - ▁uh
1339
+ - te
1340
+ - ▁up
1341
+ - ▁our
1342
+ - ▁g
1343
+ - ▁case
1344
+ - ▁w
1345
+ - ▁her
1346
+ - ▁well
1347
+ - ro
1348
+ - ▁their
1349
+ - ▁who
1350
+ - ▁been
1351
+ - ▁yeah
1352
+ - ▁said
1353
+ - ers
1354
+ - ▁mo
1355
+ - ▁other
1356
+ - ▁see
1357
+ - ive
1358
+ - tion
1359
+ - el
1360
+ - ▁more
1361
+ - ▁se
1362
+ - vi
1363
+ - ▁any
1364
+ - ▁get
1365
+ - ▁also
1366
+ - ▁fa
1367
+ - ▁here
1368
+ - ▁un
1369
+ - ▁now
1370
+ - ▁co
1371
+ - ▁people
1372
+ - age
1373
+ - ist
1374
+ - ul
1375
+ - ▁how
1376
+ - ▁going
1377
+ - h
1378
+ - ▁very
1379
+ - ▁then
1380
+ - ▁could
1381
+ - de
1382
+ - ▁ma
1383
+ - ver
1384
+ - ▁mean
1385
+ - ▁su
1386
+ - id
1387
+ - ng
1388
+ - ▁lo
1389
+ - ▁sa
1390
+ - ▁la
1391
+ - ▁sp
1392
+ - ▁us
1393
+ - ol
1394
+ - ▁pre
1395
+ - ance
1396
+ - ▁want
1397
+ - ▁work
1398
+ - ld
1399
+ - ▁did
1400
+ - ge
1401
+ - ▁state
1402
+ - ▁po
1403
+ - ▁dis
1404
+ - ity
1405
+ - able
1406
+ - ▁say
1407
+ - ▁where
1408
+ - et
1409
+ - ▁ra
1410
+ - us
1411
+ - ▁li
1412
+ - ry
1413
+ - ction
1414
+ - ▁um
1415
+ - nd
1416
+ - ne
1417
+ - ▁even
1418
+ - ▁him
1419
+ - ▁ba
1420
+ - ad
1421
+ - ▁these
1422
+ - ▁part
1423
+ - ▁really
1424
+ - lu
1425
+ - ▁them
1426
+ - ▁those
1427
+ - ▁ho
1428
+ - v
1429
+ - ▁bo
1430
+ - ▁ex
1431
+ - ect
1432
+ - ▁over
1433
+ - ▁make
1434
+ - ru
1435
+ - ▁ro
1436
+ - is
1437
+ - ▁fi
1438
+ - ▁m
1439
+ - ut
1440
+ - ke
1441
+ - ence
1442
+ - ▁bu
1443
+ - ▁ch
1444
+ - ▁comp
1445
+ - me
1446
+ - ted
1447
+ - ▁en
1448
+ - ▁fact
1449
+ - z
1450
+ - ▁into
1451
+ - ▁need
1452
+ - ure
1453
+ - ▁way
1454
+ - ies
1455
+ - end
1456
+ - ▁mi
1457
+ - ▁take
1458
+ - un
1459
+ - ▁only
1460
+ - ▁should
1461
+ - as
1462
+ - ▁look
1463
+ - ▁three
1464
+ - one
1465
+ - ▁v
1466
+ - co
1467
+ - ▁com
1468
+ - ▁man
1469
+ - ta
1470
+ - ▁good
1471
+ - ated
1472
+ - ▁back
1473
+ - ▁first
1474
+ - ▁mr
1475
+ - ▁point
1476
+ - x
1477
+ - ▁new
1478
+ - ant
1479
+ - ▁vi
1480
+ - ▁dr
1481
+ - ff
1482
+ - ▁k
1483
+ - ci
1484
+ - ▁app
1485
+ - um
1486
+ - ▁n
1487
+ - ▁exp
1488
+ - ▁under
1489
+ - ▁after
1490
+ - po
1491
+ - ight
1492
+ - ▁d
1493
+ - est
1494
+ - mp
1495
+ - ▁okay
1496
+ - ine
1497
+ - ti
1498
+ - mo
1499
+ - qu
1500
+ - ▁da
1501
+ - ▁come
1502
+ - ▁day
1503
+ - ▁oh
1504
+ - ▁issue
1505
+ - at
1506
+ - ▁question
1507
+ - ▁than
1508
+ - ▁pri
1509
+ - ho
1510
+ - ten
1511
+ - ▁ca
1512
+ - ▁off
1513
+ - ain
1514
+ - ▁di
1515
+ - ▁o
1516
+ - ▁much
1517
+ - ▁five
1518
+ - ow
1519
+ - ▁ha
1520
+ - ▁thousand
1521
+ - ▁six
1522
+ - ▁give
1523
+ - ▁let
1524
+ - ▁le
1525
+ - ary
1526
+ - ous
1527
+ - sh
1528
+ - ▁yes
1529
+ - ▁four
1530
+ - ▁before
1531
+ - ▁call
1532
+ - ▁ga
1533
+ - ▁ne
1534
+ - ▁law
1535
+ - ma
1536
+ - ▁tra
1537
+ - ish
1538
+ - ical
1539
+ - ▁something
1540
+ - iv
1541
+ - ▁why
1542
+ - ▁may
1543
+ - ▁talk
1544
+ - ▁actually
1545
+ - ig
1546
+ - ▁act
1547
+ - ▁got
1548
+ - ▁car
1549
+ - ▁through
1550
+ - ful
1551
+ - ▁made
1552
+ - ▁hundred
1553
+ - om
1554
+ - man
1555
+ - ia
1556
+ - ▁didn
1557
+ - ual
1558
+ - ▁kind
1559
+ - ▁sta
1560
+ - ting
1561
+ - ▁things
1562
+ - ell
1563
+ - ▁down
1564
+ - im
1565
+ - ▁years
1566
+ - ▁lot
1567
+ - ▁ri
1568
+ - pe
1569
+ - ard
1570
+ - he
1571
+ - ▁gra
1572
+ - ▁year
1573
+ - ▁every
1574
+ - ▁eight
1575
+ - ▁reason
1576
+ - ▁seven
1577
+ - ca
1578
+ - ise
1579
+ - ▁twenty
1580
+ - ▁many
1581
+ - ▁per
1582
+ - op
1583
+ - ▁little
1584
+ - ▁mar
1585
+ - ▁same
1586
+ - ▁find
1587
+ - ▁different
1588
+ - ▁wa
1589
+ - pp
1590
+ - ▁nine
1591
+ - if
1592
+ - ▁does
1593
+ - ie
1594
+ - ach
1595
+ - cu
1596
+ - ▁never
1597
+ - ▁thing
1598
+ - ▁whether
1599
+ - ward
1600
+ - ▁again
1601
+ - ▁put
1602
+ - ▁claim
1603
+ - ▁op
1604
+ - ugh
1605
+ - ▁show
1606
+ - tic
1607
+ - ial
1608
+ - ▁still
1609
+ - ▁government
1610
+ - ok
1611
+ - ▁th
1612
+ - ▁long
1613
+ - ▁gu
1614
+ - ▁being
1615
+ - ▁plan
1616
+ - ▁own
1617
+ - am
1618
+ - ian
1619
+ - ▁evidence
1620
+ - ot
1621
+ - mb
1622
+ - ▁most
1623
+ - ▁ru
1624
+ - mit
1625
+ - ▁thank
1626
+ - ious
1627
+ - ▁person
1628
+ - em
1629
+ - ▁use
1630
+ - na
1631
+ - ▁sha
1632
+ - ▁r
1633
+ - ▁believe
1634
+ - mi
1635
+ - ▁school
1636
+ - ▁record
1637
+ - ition
1638
+ - ▁imp
1639
+ - ound
1640
+ - ▁place
1641
+ - ities
1642
+ - ▁happen
1643
+ - ▁ph
1644
+ - ▁great
1645
+ - ative
1646
+ - ha
1647
+ - 'no'
1648
+ - ▁cha
1649
+ - ite
1650
+ - ▁clear
1651
+ - ▁help
1652
+ - dic
1653
+ - ea
1654
+ - ▁high
1655
+ - ▁judge
1656
+ - ▁al
1657
+ - ▁last
1658
+ - di
1659
+ - qui
1660
+ - ▁sure
1661
+ - tra
1662
+ - go
1663
+ - ▁sh
1664
+ - ▁both
1665
+ - ▁read
1666
+ - ving
1667
+ - ▁district
1668
+ - ▁ref
1669
+ - ▁sub
1670
+ - ▁hi
1671
+ - ▁such
1672
+ - ▁ten
1673
+ - ▁start
1674
+ - ▁doesn
1675
+ - ph
1676
+ - ▁change
1677
+ - ay
1678
+ - ▁count
1679
+ - ▁ob
1680
+ - ▁found
1681
+ - ▁du
1682
+ - ▁ki
1683
+ - ▁pay
1684
+ - ▁order
1685
+ - ▁another
1686
+ - for
1687
+ - ▁hand
1688
+ - lic
1689
+ - ▁number
1690
+ - ▁too
1691
+ - ▁ta
1692
+ - ni
1693
+ - ens
1694
+ - ▁tell
1695
+ - ven
1696
+ - port
1697
+ - ▁fun
1698
+ - ▁h
1699
+ - ▁term
1700
+ - ▁end
1701
+ - rr
1702
+ - ▁must
1703
+ - ▁home
1704
+ - ▁hu
1705
+ - ▁wi
1706
+ - side
1707
+ - ▁certain
1708
+ - ▁name
1709
+ - ▁understand
1710
+ - ▁money
1711
+ - ize
1712
+ - ▁pu
1713
+ - ▁interest
1714
+ - ip
1715
+ - ▁came
1716
+ - ▁house
1717
+ - land
1718
+ - ▁men
1719
+ - ▁require
1720
+ - ▁says
1721
+ - min
1722
+ - ▁member
1723
+ - our
1724
+ - ating
1725
+ - ber
1726
+ - ▁jo
1727
+ - ▁might
1728
+ - vo
1729
+ - ▁support
1730
+ - ries
1731
+ - son
1732
+ - ▁problem
1733
+ - ling
1734
+ - ▁old
1735
+ - ▁pe
1736
+ - ▁second
1737
+ - ▁def
1738
+ - ness
1739
+ - ▁decision
1740
+ - row
1741
+ - ▁dollars
1742
+ - ▁mu
1743
+ - ▁conf
1744
+ - use
1745
+ - ▁ah
1746
+ - ▁va
1747
+ - ▁life
1748
+ - ▁inter
1749
+ - ▁play
1750
+ - va
1751
+ - ▁saying
1752
+ - ▁pi
1753
+ - ▁far
1754
+ - ▁went
1755
+ - ▁public
1756
+ - ▁war
1757
+ - ▁trans
1758
+ - ▁live
1759
+ - ▁ask
1760
+ - ▁real
1761
+ - rate
1762
+ - ▁important
1763
+ - ▁thought
1764
+ - ▁done
1765
+ - ▁between
1766
+ - ▁argument
1767
+ - ▁rep
1768
+ - ▁consider
1769
+ - ▁anything
1770
+ - ▁direct
1771
+ - less
1772
+ - ▁ja
1773
+ - ▁doing
1774
+ - ▁course
1775
+ - line
1776
+ - ▁month
1777
+ - ▁word
1778
+ - ▁care
1779
+ - ▁provide
1780
+ - ap
1781
+ - ▁honor
1782
+ - ba
1783
+ - ator
1784
+ - j
1785
+ - ▁report
1786
+ - ▁exc
1787
+ - ▁service
1788
+ - ▁tri
1789
+ - nk
1790
+ - ▁world
1791
+ - ▁without
1792
+ - ▁ju
1793
+ - ▁against
1794
+ - ible
1795
+ - ▁used
1796
+ - ▁always
1797
+ - ▁able
1798
+ - ship
1799
+ - ▁rule
1800
+ - ▁matter
1801
+ - ▁while
1802
+ - par
1803
+ - ▁maybe
1804
+ - ▁miss
1805
+ - ▁around
1806
+ - ▁u
1807
+ - ▁today
1808
+ - ▁friend
1809
+ - ▁ti
1810
+ - ▁big
1811
+ - bu
1812
+ - ▁next
1813
+ - ju
1814
+ - ▁thirty
1815
+ - ▁pass
1816
+ - ▁y
1817
+ - ▁business
1818
+ - ▁test
1819
+ - ▁fo
1820
+ - ▁nothing
1821
+ - ▁turn
1822
+ - ▁process
1823
+ - com
1824
+ - ▁el
1825
+ - ▁better
1826
+ - ▁commission
1827
+ - ▁week
1828
+ - ▁move
1829
+ - ▁trial
1830
+ - ▁open
1831
+ - ▁each
1832
+ - ▁specific
1833
+ - ▁feel
1834
+ - ▁cap
1835
+ - ency
1836
+ - ▁particular
1837
+ - ability
1838
+ - ▁system
1839
+ - gra
1840
+ - ▁present
1841
+ - ▁looking
1842
+ - duc
1843
+ - ▁fra
1844
+ - ▁correct
1845
+ - ▁allow
1846
+ - ▁side
1847
+ - ign
1848
+ - serv
1849
+ - ▁city
1850
+ - ▁agree
1851
+ - ▁having
1852
+ - ▁defendant
1853
+ - ▁keep
1854
+ - ▁follow
1855
+ - ▁invest
1856
+ - ▁add
1857
+ - ▁information
1858
+ - ▁whole
1859
+ - ▁general
1860
+ - hi
1861
+ - ▁away
1862
+ - ▁upon
1863
+ - ▁win
1864
+ - ▁based
1865
+ - ▁program
1866
+ - ▁develop
1867
+ - ▁country
1868
+ - ified
1869
+ - ▁sort
1870
+ - ▁board
1871
+ - ▁qua
1872
+ - ex
1873
+ - ▁family
1874
+ - ▁close
1875
+ - ▁dec
1876
+ - ▁though
1877
+ - ▁hard
1878
+ - ▁answer
1879
+ - ▁guess
1880
+ - ▁job
1881
+ - ▁trying
1882
+ - ▁je
1883
+ - ▁asked
1884
+ - ▁during
1885
+ - ▁fee
1886
+ - ▁love
1887
+ - ▁already
1888
+ - ▁address
1889
+ - ▁ever
1890
+ - ▁bill
1891
+ - gue
1892
+ - ▁cause
1893
+ - ▁gen
1894
+ - ▁free
1895
+ - ▁continue
1896
+ - ▁fifty
1897
+ - ▁contract
1898
+ - ▁european
1899
+ - ▁rest
1900
+ - ▁few
1901
+ - ▁water
1902
+ - ▁line
1903
+ - ▁position
1904
+ - ▁percent
1905
+ - ▁enough
1906
+ - ▁sea
1907
+ - ▁har
1908
+ - ▁america
1909
+ - ▁full
1910
+ - ▁since
1911
+ - ▁everything
1912
+ - ▁bring
1913
+ - ▁lead
1914
+ - ▁probably
1915
+ - ▁children
1916
+ - ▁tax
1917
+ - ▁face
1918
+ - ▁best
1919
+ - ▁concern
1920
+ - ▁sign
1921
+ - ▁mhm
1922
+ - ▁minute
1923
+ - ▁bit
1924
+ - ▁stand
1925
+ - ▁situation
1926
+ - ▁speak
1927
+ - ▁hope
1928
+ - ▁yet
1929
+ - ▁discuss
1930
+ - ▁standard
1931
+ - ▁brief
1932
+ - ▁statute
1933
+ - ▁officer
1934
+ - ▁ok
1935
+ - ▁respect
1936
+ - ▁together
1937
+ - ▁agreement
1938
+ - ▁room
1939
+ - ▁least
1940
+ - ▁market
1941
+ - ▁however
1942
+ - ▁getting
1943
+ - ▁else
1944
+ - ▁true
1945
+ - ▁counsel
1946
+ - ttle
1947
+ - ▁err
1948
+ - ▁effect
1949
+ - ▁example
1950
+ - ▁less
1951
+ - ▁cost
1952
+ - ▁left
1953
+ - ▁group
1954
+ - ▁level
1955
+ - ▁policy
1956
+ - ▁town
1957
+ - ▁small
1958
+ - ▁child
1959
+ - ▁president
1960
+ - ▁large
1961
+ - ▁later
1962
+ - ▁protect
1963
+ - ▁decide
1964
+ - way
1965
+ - ▁making
1966
+ - ▁regard
1967
+ - ▁book
1968
+ - ▁pretty
1969
+ - ▁author
1970
+ - ▁hour
1971
+ - ▁idea
1972
+ - ▁include
1973
+ - ▁exactly
1974
+ - ▁power
1975
+ - ▁limit
1976
+ - ▁watch
1977
+ - ▁night
1978
+ - ▁ya
1979
+ - ▁jury
1980
+ - ▁leave
1981
+ - ▁quite
1982
+ - ▁cover
1983
+ - ▁grow
1984
+ - ▁result
1985
+ - ▁singapore
1986
+ - ▁either
1987
+ - ▁company
1988
+ - ▁stop
1989
+ - ▁community
1990
+ - ▁million
1991
+ - ▁current
1992
+ - ▁federal
1993
+ - ▁united
1994
+ - ▁mind
1995
+ - ▁step
1996
+ - ▁class
1997
+ - ▁police
1998
+ - ▁remain
1999
+ - ▁health
2000
+ - ▁final
2001
+ - ▁represent
2002
+ - ▁light
2003
+ - ▁treat
2004
+ - ▁review
2005
+ - ▁testimony
2006
+ - ▁bank
2007
+ - ▁product
2008
+ - ▁appeal
2009
+ - ▁half
2010
+ - ▁cannot
2011
+ - ▁involve
2012
+ - ▁charge
2013
+ - ▁knew
2014
+ - ▁learn
2015
+ - ▁eye
2016
+ - ▁language
2017
+ - ▁individual
2018
+ - ▁receive
2019
+ - ▁further
2020
+ - ▁mister
2021
+ - ▁view
2022
+ - ▁council
2023
+ - ▁until
2024
+ - ▁human
2025
+ - ▁stuff
2026
+ - ▁student
2027
+ - ▁become
2028
+ - ▁walk
2029
+ - ▁control
2030
+ - ▁check
2031
+ - ▁seems
2032
+ - ▁strong
2033
+ - ▁period
2034
+ - ▁please
2035
+ - ▁food
2036
+ - ▁type
2037
+ - ▁complete
2038
+ - ▁purpose
2039
+ - ▁account
2040
+ - ▁project
2041
+ - ▁sense
2042
+ - ▁national
2043
+ - ▁appear
2044
+ - ▁simply
2045
+ - ▁return
2046
+ - ▁benefit
2047
+ - ▁employ
2048
+ - ▁create
2049
+ - ▁wrong
2050
+ - ▁whatever
2051
+ - ▁morning
2052
+ - ▁suggest
2053
+ - ▁condition
2054
+ - ▁amount
2055
+ - ▁object
2056
+ - ▁third
2057
+ - ▁remember
2058
+ - ▁himself
2059
+ - ▁property
2060
+ - ▁itself
2061
+ - ▁somebody
2062
+ - ▁suppose
2063
+ - ▁future
2064
+ - ▁short
2065
+ - ▁opinion
2066
+ - ▁several
2067
+ - ▁common
2068
+ - ▁taking
2069
+ - ▁notice
2070
+ - ▁office
2071
+ - ▁almost
2072
+ - ▁father
2073
+ - ▁document
2074
+ - ▁judgment
2075
+ - ▁experience
2076
+ - ▁uhhuh
2077
+ - ▁committee
2078
+ - ▁parent
2079
+ - ▁legal
2080
+ - ▁client
2081
+ - ▁major
2082
+ - ▁congress
2083
+ - ▁rather
2084
+ - ▁proceed
2085
+ - ▁possible
2086
+ - ▁plea
2087
+ - ▁subject
2088
+ - ▁along
2089
+ - ▁mother
2090
+ - ▁increase
2091
+ - ▁meeting
2092
+ - ▁girl
2093
+ - ▁couple
2094
+ - ▁north
2095
+ - ▁local
2096
+ - ▁sometimes
2097
+ - ▁brought
2098
+ - ▁fifteen
2099
+ - ▁organ
2100
+ - ▁difference
2101
+ - ▁entire
2102
+ - ▁security
2103
+ - ▁department
2104
+ - ▁south
2105
+ - ▁vote
2106
+ - ▁basis
2107
+ - ▁twelve
2108
+ - ▁conduct
2109
+ - ▁street
2110
+ - ▁social
2111
+ - ▁wonder
2112
+ - ▁supreme
2113
+ - ▁difficult
2114
+ - ▁travel
2115
+ - ▁opportunity
2116
+ - ▁defense
2117
+ - ▁plaintiff
2118
+ - ▁white
2119
+ - ▁focus
2120
+ - ▁therefore
2121
+ - ▁challenge
2122
+ - ▁value
2123
+ - ▁except
2124
+ - ▁among
2125
+ - ▁ground
2126
+ - ▁serve
2127
+ - ▁attorney
2128
+ - ▁education
2129
+ - ▁credit
2130
+ - ▁request
2131
+ - ▁women
2132
+ - ▁basically
2133
+ - ▁political
2134
+ - ▁determine
2135
+ - ▁access
2136
+ - ▁across
2137
+ - ▁sentence
2138
+ - ▁perform
2139
+ - ▁everybody
2140
+ - ▁black
2141
+ - ▁drug
2142
+ - ▁accept
2143
+ - ▁eleven
2144
+ - ▁establish
2145
+ - ▁design
2146
+ - ▁success
2147
+ - ▁including
2148
+ - ▁john
2149
+ - ▁employee
2150
+ - ▁nice
2151
+ - ▁crime
2152
+ - ibility
2153
+ - ▁circumstance
2154
+ - ▁occur
2155
+ - ▁citizen
2156
+ - ▁woman
2157
+ - ▁witness
2158
+ - ▁explain
2159
+ - ▁additional
2160
+ - ▁history
2161
+ - ▁front
2162
+ - ▁medical
2163
+ - ▁death
2164
+ - ification
2165
+ - ▁financial
2166
+ - ▁break
2167
+ - ▁connect
2168
+ - ▁perhaps
2169
+ - ▁sorry
2170
+ - ▁especially
2171
+ - ▁trust
2172
+ - ▁insurance
2173
+ - ▁economic
2174
+ - ▁effort
2175
+ - ▁circuit
2176
+ - ▁cross
2177
+ - ▁instead
2178
+ - ▁affect
2179
+ - ▁zero
2180
+ - ▁available
2181
+ - ▁team
2182
+ - ▁application
2183
+ - ▁separate
2184
+ - ▁absolutely
2185
+ - ▁christ
2186
+ - ization
2187
+ - ▁europe
2188
+ - ▁budget
2189
+ - ▁damage
2190
+ - ▁myself
2191
+ - ▁private
2192
+ - ▁figure
2193
+ - ▁provision
2194
+ - ▁collect
2195
+ - ▁similar
2196
+ - ▁regulation
2197
+ - ▁approach
2198
+ - ▁themselves
2199
+ - ▁construct
2200
+ - ▁petition
2201
+ - ▁material
2202
+ - ▁improve
2203
+ - ▁york
2204
+ - ▁parliament
2205
+ - ▁broad
2206
+ - ▁companies
2207
+ - ▁nature
2208
+ - ▁piece
2209
+ - ▁threat
2210
+ - ▁energy
2211
+ - ▁behind
2212
+ - ▁significant
2213
+ - ▁response
2214
+ - ▁definitely
2215
+ - ▁guide
2216
+ - ▁beyond
2217
+ - ▁amendment
2218
+ - ▁measure
2219
+ - ▁criminal
2220
+ - ▁original
2221
+ - ▁brother
2222
+ - ▁impact
2223
+ - q
2224
+ decoding:
2225
+ strategy: greedy
2226
+ greedy: {}
2227
+ loss:
2228
+ loss_name: default
2229
+ warprnnt_numba_kwargs:
2230
+ fastemit_lambda: 0.005
2231
+ clamp: -1.0
2232
+ variational_noise:
2233
+ start_step: 0
2234
+ std: 0.0
2235
+ optim:
2236
+ name: adamw
2237
+ lr: 2.0
2238
+ betas:
2239
+ - 0.9
2240
+ - 0.98
2241
+ weight_decay: 0.001
2242
+ sched:
2243
+ name: NoamAnnealing
2244
+ d_model: 512
2245
+ warmup_steps: 20000
2246
+ warmup_ratio: null
2247
+ min_lr: 1.0e-06
2248
+ labels:
2249
+ - <unk>
2250
+ - s
2251
+ - ▁the
2252
+ - ▁a
2253
+ - t
2254
+ - ▁to
2255
+ - ▁and
2256
+ - ▁that
2257
+ - ''''
2258
+ - ▁of
2259
+ - ▁i
2260
+ - ▁in
2261
+ - d
2262
+ - ed
2263
+ - ing
2264
+ - ▁it
2265
+ - ▁you
2266
+ - 'n'
2267
+ - e
2268
+ - ▁is
2269
+ - 'y'
2270
+ - er
2271
+ - ▁for
2272
+ - m
2273
+ - ▁was
2274
+ - ▁we
2275
+ - re
2276
+ - ly
2277
+ - r
2278
+ - ▁he
2279
+ - ▁s
2280
+ - ▁be
2281
+ - c
2282
+ - ▁this
2283
+ - ▁so
2284
+ - ▁on
2285
+ - ▁have
2286
+ - g
2287
+ - a
2288
+ - ▁not
2289
+ - or
2290
+ - p
2291
+ - ▁they
2292
+ - ▁with
2293
+ - o
2294
+ - al
2295
+ - ▁but
2296
+ - ▁c
2297
+ - ▁are
2298
+ - ▁re
2299
+ - ▁there
2300
+ - ar
2301
+ - b
2302
+ - l
2303
+ - le
2304
+ - ▁know
2305
+ - ▁as
2306
+ - in
2307
+ - u
2308
+ - ▁what
2309
+ - ▁do
2310
+ - ▁or
2311
+ - ▁
2312
+ - ▁at
2313
+ - ation
2314
+ - w
2315
+ - ll
2316
+ - f
2317
+ - ▁one
2318
+ - ▁e
2319
+ - ▁f
2320
+ - ri
2321
+ - ▁if
2322
+ - ▁b
2323
+ - ent
2324
+ - ▁all
2325
+ - ▁de
2326
+ - ▁can
2327
+ - it
2328
+ - ▁would
2329
+ - ▁me
2330
+ - il
2331
+ - ▁like
2332
+ - ▁an
2333
+ - ce
2334
+ - ▁no
2335
+ - ▁t
2336
+ - ment
2337
+ - i
2338
+ - ve
2339
+ - k
2340
+ - ▁from
2341
+ - ▁just
2342
+ - ▁by
2343
+ - ▁his
2344
+ - ▁had
2345
+ - ion
2346
+ - ▁think
2347
+ - es
2348
+ - ch
2349
+ - 'on'
2350
+ - ▁about
2351
+ - th
2352
+ - ▁my
2353
+ - ck
2354
+ - en
2355
+ - ▁were
2356
+ - ▁your
2357
+ - ur
2358
+ - ▁will
2359
+ - ic
2360
+ - ▁go
2361
+ - ▁con
2362
+ - ▁she
2363
+ - li
2364
+ - an
2365
+ - ▁out
2366
+ - ter
2367
+ - ▁because
2368
+ - se
2369
+ - la
2370
+ - ▁when
2371
+ - ▁st
2372
+ - ▁has
2373
+ - ▁don
2374
+ - ▁which
2375
+ - ir
2376
+ - ra
2377
+ - ▁p
2378
+ - ▁two
2379
+ - ▁right
2380
+ - ▁court
2381
+ - ▁some
2382
+ - ▁pa
2383
+ - ▁pro
2384
+ - lo
2385
+ - ate
2386
+ - ▁time
2387
+ - ▁uh
2388
+ - te
2389
+ - ▁up
2390
+ - ▁our
2391
+ - ▁g
2392
+ - ▁case
2393
+ - ▁w
2394
+ - ▁her
2395
+ - ▁well
2396
+ - ro
2397
+ - ▁their
2398
+ - ▁who
2399
+ - ▁been
2400
+ - ▁yeah
2401
+ - ▁said
2402
+ - ers
2403
+ - ▁mo
2404
+ - ▁other
2405
+ - ▁see
2406
+ - ive
2407
+ - tion
2408
+ - el
2409
+ - ▁more
2410
+ - ▁se
2411
+ - vi
2412
+ - ▁any
2413
+ - ▁get
2414
+ - ▁also
2415
+ - ▁fa
2416
+ - ▁here
2417
+ - ▁un
2418
+ - ▁now
2419
+ - ▁co
2420
+ - ▁people
2421
+ - age
2422
+ - ist
2423
+ - ul
2424
+ - ▁how
2425
+ - ▁going
2426
+ - h
2427
+ - ▁very
2428
+ - ▁then
2429
+ - ▁could
2430
+ - de
2431
+ - ▁ma
2432
+ - ver
2433
+ - ▁mean
2434
+ - ▁su
2435
+ - id
2436
+ - ng
2437
+ - ▁lo
2438
+ - ▁sa
2439
+ - ▁la
2440
+ - ▁sp
2441
+ - ▁us
2442
+ - ol
2443
+ - ▁pre
2444
+ - ance
2445
+ - ▁want
2446
+ - ▁work
2447
+ - ld
2448
+ - ▁did
2449
+ - ge
2450
+ - ▁state
2451
+ - ▁po
2452
+ - ▁dis
2453
+ - ity
2454
+ - able
2455
+ - ▁say
2456
+ - ▁where
2457
+ - et
2458
+ - ▁ra
2459
+ - us
2460
+ - ▁li
2461
+ - ry
2462
+ - ction
2463
+ - ▁um
2464
+ - nd
2465
+ - ne
2466
+ - ▁even
2467
+ - ▁him
2468
+ - ▁ba
2469
+ - ad
2470
+ - ▁these
2471
+ - ▁part
2472
+ - ▁really
2473
+ - lu
2474
+ - ▁them
2475
+ - ▁those
2476
+ - ▁ho
2477
+ - v
2478
+ - ▁bo
2479
+ - ▁ex
2480
+ - ect
2481
+ - ▁over
2482
+ - ▁make
2483
+ - ru
2484
+ - ▁ro
2485
+ - is
2486
+ - ▁fi
2487
+ - ▁m
2488
+ - ut
2489
+ - ke
2490
+ - ence
2491
+ - ▁bu
2492
+ - ▁ch
2493
+ - ▁comp
2494
+ - me
2495
+ - ted
2496
+ - ▁en
2497
+ - ▁fact
2498
+ - z
2499
+ - ▁into
2500
+ - ▁need
2501
+ - ure
2502
+ - ▁way
2503
+ - ies
2504
+ - end
2505
+ - ▁mi
2506
+ - ▁take
2507
+ - un
2508
+ - ▁only
2509
+ - ▁should
2510
+ - as
2511
+ - ▁look
2512
+ - ▁three
2513
+ - one
2514
+ - ▁v
2515
+ - co
2516
+ - ▁com
2517
+ - ▁man
2518
+ - ta
2519
+ - ▁good
2520
+ - ated
2521
+ - ▁back
2522
+ - ▁first
2523
+ - ▁mr
2524
+ - ▁point
2525
+ - x
2526
+ - ▁new
2527
+ - ant
2528
+ - ▁vi
2529
+ - ▁dr
2530
+ - ff
2531
+ - ▁k
2532
+ - ci
2533
+ - ▁app
2534
+ - um
2535
+ - ▁n
2536
+ - ▁exp
2537
+ - ▁under
2538
+ - ▁after
2539
+ - po
2540
+ - ight
2541
+ - ▁d
2542
+ - est
2543
+ - mp
2544
+ - ▁okay
2545
+ - ine
2546
+ - ti
2547
+ - mo
2548
+ - qu
2549
+ - ▁da
2550
+ - ▁come
2551
+ - ▁day
2552
+ - ▁oh
2553
+ - ▁issue
2554
+ - at
2555
+ - ▁question
2556
+ - ▁than
2557
+ - ▁pri
2558
+ - ho
2559
+ - ten
2560
+ - ▁ca
2561
+ - ▁off
2562
+ - ain
2563
+ - ▁di
2564
+ - ▁o
2565
+ - ▁much
2566
+ - ▁five
2567
+ - ow
2568
+ - ▁ha
2569
+ - ▁thousand
2570
+ - ▁six
2571
+ - ▁give
2572
+ - ▁let
2573
+ - ▁le
2574
+ - ary
2575
+ - ous
2576
+ - sh
2577
+ - ▁yes
2578
+ - ▁four
2579
+ - ▁before
2580
+ - ▁call
2581
+ - ▁ga
2582
+ - ▁ne
2583
+ - ▁law
2584
+ - ma
2585
+ - ▁tra
2586
+ - ish
2587
+ - ical
2588
+ - ▁something
2589
+ - iv
2590
+ - ▁why
2591
+ - ▁may
2592
+ - ▁talk
2593
+ - ▁actually
2594
+ - ig
2595
+ - ▁act
2596
+ - ▁got
2597
+ - ▁car
2598
+ - ▁through
2599
+ - ful
2600
+ - ▁made
2601
+ - ▁hundred
2602
+ - om
2603
+ - man
2604
+ - ia
2605
+ - ▁didn
2606
+ - ual
2607
+ - ▁kind
2608
+ - ▁sta
2609
+ - ting
2610
+ - ▁things
2611
+ - ell
2612
+ - ▁down
2613
+ - im
2614
+ - ▁years
2615
+ - ▁lot
2616
+ - ▁ri
2617
+ - pe
2618
+ - ard
2619
+ - he
2620
+ - ▁gra
2621
+ - ▁year
2622
+ - ▁every
2623
+ - ▁eight
2624
+ - ▁reason
2625
+ - ▁seven
2626
+ - ca
2627
+ - ise
2628
+ - ▁twenty
2629
+ - ▁many
2630
+ - ▁per
2631
+ - op
2632
+ - ▁little
2633
+ - ▁mar
2634
+ - ▁same
2635
+ - ▁find
2636
+ - ▁different
2637
+ - ▁wa
2638
+ - pp
2639
+ - ▁nine
2640
+ - if
2641
+ - ▁does
2642
+ - ie
2643
+ - ach
2644
+ - cu
2645
+ - ▁never
2646
+ - ▁thing
2647
+ - ▁whether
2648
+ - ward
2649
+ - ▁again
2650
+ - ▁put
2651
+ - ▁claim
2652
+ - ▁op
2653
+ - ugh
2654
+ - ▁show
2655
+ - tic
2656
+ - ial
2657
+ - ▁still
2658
+ - ▁government
2659
+ - ok
2660
+ - ▁th
2661
+ - ▁long
2662
+ - ▁gu
2663
+ - ▁being
2664
+ - ▁plan
2665
+ - ▁own
2666
+ - am
2667
+ - ian
2668
+ - ▁evidence
2669
+ - ot
2670
+ - mb
2671
+ - ▁most
2672
+ - ▁ru
2673
+ - mit
2674
+ - ▁thank
2675
+ - ious
2676
+ - ▁person
2677
+ - em
2678
+ - ▁use
2679
+ - na
2680
+ - ▁sha
2681
+ - ▁r
2682
+ - ▁believe
2683
+ - mi
2684
+ - ▁school
2685
+ - ▁record
2686
+ - ition
2687
+ - ▁imp
2688
+ - ound
2689
+ - ▁place
2690
+ - ities
2691
+ - ▁happen
2692
+ - ▁ph
2693
+ - ▁great
2694
+ - ative
2695
+ - ha
2696
+ - 'no'
2697
+ - ▁cha
2698
+ - ite
2699
+ - ▁clear
2700
+ - ▁help
2701
+ - dic
2702
+ - ea
2703
+ - ▁high
2704
+ - ▁judge
2705
+ - ▁al
2706
+ - ▁last
2707
+ - di
2708
+ - qui
2709
+ - ▁sure
2710
+ - tra
2711
+ - go
2712
+ - ▁sh
2713
+ - ▁both
2714
+ - ▁read
2715
+ - ving
2716
+ - ▁district
2717
+ - ▁ref
2718
+ - ▁sub
2719
+ - ▁hi
2720
+ - ▁such
2721
+ - ▁ten
2722
+ - ▁start
2723
+ - ▁doesn
2724
+ - ph
2725
+ - ▁change
2726
+ - ay
2727
+ - ▁count
2728
+ - ▁ob
2729
+ - ▁found
2730
+ - ▁du
2731
+ - ▁ki
2732
+ - ▁pay
2733
+ - ▁order
2734
+ - ▁another
2735
+ - for
2736
+ - ▁hand
2737
+ - lic
2738
+ - ▁number
2739
+ - ▁too
2740
+ - ▁ta
2741
+ - ni
2742
+ - ens
2743
+ - ▁tell
2744
+ - ven
2745
+ - port
2746
+ - ▁fun
2747
+ - ▁h
2748
+ - ▁term
2749
+ - ▁end
2750
+ - rr
2751
+ - ▁must
2752
+ - ▁home
2753
+ - ▁hu
2754
+ - ▁wi
2755
+ - side
2756
+ - ▁certain
2757
+ - ▁name
2758
+ - ▁understand
2759
+ - ▁money
2760
+ - ize
2761
+ - ▁pu
2762
+ - ▁interest
2763
+ - ip
2764
+ - ▁came
2765
+ - ▁house
2766
+ - land
2767
+ - ▁men
2768
+ - ▁require
2769
+ - ▁says
2770
+ - min
2771
+ - ▁member
2772
+ - our
2773
+ - ating
2774
+ - ber
2775
+ - ▁jo
2776
+ - ▁might
2777
+ - vo
2778
+ - ▁support
2779
+ - ries
2780
+ - son
2781
+ - ▁problem
2782
+ - ling
2783
+ - ▁old
2784
+ - ▁pe
2785
+ - ▁second
2786
+ - ▁def
2787
+ - ness
2788
+ - ▁decision
2789
+ - row
2790
+ - ▁dollars
2791
+ - ▁mu
2792
+ - ▁conf
2793
+ - use
2794
+ - ▁ah
2795
+ - ▁va
2796
+ - ▁life
2797
+ - ▁inter
2798
+ - ▁play
2799
+ - va
2800
+ - ▁saying
2801
+ - ▁pi
2802
+ - ▁far
2803
+ - ▁went
2804
+ - ▁public
2805
+ - ▁war
2806
+ - ▁trans
2807
+ - ▁live
2808
+ - ▁ask
2809
+ - ▁real
2810
+ - rate
2811
+ - ▁important
2812
+ - ▁thought
2813
+ - ▁done
2814
+ - ▁between
2815
+ - ▁argument
2816
+ - ▁rep
2817
+ - ▁consider
2818
+ - ▁anything
2819
+ - ▁direct
2820
+ - less
2821
+ - ▁ja
2822
+ - ▁doing
2823
+ - ▁course
2824
+ - line
2825
+ - ▁month
2826
+ - ▁word
2827
+ - ▁care
2828
+ - ▁provide
2829
+ - ap
2830
+ - ▁honor
2831
+ - ba
2832
+ - ator
2833
+ - j
2834
+ - ▁report
2835
+ - ▁exc
2836
+ - ▁service
2837
+ - ▁tri
2838
+ - nk
2839
+ - ▁world
2840
+ - ▁without
2841
+ - ▁ju
2842
+ - ▁against
2843
+ - ible
2844
+ - ▁used
2845
+ - ▁always
2846
+ - ▁able
2847
+ - ship
2848
+ - ▁rule
2849
+ - ▁matter
2850
+ - ▁while
2851
+ - par
2852
+ - ▁maybe
2853
+ - ▁miss
2854
+ - ▁around
2855
+ - ▁u
2856
+ - ▁today
2857
+ - ▁friend
2858
+ - ▁ti
2859
+ - ▁big
2860
+ - bu
2861
+ - ▁next
2862
+ - ju
2863
+ - ▁thirty
2864
+ - ▁pass
2865
+ - ▁y
2866
+ - ▁business
2867
+ - ▁test
2868
+ - ▁fo
2869
+ - ▁nothing
2870
+ - ▁turn
2871
+ - ▁process
2872
+ - com
2873
+ - ▁el
2874
+ - ▁better
2875
+ - ▁commission
2876
+ - ▁week
2877
+ - ▁move
2878
+ - ▁trial
2879
+ - ▁open
2880
+ - ▁each
2881
+ - ▁specific
2882
+ - ▁feel
2883
+ - ▁cap
2884
+ - ency
2885
+ - ▁particular
2886
+ - ability
2887
+ - ▁system
2888
+ - gra
2889
+ - ▁present
2890
+ - ▁looking
2891
+ - duc
2892
+ - ▁fra
2893
+ - ▁correct
2894
+ - ▁allow
2895
+ - ▁side
2896
+ - ign
2897
+ - serv
2898
+ - ▁city
2899
+ - ▁agree
2900
+ - ▁having
2901
+ - ▁defendant
2902
+ - ▁keep
2903
+ - ▁follow
2904
+ - ▁invest
2905
+ - ▁add
2906
+ - ▁information
2907
+ - ▁whole
2908
+ - ▁general
2909
+ - hi
2910
+ - ▁away
2911
+ - ▁upon
2912
+ - ▁win
2913
+ - ▁based
2914
+ - ▁program
2915
+ - ▁develop
2916
+ - ▁country
2917
+ - ified
2918
+ - ▁sort
2919
+ - ▁board
2920
+ - ▁qua
2921
+ - ex
2922
+ - ▁family
2923
+ - ▁close
2924
+ - ▁dec
2925
+ - ▁though
2926
+ - ▁hard
2927
+ - ▁answer
2928
+ - ▁guess
2929
+ - ▁job
2930
+ - ▁trying
2931
+ - ▁je
2932
+ - ▁asked
2933
+ - ▁during
2934
+ - ▁fee
2935
+ - ▁love
2936
+ - ▁already
2937
+ - ▁address
2938
+ - ▁ever
2939
+ - ▁bill
2940
+ - gue
2941
+ - ▁cause
2942
+ - ▁gen
2943
+ - ▁free
2944
+ - ▁continue
2945
+ - ▁fifty
2946
+ - ▁contract
2947
+ - ▁european
2948
+ - ▁rest
2949
+ - ▁few
2950
+ - ▁water
2951
+ - ▁line
2952
+ - ▁position
2953
+ - ▁percent
2954
+ - ▁enough
2955
+ - ▁sea
2956
+ - ▁har
2957
+ - ▁america
2958
+ - ▁full
2959
+ - ▁since
2960
+ - ▁everything
2961
+ - ▁bring
2962
+ - ▁lead
2963
+ - ▁probably
2964
+ - ▁children
2965
+ - ▁tax
2966
+ - ▁face
2967
+ - ▁best
2968
+ - ▁concern
2969
+ - ▁sign
2970
+ - ▁mhm
2971
+ - ▁minute
2972
+ - ▁bit
2973
+ - ▁stand
2974
+ - ▁situation
2975
+ - ▁speak
2976
+ - ▁hope
2977
+ - ▁yet
2978
+ - ▁discuss
2979
+ - ▁standard
2980
+ - ▁brief
2981
+ - ▁statute
2982
+ - ▁officer
2983
+ - ▁ok
2984
+ - ▁respect
2985
+ - ▁together
2986
+ - ▁agreement
2987
+ - ▁room
2988
+ - ▁least
2989
+ - ▁market
2990
+ - ▁however
2991
+ - ▁getting
2992
+ - ▁else
2993
+ - ▁true
2994
+ - ▁counsel
2995
+ - ttle
2996
+ - ▁err
2997
+ - ▁effect
2998
+ - ▁example
2999
+ - ▁less
3000
+ - ▁cost
3001
+ - ▁left
3002
+ - ▁group
3003
+ - ▁level
3004
+ - ▁policy
3005
+ - ▁town
3006
+ - ▁small
3007
+ - ▁child
3008
+ - ▁president
3009
+ - ▁large
3010
+ - ▁later
3011
+ - ▁protect
3012
+ - ▁decide
3013
+ - way
3014
+ - ▁making
3015
+ - ▁regard
3016
+ - ▁book
3017
+ - ▁pretty
3018
+ - ▁author
3019
+ - ▁hour
3020
+ - ▁idea
3021
+ - ▁include
3022
+ - ▁exactly
3023
+ - ▁power
3024
+ - ▁limit
3025
+ - ▁watch
3026
+ - ▁night
3027
+ - ▁ya
3028
+ - ▁jury
3029
+ - ▁leave
3030
+ - ▁quite
3031
+ - ▁cover
3032
+ - ▁grow
3033
+ - ▁result
3034
+ - ▁singapore
3035
+ - ▁either
3036
+ - ▁company
3037
+ - ▁stop
3038
+ - ▁community
3039
+ - ▁million
3040
+ - ▁current
3041
+ - ▁federal
3042
+ - ▁united
3043
+ - ▁mind
3044
+ - ▁step
3045
+ - ▁class
3046
+ - ▁police
3047
+ - ▁remain
3048
+ - ▁health
3049
+ - ▁final
3050
+ - ▁represent
3051
+ - ▁light
3052
+ - ▁treat
3053
+ - ▁review
3054
+ - ▁testimony
3055
+ - ▁bank
3056
+ - ▁product
3057
+ - ▁appeal
3058
+ - ▁half
3059
+ - ▁cannot
3060
+ - ▁involve
3061
+ - ▁charge
3062
+ - ▁knew
3063
+ - ▁learn
3064
+ - ▁eye
3065
+ - ▁language
3066
+ - ▁individual
3067
+ - ▁receive
3068
+ - ▁further
3069
+ - ▁mister
3070
+ - ▁view
3071
+ - ▁council
3072
+ - ▁until
3073
+ - ▁human
3074
+ - ▁stuff
3075
+ - ▁student
3076
+ - ▁become
3077
+ - ▁walk
3078
+ - ▁control
3079
+ - ▁check
3080
+ - ▁seems
3081
+ - ▁strong
3082
+ - ▁period
3083
+ - ▁please
3084
+ - ▁food
3085
+ - ▁type
3086
+ - ▁complete
3087
+ - ▁purpose
3088
+ - ▁account
3089
+ - ▁project
3090
+ - ▁sense
3091
+ - ▁national
3092
+ - ▁appear
3093
+ - ▁simply
3094
+ - ▁return
3095
+ - ▁benefit
3096
+ - ▁employ
3097
+ - ▁create
3098
+ - ▁wrong
3099
+ - ▁whatever
3100
+ - ▁morning
3101
+ - ▁suggest
3102
+ - ▁condition
3103
+ - ▁amount
3104
+ - ▁object
3105
+ - ▁third
3106
+ - ▁remember
3107
+ - ▁himself
3108
+ - ▁property
3109
+ - ▁itself
3110
+ - ▁somebody
3111
+ - ▁suppose
3112
+ - ▁future
3113
+ - ▁short
3114
+ - ▁opinion
3115
+ - ▁several
3116
+ - ▁common
3117
+ - ▁taking
3118
+ - ▁notice
3119
+ - ▁office
3120
+ - ▁almost
3121
+ - ▁father
3122
+ - ▁document
3123
+ - ▁judgment
3124
+ - ▁experience
3125
+ - ▁uhhuh
3126
+ - ▁committee
3127
+ - ▁parent
3128
+ - ▁legal
3129
+ - ▁client
3130
+ - ▁major
3131
+ - ▁congress
3132
+ - ▁rather
3133
+ - ▁proceed
3134
+ - ▁possible
3135
+ - ▁plea
3136
+ - ▁subject
3137
+ - ▁along
3138
+ - ▁mother
3139
+ - ▁increase
3140
+ - ▁meeting
3141
+ - ▁girl
3142
+ - ▁couple
3143
+ - ▁north
3144
+ - ▁local
3145
+ - ▁sometimes
3146
+ - ▁brought
3147
+ - ▁fifteen
3148
+ - ▁organ
3149
+ - ▁difference
3150
+ - ▁entire
3151
+ - ▁security
3152
+ - ▁department
3153
+ - ▁south
3154
+ - ▁vote
3155
+ - ▁basis
3156
+ - ▁twelve
3157
+ - ▁conduct
3158
+ - ▁street
3159
+ - ▁social
3160
+ - ▁wonder
3161
+ - ▁supreme
3162
+ - ▁difficult
3163
+ - ▁travel
3164
+ - ▁opportunity
3165
+ - ▁defense
3166
+ - ▁plaintiff
3167
+ - ▁white
3168
+ - ▁focus
3169
+ - ▁therefore
3170
+ - ▁challenge
3171
+ - ▁value
3172
+ - ▁except
3173
+ - ▁among
3174
+ - ▁ground
3175
+ - ▁serve
3176
+ - ▁attorney
3177
+ - ▁education
3178
+ - ▁credit
3179
+ - ▁request
3180
+ - ▁women
3181
+ - ▁basically
3182
+ - ▁political
3183
+ - ▁determine
3184
+ - ▁access
3185
+ - ▁across
3186
+ - ▁sentence
3187
+ - ▁perform
3188
+ - ▁everybody
3189
+ - ▁black
3190
+ - ▁drug
3191
+ - ▁accept
3192
+ - ▁eleven
3193
+ - ▁establish
3194
+ - ▁design
3195
+ - ▁success
3196
+ - ▁including
3197
+ - ▁john
3198
+ - ▁employee
3199
+ - ▁nice
3200
+ - ▁crime
3201
+ - ibility
3202
+ - ▁circumstance
3203
+ - ▁occur
3204
+ - ▁citizen
3205
+ - ▁woman
3206
+ - ▁witness
3207
+ - ▁explain
3208
+ - ▁additional
3209
+ - ▁history
3210
+ - ▁front
3211
+ - ▁medical
3212
+ - ▁death
3213
+ - ification
3214
+ - ▁financial
3215
+ - ▁break
3216
+ - ▁connect
3217
+ - ▁perhaps
3218
+ - ▁sorry
3219
+ - ▁especially
3220
+ - ▁trust
3221
+ - ▁insurance
3222
+ - ▁economic
3223
+ - ▁effort
3224
+ - ▁circuit
3225
+ - ▁cross
3226
+ - ▁instead
3227
+ - ▁affect
3228
+ - ▁zero
3229
+ - ▁available
3230
+ - ▁team
3231
+ - ▁application
3232
+ - ▁separate
3233
+ - ▁absolutely
3234
+ - ▁christ
3235
+ - ization
3236
+ - ▁europe
3237
+ - ▁budget
3238
+ - ▁damage
3239
+ - ▁myself
3240
+ - ▁private
3241
+ - ▁figure
3242
+ - ▁provision
3243
+ - ▁collect
3244
+ - ▁similar
3245
+ - ▁regulation
3246
+ - ▁approach
3247
+ - ▁themselves
3248
+ - ▁construct
3249
+ - ▁petition
3250
+ - ▁material
3251
+ - ▁improve
3252
+ - ▁york
3253
+ - ▁parliament
3254
+ - ▁broad
3255
+ - ▁companies
3256
+ - ▁nature
3257
+ - ▁piece
3258
+ - ▁threat
3259
+ - ▁energy
3260
+ - ▁behind
3261
+ - ▁significant
3262
+ - ▁response
3263
+ - ▁definitely
3264
+ - ▁guide
3265
+ - ▁beyond
3266
+ - ▁amendment
3267
+ - ▁measure
3268
+ - ▁criminal
3269
+ - ▁original
3270
+ - ▁brother
3271
+ - ▁impact
3272
+ - q
3273
+ target: nemo.collections.asr.models.confidence_ensemble.ConfidenceEnsembleModel
3274
+ num_models: 0
3275
+ model_selection_block: ./payload.joblib
3276
+ nemo_version: 1.16.0rc0
model_weights.ckpt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b4625e6f5c72ddc8ebf6a11a59e93932d7ea1dd21f9abf3465d87af7557e5607
3
+ size 888
payload.joblib ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3e1373c3438db16eeea81d12f4cd15ed937dd62e9c6642a22677c89432af55e0
3
+ size 71