Arun
commited on
Commit
•
ff9a56f
1
Parent(s):
37c68dd
commit from arun
Browse files- data/bpe_unigram600/bpe.model +3 -0
- data/bpe_unigram600/bpe.vocab +600 -0
- data/bpe_unigram600/tokens.txt +600 -0
- data/bpe_unigram600/train.txt +0 -0
- exp/asr_IndEng188h-rnnt-600bpe/18epoch.pth +3 -0
- exp/asr_IndEng188h-rnnt-600bpe/19epoch.pth +3 -0
- exp/asr_IndEng188h-rnnt-600bpe/20epoch.pth +3 -0
- exp/asr_IndEng188h-rnnt-600bpe/checkpoint.pth +3 -0
- exp/asr_IndEng188h-rnnt-600bpe/config.yaml +794 -0
- exp/asr_IndEng188h-rnnt-600bpe/images/backward_time.png +0 -0
- exp/asr_IndEng188h-rnnt-600bpe/images/cer_ctc.png +0 -0
- exp/asr_IndEng188h-rnnt-600bpe/images/cer_transducer.png +0 -0
- exp/asr_IndEng188h-rnnt-600bpe/images/forward_time.png +0 -0
- exp/asr_IndEng188h-rnnt-600bpe/images/gpu_max_cached_mem_GB.png +0 -0
- exp/asr_IndEng188h-rnnt-600bpe/images/iter_time.png +0 -0
- exp/asr_IndEng188h-rnnt-600bpe/images/loss.png +0 -0
- exp/asr_IndEng188h-rnnt-600bpe/images/loss_ctc.png +0 -0
- exp/asr_IndEng188h-rnnt-600bpe/images/loss_transducer.png +0 -0
- exp/asr_IndEng188h-rnnt-600bpe/images/optim0_lr0.png +0 -0
- exp/asr_IndEng188h-rnnt-600bpe/images/optim_step_time.png +0 -0
- exp/asr_IndEng188h-rnnt-600bpe/images/train_time.png +0 -0
- exp/asr_IndEng188h-rnnt-600bpe/images/wer_transducer.png +0 -0
- exp/asr_IndEng188h-rnnt-600bpe/latest.pth +1 -0
- exp/asr_IndEng188h-rnnt-600bpe/run.sh +1 -0
- exp/asr_IndEng188h-rnnt-600bpe/train.1.log +0 -0
- exp/asr_IndEng188h-rnnt-600bpe/train.log +0 -0
- exp/asr_IndEng188h-rnnt-600bpe/valid.loss.ave.pth +1 -0
- exp/asr_IndEng188h-rnnt-600bpe/valid.loss.ave_3best.pth +3 -0
- exp/asr_IndEng188h-rnnt-600bpe/valid.loss.best.pth +1 -0
- exp/asr_stats_raw_en_bpe600/train/feats_lengths_stats.npz +3 -0
- exp/asr_stats_raw_en_bpe600/train/feats_stats.npz +3 -0
- exp/asr_stats_raw_en_bpe600/train/speech_shape +0 -0
- exp/asr_stats_raw_en_bpe600/train/text_shape +0 -0
- exp/asr_stats_raw_en_bpe600/train/text_shape.bpe +0 -0
data/bpe_unigram600/bpe.model
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:547cecfe136d2a15a04b6b7e52c76672d425dea1e54f0629ae78c167c7fef9fe
|
3 |
+
size 246712
|
data/bpe_unigram600/bpe.vocab
ADDED
@@ -0,0 +1,600 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
<unk> 0
|
2 |
+
<s> 0
|
3 |
+
</s> 0
|
4 |
+
S -3.22703
|
5 |
+
▁THE -3.61052
|
6 |
+
▁A -3.87691
|
7 |
+
E -3.88729
|
8 |
+
T -3.93521
|
9 |
+
▁TO -4.08136
|
10 |
+
I -4.11184
|
11 |
+
D -4.26252
|
12 |
+
▁P -4.32846
|
13 |
+
ING -4.4067
|
14 |
+
▁IN -4.40882
|
15 |
+
▁AT -4.42589
|
16 |
+
▁AND -4.49315
|
17 |
+
ED -4.52975
|
18 |
+
P -4.54267
|
19 |
+
▁FOR -4.54403
|
20 |
+
A -4.56166
|
21 |
+
▁M -4.56774
|
22 |
+
▁OF -4.6302
|
23 |
+
Y -4.73089
|
24 |
+
O -4.73208
|
25 |
+
ER -4.73372
|
26 |
+
M -4.7463
|
27 |
+
▁I -4.74781
|
28 |
+
U -4.77314
|
29 |
+
N -4.7761
|
30 |
+
C -4.77917
|
31 |
+
AR -4.8288
|
32 |
+
AL -4.87438
|
33 |
+
▁C -4.89035
|
34 |
+
G -4.91332
|
35 |
+
▁S -4.95958
|
36 |
+
▁ON -5.03072
|
37 |
+
R -5.08792
|
38 |
+
B -5.09664
|
39 |
+
▁F -5.10649
|
40 |
+
▁IS -5.14308
|
41 |
+
H -5.15852
|
42 |
+
L -5.20049
|
43 |
+
▁T -5.23919
|
44 |
+
F -5.24384
|
45 |
+
OR -5.25252
|
46 |
+
▁TABLE -5.25588
|
47 |
+
IN -5.25667
|
48 |
+
AN -5.28861
|
49 |
+
▁BOOK -5.32442
|
50 |
+
▁RE -5.32943
|
51 |
+
K -5.33969
|
52 |
+
▁THAT -5.35422
|
53 |
+
RE -5.39167
|
54 |
+
▁BE -5.44161
|
55 |
+
▁TWO -5.47769
|
56 |
+
IC -5.48708
|
57 |
+
IT -5.51712
|
58 |
+
LE -5.53379
|
59 |
+
EN -5.53469
|
60 |
+
V -5.54393
|
61 |
+
RA -5.57203
|
62 |
+
▁ONE -5.57621
|
63 |
+
▁G -5.58016
|
64 |
+
▁D -5.59594
|
65 |
+
UR -5.59732
|
66 |
+
▁ -5.6078
|
67 |
+
▁IT -5.6291
|
68 |
+
LY -5.63462
|
69 |
+
ES -5.64478
|
70 |
+
ON -5.6754
|
71 |
+
RI -5.68927
|
72 |
+
▁HE -5.69565
|
73 |
+
W -5.69923
|
74 |
+
▁YOU -5.70031
|
75 |
+
▁WITH -5.70058
|
76 |
+
▁B -5.71628
|
77 |
+
▁ST -5.71669
|
78 |
+
TH -5.75349
|
79 |
+
▁E -5.81036
|
80 |
+
▁WE -5.81416
|
81 |
+
AT -5.82651
|
82 |
+
▁ARE -5.83978
|
83 |
+
▁WAS -5.8475
|
84 |
+
▁NOT -5.87579
|
85 |
+
▁SO -5.87966
|
86 |
+
▁AS -5.88725
|
87 |
+
IL -5.89141
|
88 |
+
▁WANT -5.89634
|
89 |
+
▁W -5.91138
|
90 |
+
ENT -5.91178
|
91 |
+
▁K -5.93082
|
92 |
+
▁HAVE -5.93582
|
93 |
+
▁H -5.93616
|
94 |
+
CH -5.9937
|
95 |
+
▁THREE -5.9957
|
96 |
+
▁DE -5.99758
|
97 |
+
RO -6.00195
|
98 |
+
AS -6.00463
|
99 |
+
▁WILL -6.01414
|
100 |
+
▁L -6.02189
|
101 |
+
TER -6.04306
|
102 |
+
LA -6.08069
|
103 |
+
▁FOUR -6.08721
|
104 |
+
ATION -6.10588
|
105 |
+
▁HAS -6.12756
|
106 |
+
LL -6.13108
|
107 |
+
ST -6.1377
|
108 |
+
▁THIRTY -6.1444
|
109 |
+
▁CAN -6.15402
|
110 |
+
▁DO -6.16995
|
111 |
+
AD -6.17811
|
112 |
+
ERS -6.18168
|
113 |
+
EL -6.18559
|
114 |
+
▁O -6.18905
|
115 |
+
AM -6.19839
|
116 |
+
▁THIS -6.21788
|
117 |
+
UL -6.21866
|
118 |
+
IR -6.22389
|
119 |
+
▁FROM -6.22486
|
120 |
+
TION -6.23775
|
121 |
+
▁HIS -6.25902
|
122 |
+
▁AN -6.26238
|
123 |
+
ION -6.27479
|
124 |
+
▁FIVE -6.27485
|
125 |
+
▁SE -6.31218
|
126 |
+
▁V -6.31581
|
127 |
+
US -6.32137
|
128 |
+
▁ME -6.32207
|
129 |
+
▁BY -6.32378
|
130 |
+
▁LIKE -6.32739
|
131 |
+
IS -6.33515
|
132 |
+
▁SAID -6.36465
|
133 |
+
▁R -6.37404
|
134 |
+
▁EIGHT -6.38196
|
135 |
+
▁N -6.38431
|
136 |
+
▁CON -6.38792
|
137 |
+
▁PEOPLE -6.38828
|
138 |
+
LI -6.3885
|
139 |
+
▁J -6.38939
|
140 |
+
AND -6.39143
|
141 |
+
▁BUT -6.39344
|
142 |
+
J -6.39434
|
143 |
+
▁WHAT -6.3961
|
144 |
+
▁TODAY -6.39688
|
145 |
+
UN -6.40978
|
146 |
+
ATE -6.41273
|
147 |
+
▁SIX -6.4322
|
148 |
+
MENT -6.43489
|
149 |
+
▁RA -6.43868
|
150 |
+
▁OR -6.44029
|
151 |
+
UT -6.44759
|
152 |
+
▁SU -6.44849
|
153 |
+
CE -6.45804
|
154 |
+
▁WHICH -6.46112
|
155 |
+
▁CO -6.49178
|
156 |
+
▁EX -6.49982
|
157 |
+
NE -6.50057
|
158 |
+
OL -6.50792
|
159 |
+
Z -6.52752
|
160 |
+
▁SEVEN -6.5293
|
161 |
+
SE -6.53456
|
162 |
+
NG -6.53506
|
163 |
+
▁MO -6.53512
|
164 |
+
▁NEED -6.54447
|
165 |
+
VER -6.54532
|
166 |
+
▁ALL -6.5508
|
167 |
+
ITY -6.55594
|
168 |
+
VE -6.56909
|
169 |
+
▁LE -6.57438
|
170 |
+
▁GO -6.59453
|
171 |
+
▁WHO -6.60253
|
172 |
+
GE -6.60414
|
173 |
+
IVE -6.60784
|
174 |
+
▁MALL -6.62012
|
175 |
+
HI -6.63055
|
176 |
+
▁AM -6.63165
|
177 |
+
AC -6.63478
|
178 |
+
IDE -6.63538
|
179 |
+
▁WOULD -6.64317
|
180 |
+
LO -6.65078
|
181 |
+
▁NINE -6.66117
|
182 |
+
▁BA -6.67577
|
183 |
+
▁MA -6.68057
|
184 |
+
UM -6.68616
|
185 |
+
▁NEW -6.68842
|
186 |
+
▁WHEN -6.69853
|
187 |
+
▁TIME -6.71018
|
188 |
+
AY -6.71301
|
189 |
+
▁DINNER -6.71592
|
190 |
+
▁RO -6.71975
|
191 |
+
IES -6.72025
|
192 |
+
CK -6.72355
|
193 |
+
▁ALSO -6.72448
|
194 |
+
▁PRO -6.72501
|
195 |
+
▁EN -6.72688
|
196 |
+
OW -6.73337
|
197 |
+
▁VI -6.73975
|
198 |
+
OT -6.74178
|
199 |
+
▁THERE -6.74716
|
200 |
+
▁MY -6.74777
|
201 |
+
▁YEAR -6.75995
|
202 |
+
▁LA -6.76837
|
203 |
+
▁TEN -6.77342
|
204 |
+
▁ITS -6.77878
|
205 |
+
IM -6.78433
|
206 |
+
▁INDIA -6.79436
|
207 |
+
▁US -6.79702
|
208 |
+
▁ABOUT -6.79744
|
209 |
+
▁UP -6.79924
|
210 |
+
MA -6.80119
|
211 |
+
▁THOUSAND -6.80669
|
212 |
+
▁OUT -6.80809
|
213 |
+
AP -6.81835
|
214 |
+
▁BO -6.82551
|
215 |
+
AGE -6.82842
|
216 |
+
▁PLAY -6.82931
|
217 |
+
▁THEY -6.82936
|
218 |
+
X -6.83563
|
219 |
+
▁GET -6.85878
|
220 |
+
EST -6.85891
|
221 |
+
▁THEIR -6.87871
|
222 |
+
▁HAD -6.88152
|
223 |
+
IST -6.90705
|
224 |
+
▁NO -6.90895
|
225 |
+
▁SP -6.90951
|
226 |
+
▁AFTER -6.91139
|
227 |
+
▁YOUR -6.9134
|
228 |
+
TING -6.92553
|
229 |
+
CI -6.93358
|
230 |
+
▁NOVEMBER -6.9351
|
231 |
+
Q -6.93584
|
232 |
+
▁DIS -6.94836
|
233 |
+
▁BEEN -6.95192
|
234 |
+
▁APP -6.95905
|
235 |
+
▁HO -6.96008
|
236 |
+
KE -6.96533
|
237 |
+
▁UN -6.97613
|
238 |
+
OS -6.97808
|
239 |
+
ARD -6.97836
|
240 |
+
UD -6.98846
|
241 |
+
▁OVER -6.99657
|
242 |
+
▁PLEASE -7.00439
|
243 |
+
ACH -7.01272
|
244 |
+
▁OKAY -7.01452
|
245 |
+
▁PRE -7.01476
|
246 |
+
▁MORE -7.01977
|
247 |
+
▁SOME -7.03151
|
248 |
+
▁SH -7.03765
|
249 |
+
NA -7.03795
|
250 |
+
▁BR -7.04252
|
251 |
+
ANT -7.0469
|
252 |
+
▁FIRST -7.04948
|
253 |
+
▁PART -7.05024
|
254 |
+
TED -7.05628
|
255 |
+
OM -7.05745
|
256 |
+
ALLY -7.05883
|
257 |
+
▁CHA -7.07166
|
258 |
+
▁ADULTS -7.07832
|
259 |
+
RY -7.09827
|
260 |
+
URE -7.10255
|
261 |
+
IF -7.10623
|
262 |
+
▁TWENTY -7.10679
|
263 |
+
▁DAY -7.10738
|
264 |
+
AV -7.11774
|
265 |
+
▁WORK -7.11809
|
266 |
+
▁TOMORROW -7.11942
|
267 |
+
HA -7.12522
|
268 |
+
TA -7.12685
|
269 |
+
ONE -7.13045
|
270 |
+
ENCE -7.13496
|
271 |
+
▁POINT -7.13605
|
272 |
+
▁WERE -7.14235
|
273 |
+
▁HA -7.15184
|
274 |
+
▁IF -7.15203
|
275 |
+
OUS -7.15233
|
276 |
+
PE -7.16673
|
277 |
+
▁NA -7.17073
|
278 |
+
▁COM -7.17093
|
279 |
+
▁HER -7.17624
|
280 |
+
▁OTHER -7.1767
|
281 |
+
TS -7.17882
|
282 |
+
▁NOW -7.18564
|
283 |
+
HO -7.18638
|
284 |
+
▁ACT -7.20115
|
285 |
+
▁HUNDRED -7.20692
|
286 |
+
ANCE -7.22224
|
287 |
+
▁MEMBERS -7.22716
|
288 |
+
▁HOW -7.24614
|
289 |
+
▁SHE -7.24663
|
290 |
+
▁QU -7.24734
|
291 |
+
▁U -7.25298
|
292 |
+
▁SAY -7.26123
|
293 |
+
▁SC -7.26992
|
294 |
+
AK -7.28844
|
295 |
+
▁KNOW -7.30513
|
296 |
+
▁SA -7.30862
|
297 |
+
▁AR -7.31421
|
298 |
+
▁ANY -7.31752
|
299 |
+
▁THAN -7.31858
|
300 |
+
▁JO -7.32522
|
301 |
+
▁JUST -7.34454
|
302 |
+
▁ROAD -7.34694
|
303 |
+
MAN -7.35321
|
304 |
+
▁TEST -7.35526
|
305 |
+
ABLE -7.37146
|
306 |
+
HE -7.37293
|
307 |
+
▁MAN -7.38385
|
308 |
+
▁LUNCH -7.39085
|
309 |
+
▁TAKE -7.394
|
310 |
+
▁INTO -7.39728
|
311 |
+
▁LI -7.39889
|
312 |
+
▁WANTED -7.40564
|
313 |
+
AB -7.40747
|
314 |
+
UND -7.41025
|
315 |
+
GH -7.42715
|
316 |
+
MO -7.42948
|
317 |
+
▁RUN -7.43219
|
318 |
+
▁ONLY -7.43619
|
319 |
+
▁SHA -7.43835
|
320 |
+
PER -7.45224
|
321 |
+
ITE -7.45309
|
322 |
+
▁TEAM -7.47424
|
323 |
+
ISH -7.47962
|
324 |
+
▁MOST -7.48066
|
325 |
+
▁EVEN -7.4815
|
326 |
+
▁TWELVE -7.48667
|
327 |
+
▁RESERVE -7.48735
|
328 |
+
▁DID -7.49362
|
329 |
+
▁OUR -7.50195
|
330 |
+
▁HIM -7.50421
|
331 |
+
▁COMP -7.50493
|
332 |
+
ARY -7.50809
|
333 |
+
▁LOOK -7.51493
|
334 |
+
▁SEE -7.51972
|
335 |
+
▁OFF -7.52277
|
336 |
+
ATED -7.54799
|
337 |
+
▁CALL -7.55092
|
338 |
+
▁SECOND -7.55123
|
339 |
+
▁WHILE -7.57548
|
340 |
+
WAR -7.5792
|
341 |
+
▁MAKE -7.58124
|
342 |
+
▁FILM -7.58544
|
343 |
+
▁WIN -7.59518
|
344 |
+
▁WORLD -7.60713
|
345 |
+
▁RIGHT -7.61792
|
346 |
+
TEN -7.62438
|
347 |
+
▁HMM -7.63552
|
348 |
+
▁PLACE -7.63933
|
349 |
+
▁STATE -7.65324
|
350 |
+
▁INDIAN -7.65679
|
351 |
+
▁VERY -7.67008
|
352 |
+
▁REPORT -7.67447
|
353 |
+
▁COULD -7.67872
|
354 |
+
▁LAST -7.68137
|
355 |
+
▁THINK -7.68358
|
356 |
+
▁START -7.69251
|
357 |
+
▁ADD -7.69763
|
358 |
+
▁NATION -7.71825
|
359 |
+
▁WELL -7.72027
|
360 |
+
▁HIGH -7.72159
|
361 |
+
▁WHERE -7.72585
|
362 |
+
▁MADE -7.72593
|
363 |
+
▁MOVIE -7.72775
|
364 |
+
▁BACK -7.73396
|
365 |
+
▁COME -7.73618
|
366 |
+
ASH -7.74461
|
367 |
+
▁EVERY -7.74598
|
368 |
+
▁KA -7.75083
|
369 |
+
▁BEST -7.75097
|
370 |
+
▁BECAUSE -7.75918
|
371 |
+
▁ENGLAND -7.76455
|
372 |
+
ICK -7.77127
|
373 |
+
▁SHOW -7.79353
|
374 |
+
▁WAY -7.79553
|
375 |
+
COM -7.79959
|
376 |
+
▁GIVE -7.80192
|
377 |
+
▁KIND -7.81
|
378 |
+
▁CHILDREN -7.82562
|
379 |
+
▁BEFORE -7.82942
|
380 |
+
▁PAKISTAN -7.83421
|
381 |
+
▁LEAD -7.83574
|
382 |
+
▁WATCH -7.84393
|
383 |
+
▁THROUGH -7.8556
|
384 |
+
▁MUCH -7.86308
|
385 |
+
▁MANY -7.87673
|
386 |
+
▁GOOD -7.88494
|
387 |
+
▁SUNDAY -7.89062
|
388 |
+
▁FIFTY -7.89265
|
389 |
+
PORT -7.89315
|
390 |
+
▁NAGAR -7.89993
|
391 |
+
▁THIRD -7.91014
|
392 |
+
▁GREAT -7.91348
|
393 |
+
▁UNDER -7.91683
|
394 |
+
▁ADULT -7.91968
|
395 |
+
▁HELP -7.92339
|
396 |
+
▁MONDAY -7.94499
|
397 |
+
LIGHT -7.95582
|
398 |
+
▁USERS -7.96057
|
399 |
+
▁SHOULD -7.96562
|
400 |
+
▁BAN -7.98398
|
401 |
+
▁BALL -7.98559
|
402 |
+
▁PLAN -7.98578
|
403 |
+
▁VIDEO -8.00262
|
404 |
+
▁MONTH -8.00296
|
405 |
+
▁Z -8.0119
|
406 |
+
LINE -8.01271
|
407 |
+
▁NEXT -8.0174
|
408 |
+
▁DIRECT -8.02203
|
409 |
+
▁BIG -8.02322
|
410 |
+
▁YES -8.03063
|
411 |
+
▁CAME -8.0309
|
412 |
+
▁MUMBAI -8.03488
|
413 |
+
▁SEATS -8.03803
|
414 |
+
▁MATCH -8.0424
|
415 |
+
▁LOT -8.05153
|
416 |
+
▁GAME -8.05477
|
417 |
+
▁PERSON -8.05542
|
418 |
+
▁AGAINST -8.06274
|
419 |
+
▁WHY -8.06636
|
420 |
+
▁IMP -8.07545
|
421 |
+
▁OLD -8.07642
|
422 |
+
▁RESERVATION -8.08182
|
423 |
+
▁COMPANY -8.08314
|
424 |
+
▁DURING -8.0836
|
425 |
+
▁FEATURE -8.08677
|
426 |
+
▁COMMUN -8.08804
|
427 |
+
▁BETWEEN -8.10678
|
428 |
+
▁WICKET -8.10804
|
429 |
+
▁SCORE -8.1165
|
430 |
+
▁LIFE -8.11886
|
431 |
+
▁DOWN -8.11986
|
432 |
+
▁SATURDAY -8.12077
|
433 |
+
OOD -8.12441
|
434 |
+
▁TALK -8.13046
|
435 |
+
▁HOME -8.13206
|
436 |
+
▁OPEN -8.13663
|
437 |
+
▁YEAH -8.14149
|
438 |
+
▁CHANGE -8.15405
|
439 |
+
▁NINETEEN -8.15731
|
440 |
+
▁SIXTY -8.17976
|
441 |
+
▁ELECT -8.18308
|
442 |
+
▁NINETY -8.18775
|
443 |
+
▁LAUNCH -8.18837
|
444 |
+
▁TRANS -8.1884
|
445 |
+
▁FINAL -8.19708
|
446 |
+
▁CRICKET -8.20076
|
447 |
+
▁GOVERNMENT -8.20215
|
448 |
+
▁FOUND -8.21065
|
449 |
+
▁NUMBER -8.21615
|
450 |
+
▁HOWEVER -8.22472
|
451 |
+
▁ZERO -8.23194
|
452 |
+
▁SINCE -8.2349
|
453 |
+
▁DIFFERENT -8.23608
|
454 |
+
▁SERIES -8.23655
|
455 |
+
▁MUSIC -8.23932
|
456 |
+
▁DELHI -8.24631
|
457 |
+
▁SUPPORT -8.25204
|
458 |
+
▁SEVENTY -8.25296
|
459 |
+
▁AUSTRALIA -8.25935
|
460 |
+
▁SIXTEEN -8.26083
|
461 |
+
▁CHECK -8.26085
|
462 |
+
▁MAKING -8.26887
|
463 |
+
▁MINUTE -8.2817
|
464 |
+
▁THINGS -8.28175
|
465 |
+
▁CHENNAI -8.2862
|
466 |
+
▁AROUND -8.28635
|
467 |
+
▁FRIDAY -8.29226
|
468 |
+
▁LONG -8.29418
|
469 |
+
▁THURSDAY -8.29684
|
470 |
+
▁GOOGLE -8.2999
|
471 |
+
▁MINISTER -8.30143
|
472 |
+
▁BOWL -8.30317
|
473 |
+
▁SAME -8.30941
|
474 |
+
▁KHAN -8.30953
|
475 |
+
▁SPACE -8.31252
|
476 |
+
▁FACEBOOK -8.31379
|
477 |
+
▁BARB -8.3162
|
478 |
+
▁POST -8.3299
|
479 |
+
▁FOLLOW -8.33103
|
480 |
+
▁WEEK -8.33546
|
481 |
+
▁RESEARCH -8.34056
|
482 |
+
AUGHT -8.34229
|
483 |
+
▁SPEAK -8.35027
|
484 |
+
▁POLICE -8.35828
|
485 |
+
▁INNINGS -8.36175
|
486 |
+
▁BECOME -8.3667
|
487 |
+
▁KALYAN -8.36975
|
488 |
+
▁FEEL -8.37088
|
489 |
+
▁RELEASE -8.37808
|
490 |
+
▁THOSE -8.37967
|
491 |
+
▁MILLION -8.38129
|
492 |
+
▁ISSUE -8.38591
|
493 |
+
▁PHONE -8.3867
|
494 |
+
▁FIND -8.38994
|
495 |
+
▁RAHUL -8.39139
|
496 |
+
▁PROBLEM -8.39299
|
497 |
+
▁STILL -8.39881
|
498 |
+
▁COURT -8.40369
|
499 |
+
▁HAPPEN -8.40824
|
500 |
+
▁KEEP -8.40832
|
501 |
+
▁ELSE -8.4119
|
502 |
+
▁ANOTHER -8.43245
|
503 |
+
▁DEVELOP -8.43771
|
504 |
+
▁CREAT -8.43902
|
505 |
+
▁WEDNESDAY -8.43947
|
506 |
+
▁TELL -8.45327
|
507 |
+
▁CLOSE -8.45429
|
508 |
+
▁SOUTH -8.4632
|
509 |
+
▁TURN -8.46532
|
510 |
+
▁THOUGH -8.47238
|
511 |
+
▁TUESDAY -8.479
|
512 |
+
▁KOHLI -8.48821
|
513 |
+
▁SOMETHING -8.49024
|
514 |
+
▁CHILD -8.49187
|
515 |
+
▁LOVE -8.49401
|
516 |
+
▁ACCORDING -8.49564
|
517 |
+
▁DEVICE -8.49566
|
518 |
+
▁IMPORTANT -8.49938
|
519 |
+
▁GROUP -8.49938
|
520 |
+
▁CONTINUE -8.50313
|
521 |
+
▁FAMILY -8.50501
|
522 |
+
▁DRIVE -8.50672
|
523 |
+
▁COUNTRY -8.50689
|
524 |
+
▁CAPTAIN -8.51066
|
525 |
+
▁COUNTRIES -8.52208
|
526 |
+
▁WATER -8.54455
|
527 |
+
▁OWN -8.54582
|
528 |
+
▁PROJECT -8.54728
|
529 |
+
▁DECEMBER -8.54924
|
530 |
+
▁WITHOUT -8.54951
|
531 |
+
▁CENTURY -8.5671
|
532 |
+
▁FIFTEEN -8.57111
|
533 |
+
▁OFFICE -8.57391
|
534 |
+
▁ENGLISH -8.58528
|
535 |
+
▁PROGRAM -8.58937
|
536 |
+
▁FRIEND -8.59348
|
537 |
+
▁RUPEES -8.60588
|
538 |
+
▁CONSIDER -8.61006
|
539 |
+
▁BENGALURU -8.62267
|
540 |
+
▁PERFORM -8.62904
|
541 |
+
▁ACCOUNT -8.6376
|
542 |
+
▁PRIME -8.63768
|
543 |
+
▁MEDIA -8.64604
|
544 |
+
▁UNDERSTAND -8.65713
|
545 |
+
▁EXPECT -8.66592
|
546 |
+
▁RECORD -8.66815
|
547 |
+
▁EXPERIENCE -8.67257
|
548 |
+
▁OFFICIAL -8.67258
|
549 |
+
▁DECISION -8.68151
|
550 |
+
▁ELEVEN -8.68151
|
551 |
+
▁TONIGHT -8.68376
|
552 |
+
▁ATTACK -8.68837
|
553 |
+
▁BREAK -8.68841
|
554 |
+
▁AGAIN -8.68975
|
555 |
+
▁INCLUDING -8.6928
|
556 |
+
▁DESIGN -8.6928
|
557 |
+
▁ANNOUNCE -8.70651
|
558 |
+
▁COMPLETE -8.70917
|
559 |
+
▁SERVICE -8.71344
|
560 |
+
▁SEVENTEEN -8.71345
|
561 |
+
▁ALREADY -8.72509
|
562 |
+
▁TRAVEL -8.73929
|
563 |
+
▁PRODUCT -8.74165
|
564 |
+
▁PROCESS -8.74406
|
565 |
+
▁PRESIDENT -8.75605
|
566 |
+
▁KOLKATA -8.76333
|
567 |
+
▁SYSTEM -8.77067
|
568 |
+
▁ALWAYS -8.78301
|
569 |
+
▁SMARTPHONE -8.78301
|
570 |
+
▁SUCCESS -8.7855
|
571 |
+
▁MAJOR -8.78807
|
572 |
+
▁PHOTO -8.79056
|
573 |
+
▁IMPROVE -8.79552
|
574 |
+
▁AMERICA -8.79817
|
575 |
+
▁SQUARE -8.81072
|
576 |
+
▁LEAGUE -8.81841
|
577 |
+
▁INTERNATIONAL -8.82099
|
578 |
+
▁BUILD -8.82358
|
579 |
+
▁ACTUALLY -8.82361
|
580 |
+
▁SIGN -8.83142
|
581 |
+
▁CHIEF -8.83399
|
582 |
+
▁TWITTER -8.84185
|
583 |
+
1 -11.9772
|
584 |
+
0 -12.0397
|
585 |
+
3 -12.9079
|
586 |
+
4 -13.2746
|
587 |
+
9 -13.2746
|
588 |
+
2 -13.8579
|
589 |
+
i -14.3579
|
590 |
+
o -14.3579
|
591 |
+
< -14.3579
|
592 |
+
s -14.3579
|
593 |
+
' -14.3579
|
594 |
+
/ -14.3579
|
595 |
+
5 -14.3579
|
596 |
+
7 -14.3579
|
597 |
+
8 -14.3579
|
598 |
+
> -14.3579
|
599 |
+
e -14.3579
|
600 |
+
6 -15.3579
|
data/bpe_unigram600/tokens.txt
ADDED
@@ -0,0 +1,600 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
<blank>
|
2 |
+
<unk>
|
3 |
+
S
|
4 |
+
▁THE
|
5 |
+
▁A
|
6 |
+
E
|
7 |
+
T
|
8 |
+
▁TO
|
9 |
+
I
|
10 |
+
D
|
11 |
+
▁P
|
12 |
+
ING
|
13 |
+
▁IN
|
14 |
+
▁AT
|
15 |
+
▁AND
|
16 |
+
ED
|
17 |
+
P
|
18 |
+
▁FOR
|
19 |
+
A
|
20 |
+
▁M
|
21 |
+
▁OF
|
22 |
+
Y
|
23 |
+
O
|
24 |
+
ER
|
25 |
+
M
|
26 |
+
▁I
|
27 |
+
U
|
28 |
+
N
|
29 |
+
C
|
30 |
+
AR
|
31 |
+
AL
|
32 |
+
▁C
|
33 |
+
G
|
34 |
+
▁S
|
35 |
+
▁ON
|
36 |
+
R
|
37 |
+
B
|
38 |
+
▁F
|
39 |
+
▁IS
|
40 |
+
H
|
41 |
+
L
|
42 |
+
▁T
|
43 |
+
F
|
44 |
+
OR
|
45 |
+
▁TABLE
|
46 |
+
IN
|
47 |
+
AN
|
48 |
+
▁BOOK
|
49 |
+
▁RE
|
50 |
+
K
|
51 |
+
▁THAT
|
52 |
+
RE
|
53 |
+
▁BE
|
54 |
+
▁TWO
|
55 |
+
IC
|
56 |
+
IT
|
57 |
+
LE
|
58 |
+
EN
|
59 |
+
V
|
60 |
+
RA
|
61 |
+
▁ONE
|
62 |
+
▁G
|
63 |
+
▁D
|
64 |
+
UR
|
65 |
+
▁
|
66 |
+
▁IT
|
67 |
+
LY
|
68 |
+
ES
|
69 |
+
ON
|
70 |
+
RI
|
71 |
+
▁HE
|
72 |
+
W
|
73 |
+
▁YOU
|
74 |
+
▁WITH
|
75 |
+
▁B
|
76 |
+
▁ST
|
77 |
+
TH
|
78 |
+
▁E
|
79 |
+
▁WE
|
80 |
+
AT
|
81 |
+
▁ARE
|
82 |
+
▁WAS
|
83 |
+
▁NOT
|
84 |
+
▁SO
|
85 |
+
▁AS
|
86 |
+
IL
|
87 |
+
▁WANT
|
88 |
+
▁W
|
89 |
+
ENT
|
90 |
+
▁K
|
91 |
+
▁HAVE
|
92 |
+
▁H
|
93 |
+
CH
|
94 |
+
▁THREE
|
95 |
+
▁DE
|
96 |
+
RO
|
97 |
+
AS
|
98 |
+
▁WILL
|
99 |
+
▁L
|
100 |
+
TER
|
101 |
+
LA
|
102 |
+
▁FOUR
|
103 |
+
ATION
|
104 |
+
▁HAS
|
105 |
+
LL
|
106 |
+
ST
|
107 |
+
▁THIRTY
|
108 |
+
▁CAN
|
109 |
+
▁DO
|
110 |
+
AD
|
111 |
+
ERS
|
112 |
+
EL
|
113 |
+
▁O
|
114 |
+
AM
|
115 |
+
▁THIS
|
116 |
+
UL
|
117 |
+
IR
|
118 |
+
▁FROM
|
119 |
+
TION
|
120 |
+
▁HIS
|
121 |
+
▁AN
|
122 |
+
ION
|
123 |
+
▁FIVE
|
124 |
+
▁SE
|
125 |
+
▁V
|
126 |
+
US
|
127 |
+
▁ME
|
128 |
+
▁BY
|
129 |
+
▁LIKE
|
130 |
+
IS
|
131 |
+
▁SAID
|
132 |
+
▁R
|
133 |
+
▁EIGHT
|
134 |
+
▁N
|
135 |
+
▁CON
|
136 |
+
▁PEOPLE
|
137 |
+
LI
|
138 |
+
▁J
|
139 |
+
AND
|
140 |
+
▁BUT
|
141 |
+
J
|
142 |
+
▁WHAT
|
143 |
+
▁TODAY
|
144 |
+
UN
|
145 |
+
ATE
|
146 |
+
▁SIX
|
147 |
+
MENT
|
148 |
+
▁RA
|
149 |
+
▁OR
|
150 |
+
UT
|
151 |
+
▁SU
|
152 |
+
CE
|
153 |
+
▁WHICH
|
154 |
+
▁CO
|
155 |
+
▁EX
|
156 |
+
NE
|
157 |
+
OL
|
158 |
+
Z
|
159 |
+
▁SEVEN
|
160 |
+
SE
|
161 |
+
NG
|
162 |
+
▁MO
|
163 |
+
▁NEED
|
164 |
+
VER
|
165 |
+
▁ALL
|
166 |
+
ITY
|
167 |
+
VE
|
168 |
+
▁LE
|
169 |
+
▁GO
|
170 |
+
▁WHO
|
171 |
+
GE
|
172 |
+
IVE
|
173 |
+
▁MALL
|
174 |
+
HI
|
175 |
+
▁AM
|
176 |
+
AC
|
177 |
+
IDE
|
178 |
+
▁WOULD
|
179 |
+
LO
|
180 |
+
▁NINE
|
181 |
+
▁BA
|
182 |
+
▁MA
|
183 |
+
UM
|
184 |
+
▁NEW
|
185 |
+
▁WHEN
|
186 |
+
▁TIME
|
187 |
+
AY
|
188 |
+
▁DINNER
|
189 |
+
▁RO
|
190 |
+
IES
|
191 |
+
CK
|
192 |
+
▁ALSO
|
193 |
+
▁PRO
|
194 |
+
▁EN
|
195 |
+
OW
|
196 |
+
▁VI
|
197 |
+
OT
|
198 |
+
▁THERE
|
199 |
+
▁MY
|
200 |
+
▁YEAR
|
201 |
+
▁LA
|
202 |
+
▁TEN
|
203 |
+
▁ITS
|
204 |
+
IM
|
205 |
+
▁INDIA
|
206 |
+
▁US
|
207 |
+
▁ABOUT
|
208 |
+
▁UP
|
209 |
+
MA
|
210 |
+
▁THOUSAND
|
211 |
+
▁OUT
|
212 |
+
AP
|
213 |
+
▁BO
|
214 |
+
AGE
|
215 |
+
▁PLAY
|
216 |
+
▁THEY
|
217 |
+
X
|
218 |
+
▁GET
|
219 |
+
EST
|
220 |
+
▁THEIR
|
221 |
+
▁HAD
|
222 |
+
IST
|
223 |
+
▁NO
|
224 |
+
▁SP
|
225 |
+
▁AFTER
|
226 |
+
▁YOUR
|
227 |
+
TING
|
228 |
+
CI
|
229 |
+
▁NOVEMBER
|
230 |
+
Q
|
231 |
+
▁DIS
|
232 |
+
▁BEEN
|
233 |
+
▁APP
|
234 |
+
▁HO
|
235 |
+
KE
|
236 |
+
▁UN
|
237 |
+
OS
|
238 |
+
ARD
|
239 |
+
UD
|
240 |
+
▁OVER
|
241 |
+
▁PLEASE
|
242 |
+
ACH
|
243 |
+
▁OKAY
|
244 |
+
▁PRE
|
245 |
+
▁MORE
|
246 |
+
▁SOME
|
247 |
+
▁SH
|
248 |
+
NA
|
249 |
+
▁BR
|
250 |
+
ANT
|
251 |
+
▁FIRST
|
252 |
+
▁PART
|
253 |
+
TED
|
254 |
+
OM
|
255 |
+
ALLY
|
256 |
+
▁CHA
|
257 |
+
▁ADULTS
|
258 |
+
RY
|
259 |
+
URE
|
260 |
+
IF
|
261 |
+
▁TWENTY
|
262 |
+
▁DAY
|
263 |
+
AV
|
264 |
+
▁WORK
|
265 |
+
▁TOMORROW
|
266 |
+
HA
|
267 |
+
TA
|
268 |
+
ONE
|
269 |
+
ENCE
|
270 |
+
▁POINT
|
271 |
+
▁WERE
|
272 |
+
▁HA
|
273 |
+
▁IF
|
274 |
+
OUS
|
275 |
+
PE
|
276 |
+
▁NA
|
277 |
+
▁COM
|
278 |
+
▁HER
|
279 |
+
▁OTHER
|
280 |
+
TS
|
281 |
+
▁NOW
|
282 |
+
HO
|
283 |
+
▁ACT
|
284 |
+
▁HUNDRED
|
285 |
+
ANCE
|
286 |
+
▁MEMBERS
|
287 |
+
▁HOW
|
288 |
+
▁SHE
|
289 |
+
▁QU
|
290 |
+
▁U
|
291 |
+
▁SAY
|
292 |
+
▁SC
|
293 |
+
AK
|
294 |
+
▁KNOW
|
295 |
+
▁SA
|
296 |
+
▁AR
|
297 |
+
▁ANY
|
298 |
+
▁THAN
|
299 |
+
▁JO
|
300 |
+
▁JUST
|
301 |
+
▁ROAD
|
302 |
+
MAN
|
303 |
+
▁TEST
|
304 |
+
ABLE
|
305 |
+
HE
|
306 |
+
▁MAN
|
307 |
+
▁LUNCH
|
308 |
+
▁TAKE
|
309 |
+
▁INTO
|
310 |
+
▁LI
|
311 |
+
▁WANTED
|
312 |
+
AB
|
313 |
+
UND
|
314 |
+
GH
|
315 |
+
MO
|
316 |
+
▁RUN
|
317 |
+
▁ONLY
|
318 |
+
▁SHA
|
319 |
+
PER
|
320 |
+
ITE
|
321 |
+
▁TEAM
|
322 |
+
ISH
|
323 |
+
▁MOST
|
324 |
+
▁EVEN
|
325 |
+
▁TWELVE
|
326 |
+
▁RESERVE
|
327 |
+
▁DID
|
328 |
+
▁OUR
|
329 |
+
▁HIM
|
330 |
+
▁COMP
|
331 |
+
ARY
|
332 |
+
▁LOOK
|
333 |
+
▁SEE
|
334 |
+
▁OFF
|
335 |
+
ATED
|
336 |
+
▁CALL
|
337 |
+
▁SECOND
|
338 |
+
▁WHILE
|
339 |
+
WAR
|
340 |
+
▁MAKE
|
341 |
+
▁FILM
|
342 |
+
▁WIN
|
343 |
+
▁WORLD
|
344 |
+
▁RIGHT
|
345 |
+
TEN
|
346 |
+
▁HMM
|
347 |
+
▁PLACE
|
348 |
+
▁STATE
|
349 |
+
▁INDIAN
|
350 |
+
▁VERY
|
351 |
+
▁REPORT
|
352 |
+
▁COULD
|
353 |
+
▁LAST
|
354 |
+
▁THINK
|
355 |
+
▁START
|
356 |
+
▁ADD
|
357 |
+
▁NATION
|
358 |
+
▁WELL
|
359 |
+
▁HIGH
|
360 |
+
▁WHERE
|
361 |
+
▁MADE
|
362 |
+
▁MOVIE
|
363 |
+
▁BACK
|
364 |
+
▁COME
|
365 |
+
ASH
|
366 |
+
▁EVERY
|
367 |
+
▁KA
|
368 |
+
▁BEST
|
369 |
+
▁BECAUSE
|
370 |
+
▁ENGLAND
|
371 |
+
ICK
|
372 |
+
▁SHOW
|
373 |
+
▁WAY
|
374 |
+
COM
|
375 |
+
▁GIVE
|
376 |
+
▁KIND
|
377 |
+
▁CHILDREN
|
378 |
+
▁BEFORE
|
379 |
+
▁PAKISTAN
|
380 |
+
▁LEAD
|
381 |
+
▁WATCH
|
382 |
+
▁THROUGH
|
383 |
+
▁MUCH
|
384 |
+
▁MANY
|
385 |
+
▁GOOD
|
386 |
+
▁SUNDAY
|
387 |
+
▁FIFTY
|
388 |
+
PORT
|
389 |
+
▁NAGAR
|
390 |
+
▁THIRD
|
391 |
+
▁GREAT
|
392 |
+
▁UNDER
|
393 |
+
▁ADULT
|
394 |
+
▁HELP
|
395 |
+
▁MONDAY
|
396 |
+
LIGHT
|
397 |
+
▁USERS
|
398 |
+
▁SHOULD
|
399 |
+
▁BAN
|
400 |
+
▁BALL
|
401 |
+
▁PLAN
|
402 |
+
▁VIDEO
|
403 |
+
▁MONTH
|
404 |
+
▁Z
|
405 |
+
LINE
|
406 |
+
▁NEXT
|
407 |
+
▁DIRECT
|
408 |
+
▁BIG
|
409 |
+
▁YES
|
410 |
+
▁CAME
|
411 |
+
▁MUMBAI
|
412 |
+
▁SEATS
|
413 |
+
▁MATCH
|
414 |
+
▁LOT
|
415 |
+
▁GAME
|
416 |
+
▁PERSON
|
417 |
+
▁AGAINST
|
418 |
+
▁WHY
|
419 |
+
▁IMP
|
420 |
+
▁OLD
|
421 |
+
▁RESERVATION
|
422 |
+
▁COMPANY
|
423 |
+
▁DURING
|
424 |
+
▁FEATURE
|
425 |
+
▁COMMUN
|
426 |
+
▁BETWEEN
|
427 |
+
▁WICKET
|
428 |
+
▁SCORE
|
429 |
+
▁LIFE
|
430 |
+
▁DOWN
|
431 |
+
▁SATURDAY
|
432 |
+
OOD
|
433 |
+
▁TALK
|
434 |
+
▁HOME
|
435 |
+
▁OPEN
|
436 |
+
▁YEAH
|
437 |
+
▁CHANGE
|
438 |
+
▁NINETEEN
|
439 |
+
▁SIXTY
|
440 |
+
▁ELECT
|
441 |
+
▁NINETY
|
442 |
+
▁LAUNCH
|
443 |
+
▁TRANS
|
444 |
+
▁FINAL
|
445 |
+
▁CRICKET
|
446 |
+
▁GOVERNMENT
|
447 |
+
▁FOUND
|
448 |
+
▁NUMBER
|
449 |
+
▁HOWEVER
|
450 |
+
▁ZERO
|
451 |
+
▁SINCE
|
452 |
+
▁DIFFERENT
|
453 |
+
▁SERIES
|
454 |
+
▁MUSIC
|
455 |
+
▁DELHI
|
456 |
+
▁SUPPORT
|
457 |
+
▁SEVENTY
|
458 |
+
▁AUSTRALIA
|
459 |
+
▁SIXTEEN
|
460 |
+
▁CHECK
|
461 |
+
▁MAKING
|
462 |
+
▁MINUTE
|
463 |
+
▁THINGS
|
464 |
+
▁CHENNAI
|
465 |
+
▁AROUND
|
466 |
+
▁FRIDAY
|
467 |
+
▁LONG
|
468 |
+
▁THURSDAY
|
469 |
+
▁GOOGLE
|
470 |
+
▁MINISTER
|
471 |
+
▁BOWL
|
472 |
+
▁SAME
|
473 |
+
▁KHAN
|
474 |
+
▁SPACE
|
475 |
+
▁FACEBOOK
|
476 |
+
▁BARB
|
477 |
+
▁POST
|
478 |
+
▁FOLLOW
|
479 |
+
▁WEEK
|
480 |
+
▁RESEARCH
|
481 |
+
AUGHT
|
482 |
+
▁SPEAK
|
483 |
+
▁POLICE
|
484 |
+
▁INNINGS
|
485 |
+
▁BECOME
|
486 |
+
▁KALYAN
|
487 |
+
▁FEEL
|
488 |
+
▁RELEASE
|
489 |
+
▁THOSE
|
490 |
+
▁MILLION
|
491 |
+
▁ISSUE
|
492 |
+
▁PHONE
|
493 |
+
▁FIND
|
494 |
+
▁RAHUL
|
495 |
+
▁PROBLEM
|
496 |
+
▁STILL
|
497 |
+
▁COURT
|
498 |
+
▁HAPPEN
|
499 |
+
▁KEEP
|
500 |
+
▁ELSE
|
501 |
+
▁ANOTHER
|
502 |
+
▁DEVELOP
|
503 |
+
▁CREAT
|
504 |
+
▁WEDNESDAY
|
505 |
+
▁TELL
|
506 |
+
▁CLOSE
|
507 |
+
▁SOUTH
|
508 |
+
▁TURN
|
509 |
+
▁THOUGH
|
510 |
+
▁TUESDAY
|
511 |
+
▁KOHLI
|
512 |
+
▁SOMETHING
|
513 |
+
▁CHILD
|
514 |
+
▁LOVE
|
515 |
+
▁ACCORDING
|
516 |
+
▁DEVICE
|
517 |
+
▁IMPORTANT
|
518 |
+
▁GROUP
|
519 |
+
▁CONTINUE
|
520 |
+
▁FAMILY
|
521 |
+
▁DRIVE
|
522 |
+
▁COUNTRY
|
523 |
+
▁CAPTAIN
|
524 |
+
▁COUNTRIES
|
525 |
+
▁WATER
|
526 |
+
▁OWN
|
527 |
+
▁PROJECT
|
528 |
+
▁DECEMBER
|
529 |
+
▁WITHOUT
|
530 |
+
▁CENTURY
|
531 |
+
▁FIFTEEN
|
532 |
+
▁OFFICE
|
533 |
+
▁ENGLISH
|
534 |
+
▁PROGRAM
|
535 |
+
▁FRIEND
|
536 |
+
▁RUPEES
|
537 |
+
▁CONSIDER
|
538 |
+
▁BENGALURU
|
539 |
+
▁PERFORM
|
540 |
+
▁ACCOUNT
|
541 |
+
▁PRIME
|
542 |
+
▁MEDIA
|
543 |
+
▁UNDERSTAND
|
544 |
+
▁EXPECT
|
545 |
+
▁RECORD
|
546 |
+
▁EXPERIENCE
|
547 |
+
▁OFFICIAL
|
548 |
+
▁DECISION
|
549 |
+
▁ELEVEN
|
550 |
+
▁TONIGHT
|
551 |
+
▁ATTACK
|
552 |
+
▁BREAK
|
553 |
+
▁AGAIN
|
554 |
+
▁INCLUDING
|
555 |
+
▁DESIGN
|
556 |
+
▁ANNOUNCE
|
557 |
+
▁COMPLETE
|
558 |
+
▁SERVICE
|
559 |
+
▁SEVENTEEN
|
560 |
+
▁ALREADY
|
561 |
+
▁TRAVEL
|
562 |
+
▁PRODUCT
|
563 |
+
▁PROCESS
|
564 |
+
▁PRESIDENT
|
565 |
+
▁KOLKATA
|
566 |
+
▁SYSTEM
|
567 |
+
▁ALWAYS
|
568 |
+
▁SMARTPHONE
|
569 |
+
▁SUCCESS
|
570 |
+
▁MAJOR
|
571 |
+
▁PHOTO
|
572 |
+
▁IMPROVE
|
573 |
+
▁AMERICA
|
574 |
+
▁SQUARE
|
575 |
+
▁LEAGUE
|
576 |
+
▁INTERNATIONAL
|
577 |
+
▁BUILD
|
578 |
+
▁ACTUALLY
|
579 |
+
▁SIGN
|
580 |
+
▁CHIEF
|
581 |
+
▁TWITTER
|
582 |
+
1
|
583 |
+
0
|
584 |
+
3
|
585 |
+
4
|
586 |
+
9
|
587 |
+
2
|
588 |
+
i
|
589 |
+
o
|
590 |
+
<
|
591 |
+
s
|
592 |
+
'
|
593 |
+
/
|
594 |
+
5
|
595 |
+
7
|
596 |
+
8
|
597 |
+
>
|
598 |
+
e
|
599 |
+
6
|
600 |
+
<sos/eos>
|
data/bpe_unigram600/train.txt
ADDED
The diff for this file is too large to render.
See raw diff
|
|
exp/asr_IndEng188h-rnnt-600bpe/18epoch.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:16554e7f89c3b981e0a8fe8e56dec1c0fb5aa8a27ac195e2308396bf2ffc62c9
|
3 |
+
size 347885143
|
exp/asr_IndEng188h-rnnt-600bpe/19epoch.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:73b4cee9401cbe722f836c9417c6a652ac8841d758d3890db2c72896a2cde2ec
|
3 |
+
size 347885143
|
exp/asr_IndEng188h-rnnt-600bpe/20epoch.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8017bfbba23e95fb853756480539e86930307e8e13a020ecb9f1c792b5f18941
|
3 |
+
size 347885143
|
exp/asr_IndEng188h-rnnt-600bpe/checkpoint.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7d93a74b924ad74c8aa240f71ecba3fedbf03c3650f7fa4110b0a22e523b4c8c
|
3 |
+
size 1043304322
|
exp/asr_IndEng188h-rnnt-600bpe/config.yaml
ADDED
@@ -0,0 +1,794 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
config: conf/tuning/train_conformer-rnn_transducer.yaml
|
2 |
+
print_config: false
|
3 |
+
log_level: INFO
|
4 |
+
dry_run: false
|
5 |
+
iterator_type: sequence
|
6 |
+
output_dir: exp/asr_IndEng188h-rnnt-600bpe
|
7 |
+
ngpu: 1
|
8 |
+
seed: 0
|
9 |
+
num_workers: 1
|
10 |
+
num_att_plot: 0
|
11 |
+
dist_backend: nccl
|
12 |
+
dist_init_method: env://
|
13 |
+
dist_world_size: null
|
14 |
+
dist_rank: null
|
15 |
+
local_rank: 0
|
16 |
+
dist_master_addr: null
|
17 |
+
dist_master_port: null
|
18 |
+
dist_launcher: null
|
19 |
+
multiprocessing_distributed: false
|
20 |
+
unused_parameters: false
|
21 |
+
sharded_ddp: false
|
22 |
+
cudnn_enabled: true
|
23 |
+
cudnn_benchmark: false
|
24 |
+
cudnn_deterministic: true
|
25 |
+
collect_stats: false
|
26 |
+
write_collected_feats: false
|
27 |
+
max_epoch: 20
|
28 |
+
patience: null
|
29 |
+
val_scheduler_criterion:
|
30 |
+
- valid
|
31 |
+
- loss
|
32 |
+
early_stopping_criterion:
|
33 |
+
- valid
|
34 |
+
- loss
|
35 |
+
- min
|
36 |
+
best_model_criterion:
|
37 |
+
- - valid
|
38 |
+
- loss
|
39 |
+
- min
|
40 |
+
keep_nbest_models: 3
|
41 |
+
nbest_averaging_interval: 0
|
42 |
+
grad_clip: 5.0
|
43 |
+
grad_clip_type: 2.0
|
44 |
+
grad_noise: false
|
45 |
+
accum_grad: 4
|
46 |
+
no_forward_run: false
|
47 |
+
resume: true
|
48 |
+
train_dtype: float32
|
49 |
+
use_amp: false
|
50 |
+
log_interval: null
|
51 |
+
use_matplotlib: true
|
52 |
+
use_tensorboard: true
|
53 |
+
use_wandb: false
|
54 |
+
wandb_project: null
|
55 |
+
wandb_id: null
|
56 |
+
wandb_entity: null
|
57 |
+
wandb_name: null
|
58 |
+
wandb_model_log_interval: -1
|
59 |
+
detect_anomaly: false
|
60 |
+
pretrain_path: null
|
61 |
+
init_param: []
|
62 |
+
ignore_init_mismatch: false
|
63 |
+
freeze_param: []
|
64 |
+
num_iters_per_epoch: null
|
65 |
+
batch_size: 20
|
66 |
+
valid_batch_size: null
|
67 |
+
batch_bins: 64
|
68 |
+
valid_batch_bins: null
|
69 |
+
train_shape_file:
|
70 |
+
- exp/asr_stats_raw_en_bpe600/train/speech_shape
|
71 |
+
- exp/asr_stats_raw_en_bpe600/train/text_shape.bpe
|
72 |
+
valid_shape_file:
|
73 |
+
- exp/asr_stats_raw_en_bpe600/valid/speech_shape
|
74 |
+
- exp/asr_stats_raw_en_bpe600/valid/text_shape.bpe
|
75 |
+
batch_type: folded
|
76 |
+
valid_batch_type: null
|
77 |
+
fold_length:
|
78 |
+
- 80000
|
79 |
+
- 150
|
80 |
+
sort_in_batch: descending
|
81 |
+
sort_batch: descending
|
82 |
+
multiple_iterator: false
|
83 |
+
chunk_length: 500
|
84 |
+
chunk_shift_ratio: 0.5
|
85 |
+
num_cache_chunks: 1024
|
86 |
+
train_data_path_and_name_and_type:
|
87 |
+
- - dump/raw/train_English/wav.scp
|
88 |
+
- speech
|
89 |
+
- sound
|
90 |
+
- - dump/raw/train_English/text
|
91 |
+
- text
|
92 |
+
- text
|
93 |
+
valid_data_path_and_name_and_type:
|
94 |
+
- - dump/raw/dev_English/wav.scp
|
95 |
+
- speech
|
96 |
+
- sound
|
97 |
+
- - dump/raw/dev_English/text
|
98 |
+
- text
|
99 |
+
- text
|
100 |
+
allow_variable_data_keys: false
|
101 |
+
max_cache_size: 0.0
|
102 |
+
max_cache_fd: 32
|
103 |
+
valid_max_cache_size: null
|
104 |
+
optim: adam
|
105 |
+
optim_conf:
|
106 |
+
lr: 0.0008
|
107 |
+
weight_decay: 1.0e-06
|
108 |
+
scheduler: warmuplr
|
109 |
+
scheduler_conf:
|
110 |
+
warmup_steps: 25000
|
111 |
+
token_list:
|
112 |
+
- <blank>
|
113 |
+
- <unk>
|
114 |
+
- S
|
115 |
+
- ▁THE
|
116 |
+
- ▁A
|
117 |
+
- E
|
118 |
+
- T
|
119 |
+
- ▁TO
|
120 |
+
- I
|
121 |
+
- D
|
122 |
+
- ▁P
|
123 |
+
- ING
|
124 |
+
- ▁IN
|
125 |
+
- ▁AT
|
126 |
+
- ▁AND
|
127 |
+
- ED
|
128 |
+
- P
|
129 |
+
- ▁FOR
|
130 |
+
- A
|
131 |
+
- ▁M
|
132 |
+
- ▁OF
|
133 |
+
- Y
|
134 |
+
- O
|
135 |
+
- ER
|
136 |
+
- M
|
137 |
+
- ▁I
|
138 |
+
- U
|
139 |
+
- N
|
140 |
+
- C
|
141 |
+
- AR
|
142 |
+
- AL
|
143 |
+
- ▁C
|
144 |
+
- G
|
145 |
+
- ▁S
|
146 |
+
- ▁ON
|
147 |
+
- R
|
148 |
+
- B
|
149 |
+
- ▁F
|
150 |
+
- ▁IS
|
151 |
+
- H
|
152 |
+
- L
|
153 |
+
- ▁T
|
154 |
+
- F
|
155 |
+
- OR
|
156 |
+
- ▁TABLE
|
157 |
+
- IN
|
158 |
+
- AN
|
159 |
+
- ▁BOOK
|
160 |
+
- ▁RE
|
161 |
+
- K
|
162 |
+
- ▁THAT
|
163 |
+
- RE
|
164 |
+
- ▁BE
|
165 |
+
- ▁TWO
|
166 |
+
- IC
|
167 |
+
- IT
|
168 |
+
- LE
|
169 |
+
- EN
|
170 |
+
- V
|
171 |
+
- RA
|
172 |
+
- ▁ONE
|
173 |
+
- ▁G
|
174 |
+
- ▁D
|
175 |
+
- UR
|
176 |
+
- ▁
|
177 |
+
- ▁IT
|
178 |
+
- LY
|
179 |
+
- ES
|
180 |
+
- 'ON'
|
181 |
+
- RI
|
182 |
+
- ▁HE
|
183 |
+
- W
|
184 |
+
- ▁YOU
|
185 |
+
- ▁WITH
|
186 |
+
- ▁B
|
187 |
+
- ▁ST
|
188 |
+
- TH
|
189 |
+
- ▁E
|
190 |
+
- ▁WE
|
191 |
+
- AT
|
192 |
+
- ▁ARE
|
193 |
+
- ▁WAS
|
194 |
+
- ▁NOT
|
195 |
+
- ▁SO
|
196 |
+
- ▁AS
|
197 |
+
- IL
|
198 |
+
- ▁WANT
|
199 |
+
- ▁W
|
200 |
+
- ENT
|
201 |
+
- ▁K
|
202 |
+
- ▁HAVE
|
203 |
+
- ▁H
|
204 |
+
- CH
|
205 |
+
- ▁THREE
|
206 |
+
- ▁DE
|
207 |
+
- RO
|
208 |
+
- AS
|
209 |
+
- ▁WILL
|
210 |
+
- ▁L
|
211 |
+
- TER
|
212 |
+
- LA
|
213 |
+
- ▁FOUR
|
214 |
+
- ATION
|
215 |
+
- ▁HAS
|
216 |
+
- LL
|
217 |
+
- ST
|
218 |
+
- ▁THIRTY
|
219 |
+
- ▁CAN
|
220 |
+
- ▁DO
|
221 |
+
- AD
|
222 |
+
- ERS
|
223 |
+
- EL
|
224 |
+
- ▁O
|
225 |
+
- AM
|
226 |
+
- ▁THIS
|
227 |
+
- UL
|
228 |
+
- IR
|
229 |
+
- ▁FROM
|
230 |
+
- TION
|
231 |
+
- ▁HIS
|
232 |
+
- ▁AN
|
233 |
+
- ION
|
234 |
+
- ▁FIVE
|
235 |
+
- ▁SE
|
236 |
+
- ▁V
|
237 |
+
- US
|
238 |
+
- ▁ME
|
239 |
+
- ▁BY
|
240 |
+
- ▁LIKE
|
241 |
+
- IS
|
242 |
+
- ▁SAID
|
243 |
+
- ▁R
|
244 |
+
- ▁EIGHT
|
245 |
+
- ▁N
|
246 |
+
- ▁CON
|
247 |
+
- ▁PEOPLE
|
248 |
+
- LI
|
249 |
+
- ▁J
|
250 |
+
- AND
|
251 |
+
- ▁BUT
|
252 |
+
- J
|
253 |
+
- ▁WHAT
|
254 |
+
- ▁TODAY
|
255 |
+
- UN
|
256 |
+
- ATE
|
257 |
+
- ▁SIX
|
258 |
+
- MENT
|
259 |
+
- ▁RA
|
260 |
+
- ▁OR
|
261 |
+
- UT
|
262 |
+
- ▁SU
|
263 |
+
- CE
|
264 |
+
- ▁WHICH
|
265 |
+
- ▁CO
|
266 |
+
- ▁EX
|
267 |
+
- NE
|
268 |
+
- OL
|
269 |
+
- Z
|
270 |
+
- ▁SEVEN
|
271 |
+
- SE
|
272 |
+
- NG
|
273 |
+
- ▁MO
|
274 |
+
- ▁NEED
|
275 |
+
- VER
|
276 |
+
- ▁ALL
|
277 |
+
- ITY
|
278 |
+
- VE
|
279 |
+
- ▁LE
|
280 |
+
- ▁GO
|
281 |
+
- ▁WHO
|
282 |
+
- GE
|
283 |
+
- IVE
|
284 |
+
- ▁MALL
|
285 |
+
- HI
|
286 |
+
- ▁AM
|
287 |
+
- AC
|
288 |
+
- IDE
|
289 |
+
- ▁WOULD
|
290 |
+
- LO
|
291 |
+
- ▁NINE
|
292 |
+
- ▁BA
|
293 |
+
- ▁MA
|
294 |
+
- UM
|
295 |
+
- ▁NEW
|
296 |
+
- ▁WHEN
|
297 |
+
- ▁TIME
|
298 |
+
- AY
|
299 |
+
- ▁DINNER
|
300 |
+
- ▁RO
|
301 |
+
- IES
|
302 |
+
- CK
|
303 |
+
- ▁ALSO
|
304 |
+
- ▁PRO
|
305 |
+
- ▁EN
|
306 |
+
- OW
|
307 |
+
- ▁VI
|
308 |
+
- OT
|
309 |
+
- ▁THERE
|
310 |
+
- ▁MY
|
311 |
+
- ▁YEAR
|
312 |
+
- ▁LA
|
313 |
+
- ▁TEN
|
314 |
+
- ▁ITS
|
315 |
+
- IM
|
316 |
+
- ▁INDIA
|
317 |
+
- ▁US
|
318 |
+
- ▁ABOUT
|
319 |
+
- ▁UP
|
320 |
+
- MA
|
321 |
+
- ▁THOUSAND
|
322 |
+
- ▁OUT
|
323 |
+
- AP
|
324 |
+
- ▁BO
|
325 |
+
- AGE
|
326 |
+
- ▁PLAY
|
327 |
+
- ▁THEY
|
328 |
+
- X
|
329 |
+
- ▁GET
|
330 |
+
- EST
|
331 |
+
- ▁THEIR
|
332 |
+
- ▁HAD
|
333 |
+
- IST
|
334 |
+
- ▁NO
|
335 |
+
- ▁SP
|
336 |
+
- ▁AFTER
|
337 |
+
- ▁YOUR
|
338 |
+
- TING
|
339 |
+
- CI
|
340 |
+
- ▁NOVEMBER
|
341 |
+
- Q
|
342 |
+
- ▁DIS
|
343 |
+
- ▁BEEN
|
344 |
+
- ▁APP
|
345 |
+
- ▁HO
|
346 |
+
- KE
|
347 |
+
- ▁UN
|
348 |
+
- OS
|
349 |
+
- ARD
|
350 |
+
- UD
|
351 |
+
- ▁OVER
|
352 |
+
- ▁PLEASE
|
353 |
+
- ACH
|
354 |
+
- ▁OKAY
|
355 |
+
- ▁PRE
|
356 |
+
- ▁MORE
|
357 |
+
- ▁SOME
|
358 |
+
- ▁SH
|
359 |
+
- NA
|
360 |
+
- ▁BR
|
361 |
+
- ANT
|
362 |
+
- ▁FIRST
|
363 |
+
- ▁PART
|
364 |
+
- TED
|
365 |
+
- OM
|
366 |
+
- ALLY
|
367 |
+
- ▁CHA
|
368 |
+
- ▁ADULTS
|
369 |
+
- RY
|
370 |
+
- URE
|
371 |
+
- IF
|
372 |
+
- ▁TWENTY
|
373 |
+
- ▁DAY
|
374 |
+
- AV
|
375 |
+
- ▁WORK
|
376 |
+
- ▁TOMORROW
|
377 |
+
- HA
|
378 |
+
- TA
|
379 |
+
- ONE
|
380 |
+
- ENCE
|
381 |
+
- ▁POINT
|
382 |
+
- ▁WERE
|
383 |
+
- ▁HA
|
384 |
+
- ▁IF
|
385 |
+
- OUS
|
386 |
+
- PE
|
387 |
+
- ▁NA
|
388 |
+
- ▁COM
|
389 |
+
- ▁HER
|
390 |
+
- ▁OTHER
|
391 |
+
- TS
|
392 |
+
- ▁NOW
|
393 |
+
- HO
|
394 |
+
- ▁ACT
|
395 |
+
- ▁HUNDRED
|
396 |
+
- ANCE
|
397 |
+
- ▁MEMBERS
|
398 |
+
- ▁HOW
|
399 |
+
- ▁SHE
|
400 |
+
- ▁QU
|
401 |
+
- ▁U
|
402 |
+
- ▁SAY
|
403 |
+
- ▁SC
|
404 |
+
- AK
|
405 |
+
- ▁KNOW
|
406 |
+
- ▁SA
|
407 |
+
- ▁AR
|
408 |
+
- ▁ANY
|
409 |
+
- ▁THAN
|
410 |
+
- ▁JO
|
411 |
+
- ▁JUST
|
412 |
+
- ▁ROAD
|
413 |
+
- MAN
|
414 |
+
- ▁TEST
|
415 |
+
- ABLE
|
416 |
+
- HE
|
417 |
+
- ▁MAN
|
418 |
+
- ▁LUNCH
|
419 |
+
- ▁TAKE
|
420 |
+
- ▁INTO
|
421 |
+
- ▁LI
|
422 |
+
- ▁WANTED
|
423 |
+
- AB
|
424 |
+
- UND
|
425 |
+
- GH
|
426 |
+
- MO
|
427 |
+
- ▁RUN
|
428 |
+
- ▁ONLY
|
429 |
+
- ▁SHA
|
430 |
+
- PER
|
431 |
+
- ITE
|
432 |
+
- ▁TEAM
|
433 |
+
- ISH
|
434 |
+
- ▁MOST
|
435 |
+
- ▁EVEN
|
436 |
+
- ▁TWELVE
|
437 |
+
- ▁RESERVE
|
438 |
+
- ▁DID
|
439 |
+
- ▁OUR
|
440 |
+
- ▁HIM
|
441 |
+
- ▁COMP
|
442 |
+
- ARY
|
443 |
+
- ▁LOOK
|
444 |
+
- ▁SEE
|
445 |
+
- ▁OFF
|
446 |
+
- ATED
|
447 |
+
- ▁CALL
|
448 |
+
- ▁SECOND
|
449 |
+
- ▁WHILE
|
450 |
+
- WAR
|
451 |
+
- ▁MAKE
|
452 |
+
- ▁FILM
|
453 |
+
- ▁WIN
|
454 |
+
- ▁WORLD
|
455 |
+
- ▁RIGHT
|
456 |
+
- TEN
|
457 |
+
- ▁HMM
|
458 |
+
- ▁PLACE
|
459 |
+
- ▁STATE
|
460 |
+
- ▁INDIAN
|
461 |
+
- ▁VERY
|
462 |
+
- ▁REPORT
|
463 |
+
- ▁COULD
|
464 |
+
- ▁LAST
|
465 |
+
- ▁THINK
|
466 |
+
- ▁START
|
467 |
+
- ▁ADD
|
468 |
+
- ▁NATION
|
469 |
+
- ▁WELL
|
470 |
+
- ▁HIGH
|
471 |
+
- ▁WHERE
|
472 |
+
- ▁MADE
|
473 |
+
- ▁MOVIE
|
474 |
+
- ▁BACK
|
475 |
+
- ▁COME
|
476 |
+
- ASH
|
477 |
+
- ▁EVERY
|
478 |
+
- ▁KA
|
479 |
+
- ▁BEST
|
480 |
+
- ▁BECAUSE
|
481 |
+
- ▁ENGLAND
|
482 |
+
- ICK
|
483 |
+
- ▁SHOW
|
484 |
+
- ▁WAY
|
485 |
+
- COM
|
486 |
+
- ▁GIVE
|
487 |
+
- ▁KIND
|
488 |
+
- ▁CHILDREN
|
489 |
+
- ▁BEFORE
|
490 |
+
- ▁PAKISTAN
|
491 |
+
- ▁LEAD
|
492 |
+
- ▁WATCH
|
493 |
+
- ▁THROUGH
|
494 |
+
- ▁MUCH
|
495 |
+
- ▁MANY
|
496 |
+
- ▁GOOD
|
497 |
+
- ▁SUNDAY
|
498 |
+
- ▁FIFTY
|
499 |
+
- PORT
|
500 |
+
- ▁NAGAR
|
501 |
+
- ▁THIRD
|
502 |
+
- ▁GREAT
|
503 |
+
- ▁UNDER
|
504 |
+
- ▁ADULT
|
505 |
+
- ▁HELP
|
506 |
+
- ▁MONDAY
|
507 |
+
- LIGHT
|
508 |
+
- ▁USERS
|
509 |
+
- ▁SHOULD
|
510 |
+
- ▁BAN
|
511 |
+
- ▁BALL
|
512 |
+
- ▁PLAN
|
513 |
+
- ▁VIDEO
|
514 |
+
- ▁MONTH
|
515 |
+
- ▁Z
|
516 |
+
- LINE
|
517 |
+
- ▁NEXT
|
518 |
+
- ▁DIRECT
|
519 |
+
- ▁BIG
|
520 |
+
- ▁YES
|
521 |
+
- ▁CAME
|
522 |
+
- ▁MUMBAI
|
523 |
+
- ▁SEATS
|
524 |
+
- ▁MATCH
|
525 |
+
- ▁LOT
|
526 |
+
- ▁GAME
|
527 |
+
- ▁PERSON
|
528 |
+
- ▁AGAINST
|
529 |
+
- ▁WHY
|
530 |
+
- ▁IMP
|
531 |
+
- ▁OLD
|
532 |
+
- ▁RESERVATION
|
533 |
+
- ▁COMPANY
|
534 |
+
- ▁DURING
|
535 |
+
- ▁FEATURE
|
536 |
+
- ▁COMMUN
|
537 |
+
- ▁BETWEEN
|
538 |
+
- ▁WICKET
|
539 |
+
- ▁SCORE
|
540 |
+
- ▁LIFE
|
541 |
+
- ▁DOWN
|
542 |
+
- ▁SATURDAY
|
543 |
+
- OOD
|
544 |
+
- ▁TALK
|
545 |
+
- ▁HOME
|
546 |
+
- ▁OPEN
|
547 |
+
- ▁YEAH
|
548 |
+
- ▁CHANGE
|
549 |
+
- ▁NINETEEN
|
550 |
+
- ▁SIXTY
|
551 |
+
- ▁ELECT
|
552 |
+
- ▁NINETY
|
553 |
+
- ▁LAUNCH
|
554 |
+
- ▁TRANS
|
555 |
+
- ▁FINAL
|
556 |
+
- ▁CRICKET
|
557 |
+
- ▁GOVERNMENT
|
558 |
+
- ▁FOUND
|
559 |
+
- ▁NUMBER
|
560 |
+
- ▁HOWEVER
|
561 |
+
- ▁ZERO
|
562 |
+
- ▁SINCE
|
563 |
+
- ▁DIFFERENT
|
564 |
+
- ▁SERIES
|
565 |
+
- ▁MUSIC
|
566 |
+
- ▁DELHI
|
567 |
+
- ▁SUPPORT
|
568 |
+
- ▁SEVENTY
|
569 |
+
- ▁AUSTRALIA
|
570 |
+
- ▁SIXTEEN
|
571 |
+
- ▁CHECK
|
572 |
+
- ▁MAKING
|
573 |
+
- ▁MINUTE
|
574 |
+
- ▁THINGS
|
575 |
+
- ▁CHENNAI
|
576 |
+
- ▁AROUND
|
577 |
+
- ▁FRIDAY
|
578 |
+
- ▁LONG
|
579 |
+
- ▁THURSDAY
|
580 |
+
- ▁GOOGLE
|
581 |
+
- ▁MINISTER
|
582 |
+
- ▁BOWL
|
583 |
+
- ▁SAME
|
584 |
+
- ▁KHAN
|
585 |
+
- ▁SPACE
|
586 |
+
- ▁FACEBOOK
|
587 |
+
- ▁BARB
|
588 |
+
- ▁POST
|
589 |
+
- ▁FOLLOW
|
590 |
+
- ▁WEEK
|
591 |
+
- ▁RESEARCH
|
592 |
+
- AUGHT
|
593 |
+
- ▁SPEAK
|
594 |
+
- ▁POLICE
|
595 |
+
- ▁INNINGS
|
596 |
+
- ▁BECOME
|
597 |
+
- ▁KALYAN
|
598 |
+
- ▁FEEL
|
599 |
+
- ▁RELEASE
|
600 |
+
- ▁THOSE
|
601 |
+
- ▁MILLION
|
602 |
+
- ▁ISSUE
|
603 |
+
- ▁PHONE
|
604 |
+
- ▁FIND
|
605 |
+
- ▁RAHUL
|
606 |
+
- ▁PROBLEM
|
607 |
+
- ▁STILL
|
608 |
+
- ▁COURT
|
609 |
+
- ▁HAPPEN
|
610 |
+
- ▁KEEP
|
611 |
+
- ▁ELSE
|
612 |
+
- ▁ANOTHER
|
613 |
+
- ▁DEVELOP
|
614 |
+
- ▁CREAT
|
615 |
+
- ▁WEDNESDAY
|
616 |
+
- ▁TELL
|
617 |
+
- ▁CLOSE
|
618 |
+
- ▁SOUTH
|
619 |
+
- ▁TURN
|
620 |
+
- ▁THOUGH
|
621 |
+
- ▁TUESDAY
|
622 |
+
- ▁KOHLI
|
623 |
+
- ▁SOMETHING
|
624 |
+
- ▁CHILD
|
625 |
+
- ▁LOVE
|
626 |
+
- ▁ACCORDING
|
627 |
+
- ▁DEVICE
|
628 |
+
- ▁IMPORTANT
|
629 |
+
- ▁GROUP
|
630 |
+
- ▁CONTINUE
|
631 |
+
- ▁FAMILY
|
632 |
+
- ▁DRIVE
|
633 |
+
- ▁COUNTRY
|
634 |
+
- ▁CAPTAIN
|
635 |
+
- ▁COUNTRIES
|
636 |
+
- ▁WATER
|
637 |
+
- ▁OWN
|
638 |
+
- ▁PROJECT
|
639 |
+
- ▁DECEMBER
|
640 |
+
- ▁WITHOUT
|
641 |
+
- ▁CENTURY
|
642 |
+
- ▁FIFTEEN
|
643 |
+
- ▁OFFICE
|
644 |
+
- ▁ENGLISH
|
645 |
+
- ▁PROGRAM
|
646 |
+
- ▁FRIEND
|
647 |
+
- ▁RUPEES
|
648 |
+
- ▁CONSIDER
|
649 |
+
- ▁BENGALURU
|
650 |
+
- ▁PERFORM
|
651 |
+
- ▁ACCOUNT
|
652 |
+
- ▁PRIME
|
653 |
+
- ▁MEDIA
|
654 |
+
- ▁UNDERSTAND
|
655 |
+
- ▁EXPECT
|
656 |
+
- ▁RECORD
|
657 |
+
- ▁EXPERIENCE
|
658 |
+
- ▁OFFICIAL
|
659 |
+
- ▁DECISION
|
660 |
+
- ▁ELEVEN
|
661 |
+
- ▁TONIGHT
|
662 |
+
- ▁ATTACK
|
663 |
+
- ▁BREAK
|
664 |
+
- ▁AGAIN
|
665 |
+
- ▁INCLUDING
|
666 |
+
- ▁DESIGN
|
667 |
+
- ▁ANNOUNCE
|
668 |
+
- ▁COMPLETE
|
669 |
+
- ▁SERVICE
|
670 |
+
- ▁SEVENTEEN
|
671 |
+
- ▁ALREADY
|
672 |
+
- ▁TRAVEL
|
673 |
+
- ▁PRODUCT
|
674 |
+
- ▁PROCESS
|
675 |
+
- ▁PRESIDENT
|
676 |
+
- ▁KOLKATA
|
677 |
+
- ▁SYSTEM
|
678 |
+
- ▁ALWAYS
|
679 |
+
- ▁SMARTPHONE
|
680 |
+
- ▁SUCCESS
|
681 |
+
- ▁MAJOR
|
682 |
+
- ▁PHOTO
|
683 |
+
- ▁IMPROVE
|
684 |
+
- ▁AMERICA
|
685 |
+
- ▁SQUARE
|
686 |
+
- ▁LEAGUE
|
687 |
+
- ▁INTERNATIONAL
|
688 |
+
- ▁BUILD
|
689 |
+
- ▁ACTUALLY
|
690 |
+
- ▁SIGN
|
691 |
+
- ▁CHIEF
|
692 |
+
- ▁TWITTER
|
693 |
+
- '1'
|
694 |
+
- '0'
|
695 |
+
- '3'
|
696 |
+
- '4'
|
697 |
+
- '9'
|
698 |
+
- '2'
|
699 |
+
- i
|
700 |
+
- o
|
701 |
+
- <
|
702 |
+
- s
|
703 |
+
- ''''
|
704 |
+
- /
|
705 |
+
- '5'
|
706 |
+
- '7'
|
707 |
+
- '8'
|
708 |
+
- '>'
|
709 |
+
- e
|
710 |
+
- '6'
|
711 |
+
- <sos/eos>
|
712 |
+
init: null
|
713 |
+
input_size: null
|
714 |
+
ctc_conf:
|
715 |
+
dropout_rate: 0.0
|
716 |
+
ctc_type: builtin
|
717 |
+
reduce: true
|
718 |
+
ignore_nan_grad: true
|
719 |
+
joint_net_conf:
|
720 |
+
joint_space_size: 640
|
721 |
+
use_preprocessor: true
|
722 |
+
token_type: bpe
|
723 |
+
bpemodel: data/en_token_list/bpe_unigram600/bpe.model
|
724 |
+
non_linguistic_symbols: null
|
725 |
+
cleaner: null
|
726 |
+
g2p: null
|
727 |
+
speech_volume_normalize: null
|
728 |
+
rir_scp: null
|
729 |
+
rir_apply_prob: 1.0
|
730 |
+
noise_scp: null
|
731 |
+
noise_apply_prob: 1.0
|
732 |
+
noise_db_range: '13_15'
|
733 |
+
frontend: default
|
734 |
+
frontend_conf:
|
735 |
+
n_fft: 512
|
736 |
+
hop_length: 160
|
737 |
+
fs: 16k
|
738 |
+
specaug: specaug
|
739 |
+
specaug_conf:
|
740 |
+
apply_time_warp: true
|
741 |
+
time_warp_window: 5
|
742 |
+
time_warp_mode: bicubic
|
743 |
+
apply_freq_mask: true
|
744 |
+
freq_mask_width_range:
|
745 |
+
- 0
|
746 |
+
- 30
|
747 |
+
num_freq_mask: 2
|
748 |
+
apply_time_mask: true
|
749 |
+
time_mask_width_range:
|
750 |
+
- 0
|
751 |
+
- 40
|
752 |
+
num_time_mask: 2
|
753 |
+
normalize: global_mvn
|
754 |
+
normalize_conf:
|
755 |
+
stats_file: exp/asr_stats_raw_en_bpe600/train/feats_stats.npz
|
756 |
+
model: espnet
|
757 |
+
model_conf:
|
758 |
+
ctc_weight: 0.3
|
759 |
+
report_cer: false
|
760 |
+
report_wer: false
|
761 |
+
preencoder: null
|
762 |
+
preencoder_conf: {}
|
763 |
+
encoder: conformer
|
764 |
+
encoder_conf:
|
765 |
+
output_size: 512
|
766 |
+
attention_heads: 8
|
767 |
+
linear_units: 2048
|
768 |
+
num_blocks: 12
|
769 |
+
dropout_rate: 0.1
|
770 |
+
positional_dropout_rate: 0.1
|
771 |
+
attention_dropout_rate: 0.1
|
772 |
+
input_layer: conv2d
|
773 |
+
normalize_before: true
|
774 |
+
macaron_style: true
|
775 |
+
rel_pos_type: latest
|
776 |
+
pos_enc_layer_type: rel_pos
|
777 |
+
selfattention_layer_type: rel_selfattn
|
778 |
+
activation_type: swish
|
779 |
+
use_cnn_module: true
|
780 |
+
cnn_module_kernel: 15
|
781 |
+
postencoder: null
|
782 |
+
postencoder_conf: {}
|
783 |
+
decoder: transducer
|
784 |
+
decoder_conf:
|
785 |
+
rnn_type: lstm
|
786 |
+
num_layers: 1
|
787 |
+
hidden_size: 512
|
788 |
+
dropout: 0.1
|
789 |
+
dropout_embed: 0.2
|
790 |
+
required:
|
791 |
+
- output_dir
|
792 |
+
- token_list
|
793 |
+
version: '202205'
|
794 |
+
distributed: false
|
exp/asr_IndEng188h-rnnt-600bpe/images/backward_time.png
ADDED
exp/asr_IndEng188h-rnnt-600bpe/images/cer_ctc.png
ADDED
exp/asr_IndEng188h-rnnt-600bpe/images/cer_transducer.png
ADDED
exp/asr_IndEng188h-rnnt-600bpe/images/forward_time.png
ADDED
exp/asr_IndEng188h-rnnt-600bpe/images/gpu_max_cached_mem_GB.png
ADDED
exp/asr_IndEng188h-rnnt-600bpe/images/iter_time.png
ADDED
exp/asr_IndEng188h-rnnt-600bpe/images/loss.png
ADDED
exp/asr_IndEng188h-rnnt-600bpe/images/loss_ctc.png
ADDED
exp/asr_IndEng188h-rnnt-600bpe/images/loss_transducer.png
ADDED
exp/asr_IndEng188h-rnnt-600bpe/images/optim0_lr0.png
ADDED
exp/asr_IndEng188h-rnnt-600bpe/images/optim_step_time.png
ADDED
exp/asr_IndEng188h-rnnt-600bpe/images/train_time.png
ADDED
exp/asr_IndEng188h-rnnt-600bpe/images/wer_transducer.png
ADDED
exp/asr_IndEng188h-rnnt-600bpe/latest.pth
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
20epoch.pth
|
exp/asr_IndEng188h-rnnt-600bpe/run.sh
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
./asr.sh --stage 9 --stop_stage 10 --lang en --ngpu 1 --max_wav_duration 30 --asr_config conf/tuning/train_conformer-rnn_transducer.yaml --lm_config conf/tuning/train_lm_transformer2.yaml --inference_config conf/tuning/decode-conformer-rnnt-transducer.yaml --train_set train_English --valid_set dev_English --test_sets eval_English --stage 10 "$@"; exit $?
|
exp/asr_IndEng188h-rnnt-600bpe/train.1.log
ADDED
The diff for this file is too large to render.
See raw diff
|
|
exp/asr_IndEng188h-rnnt-600bpe/train.log
ADDED
The diff for this file is too large to render.
See raw diff
|
|
exp/asr_IndEng188h-rnnt-600bpe/valid.loss.ave.pth
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
valid.loss.ave_3best.pth
|
exp/asr_IndEng188h-rnnt-600bpe/valid.loss.ave_3best.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:85fd516bbe7465276b39d7b069b1c4c5872398f41f291f63b2cea382c974cd80
|
3 |
+
size 347878673
|
exp/asr_IndEng188h-rnnt-600bpe/valid.loss.best.pth
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
20epoch.pth
|
exp/asr_stats_raw_en_bpe600/train/feats_lengths_stats.npz
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:2867dc18918c7a259303cad48e8bce8d95163d8e6e783034812cf6d631918e83
|
3 |
+
size 778
|
exp/asr_stats_raw_en_bpe600/train/feats_stats.npz
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:071db33036bc4c90734a2e921e7ce0532b7743b394df96c3154c9f74aa4dd0c8
|
3 |
+
size 1402
|
exp/asr_stats_raw_en_bpe600/train/speech_shape
ADDED
The diff for this file is too large to render.
See raw diff
|
|
exp/asr_stats_raw_en_bpe600/train/text_shape
ADDED
The diff for this file is too large to render.
See raw diff
|
|
exp/asr_stats_raw_en_bpe600/train/text_shape.bpe
ADDED
The diff for this file is too large to render.
See raw diff
|
|