Hugging Face's logo

marcoyang
/
icefall-zipformer-M-librispeech-bf16

TensorBoard
Model card Files
xet
 Metrics Community
yangxiaoyu6 commited on
Commit
0d55c0f
·
1 Parent(s): dfb8a62

add files

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. data/lang_bpe_500/bpe.model +3 -0
  2. data/lang_bpe_500/tokens.txt +502 -0
  3. decoding_results/greedy_search/errs-test-clean-greedy_search-epoch-30-avg-11-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  4. decoding_results/greedy_search/errs-test-clean-greedy_search-epoch-30-avg-13-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  5. decoding_results/greedy_search/errs-test-clean-greedy_search-epoch-30-avg-15-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  6. decoding_results/greedy_search/errs-test-clean-greedy_search-epoch-30-avg-7-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  7. decoding_results/greedy_search/errs-test-clean-greedy_search-epoch-30-avg-9-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  8. decoding_results/greedy_search/errs-test-clean-greedy_search-epoch-40-avg-11-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  9. decoding_results/greedy_search/errs-test-clean-greedy_search-epoch-40-avg-13-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  10. decoding_results/greedy_search/errs-test-clean-greedy_search-epoch-40-avg-15-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  11. decoding_results/greedy_search/errs-test-clean-greedy_search-epoch-40-avg-17-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  12. decoding_results/greedy_search/errs-test-clean-greedy_search-epoch-40-avg-19-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  13. decoding_results/greedy_search/errs-test-clean-greedy_search-epoch-40-avg-7-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  14. decoding_results/greedy_search/errs-test-clean-greedy_search-epoch-40-avg-9-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  15. decoding_results/greedy_search/errs-test-other-greedy_search-epoch-30-avg-11-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  16. decoding_results/greedy_search/errs-test-other-greedy_search-epoch-30-avg-13-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  17. decoding_results/greedy_search/errs-test-other-greedy_search-epoch-30-avg-15-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  18. decoding_results/greedy_search/errs-test-other-greedy_search-epoch-30-avg-7-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  19. decoding_results/greedy_search/errs-test-other-greedy_search-epoch-30-avg-9-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  20. decoding_results/greedy_search/errs-test-other-greedy_search-epoch-40-avg-11-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  21. decoding_results/greedy_search/errs-test-other-greedy_search-epoch-40-avg-13-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  22. decoding_results/greedy_search/errs-test-other-greedy_search-epoch-40-avg-15-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  23. decoding_results/greedy_search/errs-test-other-greedy_search-epoch-40-avg-17-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  24. decoding_results/greedy_search/errs-test-other-greedy_search-epoch-40-avg-19-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  25. decoding_results/greedy_search/errs-test-other-greedy_search-epoch-40-avg-7-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  26. decoding_results/greedy_search/errs-test-other-greedy_search-epoch-40-avg-9-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  27. decoding_results/greedy_search/log-decode-epoch-30-avg-11-context-2-max-sym-per-frame-1-use-averaged-model-2024-08-08-10-10-56 +25 -0
  28. decoding_results/greedy_search/log-decode-epoch-30-avg-13-context-2-max-sym-per-frame-1-use-averaged-model-2024-08-08-10-09-48 +27 -0
  29. decoding_results/greedy_search/log-decode-epoch-30-avg-15-context-2-max-sym-per-frame-1-use-averaged-model-2024-08-07-10-39-22 +28 -0
  30. decoding_results/greedy_search/log-decode-epoch-30-avg-15-context-2-max-sym-per-frame-1-use-averaged-model-2024-08-08-10-08-02 +26 -0
  31. decoding_results/greedy_search/log-decode-epoch-30-avg-7-context-2-max-sym-per-frame-1-use-averaged-model-2024-08-08-10-13-10 +28 -0
  32. decoding_results/greedy_search/log-decode-epoch-30-avg-9-context-2-max-sym-per-frame-1-use-averaged-model-2024-08-08-10-12-02 +25 -0
  33. decoding_results/greedy_search/log-decode-epoch-40-avg-11-context-2-max-sym-per-frame-1-use-averaged-model-2024-08-13-10-45-34 +26 -0
  34. decoding_results/greedy_search/log-decode-epoch-40-avg-13-context-2-max-sym-per-frame-1-use-averaged-model-2024-08-13-10-44-27 +26 -0
  35. decoding_results/greedy_search/log-decode-epoch-40-avg-15-context-2-max-sym-per-frame-1-use-averaged-model-2024-08-13-10-43-19 +26 -0
  36. decoding_results/greedy_search/log-decode-epoch-40-avg-17-context-2-max-sym-per-frame-1-use-averaged-model-2024-08-13-10-42-13 +26 -0
  37. decoding_results/greedy_search/log-decode-epoch-40-avg-19-context-2-max-sym-per-frame-1-use-averaged-model-2024-08-13-10-40-58 +25 -0
  38. decoding_results/greedy_search/log-decode-epoch-40-avg-7-context-2-max-sym-per-frame-1-use-averaged-model-2024-08-13-10-47-51 +27 -0
  39. decoding_results/greedy_search/log-decode-epoch-40-avg-9-context-2-max-sym-per-frame-1-use-averaged-model-2024-08-13-10-46-41 +27 -0
  40. decoding_results/greedy_search/recogs-test-clean-greedy_search-epoch-30-avg-11-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  41. decoding_results/greedy_search/recogs-test-clean-greedy_search-epoch-30-avg-13-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  42. decoding_results/greedy_search/recogs-test-clean-greedy_search-epoch-30-avg-15-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  43. decoding_results/greedy_search/recogs-test-clean-greedy_search-epoch-30-avg-7-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  44. decoding_results/greedy_search/recogs-test-clean-greedy_search-epoch-30-avg-9-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  45. decoding_results/greedy_search/recogs-test-clean-greedy_search-epoch-40-avg-11-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  46. decoding_results/greedy_search/recogs-test-clean-greedy_search-epoch-40-avg-13-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  47. decoding_results/greedy_search/recogs-test-clean-greedy_search-epoch-40-avg-15-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  48. decoding_results/greedy_search/recogs-test-clean-greedy_search-epoch-40-avg-17-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  49. decoding_results/greedy_search/recogs-test-clean-greedy_search-epoch-40-avg-19-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
  50. decoding_results/greedy_search/recogs-test-clean-greedy_search-epoch-40-avg-7-context-2-max-sym-per-frame-1-use-averaged-model.txt +0 -0
data/lang_bpe_500/bpe.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c53433de083c4a6ad12d034550ef22de68cec62c4f58932a7b6b8b2f1e743fa5
3
+ size 244865
data/lang_bpe_500/tokens.txt ADDED
@@ -0,0 +1,502 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ <blk> 0
2
+ <sos/eos> 1
3
+ <unk> 2
4
+ S 3
5
+ ▁THE 4
6
+ ▁A 5
7
+ T 6
8
+ ▁AND 7
9
+ ED 8
10
+ ▁OF 9
11
+ ▁TO 10
12
+ E 11
13
+ D 12
14
+ N 13
15
+ ING 14
16
+ ▁IN 15
17
+ Y 16
18
+ M 17
19
+ C 18
20
+ ▁I 19
21
+ A 20
22
+ P 21
23
+ ▁HE 22
24
+ R 23
25
+ O 24
26
+ L 25
27
+ RE 26
28
+ I 27
29
+ U 28
30
+ ER 29
31
+ ▁IT 30
32
+ LY 31
33
+ ▁THAT 32
34
+ ▁WAS 33
35
+ ▁ 34
36
+ ▁S 35
37
+ AR 36
38
+ ▁BE 37
39
+ F 38
40
+ ▁C 39
41
+ IN 40
42
+ B 41
43
+ ▁FOR 42
44
+ OR 43
45
+ LE 44
46
+ ' 45
47
+ ▁HIS 46
48
+ ▁YOU 47
49
+ AL 48
50
+ ▁RE 49
51
+ V 50
52
+ ▁B 51
53
+ G 52
54
+ RI 53
55
+ ▁E 54
56
+ ▁WITH 55
57
+ ▁T 56
58
+ ▁AS 57
59
+ LL 58
60
+ ▁P 59
61
+ ▁HER 60
62
+ ST 61
63
+ ▁HAD 62
64
+ ▁SO 63
65
+ ▁F 64
66
+ W 65
67
+ CE 66
68
+ ▁IS 67
69
+ ND 68
70
+ ▁NOT 69
71
+ TH 70
72
+ ▁BUT 71
73
+ EN 72
74
+ ▁SHE 73
75
+ ▁ON 74
76
+ VE 75
77
+ ON 76
78
+ SE 77
79
+ ▁DE 78
80
+ UR 79
81
+ ▁G 80
82
+ CH 81
83
+ K 82
84
+ TER 83
85
+ ▁AT 84
86
+ IT 85
87
+ ▁ME 86
88
+ RO 87
89
+ NE 88
90
+ RA 89
91
+ ES 90
92
+ IL 91
93
+ NG 92
94
+ IC 93
95
+ ▁NO 94
96
+ ▁HIM 95
97
+ ENT 96
98
+ IR 97
99
+ ▁WE 98
100
+ H 99
101
+ ▁DO 100
102
+ ▁ALL 101
103
+ ▁HAVE 102
104
+ LO 103
105
+ ▁BY 104
106
+ ▁MY 105
107
+ ▁MO 106
108
+ ▁THIS 107
109
+ LA 108
110
+ ▁ST 109
111
+ ▁WHICH 110
112
+ ▁CON 111
113
+ ▁THEY 112
114
+ CK 113
115
+ TE 114
116
+ ▁SAID 115
117
+ ▁FROM 116
118
+ ▁GO 117
119
+ ▁WHO 118
120
+ ▁TH 119
121
+ ▁OR 120
122
+ ▁D 121
123
+ ▁W 122
124
+ VER 123
125
+ LI 124
126
+ ▁SE 125
127
+ ▁ONE 126
128
+ ▁CA 127
129
+ ▁AN 128
130
+ ▁LA 129
131
+ ▁WERE 130
132
+ EL 131
133
+ ▁HA 132
134
+ ▁MAN 133
135
+ ▁FA 134
136
+ ▁EX 135
137
+ AD 136
138
+ ▁SU 137
139
+ RY 138
140
+ ▁MI 139
141
+ AT 140
142
+ ▁BO 141
143
+ ▁WHEN 142
144
+ AN 143
145
+ THER 144
146
+ PP 145
147
+ ATION 146
148
+ ▁FI 147
149
+ ▁WOULD 148
150
+ ▁PRO 149
151
+ OW 150
152
+ ET 151
153
+ ▁O 152
154
+ ▁THERE 153
155
+ ▁HO 154
156
+ ION 155
157
+ ▁WHAT 156
158
+ ▁FE 157
159
+ ▁PA 158
160
+ US 159
161
+ MENT 160
162
+ ▁MA 161
163
+ UT 162
164
+ ▁OUT 163
165
+ ▁THEIR 164
166
+ ▁IF 165
167
+ ▁LI 166
168
+ ▁K 167
169
+ ▁WILL 168
170
+ ▁ARE 169
171
+ ID 170
172
+ ▁RO 171
173
+ DE 172
174
+ TION 173
175
+ ▁WA 174
176
+ PE 175
177
+ ▁UP 176
178
+ ▁SP 177
179
+ ▁PO 178
180
+ IGHT 179
181
+ ▁UN 180
182
+ RU 181
183
+ ▁LO 182
184
+ AS 183
185
+ OL 184
186
+ ▁LE 185
187
+ ▁BEEN 186
188
+ ▁SH 187
189
+ ▁RA 188
190
+ ▁SEE 189
191
+ KE 190
192
+ UL 191
193
+ TED 192
194
+ ▁SA 193
195
+ UN 194
196
+ UND 195
197
+ ANT 196
198
+ ▁NE 197
199
+ IS 198
200
+ ▁THEM 199
201
+ CI 200
202
+ GE 201
203
+ ▁COULD 202
204
+ ▁DIS 203
205
+ OM 204
206
+ ISH 205
207
+ HE 206
208
+ EST 207
209
+ ▁SOME 208
210
+ ENCE 209
211
+ ITY 210
212
+ IVE 211
213
+ ▁US 212
214
+ ▁MORE 213
215
+ ▁EN 214
216
+ ARD 215
217
+ ATE 216
218
+ ▁YOUR 217
219
+ ▁INTO 218
220
+ ▁KNOW 219
221
+ ▁CO 220
222
+ ANCE 221
223
+ ▁TIME 222
224
+ ▁WI 223
225
+ ▁YE 224
226
+ AGE 225
227
+ ▁NOW 226
228
+ TI 227
229
+ FF 228
230
+ ABLE 229
231
+ ▁VERY 230
232
+ ▁LIKE 231
233
+ AM 232
234
+ HI 233
235
+ Z 234
236
+ ▁OTHER 235
237
+ ▁THAN 236
238
+ ▁LITTLE 237
239
+ ▁DID 238
240
+ ▁LOOK 239
241
+ TY 240
242
+ ERS 241
243
+ ▁CAN 242
244
+ ▁CHA 243
245
+ ▁AR 244
246
+ X 245
247
+ FUL 246
248
+ UGH 247
249
+ ▁BA 248
250
+ ▁DAY 249
251
+ ▁ABOUT 250
252
+ TEN 251
253
+ IM 252
254
+ ▁ANY 253
255
+ ▁PRE 254
256
+ ▁OVER 255
257
+ IES 256
258
+ NESS 257
259
+ ME 258
260
+ BLE 259
261
+ ▁M 260
262
+ ROW 261
263
+ ▁HAS 262
264
+ ▁GREAT 263
265
+ ▁VI 264
266
+ TA 265
267
+ ▁AFTER 266
268
+ PER 267
269
+ ▁AGAIN 268
270
+ HO 269
271
+ SH 270
272
+ ▁UPON 271
273
+ ▁DI 272
274
+ ▁HAND 273
275
+ ▁COM 274
276
+ IST 275
277
+ TURE 276
278
+ ▁STA 277
279
+ ▁THEN 278
280
+ ▁SHOULD 279
281
+ ▁GA 280
282
+ OUS 281
283
+ OUR 282
284
+ ▁WELL 283
285
+ ▁ONLY 284
286
+ MAN 285
287
+ ▁GOOD 286
288
+ ▁TWO 287
289
+ ▁MAR 288
290
+ ▁SAY 289
291
+ ▁HU 290
292
+ TING 291
293
+ ▁OUR 292
294
+ RESS 293
295
+ ▁DOWN 294
296
+ IOUS 295
297
+ ▁BEFORE 296
298
+ ▁DA 297
299
+ ▁NA 298
300
+ QUI 299
301
+ ▁MADE 300
302
+ ▁EVERY 301
303
+ ▁OLD 302
304
+ ▁EVEN 303
305
+ IG 304
306
+ ▁COME 305
307
+ ▁GRA 306
308
+ ▁RI 307
309
+ ▁LONG 308
310
+ OT 309
311
+ SIDE 310
312
+ WARD 311
313
+ ▁FO 312
314
+ ▁WHERE 313
315
+ MO 314
316
+ LESS 315
317
+ ▁SC 316
318
+ ▁MUST 317
319
+ ▁NEVER 318
320
+ ▁HOW 319
321
+ ▁CAME 320
322
+ ▁SUCH 321
323
+ ▁RU 322
324
+ ▁TAKE 323
325
+ ▁WO 324
326
+ ▁CAR 325
327
+ UM 326
328
+ AK 327
329
+ ▁THINK 328
330
+ ▁MUCH 329
331
+ ▁MISTER 330
332
+ ▁MAY 331
333
+ ▁JO 332
334
+ ▁WAY 333
335
+ ▁COMP 334
336
+ ▁THOUGHT 335
337
+ ▁STO 336
338
+ ▁MEN 337
339
+ ▁BACK 338
340
+ ▁DON 339
341
+ J 340
342
+ ▁LET 341
343
+ ▁TRA 342
344
+ ▁FIRST 343
345
+ ▁JUST 344
346
+ ▁VA 345
347
+ ▁OWN 346
348
+ ▁PLA 347
349
+ ▁MAKE 348
350
+ ATED 349
351
+ ▁HIMSELF 350
352
+ ▁WENT 351
353
+ ▁PI 352
354
+ GG 353
355
+ RING 354
356
+ ▁DU 355
357
+ ▁MIGHT 356
358
+ ▁PART 357
359
+ ▁GIVE 358
360
+ ▁IMP 359
361
+ ▁BU 360
362
+ ▁PER 361
363
+ ▁PLACE 362
364
+ ▁HOUSE 363
365
+ ▁THROUGH 364
366
+ IAN 365
367
+ ▁SW 366
368
+ ▁UNDER 367
369
+ QUE 368
370
+ ▁AWAY 369
371
+ ▁LOVE 370
372
+ QUA 371
373
+ ▁LIFE 372
374
+ ▁GET 373
375
+ ▁WITHOUT 374
376
+ ▁PASS 375
377
+ ▁TURN 376
378
+ IGN 377
379
+ ▁HEAD 378
380
+ ▁MOST 379
381
+ ▁THOSE 380
382
+ ▁SHALL 381
383
+ ▁EYES 382
384
+ ▁COL 383
385
+ ▁STILL 384
386
+ ▁NIGHT 385
387
+ ▁NOTHING 386
388
+ ITION 387
389
+ HA 388
390
+ ▁TELL 389
391
+ ▁WORK 390
392
+ ▁LAST 391
393
+ ▁NEW 392
394
+ ▁FACE 393
395
+ ▁HI 394
396
+ ▁WORD 395
397
+ ▁FOUND 396
398
+ ▁COUNT 397
399
+ ▁OB 398
400
+ ▁WHILE 399
401
+ ▁SHA 400
402
+ ▁MEAN 401
403
+ ▁SAW 402
404
+ ▁PEOPLE 403
405
+ ▁FRIEND 404
406
+ ▁THREE 405
407
+ ▁ROOM 406
408
+ ▁SAME 407
409
+ ▁THOUGH 408
410
+ ▁RIGHT 409
411
+ ▁CHILD 410
412
+ ▁FATHER 411
413
+ ▁ANOTHER 412
414
+ ▁HEART 413
415
+ ▁WANT 414
416
+ ▁TOOK 415
417
+ OOK 416
418
+ ▁LIGHT 417
419
+ ▁MISSUS 418
420
+ ▁OPEN 419
421
+ ▁JU 420
422
+ ▁ASKED 421
423
+ PORT 422
424
+ ▁LEFT 423
425
+ ▁JA 424
426
+ ▁WORLD 425
427
+ ▁HOME 426
428
+ ▁WHY 427
429
+ ▁ALWAYS 428
430
+ ▁ANSWER 429
431
+ ▁SEEMED 430
432
+ ▁SOMETHING 431
433
+ ▁GIRL 432
434
+ ▁BECAUSE 433
435
+ ▁NAME 434
436
+ ▁TOLD 435
437
+ ▁NI 436
438
+ ▁HIGH 437
439
+ IZE 438
440
+ ▁WOMAN 439
441
+ ▁FOLLOW 440
442
+ ▁RETURN 441
443
+ ▁KNEW 442
444
+ ▁EACH 443
445
+ ▁KIND 444
446
+ ▁JE 445
447
+ ▁ACT 446
448
+ ▁LU 447
449
+ ▁CERTAIN 448
450
+ ▁YEARS 449
451
+ ▁QUITE 450
452
+ ▁APPEAR 451
453
+ ▁BETTER 452
454
+ ▁HALF 453
455
+ ▁PRESENT 454
456
+ ▁PRINCE 455
457
+ SHIP 456
458
+ ▁ALSO 457
459
+ ▁BEGAN 458
460
+ ▁HAVING 459
461
+ ▁ENOUGH 460
462
+ ▁PERSON 461
463
+ ▁LADY 462
464
+ ▁WHITE 463
465
+ ▁COURSE 464
466
+ ▁VOICE 465
467
+ ▁SPEAK 466
468
+ ▁POWER 467
469
+ ▁MORNING 468
470
+ ▁BETWEEN 469
471
+ ▁AMONG 470
472
+ ▁KEEP 471
473
+ ▁WALK 472
474
+ ▁MATTER 473
475
+ ▁TEA 474
476
+ ▁BELIEVE 475
477
+ ▁SMALL 476
478
+ ▁TALK 477
479
+ ▁FELT 478
480
+ ▁HORSE 479
481
+ ▁MYSELF 480
482
+ ▁SIX 481
483
+ ▁HOWEVER 482
484
+ ▁FULL 483
485
+ ▁HERSELF 484
486
+ ▁POINT 485
487
+ ▁STOOD 486
488
+ ▁HUNDRED 487
489
+ ▁ALMOST 488
490
+ ▁SINCE 489
491
+ ▁LARGE 490
492
+ ▁LEAVE 491
493
+ ▁PERHAPS 492
494
+ ▁DARK 493
495
+ ▁SUDDEN 494
496
+ ▁REPLIED 495
497
+ ▁ANYTHING 496
498
+ ▁WONDER 497
499
+ ▁UNTIL 498
500
+ Q 499
501
+ #0 500
502
+ #1 501
decoding_results/greedy_search/errs-test-clean-greedy_search-epoch-30-avg-11-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/errs-test-clean-greedy_search-epoch-30-avg-13-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/errs-test-clean-greedy_search-epoch-30-avg-15-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/errs-test-clean-greedy_search-epoch-30-avg-7-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/errs-test-clean-greedy_search-epoch-30-avg-9-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/errs-test-clean-greedy_search-epoch-40-avg-11-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/errs-test-clean-greedy_search-epoch-40-avg-13-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/errs-test-clean-greedy_search-epoch-40-avg-15-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/errs-test-clean-greedy_search-epoch-40-avg-17-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/errs-test-clean-greedy_search-epoch-40-avg-19-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/errs-test-clean-greedy_search-epoch-40-avg-7-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/errs-test-clean-greedy_search-epoch-40-avg-9-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/errs-test-other-greedy_search-epoch-30-avg-11-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/errs-test-other-greedy_search-epoch-30-avg-13-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/errs-test-other-greedy_search-epoch-30-avg-15-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/errs-test-other-greedy_search-epoch-30-avg-7-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/errs-test-other-greedy_search-epoch-30-avg-9-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/errs-test-other-greedy_search-epoch-40-avg-11-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/errs-test-other-greedy_search-epoch-40-avg-13-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/errs-test-other-greedy_search-epoch-40-avg-15-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/errs-test-other-greedy_search-epoch-40-avg-17-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/errs-test-other-greedy_search-epoch-40-avg-19-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/errs-test-other-greedy_search-epoch-40-avg-7-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/errs-test-other-greedy_search-epoch-40-avg-9-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/log-decode-epoch-30-avg-11-context-2-max-sym-per-frame-1-use-averaged-model-2024-08-08-10-10-56 ADDED
@@ -0,0 +1,25 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2024-08-08 10:10:56,739 INFO [decode.py:833] Decoding started
2
+ 2024-08-08 10:10:56,740 INFO [decode.py:839] Device: cuda:0
3
+ 2024-08-08 10:10:56,743 INFO [decode.py:849] {'best_train_loss': inf, 'best_valid_loss': inf, 'best_train_epoch': -1, 'best_valid_epoch': -1, 'batch_idx_train': 0, 'log_interval': 50, 'reset_interval': 200, 'valid_interval': 3000, 'feature_dim': 80, 'subsampling_factor': 4, 'ignore_id': -1, 'label_smoothing': 0.1, 'warm_step': 2000, 'env_info': {'k2-version': '1.24.4', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': 'ff1d435a8d3c4eaa15828a84a7240678a70539a7', 'k2-git-date': 'Fri Feb 23 01:48:38 2024', 'lhotse-version': '1.26.0', 'torch-version': '2.0.1+cu117', 'torch-cuda-available': True, 'torch-cuda-version': '11.7', 'python-version': '3.10', 'icefall-git-branch': 'zipformer_bf16', 'icefall-git-sha1': 'd1974bef-dirty', 'icefall-git-date': 'Thu Aug 8 10:10:22 2024', 'icefall-path': '/star-xy/softwares/icefall_development/icefall_bf16', 'k2-path': '/star-xy/softwares/pyenvs/k2_cuda11/k2_cuda11/lib/python3.10/site-packages/k2/__init__.py', 'lhotse-path': '/star-xy/softwares/pyenvs/k2_cuda11/k2_cuda11/lib/python3.10/site-packages/lhotse/__init__.py', 'hostname': 'de-74279-k2-train-2-1207150844-f49d8c4f4-c49d5', 'IP address': '10.177.22.19'}, 'epoch': 30, 'iter': 0, 'avg': 11, 'use_averaged_model': True, 'exp_dir': PosixPath('zipformer/exp-full-liri-full-bf16-md1000'), 'bpe_model': 'data/lang_bpe_500/bpe.model', 'lang_dir': PosixPath('data/lang_bpe_500'), 'decoding_method': 'greedy_search', 'beam_size': 4, 'beam': 20.0, 'ngram_lm_scale': 0.01, 'max_contexts': 8, 'max_states': 64, 'context_size': 2, 'max_sym_per_frame': 1, 'num_paths': 200, 'nbest_scale': 0.5, 'use_shallow_fusion': False, 'lm_type': 'rnn', 'lm_scale': 0.3, 'tokens_ngram': 2, 'backoff_id': 500, 'context_score': 2, 'context_file': '', 'num_encoder_layers': '2,2,3,4,3,2', 'downsampling_factor': '1,2,4,8,4,2', 'feedforward_dim': '512,768,1024,1536,1024,768', 'num_heads': '4,4,4,8,4,4', 'encoder_dim': '192,256,384,512,384,256', 'query_head_dim': '32', 'value_head_dim': '12', 'pos_head_dim': '4', 'pos_dim': 48, 'encoder_unmasked_dim': '192,192,256,256,256,192', 'cnn_module_kernel': '31,31,15,15,15,31', 'decoder_dim': 512, 'joiner_dim': 512, 'attention_decoder_dim': 512, 'attention_decoder_num_layers': 6, 'attention_decoder_attention_dim': 512, 'attention_decoder_num_heads': 8, 'attention_decoder_feedforward_dim': 2048, 'causal': False, 'chunk_size': '16,32,64,-1', 'left_context_frames': '64,128,256,-1', 'use_transducer': True, 'use_ctc': False, 'use_attention_decoder': False, 'full_libri': True, 'mini_libri': False, 'manifest_dir': PosixPath('data/fbank'), 'max_duration': 600, 'bucketing_sampler': True, 'num_buckets': 30, 'concatenate_cuts': False, 'duration_factor': 1.0, 'gap': 1.0, 'on_the_fly_feats': False, 'shuffle': True, 'drop_last': True, 'return_cuts': True, 'num_workers': 2, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'input_strategy': 'PrecomputedFeatures', 'lm_vocab_size': 500, 'lm_epoch': 7, 'lm_avg': 1, 'lm_exp_dir': None, 'rnn_lm_embedding_dim': 2048, 'rnn_lm_hidden_dim': 2048, 'rnn_lm_num_layers': 3, 'rnn_lm_tie_weights': True, 'transformer_lm_exp_dir': None, 'transformer_lm_dim_feedforward': 2048, 'transformer_lm_encoder_dim': 768, 'transformer_lm_embedding_dim': 768, 'transformer_lm_nhead': 8, 'transformer_lm_num_layers': 16, 'transformer_lm_tie_weights': True, 'res_dir': PosixPath('zipformer/exp-full-liri-full-bf16-md1000/greedy_search'), 'has_contexts': False, 'suffix': 'epoch-30-avg-11-context-2-max-sym-per-frame-1-use-averaged-model', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
4
+ 2024-08-08 10:10:56,743 INFO [decode.py:851] About to create model
5
+ 2024-08-08 10:10:57,476 INFO [decode.py:918] Calculating the averaged model over epoch range from 19 (excluded) to 30
6
+ 2024-08-08 10:11:02,633 INFO [decode.py:1012] Number of model parameters: 65549011
7
+ 2024-08-08 10:11:02,633 INFO [asr_datamodule.py:467] About to get test-clean cuts
8
+ 2024-08-08 10:11:02,654 INFO [asr_datamodule.py:474] About to get test-other cuts
9
+ 2024-08-08 10:11:05,912 INFO [decode.py:706] batch 0/?, cuts processed until now is 36
10
+ 2024-08-08 10:11:31,134 INFO [decode.py:722] The transcripts are stored in zipformer/exp-full-liri-full-bf16-md1000/greedy_search/recogs-test-clean-greedy_search-epoch-30-avg-11-context-2-max-sym-per-frame-1-use-averaged-model.txt
11
+ 2024-08-08 10:11:31,246 INFO [utils.py:657] [test-clean-greedy_search] %WER 2.42% [1274 / 52576, 152 ins, 109 del, 1013 sub ]
12
+ 2024-08-08 10:11:31,499 INFO [decode.py:735] Wrote detailed error stats to zipformer/exp-full-liri-full-bf16-md1000/greedy_search/errs-test-clean-greedy_search-epoch-30-avg-11-context-2-max-sym-per-frame-1-use-averaged-model.txt
13
+ 2024-08-08 10:11:31,503 INFO [decode.py:751]
14
+ For test-clean, WER of different settings are:
15
+ greedy_search 2.42 best for test-clean
16
+
17
+ 2024-08-08 10:11:32,765 INFO [decode.py:706] batch 0/?, cuts processed until now is 43
18
+ 2024-08-08 10:11:56,085 INFO [decode.py:722] The transcripts are stored in zipformer/exp-full-liri-full-bf16-md1000/greedy_search/recogs-test-other-greedy_search-epoch-30-avg-11-context-2-max-sym-per-frame-1-use-averaged-model.txt
19
+ 2024-08-08 10:11:56,197 INFO [utils.py:657] [test-other-greedy_search] %WER 5.42% [2837 / 52343, 283 ins, 249 del, 2305 sub ]
20
+ 2024-08-08 10:11:56,482 INFO [decode.py:735] Wrote detailed error stats to zipformer/exp-full-liri-full-bf16-md1000/greedy_search/errs-test-other-greedy_search-epoch-30-avg-11-context-2-max-sym-per-frame-1-use-averaged-model.txt
21
+ 2024-08-08 10:11:56,485 INFO [decode.py:751]
22
+ For test-other, WER of different settings are:
23
+ greedy_search 5.42 best for test-other
24
+
25
+ 2024-08-08 10:11:56,485 INFO [decode.py:1047] Done!
decoding_results/greedy_search/log-decode-epoch-30-avg-13-context-2-max-sym-per-frame-1-use-averaged-model-2024-08-08-10-09-48 ADDED
@@ -0,0 +1,27 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2024-08-08 10:09:48,524 INFO [decode.py:833] Decoding started
2
+ 2024-08-08 10:09:48,524 INFO [decode.py:839] Device: cuda:0
3
+ 2024-08-08 10:09:48,531 INFO [decode.py:849] {'best_train_loss': inf, 'best_valid_loss': inf, 'best_train_epoch': -1, 'best_valid_epoch': -1, 'batch_idx_train': 0, 'log_interval': 50, 'reset_interval': 200, 'valid_interval': 3000, 'feature_dim': 80, 'subsampling_factor': 4, 'ignore_id': -1, 'label_smoothing': 0.1, 'warm_step': 2000, 'env_info': {'k2-version': '1.24.4', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': 'ff1d435a8d3c4eaa15828a84a7240678a70539a7', 'k2-git-date': 'Fri Feb 23 01:48:38 2024', 'lhotse-version': '1.26.0', 'torch-version': '2.0.1+cu117', 'torch-cuda-available': True, 'torch-cuda-version': '11.7', 'python-version': '3.10', 'icefall-git-branch': 'zipformer_bf16', 'icefall-git-sha1': '7692bd4d-dirty', 'icefall-git-date': 'Wed Jul 24 15:06:10 2024', 'icefall-path': '/star-xy/softwares/icefall_development/icefall_bf16', 'k2-path': '/star-xy/softwares/pyenvs/k2_cuda11/k2_cuda11/lib/python3.10/site-packages/k2/__init__.py', 'lhotse-path': '/star-xy/softwares/pyenvs/k2_cuda11/k2_cuda11/lib/python3.10/site-packages/lhotse/__init__.py', 'hostname': 'de-74279-k2-train-2-1207150844-f49d8c4f4-c49d5', 'IP address': '10.177.22.19'}, 'epoch': 30, 'iter': 0, 'avg': 13, 'use_averaged_model': True, 'exp_dir': PosixPath('zipformer/exp-full-liri-full-bf16-md1000'), 'bpe_model': 'data/lang_bpe_500/bpe.model', 'lang_dir': PosixPath('data/lang_bpe_500'), 'decoding_method': 'greedy_search', 'beam_size': 4, 'beam': 20.0, 'ngram_lm_scale': 0.01, 'max_contexts': 8, 'max_states': 64, 'context_size': 2, 'max_sym_per_frame': 1, 'num_paths': 200, 'nbest_scale': 0.5, 'use_shallow_fusion': False, 'lm_type': 'rnn', 'lm_scale': 0.3, 'tokens_ngram': 2, 'backoff_id': 500, 'context_score': 2, 'context_file': '', 'num_encoder_layers': '2,2,3,4,3,2', 'downsampling_factor': '1,2,4,8,4,2', 'feedforward_dim': '512,768,1024,1536,1024,768', 'num_heads': '4,4,4,8,4,4', 'encoder_dim': '192,256,384,512,384,256', 'query_head_dim': '32', 'value_head_dim': '12', 'pos_head_dim': '4', 'pos_dim': 48, 'encoder_unmasked_dim': '192,192,256,256,256,192', 'cnn_module_kernel': '31,31,15,15,15,31', 'decoder_dim': 512, 'joiner_dim': 512, 'attention_decoder_dim': 512, 'attention_decoder_num_layers': 6, 'attention_decoder_attention_dim': 512, 'attention_decoder_num_heads': 8, 'attention_decoder_feedforward_dim': 2048, 'causal': False, 'chunk_size': '16,32,64,-1', 'left_context_frames': '64,128,256,-1', 'use_transducer': True, 'use_ctc': False, 'use_attention_decoder': False, 'full_libri': True, 'mini_libri': False, 'manifest_dir': PosixPath('data/fbank'), 'max_duration': 600, 'bucketing_sampler': True, 'num_buckets': 30, 'concatenate_cuts': False, 'duration_factor': 1.0, 'gap': 1.0, 'on_the_fly_feats': False, 'shuffle': True, 'drop_last': True, 'return_cuts': True, 'num_workers': 2, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'input_strategy': 'PrecomputedFeatures', 'lm_vocab_size': 500, 'lm_epoch': 7, 'lm_avg': 1, 'lm_exp_dir': None, 'rnn_lm_embedding_dim': 2048, 'rnn_lm_hidden_dim': 2048, 'rnn_lm_num_layers': 3, 'rnn_lm_tie_weights': True, 'transformer_lm_exp_dir': None, 'transformer_lm_dim_feedforward': 2048, 'transformer_lm_encoder_dim': 768, 'transformer_lm_embedding_dim': 768, 'transformer_lm_nhead': 8, 'transformer_lm_num_layers': 16, 'transformer_lm_tie_weights': True, 'res_dir': PosixPath('zipformer/exp-full-liri-full-bf16-md1000/greedy_search'), 'has_contexts': False, 'suffix': 'epoch-30-avg-13-context-2-max-sym-per-frame-1-use-averaged-model', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
4
+ 2024-08-08 10:09:48,531 INFO [decode.py:851] About to create model
5
+ 2024-08-08 10:09:49,284 INFO [decode.py:918] Calculating the averaged model over epoch range from 17 (excluded) to 30
6
+ 2024-08-08 10:09:54,635 INFO [decode.py:1012] Number of model parameters: 65549011
7
+ 2024-08-08 10:09:54,636 INFO [asr_datamodule.py:467] About to get test-clean cuts
8
+ 2024-08-08 10:09:54,673 INFO [asr_datamodule.py:474] About to get test-other cuts
9
+ 2024-08-08 10:09:57,847 INFO [decode.py:706] batch 0/?, cuts processed until now is 36
10
+ 2024-08-08 10:09:59,108 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([1.8040, 3.0072, 2.6733, 3.1166], device='cuda:0')
11
+ 2024-08-08 10:10:22,145 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([4.1837, 4.5047, 3.0808, 2.3496], device='cuda:0')
12
+ 2024-08-08 10:10:24,049 INFO [decode.py:722] The transcripts are stored in zipformer/exp-full-liri-full-bf16-md1000/greedy_search/recogs-test-clean-greedy_search-epoch-30-avg-13-context-2-max-sym-per-frame-1-use-averaged-model.txt
13
+ 2024-08-08 10:10:24,169 INFO [utils.py:657] [test-clean-greedy_search] %WER 2.42% [1274 / 52576, 149 ins, 109 del, 1016 sub ]
14
+ 2024-08-08 10:10:24,445 INFO [decode.py:735] Wrote detailed error stats to zipformer/exp-full-liri-full-bf16-md1000/greedy_search/errs-test-clean-greedy_search-epoch-30-avg-13-context-2-max-sym-per-frame-1-use-averaged-model.txt
15
+ 2024-08-08 10:10:24,452 INFO [decode.py:751]
16
+ For test-clean, WER of different settings are:
17
+ greedy_search 2.42 best for test-clean
18
+
19
+ 2024-08-08 10:10:25,705 INFO [decode.py:706] batch 0/?, cuts processed until now is 43
20
+ 2024-08-08 10:10:49,657 INFO [decode.py:722] The transcripts are stored in zipformer/exp-full-liri-full-bf16-md1000/greedy_search/recogs-test-other-greedy_search-epoch-30-avg-13-context-2-max-sym-per-frame-1-use-averaged-model.txt
21
+ 2024-08-08 10:10:49,776 INFO [utils.py:657] [test-other-greedy_search] %WER 5.41% [2830 / 52343, 278 ins, 242 del, 2310 sub ]
22
+ 2024-08-08 10:10:50,053 INFO [decode.py:735] Wrote detailed error stats to zipformer/exp-full-liri-full-bf16-md1000/greedy_search/errs-test-other-greedy_search-epoch-30-avg-13-context-2-max-sym-per-frame-1-use-averaged-model.txt
23
+ 2024-08-08 10:10:50,056 INFO [decode.py:751]
24
+ For test-other, WER of different settings are:
25
+ greedy_search 5.41 best for test-other
26
+
27
+ 2024-08-08 10:10:50,056 INFO [decode.py:1047] Done!
decoding_results/greedy_search/log-decode-epoch-30-avg-15-context-2-max-sym-per-frame-1-use-averaged-model-2024-08-07-10-39-22 ADDED
@@ -0,0 +1,28 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2024-08-07 10:39:22,792 INFO [decode_bf16.py:833] Decoding started
2
+ 2024-08-07 10:39:22,793 INFO [decode_bf16.py:839] Device: cuda:0
3
+ 2024-08-07 10:39:22,801 INFO [decode_bf16.py:849] {'best_train_loss': inf, 'best_valid_loss': inf, 'best_train_epoch': -1, 'best_valid_epoch': -1, 'batch_idx_train': 0, 'log_interval': 50, 'reset_interval': 200, 'valid_interval': 3000, 'feature_dim': 80, 'subsampling_factor': 4, 'ignore_id': -1, 'label_smoothing': 0.1, 'warm_step': 2000, 'env_info': {'k2-version': '1.24.4', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': 'ff1d435a8d3c4eaa15828a84a7240678a70539a7', 'k2-git-date': 'Fri Feb 23 01:48:38 2024', 'lhotse-version': '1.26.0', 'torch-version': '2.0.1+cu117', 'torch-cuda-available': True, 'torch-cuda-version': '11.7', 'python-version': '3.10', 'icefall-git-branch': 'zipformer_bf16', 'icefall-git-sha1': '7692bd4d-dirty', 'icefall-git-date': 'Wed Jul 24 15:06:10 2024', 'icefall-path': '/star-xy/softwares/icefall_development/icefall_bf16', 'k2-path': '/star-xy/softwares/pyenvs/k2_cuda11/k2_cuda11/lib/python3.10/site-packages/k2/__init__.py', 'lhotse-path': '/star-xy/softwares/pyenvs/k2_cuda11/k2_cuda11/lib/python3.10/site-packages/lhotse/__init__.py', 'hostname': 'de-74279-k2-train-9-0208143539-7dbf569d4f-r7nrb', 'IP address': '10.177.13.150'}, 'epoch': 30, 'iter': 0, 'avg': 15, 'use_averaged_model': True, 'exp_dir': PosixPath('zipformer/exp-full-liri-full-bf16-md1000'), 'bpe_model': 'data/lang_bpe_500/bpe.model', 'lang_dir': PosixPath('data/lang_bpe_500'), 'decoding_method': 'greedy_search', 'beam_size': 4, 'beam': 20.0, 'ngram_lm_scale': 0.01, 'max_contexts': 8, 'max_states': 64, 'context_size': 2, 'max_sym_per_frame': 1, 'num_paths': 200, 'nbest_scale': 0.5, 'use_shallow_fusion': False, 'lm_type': 'rnn', 'lm_scale': 0.3, 'tokens_ngram': 2, 'backoff_id': 500, 'context_score': 2, 'context_file': '', 'num_encoder_layers': '2,2,3,4,3,2', 'downsampling_factor': '1,2,4,8,4,2', 'feedforward_dim': '512,768,1024,1536,1024,768', 'num_heads': '4,4,4,8,4,4', 'encoder_dim': '192,256,384,512,384,256', 'query_head_dim': '32', 'value_head_dim': '12', 'pos_head_dim': '4', 'pos_dim': 48, 'encoder_unmasked_dim': '192,192,256,256,256,192', 'cnn_module_kernel': '31,31,15,15,15,31', 'decoder_dim': 512, 'joiner_dim': 512, 'attention_decoder_dim': 512, 'attention_decoder_num_layers': 6, 'attention_decoder_attention_dim': 512, 'attention_decoder_num_heads': 8, 'attention_decoder_feedforward_dim': 2048, 'causal': False, 'chunk_size': '16,32,64,-1', 'left_context_frames': '64,128,256,-1', 'use_transducer': True, 'use_ctc': False, 'use_attention_decoder': False, 'full_libri': True, 'mini_libri': False, 'manifest_dir': PosixPath('data/fbank'), 'max_duration': 600, 'bucketing_sampler': True, 'num_buckets': 30, 'concatenate_cuts': False, 'duration_factor': 1.0, 'gap': 1.0, 'on_the_fly_feats': False, 'shuffle': True, 'drop_last': True, 'return_cuts': True, 'num_workers': 2, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'input_strategy': 'PrecomputedFeatures', 'lm_vocab_size': 500, 'lm_epoch': 7, 'lm_avg': 1, 'lm_exp_dir': None, 'rnn_lm_embedding_dim': 2048, 'rnn_lm_hidden_dim': 2048, 'rnn_lm_num_layers': 3, 'rnn_lm_tie_weights': True, 'transformer_lm_exp_dir': None, 'transformer_lm_dim_feedforward': 2048, 'transformer_lm_encoder_dim': 768, 'transformer_lm_embedding_dim': 768, 'transformer_lm_nhead': 8, 'transformer_lm_num_layers': 16, 'transformer_lm_tie_weights': True, 'res_dir': PosixPath('zipformer/exp-full-liri-full-bf16-md1000/greedy_search'), 'has_contexts': False, 'suffix': 'epoch-30-avg-15-context-2-max-sym-per-frame-1-use-averaged-model', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
4
+ 2024-08-07 10:39:22,802 INFO [decode_bf16.py:851] About to create model
5
+ 2024-08-07 10:39:23,927 INFO [decode_bf16.py:918] Calculating the averaged model over epoch range from 15 (excluded) to 30
6
+ 2024-08-07 10:39:37,680 INFO [decode_bf16.py:1012] Number of model parameters: 65549011
7
+ 2024-08-07 10:39:37,681 INFO [asr_datamodule.py:467] About to get test-clean cuts
8
+ 2024-08-07 10:39:37,692 INFO [asr_datamodule.py:474] About to get test-other cuts
9
+ 2024-08-07 10:39:46,632 INFO [decode_bf16.py:706] batch 0/?, cuts processed until now is 36
10
+ 2024-08-07 10:40:00,307 INFO [zipformer_full_bf16.py:1858] name=None, attn_weights_entropy = tensor([5.6690, 4.9965, 5.1595, 5.4509], device='cuda:0')
11
+ 2024-08-07 10:40:18,871 INFO [zipformer_full_bf16.py:1858] name=None, attn_weights_entropy = tensor([5.6741, 5.0880, 5.3362, 4.9879], device='cuda:0')
12
+ 2024-08-07 10:40:57,742 INFO [decode_bf16.py:722] The transcripts are stored in zipformer/exp-full-liri-full-bf16-md1000/greedy_search/recogs-test-clean-greedy_search-epoch-30-avg-15-context-2-max-sym-per-frame-1-use-averaged-model.txt
13
+ 2024-08-07 10:40:57,930 INFO [utils.py:657] [test-clean-greedy_search] %WER 2.42% [1270 / 52576, 147 ins, 109 del, 1014 sub ]
14
+ 2024-08-07 10:40:58,368 INFO [decode_bf16.py:735] Wrote detailed error stats to zipformer/exp-full-liri-full-bf16-md1000/greedy_search/errs-test-clean-greedy_search-epoch-30-avg-15-context-2-max-sym-per-frame-1-use-averaged-model.txt
15
+ 2024-08-07 10:40:58,372 INFO [decode_bf16.py:751]
16
+ For test-clean, WER of different settings are:
17
+ greedy_search 2.42 best for test-clean
18
+
19
+ 2024-08-07 10:41:00,192 INFO [decode_bf16.py:706] batch 0/?, cuts processed until now is 43
20
+ 2024-08-07 10:41:05,234 INFO [zipformer_full_bf16.py:1858] name=None, attn_weights_entropy = tensor([5.7368, 5.1922, 5.5282, 5.1211], device='cuda:0')
21
+ 2024-08-07 10:41:48,803 INFO [decode_bf16.py:722] The transcripts are stored in zipformer/exp-full-liri-full-bf16-md1000/greedy_search/recogs-test-other-greedy_search-epoch-30-avg-15-context-2-max-sym-per-frame-1-use-averaged-model.txt
22
+ 2024-08-07 10:41:48,999 INFO [utils.py:657] [test-other-greedy_search] %WER 5.44% [2847 / 52343, 277 ins, 249 del, 2321 sub ]
23
+ 2024-08-07 10:41:49,461 INFO [decode_bf16.py:735] Wrote detailed error stats to zipformer/exp-full-liri-full-bf16-md1000/greedy_search/errs-test-other-greedy_search-epoch-30-avg-15-context-2-max-sym-per-frame-1-use-averaged-model.txt
24
+ 2024-08-07 10:41:49,464 INFO [decode_bf16.py:751]
25
+ For test-other, WER of different settings are:
26
+ greedy_search 5.44 best for test-other
27
+
28
+ 2024-08-07 10:41:49,464 INFO [decode_bf16.py:1047] Done!
decoding_results/greedy_search/log-decode-epoch-30-avg-15-context-2-max-sym-per-frame-1-use-averaged-model-2024-08-08-10-08-02 ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2024-08-08 10:08:02,252 INFO [decode.py:833] Decoding started
2
+ 2024-08-08 10:08:02,253 INFO [decode.py:839] Device: cuda:0
3
+ 2024-08-08 10:08:02,260 INFO [decode.py:849] {'best_train_loss': inf, 'best_valid_loss': inf, 'best_train_epoch': -1, 'best_valid_epoch': -1, 'batch_idx_train': 0, 'log_interval': 50, 'reset_interval': 200, 'valid_interval': 3000, 'feature_dim': 80, 'subsampling_factor': 4, 'ignore_id': -1, 'label_smoothing': 0.1, 'warm_step': 2000, 'env_info': {'k2-version': '1.24.4', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': 'ff1d435a8d3c4eaa15828a84a7240678a70539a7', 'k2-git-date': 'Fri Feb 23 01:48:38 2024', 'lhotse-version': '1.26.0', 'torch-version': '2.0.1+cu117', 'torch-cuda-available': True, 'torch-cuda-version': '11.7', 'python-version': '3.10', 'icefall-git-branch': 'zipformer_bf16', 'icefall-git-sha1': '7692bd4d-dirty', 'icefall-git-date': 'Wed Jul 24 15:06:10 2024', 'icefall-path': '/star-xy/softwares/icefall_development/icefall_bf16', 'k2-path': '/star-xy/softwares/pyenvs/k2_cuda11/k2_cuda11/lib/python3.10/site-packages/k2/__init__.py', 'lhotse-path': '/star-xy/softwares/pyenvs/k2_cuda11/k2_cuda11/lib/python3.10/site-packages/lhotse/__init__.py', 'hostname': 'de-74279-k2-train-2-1207150844-f49d8c4f4-c49d5', 'IP address': '10.177.22.19'}, 'epoch': 30, 'iter': 0, 'avg': 15, 'use_averaged_model': True, 'exp_dir': PosixPath('zipformer/exp-full-liri-full-bf16-md1000'), 'bpe_model': 'data/lang_bpe_500/bpe.model', 'lang_dir': PosixPath('data/lang_bpe_500'), 'decoding_method': 'greedy_search', 'beam_size': 4, 'beam': 20.0, 'ngram_lm_scale': 0.01, 'max_contexts': 8, 'max_states': 64, 'context_size': 2, 'max_sym_per_frame': 1, 'num_paths': 200, 'nbest_scale': 0.5, 'use_shallow_fusion': False, 'lm_type': 'rnn', 'lm_scale': 0.3, 'tokens_ngram': 2, 'backoff_id': 500, 'context_score': 2, 'context_file': '', 'num_encoder_layers': '2,2,3,4,3,2', 'downsampling_factor': '1,2,4,8,4,2', 'feedforward_dim': '512,768,1024,1536,1024,768', 'num_heads': '4,4,4,8,4,4', 'encoder_dim': '192,256,384,512,384,256', 'query_head_dim': '32', 'value_head_dim': '12', 'pos_head_dim': '4', 'pos_dim': 48, 'encoder_unmasked_dim': '192,192,256,256,256,192', 'cnn_module_kernel': '31,31,15,15,15,31', 'decoder_dim': 512, 'joiner_dim': 512, 'attention_decoder_dim': 512, 'attention_decoder_num_layers': 6, 'attention_decoder_attention_dim': 512, 'attention_decoder_num_heads': 8, 'attention_decoder_feedforward_dim': 2048, 'causal': False, 'chunk_size': '16,32,64,-1', 'left_context_frames': '64,128,256,-1', 'use_transducer': True, 'use_ctc': False, 'use_attention_decoder': False, 'full_libri': True, 'mini_libri': False, 'manifest_dir': PosixPath('data/fbank'), 'max_duration': 600, 'bucketing_sampler': True, 'num_buckets': 30, 'concatenate_cuts': False, 'duration_factor': 1.0, 'gap': 1.0, 'on_the_fly_feats': False, 'shuffle': True, 'drop_last': True, 'return_cuts': True, 'num_workers': 2, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'input_strategy': 'PrecomputedFeatures', 'lm_vocab_size': 500, 'lm_epoch': 7, 'lm_avg': 1, 'lm_exp_dir': None, 'rnn_lm_embedding_dim': 2048, 'rnn_lm_hidden_dim': 2048, 'rnn_lm_num_layers': 3, 'rnn_lm_tie_weights': True, 'transformer_lm_exp_dir': None, 'transformer_lm_dim_feedforward': 2048, 'transformer_lm_encoder_dim': 768, 'transformer_lm_embedding_dim': 768, 'transformer_lm_nhead': 8, 'transformer_lm_num_layers': 16, 'transformer_lm_tie_weights': True, 'res_dir': PosixPath('zipformer/exp-full-liri-full-bf16-md1000/greedy_search'), 'has_contexts': False, 'suffix': 'epoch-30-avg-15-context-2-max-sym-per-frame-1-use-averaged-model', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
4
+ 2024-08-08 10:08:02,260 INFO [decode.py:851] About to create model
5
+ 2024-08-08 10:08:03,018 INFO [decode.py:918] Calculating the averaged model over epoch range from 15 (excluded) to 30
6
+ 2024-08-08 10:08:09,307 INFO [decode.py:1012] Number of model parameters: 65549011
7
+ 2024-08-08 10:08:09,308 INFO [asr_datamodule.py:467] About to get test-clean cuts
8
+ 2024-08-08 10:08:09,323 INFO [asr_datamodule.py:474] About to get test-other cuts
9
+ 2024-08-08 10:08:12,453 INFO [decode.py:706] batch 0/?, cuts processed until now is 36
10
+ 2024-08-08 10:08:27,723 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([3.9975, 4.9900, 4.9416, 2.7579], device='cuda:0')
11
+ 2024-08-08 10:08:36,998 INFO [decode.py:722] The transcripts are stored in zipformer/exp-full-liri-full-bf16-md1000/greedy_search/recogs-test-clean-greedy_search-epoch-30-avg-15-context-2-max-sym-per-frame-1-use-averaged-model.txt
12
+ 2024-08-08 10:08:37,120 INFO [utils.py:657] [test-clean-greedy_search] %WER 2.42% [1270 / 52576, 147 ins, 109 del, 1014 sub ]
13
+ 2024-08-08 10:08:37,381 INFO [decode.py:735] Wrote detailed error stats to zipformer/exp-full-liri-full-bf16-md1000/greedy_search/errs-test-clean-greedy_search-epoch-30-avg-15-context-2-max-sym-per-frame-1-use-averaged-model.txt
14
+ 2024-08-08 10:08:37,384 INFO [decode.py:751]
15
+ For test-clean, WER of different settings are:
16
+ greedy_search 2.42 best for test-clean
17
+
18
+ 2024-08-08 10:08:38,543 INFO [decode.py:706] batch 0/?, cuts processed until now is 43
19
+ 2024-08-08 10:09:01,258 INFO [decode.py:722] The transcripts are stored in zipformer/exp-full-liri-full-bf16-md1000/greedy_search/recogs-test-other-greedy_search-epoch-30-avg-15-context-2-max-sym-per-frame-1-use-averaged-model.txt
20
+ 2024-08-08 10:09:01,375 INFO [utils.py:657] [test-other-greedy_search] %WER 5.44% [2847 / 52343, 277 ins, 249 del, 2321 sub ]
21
+ 2024-08-08 10:09:01,638 INFO [decode.py:735] Wrote detailed error stats to zipformer/exp-full-liri-full-bf16-md1000/greedy_search/errs-test-other-greedy_search-epoch-30-avg-15-context-2-max-sym-per-frame-1-use-averaged-model.txt
22
+ 2024-08-08 10:09:01,641 INFO [decode.py:751]
23
+ For test-other, WER of different settings are:
24
+ greedy_search 5.44 best for test-other
25
+
26
+ 2024-08-08 10:09:01,641 INFO [decode.py:1047] Done!
decoding_results/greedy_search/log-decode-epoch-30-avg-7-context-2-max-sym-per-frame-1-use-averaged-model-2024-08-08-10-13-10 ADDED
@@ -0,0 +1,28 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2024-08-08 10:13:10,193 INFO [decode.py:833] Decoding started
2
+ 2024-08-08 10:13:10,193 INFO [decode.py:839] Device: cuda:0
3
+ 2024-08-08 10:13:10,200 INFO [decode.py:849] {'best_train_loss': inf, 'best_valid_loss': inf, 'best_train_epoch': -1, 'best_valid_epoch': -1, 'batch_idx_train': 0, 'log_interval': 50, 'reset_interval': 200, 'valid_interval': 3000, 'feature_dim': 80, 'subsampling_factor': 4, 'ignore_id': -1, 'label_smoothing': 0.1, 'warm_step': 2000, 'env_info': {'k2-version': '1.24.4', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': 'ff1d435a8d3c4eaa15828a84a7240678a70539a7', 'k2-git-date': 'Fri Feb 23 01:48:38 2024', 'lhotse-version': '1.26.0', 'torch-version': '2.0.1+cu117', 'torch-cuda-available': True, 'torch-cuda-version': '11.7', 'python-version': '3.10', 'icefall-git-branch': 'zipformer_bf16', 'icefall-git-sha1': 'd1974bef-dirty', 'icefall-git-date': 'Thu Aug 8 10:10:22 2024', 'icefall-path': '/star-xy/softwares/icefall_development/icefall_bf16', 'k2-path': '/star-xy/softwares/pyenvs/k2_cuda11/k2_cuda11/lib/python3.10/site-packages/k2/__init__.py', 'lhotse-path': '/star-xy/softwares/pyenvs/k2_cuda11/k2_cuda11/lib/python3.10/site-packages/lhotse/__init__.py', 'hostname': 'de-74279-k2-train-2-1207150844-f49d8c4f4-c49d5', 'IP address': '10.177.22.19'}, 'epoch': 30, 'iter': 0, 'avg': 7, 'use_averaged_model': True, 'exp_dir': PosixPath('zipformer/exp-full-liri-full-bf16-md1000'), 'bpe_model': 'data/lang_bpe_500/bpe.model', 'lang_dir': PosixPath('data/lang_bpe_500'), 'decoding_method': 'greedy_search', 'beam_size': 4, 'beam': 20.0, 'ngram_lm_scale': 0.01, 'max_contexts': 8, 'max_states': 64, 'context_size': 2, 'max_sym_per_frame': 1, 'num_paths': 200, 'nbest_scale': 0.5, 'use_shallow_fusion': False, 'lm_type': 'rnn', 'lm_scale': 0.3, 'tokens_ngram': 2, 'backoff_id': 500, 'context_score': 2, 'context_file': '', 'num_encoder_layers': '2,2,3,4,3,2', 'downsampling_factor': '1,2,4,8,4,2', 'feedforward_dim': '512,768,1024,1536,1024,768', 'num_heads': '4,4,4,8,4,4', 'encoder_dim': '192,256,384,512,384,256', 'query_head_dim': '32', 'value_head_dim': '12', 'pos_head_dim': '4', 'pos_dim': 48, 'encoder_unmasked_dim': '192,192,256,256,256,192', 'cnn_module_kernel': '31,31,15,15,15,31', 'decoder_dim': 512, 'joiner_dim': 512, 'attention_decoder_dim': 512, 'attention_decoder_num_layers': 6, 'attention_decoder_attention_dim': 512, 'attention_decoder_num_heads': 8, 'attention_decoder_feedforward_dim': 2048, 'causal': False, 'chunk_size': '16,32,64,-1', 'left_context_frames': '64,128,256,-1', 'use_transducer': True, 'use_ctc': False, 'use_attention_decoder': False, 'full_libri': True, 'mini_libri': False, 'manifest_dir': PosixPath('data/fbank'), 'max_duration': 600, 'bucketing_sampler': True, 'num_buckets': 30, 'concatenate_cuts': False, 'duration_factor': 1.0, 'gap': 1.0, 'on_the_fly_feats': False, 'shuffle': True, 'drop_last': True, 'return_cuts': True, 'num_workers': 2, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'input_strategy': 'PrecomputedFeatures', 'lm_vocab_size': 500, 'lm_epoch': 7, 'lm_avg': 1, 'lm_exp_dir': None, 'rnn_lm_embedding_dim': 2048, 'rnn_lm_hidden_dim': 2048, 'rnn_lm_num_layers': 3, 'rnn_lm_tie_weights': True, 'transformer_lm_exp_dir': None, 'transformer_lm_dim_feedforward': 2048, 'transformer_lm_encoder_dim': 768, 'transformer_lm_embedding_dim': 768, 'transformer_lm_nhead': 8, 'transformer_lm_num_layers': 16, 'transformer_lm_tie_weights': True, 'res_dir': PosixPath('zipformer/exp-full-liri-full-bf16-md1000/greedy_search'), 'has_contexts': False, 'suffix': 'epoch-30-avg-7-context-2-max-sym-per-frame-1-use-averaged-model', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
4
+ 2024-08-08 10:13:10,200 INFO [decode.py:851] About to create model
5
+ 2024-08-08 10:13:10,953 INFO [decode.py:918] Calculating the averaged model over epoch range from 23 (excluded) to 30
6
+ 2024-08-08 10:13:15,869 INFO [decode.py:1012] Number of model parameters: 65549011
7
+ 2024-08-08 10:13:15,869 INFO [asr_datamodule.py:467] About to get test-clean cuts
8
+ 2024-08-08 10:13:15,874 INFO [asr_datamodule.py:474] About to get test-other cuts
9
+ 2024-08-08 10:13:19,051 INFO [decode.py:706] batch 0/?, cuts processed until now is 36
10
+ 2024-08-08 10:13:34,331 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([2.9358, 3.8801, 3.6067, 2.8585], device='cuda:0')
11
+ 2024-08-08 10:13:39,753 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([6.4703, 5.9052, 6.2376, 5.8522], device='cuda:0')
12
+ 2024-08-08 10:13:44,702 INFO [decode.py:722] The transcripts are stored in zipformer/exp-full-liri-full-bf16-md1000/greedy_search/recogs-test-clean-greedy_search-epoch-30-avg-7-context-2-max-sym-per-frame-1-use-averaged-model.txt
13
+ 2024-08-08 10:13:44,814 INFO [utils.py:657] [test-clean-greedy_search] %WER 2.39% [1257 / 52576, 141 ins, 102 del, 1014 sub ]
14
+ 2024-08-08 10:13:45,082 INFO [decode.py:735] Wrote detailed error stats to zipformer/exp-full-liri-full-bf16-md1000/greedy_search/errs-test-clean-greedy_search-epoch-30-avg-7-context-2-max-sym-per-frame-1-use-averaged-model.txt
15
+ 2024-08-08 10:13:45,086 INFO [decode.py:751]
16
+ For test-clean, WER of different settings are:
17
+ greedy_search 2.39 best for test-clean
18
+
19
+ 2024-08-08 10:13:46,330 INFO [decode.py:706] batch 0/?, cuts processed until now is 43
20
+ 2024-08-08 10:13:58,861 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([2.1300, 4.5340, 4.2289, 3.3131], device='cuda:0')
21
+ 2024-08-08 10:14:10,040 INFO [decode.py:722] The transcripts are stored in zipformer/exp-full-liri-full-bf16-md1000/greedy_search/recogs-test-other-greedy_search-epoch-30-avg-7-context-2-max-sym-per-frame-1-use-averaged-model.txt
22
+ 2024-08-08 10:14:10,161 INFO [utils.py:657] [test-other-greedy_search] %WER 5.47% [2861 / 52343, 283 ins, 246 del, 2332 sub ]
23
+ 2024-08-08 10:14:10,430 INFO [decode.py:735] Wrote detailed error stats to zipformer/exp-full-liri-full-bf16-md1000/greedy_search/errs-test-other-greedy_search-epoch-30-avg-7-context-2-max-sym-per-frame-1-use-averaged-model.txt
24
+ 2024-08-08 10:14:10,433 INFO [decode.py:751]
25
+ For test-other, WER of different settings are:
26
+ greedy_search 5.47 best for test-other
27
+
28
+ 2024-08-08 10:14:10,434 INFO [decode.py:1047] Done!
decoding_results/greedy_search/log-decode-epoch-30-avg-9-context-2-max-sym-per-frame-1-use-averaged-model-2024-08-08-10-12-02 ADDED
@@ -0,0 +1,25 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2024-08-08 10:12:02,841 INFO [decode.py:833] Decoding started
2
+ 2024-08-08 10:12:02,842 INFO [decode.py:839] Device: cuda:0
3
+ 2024-08-08 10:12:02,848 INFO [decode.py:849] {'best_train_loss': inf, 'best_valid_loss': inf, 'best_train_epoch': -1, 'best_valid_epoch': -1, 'batch_idx_train': 0, 'log_interval': 50, 'reset_interval': 200, 'valid_interval': 3000, 'feature_dim': 80, 'subsampling_factor': 4, 'ignore_id': -1, 'label_smoothing': 0.1, 'warm_step': 2000, 'env_info': {'k2-version': '1.24.4', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': 'ff1d435a8d3c4eaa15828a84a7240678a70539a7', 'k2-git-date': 'Fri Feb 23 01:48:38 2024', 'lhotse-version': '1.26.0', 'torch-version': '2.0.1+cu117', 'torch-cuda-available': True, 'torch-cuda-version': '11.7', 'python-version': '3.10', 'icefall-git-branch': 'zipformer_bf16', 'icefall-git-sha1': 'd1974bef-dirty', 'icefall-git-date': 'Thu Aug 8 10:10:22 2024', 'icefall-path': '/star-xy/softwares/icefall_development/icefall_bf16', 'k2-path': '/star-xy/softwares/pyenvs/k2_cuda11/k2_cuda11/lib/python3.10/site-packages/k2/__init__.py', 'lhotse-path': '/star-xy/softwares/pyenvs/k2_cuda11/k2_cuda11/lib/python3.10/site-packages/lhotse/__init__.py', 'hostname': 'de-74279-k2-train-2-1207150844-f49d8c4f4-c49d5', 'IP address': '10.177.22.19'}, 'epoch': 30, 'iter': 0, 'avg': 9, 'use_averaged_model': True, 'exp_dir': PosixPath('zipformer/exp-full-liri-full-bf16-md1000'), 'bpe_model': 'data/lang_bpe_500/bpe.model', 'lang_dir': PosixPath('data/lang_bpe_500'), 'decoding_method': 'greedy_search', 'beam_size': 4, 'beam': 20.0, 'ngram_lm_scale': 0.01, 'max_contexts': 8, 'max_states': 64, 'context_size': 2, 'max_sym_per_frame': 1, 'num_paths': 200, 'nbest_scale': 0.5, 'use_shallow_fusion': False, 'lm_type': 'rnn', 'lm_scale': 0.3, 'tokens_ngram': 2, 'backoff_id': 500, 'context_score': 2, 'context_file': '', 'num_encoder_layers': '2,2,3,4,3,2', 'downsampling_factor': '1,2,4,8,4,2', 'feedforward_dim': '512,768,1024,1536,1024,768', 'num_heads': '4,4,4,8,4,4', 'encoder_dim': '192,256,384,512,384,256', 'query_head_dim': '32', 'value_head_dim': '12', 'pos_head_dim': '4', 'pos_dim': 48, 'encoder_unmasked_dim': '192,192,256,256,256,192', 'cnn_module_kernel': '31,31,15,15,15,31', 'decoder_dim': 512, 'joiner_dim': 512, 'attention_decoder_dim': 512, 'attention_decoder_num_layers': 6, 'attention_decoder_attention_dim': 512, 'attention_decoder_num_heads': 8, 'attention_decoder_feedforward_dim': 2048, 'causal': False, 'chunk_size': '16,32,64,-1', 'left_context_frames': '64,128,256,-1', 'use_transducer': True, 'use_ctc': False, 'use_attention_decoder': False, 'full_libri': True, 'mini_libri': False, 'manifest_dir': PosixPath('data/fbank'), 'max_duration': 600, 'bucketing_sampler': True, 'num_buckets': 30, 'concatenate_cuts': False, 'duration_factor': 1.0, 'gap': 1.0, 'on_the_fly_feats': False, 'shuffle': True, 'drop_last': True, 'return_cuts': True, 'num_workers': 2, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'input_strategy': 'PrecomputedFeatures', 'lm_vocab_size': 500, 'lm_epoch': 7, 'lm_avg': 1, 'lm_exp_dir': None, 'rnn_lm_embedding_dim': 2048, 'rnn_lm_hidden_dim': 2048, 'rnn_lm_num_layers': 3, 'rnn_lm_tie_weights': True, 'transformer_lm_exp_dir': None, 'transformer_lm_dim_feedforward': 2048, 'transformer_lm_encoder_dim': 768, 'transformer_lm_embedding_dim': 768, 'transformer_lm_nhead': 8, 'transformer_lm_num_layers': 16, 'transformer_lm_tie_weights': True, 'res_dir': PosixPath('zipformer/exp-full-liri-full-bf16-md1000/greedy_search'), 'has_contexts': False, 'suffix': 'epoch-30-avg-9-context-2-max-sym-per-frame-1-use-averaged-model', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
4
+ 2024-08-08 10:12:02,848 INFO [decode.py:851] About to create model
5
+ 2024-08-08 10:12:03,572 INFO [decode.py:918] Calculating the averaged model over epoch range from 21 (excluded) to 30
6
+ 2024-08-08 10:12:08,825 INFO [decode.py:1012] Number of model parameters: 65549011
7
+ 2024-08-08 10:12:08,826 INFO [asr_datamodule.py:467] About to get test-clean cuts
8
+ 2024-08-08 10:12:08,834 INFO [asr_datamodule.py:474] About to get test-other cuts
9
+ 2024-08-08 10:12:11,932 INFO [decode.py:706] batch 0/?, cuts processed until now is 36
10
+ 2024-08-08 10:12:37,537 INFO [decode.py:722] The transcripts are stored in zipformer/exp-full-liri-full-bf16-md1000/greedy_search/recogs-test-clean-greedy_search-epoch-30-avg-9-context-2-max-sym-per-frame-1-use-averaged-model.txt
11
+ 2024-08-08 10:12:37,650 INFO [utils.py:657] [test-clean-greedy_search] %WER 2.39% [1259 / 52576, 145 ins, 105 del, 1009 sub ]
12
+ 2024-08-08 10:12:37,918 INFO [decode.py:735] Wrote detailed error stats to zipformer/exp-full-liri-full-bf16-md1000/greedy_search/errs-test-clean-greedy_search-epoch-30-avg-9-context-2-max-sym-per-frame-1-use-averaged-model.txt
13
+ 2024-08-08 10:12:37,922 INFO [decode.py:751]
14
+ For test-clean, WER of different settings are:
15
+ greedy_search 2.39 best for test-clean
16
+
17
+ 2024-08-08 10:12:39,182 INFO [decode.py:706] batch 0/?, cuts processed until now is 43
18
+ 2024-08-08 10:13:03,213 INFO [decode.py:722] The transcripts are stored in zipformer/exp-full-liri-full-bf16-md1000/greedy_search/recogs-test-other-greedy_search-epoch-30-avg-9-context-2-max-sym-per-frame-1-use-averaged-model.txt
19
+ 2024-08-08 10:13:03,335 INFO [utils.py:657] [test-other-greedy_search] %WER 5.42% [2836 / 52343, 282 ins, 245 del, 2309 sub ]
20
+ 2024-08-08 10:13:03,605 INFO [decode.py:735] Wrote detailed error stats to zipformer/exp-full-liri-full-bf16-md1000/greedy_search/errs-test-other-greedy_search-epoch-30-avg-9-context-2-max-sym-per-frame-1-use-averaged-model.txt
21
+ 2024-08-08 10:13:03,608 INFO [decode.py:751]
22
+ For test-other, WER of different settings are:
23
+ greedy_search 5.42 best for test-other
24
+
25
+ 2024-08-08 10:13:03,608 INFO [decode.py:1047] Done!
decoding_results/greedy_search/log-decode-epoch-40-avg-11-context-2-max-sym-per-frame-1-use-averaged-model-2024-08-13-10-45-34 ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2024-08-13 10:45:34,039 INFO [decode.py:833] Decoding started
2
+ 2024-08-13 10:45:34,040 INFO [decode.py:839] Device: cuda:0
3
+ 2024-08-13 10:45:34,045 INFO [decode.py:849] {'best_train_loss': inf, 'best_valid_loss': inf, 'best_train_epoch': -1, 'best_valid_epoch': -1, 'batch_idx_train': 0, 'log_interval': 50, 'reset_interval': 200, 'valid_interval': 3000, 'feature_dim': 80, 'subsampling_factor': 4, 'ignore_id': -1, 'label_smoothing': 0.1, 'warm_step': 2000, 'env_info': {'k2-version': '1.24.4', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': 'ff1d435a8d3c4eaa15828a84a7240678a70539a7', 'k2-git-date': 'Fri Feb 23 01:48:38 2024', 'lhotse-version': '1.26.0', 'torch-version': '2.0.1+cu117', 'torch-cuda-available': True, 'torch-cuda-version': '11.7', 'python-version': '3.10', 'icefall-git-branch': 'zipformer_bf16', 'icefall-git-sha1': 'a288d412-clean', 'icefall-git-date': 'Mon Aug 12 11:30:21 2024', 'icefall-path': '/star-xy/softwares/icefall_development/icefall_bf16', 'k2-path': '/star-xy/softwares/pyenvs/k2_cuda11/k2_cuda11/lib/python3.10/site-packages/k2/__init__.py', 'lhotse-path': '/star-xy/softwares/pyenvs/k2_cuda11/k2_cuda11/lib/python3.10/site-packages/lhotse/__init__.py', 'hostname': 'de-74279-k2-train-2-1207150844-f49d8c4f4-c49d5', 'IP address': '10.177.22.19'}, 'epoch': 40, 'iter': 0, 'avg': 11, 'use_averaged_model': True, 'exp_dir': PosixPath('zipformer/exp-full-libri-full-bf16-md1000'), 'bpe_model': 'data/lang_bpe_500/bpe.model', 'lang_dir': PosixPath('data/lang_bpe_500'), 'decoding_method': 'greedy_search', 'beam_size': 4, 'beam': 20.0, 'ngram_lm_scale': 0.01, 'max_contexts': 8, 'max_states': 64, 'context_size': 2, 'max_sym_per_frame': 1, 'num_paths': 200, 'nbest_scale': 0.5, 'use_shallow_fusion': False, 'lm_type': 'rnn', 'lm_scale': 0.3, 'tokens_ngram': 2, 'backoff_id': 500, 'context_score': 2, 'context_file': '', 'num_encoder_layers': '2,2,3,4,3,2', 'downsampling_factor': '1,2,4,8,4,2', 'feedforward_dim': '512,768,1024,1536,1024,768', 'num_heads': '4,4,4,8,4,4', 'encoder_dim': '192,256,384,512,384,256', 'query_head_dim': '32', 'value_head_dim': '12', 'pos_head_dim': '4', 'pos_dim': 48, 'encoder_unmasked_dim': '192,192,256,256,256,192', 'cnn_module_kernel': '31,31,15,15,15,31', 'decoder_dim': 512, 'joiner_dim': 512, 'attention_decoder_dim': 512, 'attention_decoder_num_layers': 6, 'attention_decoder_attention_dim': 512, 'attention_decoder_num_heads': 8, 'attention_decoder_feedforward_dim': 2048, 'causal': False, 'chunk_size': '16,32,64,-1', 'left_context_frames': '64,128,256,-1', 'use_transducer': True, 'use_ctc': False, 'use_attention_decoder': False, 'full_libri': True, 'mini_libri': False, 'manifest_dir': PosixPath('data/fbank'), 'max_duration': 600, 'bucketing_sampler': True, 'num_buckets': 30, 'concatenate_cuts': False, 'duration_factor': 1.0, 'gap': 1.0, 'on_the_fly_feats': False, 'shuffle': True, 'drop_last': True, 'return_cuts': True, 'num_workers': 2, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'input_strategy': 'PrecomputedFeatures', 'lm_vocab_size': 500, 'lm_epoch': 7, 'lm_avg': 1, 'lm_exp_dir': None, 'rnn_lm_embedding_dim': 2048, 'rnn_lm_hidden_dim': 2048, 'rnn_lm_num_layers': 3, 'rnn_lm_tie_weights': True, 'transformer_lm_exp_dir': None, 'transformer_lm_dim_feedforward': 2048, 'transformer_lm_encoder_dim': 768, 'transformer_lm_embedding_dim': 768, 'transformer_lm_nhead': 8, 'transformer_lm_num_layers': 16, 'transformer_lm_tie_weights': True, 'res_dir': PosixPath('zipformer/exp-full-libri-full-bf16-md1000/greedy_search'), 'has_contexts': False, 'suffix': 'epoch-40-avg-11-context-2-max-sym-per-frame-1-use-averaged-model', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
4
+ 2024-08-13 10:45:34,045 INFO [decode.py:851] About to create model
5
+ 2024-08-13 10:45:34,809 INFO [decode.py:918] Calculating the averaged model over epoch range from 29 (excluded) to 40
6
+ 2024-08-13 10:45:40,202 INFO [decode.py:1012] Number of model parameters: 65549011
7
+ 2024-08-13 10:45:40,202 INFO [asr_datamodule.py:467] About to get test-clean cuts
8
+ 2024-08-13 10:45:40,212 INFO [asr_datamodule.py:474] About to get test-other cuts
9
+ 2024-08-13 10:45:43,555 INFO [decode.py:706] batch 0/?, cuts processed until now is 36
10
+ 2024-08-13 10:45:43,677 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([3.8951, 3.0322, 3.3314, 3.2371], device='cuda:0')
11
+ 2024-08-13 10:46:09,265 INFO [decode.py:722] The transcripts are stored in zipformer/exp-full-libri-full-bf16-md1000/greedy_search/recogs-test-clean-greedy_search-epoch-40-avg-11-context-2-max-sym-per-frame-1-use-averaged-model.txt
12
+ 2024-08-13 10:46:09,376 INFO [utils.py:657] [test-clean-greedy_search] %WER 2.41% [1266 / 52576, 141 ins, 107 del, 1018 sub ]
13
+ 2024-08-13 10:46:09,648 INFO [decode.py:735] Wrote detailed error stats to zipformer/exp-full-libri-full-bf16-md1000/greedy_search/errs-test-clean-greedy_search-epoch-40-avg-11-context-2-max-sym-per-frame-1-use-averaged-model.txt
14
+ 2024-08-13 10:46:09,653 INFO [decode.py:751]
15
+ For test-clean, WER of different settings are:
16
+ greedy_search 2.41 best for test-clean
17
+
18
+ 2024-08-13 10:46:10,929 INFO [decode.py:706] batch 0/?, cuts processed until now is 43
19
+ 2024-08-13 10:46:34,865 INFO [decode.py:722] The transcripts are stored in zipformer/exp-full-libri-full-bf16-md1000/greedy_search/recogs-test-other-greedy_search-epoch-40-avg-11-context-2-max-sym-per-frame-1-use-averaged-model.txt
20
+ 2024-08-13 10:46:34,987 INFO [utils.py:657] [test-other-greedy_search] %WER 5.41% [2830 / 52343, 273 ins, 275 del, 2282 sub ]
21
+ 2024-08-13 10:46:35,255 INFO [decode.py:735] Wrote detailed error stats to zipformer/exp-full-libri-full-bf16-md1000/greedy_search/errs-test-other-greedy_search-epoch-40-avg-11-context-2-max-sym-per-frame-1-use-averaged-model.txt
22
+ 2024-08-13 10:46:35,258 INFO [decode.py:751]
23
+ For test-other, WER of different settings are:
24
+ greedy_search 5.41 best for test-other
25
+
26
+ 2024-08-13 10:46:35,258 INFO [decode.py:1047] Done!
decoding_results/greedy_search/log-decode-epoch-40-avg-13-context-2-max-sym-per-frame-1-use-averaged-model-2024-08-13-10-44-27 ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2024-08-13 10:44:27,327 INFO [decode.py:833] Decoding started
2
+ 2024-08-13 10:44:27,327 INFO [decode.py:839] Device: cuda:0
3
+ 2024-08-13 10:44:27,334 INFO [decode.py:849] {'best_train_loss': inf, 'best_valid_loss': inf, 'best_train_epoch': -1, 'best_valid_epoch': -1, 'batch_idx_train': 0, 'log_interval': 50, 'reset_interval': 200, 'valid_interval': 3000, 'feature_dim': 80, 'subsampling_factor': 4, 'ignore_id': -1, 'label_smoothing': 0.1, 'warm_step': 2000, 'env_info': {'k2-version': '1.24.4', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': 'ff1d435a8d3c4eaa15828a84a7240678a70539a7', 'k2-git-date': 'Fri Feb 23 01:48:38 2024', 'lhotse-version': '1.26.0', 'torch-version': '2.0.1+cu117', 'torch-cuda-available': True, 'torch-cuda-version': '11.7', 'python-version': '3.10', 'icefall-git-branch': 'zipformer_bf16', 'icefall-git-sha1': 'a288d412-clean', 'icefall-git-date': 'Mon Aug 12 11:30:21 2024', 'icefall-path': '/star-xy/softwares/icefall_development/icefall_bf16', 'k2-path': '/star-xy/softwares/pyenvs/k2_cuda11/k2_cuda11/lib/python3.10/site-packages/k2/__init__.py', 'lhotse-path': '/star-xy/softwares/pyenvs/k2_cuda11/k2_cuda11/lib/python3.10/site-packages/lhotse/__init__.py', 'hostname': 'de-74279-k2-train-2-1207150844-f49d8c4f4-c49d5', 'IP address': '10.177.22.19'}, 'epoch': 40, 'iter': 0, 'avg': 13, 'use_averaged_model': True, 'exp_dir': PosixPath('zipformer/exp-full-libri-full-bf16-md1000'), 'bpe_model': 'data/lang_bpe_500/bpe.model', 'lang_dir': PosixPath('data/lang_bpe_500'), 'decoding_method': 'greedy_search', 'beam_size': 4, 'beam': 20.0, 'ngram_lm_scale': 0.01, 'max_contexts': 8, 'max_states': 64, 'context_size': 2, 'max_sym_per_frame': 1, 'num_paths': 200, 'nbest_scale': 0.5, 'use_shallow_fusion': False, 'lm_type': 'rnn', 'lm_scale': 0.3, 'tokens_ngram': 2, 'backoff_id': 500, 'context_score': 2, 'context_file': '', 'num_encoder_layers': '2,2,3,4,3,2', 'downsampling_factor': '1,2,4,8,4,2', 'feedforward_dim': '512,768,1024,1536,1024,768', 'num_heads': '4,4,4,8,4,4', 'encoder_dim': '192,256,384,512,384,256', 'query_head_dim': '32', 'value_head_dim': '12', 'pos_head_dim': '4', 'pos_dim': 48, 'encoder_unmasked_dim': '192,192,256,256,256,192', 'cnn_module_kernel': '31,31,15,15,15,31', 'decoder_dim': 512, 'joiner_dim': 512, 'attention_decoder_dim': 512, 'attention_decoder_num_layers': 6, 'attention_decoder_attention_dim': 512, 'attention_decoder_num_heads': 8, 'attention_decoder_feedforward_dim': 2048, 'causal': False, 'chunk_size': '16,32,64,-1', 'left_context_frames': '64,128,256,-1', 'use_transducer': True, 'use_ctc': False, 'use_attention_decoder': False, 'full_libri': True, 'mini_libri': False, 'manifest_dir': PosixPath('data/fbank'), 'max_duration': 600, 'bucketing_sampler': True, 'num_buckets': 30, 'concatenate_cuts': False, 'duration_factor': 1.0, 'gap': 1.0, 'on_the_fly_feats': False, 'shuffle': True, 'drop_last': True, 'return_cuts': True, 'num_workers': 2, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'input_strategy': 'PrecomputedFeatures', 'lm_vocab_size': 500, 'lm_epoch': 7, 'lm_avg': 1, 'lm_exp_dir': None, 'rnn_lm_embedding_dim': 2048, 'rnn_lm_hidden_dim': 2048, 'rnn_lm_num_layers': 3, 'rnn_lm_tie_weights': True, 'transformer_lm_exp_dir': None, 'transformer_lm_dim_feedforward': 2048, 'transformer_lm_encoder_dim': 768, 'transformer_lm_embedding_dim': 768, 'transformer_lm_nhead': 8, 'transformer_lm_num_layers': 16, 'transformer_lm_tie_weights': True, 'res_dir': PosixPath('zipformer/exp-full-libri-full-bf16-md1000/greedy_search'), 'has_contexts': False, 'suffix': 'epoch-40-avg-13-context-2-max-sym-per-frame-1-use-averaged-model', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
4
+ 2024-08-13 10:44:27,334 INFO [decode.py:851] About to create model
5
+ 2024-08-13 10:44:28,056 INFO [decode.py:918] Calculating the averaged model over epoch range from 27 (excluded) to 40
6
+ 2024-08-13 10:44:33,483 INFO [decode.py:1012] Number of model parameters: 65549011
7
+ 2024-08-13 10:44:33,483 INFO [asr_datamodule.py:467] About to get test-clean cuts
8
+ 2024-08-13 10:44:33,493 INFO [asr_datamodule.py:474] About to get test-other cuts
9
+ 2024-08-13 10:44:36,667 INFO [decode.py:706] batch 0/?, cuts processed until now is 36
10
+ 2024-08-13 10:45:01,929 INFO [decode.py:722] The transcripts are stored in zipformer/exp-full-libri-full-bf16-md1000/greedy_search/recogs-test-clean-greedy_search-epoch-40-avg-13-context-2-max-sym-per-frame-1-use-averaged-model.txt
11
+ 2024-08-13 10:45:02,049 INFO [utils.py:657] [test-clean-greedy_search] %WER 2.40% [1264 / 52576, 141 ins, 105 del, 1018 sub ]
12
+ 2024-08-13 10:45:02,312 INFO [decode.py:735] Wrote detailed error stats to zipformer/exp-full-libri-full-bf16-md1000/greedy_search/errs-test-clean-greedy_search-epoch-40-avg-13-context-2-max-sym-per-frame-1-use-averaged-model.txt
13
+ 2024-08-13 10:45:02,315 INFO [decode.py:751]
14
+ For test-clean, WER of different settings are:
15
+ greedy_search 2.4 best for test-clean
16
+
17
+ 2024-08-13 10:45:03,524 INFO [decode.py:706] batch 0/?, cuts processed until now is 43
18
+ 2024-08-13 10:45:09,888 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([5.4063, 5.6510, 4.5869, 3.7669], device='cuda:0')
19
+ 2024-08-13 10:45:26,884 INFO [decode.py:722] The transcripts are stored in zipformer/exp-full-libri-full-bf16-md1000/greedy_search/recogs-test-other-greedy_search-epoch-40-avg-13-context-2-max-sym-per-frame-1-use-averaged-model.txt
20
+ 2024-08-13 10:45:27,000 INFO [utils.py:657] [test-other-greedy_search] %WER 5.36% [2807 / 52343, 269 ins, 263 del, 2275 sub ]
21
+ 2024-08-13 10:45:27,268 INFO [decode.py:735] Wrote detailed error stats to zipformer/exp-full-libri-full-bf16-md1000/greedy_search/errs-test-other-greedy_search-epoch-40-avg-13-context-2-max-sym-per-frame-1-use-averaged-model.txt
22
+ 2024-08-13 10:45:27,271 INFO [decode.py:751]
23
+ For test-other, WER of different settings are:
24
+ greedy_search 5.36 best for test-other
25
+
26
+ 2024-08-13 10:45:27,271 INFO [decode.py:1047] Done!
decoding_results/greedy_search/log-decode-epoch-40-avg-15-context-2-max-sym-per-frame-1-use-averaged-model-2024-08-13-10-43-19 ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2024-08-13 10:43:19,424 INFO [decode.py:833] Decoding started
2
+ 2024-08-13 10:43:19,424 INFO [decode.py:839] Device: cuda:0
3
+ 2024-08-13 10:43:19,427 INFO [decode.py:849] {'best_train_loss': inf, 'best_valid_loss': inf, 'best_train_epoch': -1, 'best_valid_epoch': -1, 'batch_idx_train': 0, 'log_interval': 50, 'reset_interval': 200, 'valid_interval': 3000, 'feature_dim': 80, 'subsampling_factor': 4, 'ignore_id': -1, 'label_smoothing': 0.1, 'warm_step': 2000, 'env_info': {'k2-version': '1.24.4', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': 'ff1d435a8d3c4eaa15828a84a7240678a70539a7', 'k2-git-date': 'Fri Feb 23 01:48:38 2024', 'lhotse-version': '1.26.0', 'torch-version': '2.0.1+cu117', 'torch-cuda-available': True, 'torch-cuda-version': '11.7', 'python-version': '3.10', 'icefall-git-branch': 'zipformer_bf16', 'icefall-git-sha1': 'a288d412-clean', 'icefall-git-date': 'Mon Aug 12 11:30:21 2024', 'icefall-path': '/star-xy/softwares/icefall_development/icefall_bf16', 'k2-path': '/star-xy/softwares/pyenvs/k2_cuda11/k2_cuda11/lib/python3.10/site-packages/k2/__init__.py', 'lhotse-path': '/star-xy/softwares/pyenvs/k2_cuda11/k2_cuda11/lib/python3.10/site-packages/lhotse/__init__.py', 'hostname': 'de-74279-k2-train-2-1207150844-f49d8c4f4-c49d5', 'IP address': '10.177.22.19'}, 'epoch': 40, 'iter': 0, 'avg': 15, 'use_averaged_model': True, 'exp_dir': PosixPath('zipformer/exp-full-libri-full-bf16-md1000'), 'bpe_model': 'data/lang_bpe_500/bpe.model', 'lang_dir': PosixPath('data/lang_bpe_500'), 'decoding_method': 'greedy_search', 'beam_size': 4, 'beam': 20.0, 'ngram_lm_scale': 0.01, 'max_contexts': 8, 'max_states': 64, 'context_size': 2, 'max_sym_per_frame': 1, 'num_paths': 200, 'nbest_scale': 0.5, 'use_shallow_fusion': False, 'lm_type': 'rnn', 'lm_scale': 0.3, 'tokens_ngram': 2, 'backoff_id': 500, 'context_score': 2, 'context_file': '', 'num_encoder_layers': '2,2,3,4,3,2', 'downsampling_factor': '1,2,4,8,4,2', 'feedforward_dim': '512,768,1024,1536,1024,768', 'num_heads': '4,4,4,8,4,4', 'encoder_dim': '192,256,384,512,384,256', 'query_head_dim': '32', 'value_head_dim': '12', 'pos_head_dim': '4', 'pos_dim': 48, 'encoder_unmasked_dim': '192,192,256,256,256,192', 'cnn_module_kernel': '31,31,15,15,15,31', 'decoder_dim': 512, 'joiner_dim': 512, 'attention_decoder_dim': 512, 'attention_decoder_num_layers': 6, 'attention_decoder_attention_dim': 512, 'attention_decoder_num_heads': 8, 'attention_decoder_feedforward_dim': 2048, 'causal': False, 'chunk_size': '16,32,64,-1', 'left_context_frames': '64,128,256,-1', 'use_transducer': True, 'use_ctc': False, 'use_attention_decoder': False, 'full_libri': True, 'mini_libri': False, 'manifest_dir': PosixPath('data/fbank'), 'max_duration': 600, 'bucketing_sampler': True, 'num_buckets': 30, 'concatenate_cuts': False, 'duration_factor': 1.0, 'gap': 1.0, 'on_the_fly_feats': False, 'shuffle': True, 'drop_last': True, 'return_cuts': True, 'num_workers': 2, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'input_strategy': 'PrecomputedFeatures', 'lm_vocab_size': 500, 'lm_epoch': 7, 'lm_avg': 1, 'lm_exp_dir': None, 'rnn_lm_embedding_dim': 2048, 'rnn_lm_hidden_dim': 2048, 'rnn_lm_num_layers': 3, 'rnn_lm_tie_weights': True, 'transformer_lm_exp_dir': None, 'transformer_lm_dim_feedforward': 2048, 'transformer_lm_encoder_dim': 768, 'transformer_lm_embedding_dim': 768, 'transformer_lm_nhead': 8, 'transformer_lm_num_layers': 16, 'transformer_lm_tie_weights': True, 'res_dir': PosixPath('zipformer/exp-full-libri-full-bf16-md1000/greedy_search'), 'has_contexts': False, 'suffix': 'epoch-40-avg-15-context-2-max-sym-per-frame-1-use-averaged-model', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
4
+ 2024-08-13 10:43:19,428 INFO [decode.py:851] About to create model
5
+ 2024-08-13 10:43:20,166 INFO [decode.py:918] Calculating the averaged model over epoch range from 25 (excluded) to 40
6
+ 2024-08-13 10:43:26,038 INFO [decode.py:1012] Number of model parameters: 65549011
7
+ 2024-08-13 10:43:26,038 INFO [asr_datamodule.py:467] About to get test-clean cuts
8
+ 2024-08-13 10:43:26,051 INFO [asr_datamodule.py:474] About to get test-other cuts
9
+ 2024-08-13 10:43:29,362 INFO [decode.py:706] batch 0/?, cuts processed until now is 36
10
+ 2024-08-13 10:43:45,923 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([2.2681, 4.7385, 4.4889, 3.5183], device='cuda:0')
11
+ 2024-08-13 10:43:54,940 INFO [decode.py:722] The transcripts are stored in zipformer/exp-full-libri-full-bf16-md1000/greedy_search/recogs-test-clean-greedy_search-epoch-40-avg-15-context-2-max-sym-per-frame-1-use-averaged-model.txt
12
+ 2024-08-13 10:43:55,060 INFO [utils.py:657] [test-clean-greedy_search] %WER 2.39% [1258 / 52576, 135 ins, 108 del, 1015 sub ]
13
+ 2024-08-13 10:43:55,374 INFO [decode.py:735] Wrote detailed error stats to zipformer/exp-full-libri-full-bf16-md1000/greedy_search/errs-test-clean-greedy_search-epoch-40-avg-15-context-2-max-sym-per-frame-1-use-averaged-model.txt
14
+ 2024-08-13 10:43:55,377 INFO [decode.py:751]
15
+ For test-clean, WER of different settings are:
16
+ greedy_search 2.39 best for test-clean
17
+
18
+ 2024-08-13 10:43:56,709 INFO [decode.py:706] batch 0/?, cuts processed until now is 43
19
+ 2024-08-13 10:44:20,248 INFO [decode.py:722] The transcripts are stored in zipformer/exp-full-libri-full-bf16-md1000/greedy_search/recogs-test-other-greedy_search-epoch-40-avg-15-context-2-max-sym-per-frame-1-use-averaged-model.txt
20
+ 2024-08-13 10:44:20,362 INFO [utils.py:657] [test-other-greedy_search] %WER 5.42% [2835 / 52343, 279 ins, 264 del, 2292 sub ]
21
+ 2024-08-13 10:44:20,630 INFO [decode.py:735] Wrote detailed error stats to zipformer/exp-full-libri-full-bf16-md1000/greedy_search/errs-test-other-greedy_search-epoch-40-avg-15-context-2-max-sym-per-frame-1-use-averaged-model.txt
22
+ 2024-08-13 10:44:20,633 INFO [decode.py:751]
23
+ For test-other, WER of different settings are:
24
+ greedy_search 5.42 best for test-other
25
+
26
+ 2024-08-13 10:44:20,634 INFO [decode.py:1047] Done!
decoding_results/greedy_search/log-decode-epoch-40-avg-17-context-2-max-sym-per-frame-1-use-averaged-model-2024-08-13-10-42-13 ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2024-08-13 10:42:13,088 INFO [decode.py:833] Decoding started
2
+ 2024-08-13 10:42:13,088 INFO [decode.py:839] Device: cuda:0
3
+ 2024-08-13 10:42:13,092 INFO [decode.py:849] {'best_train_loss': inf, 'best_valid_loss': inf, 'best_train_epoch': -1, 'best_valid_epoch': -1, 'batch_idx_train': 0, 'log_interval': 50, 'reset_interval': 200, 'valid_interval': 3000, 'feature_dim': 80, 'subsampling_factor': 4, 'ignore_id': -1, 'label_smoothing': 0.1, 'warm_step': 2000, 'env_info': {'k2-version': '1.24.4', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': 'ff1d435a8d3c4eaa15828a84a7240678a70539a7', 'k2-git-date': 'Fri Feb 23 01:48:38 2024', 'lhotse-version': '1.26.0', 'torch-version': '2.0.1+cu117', 'torch-cuda-available': True, 'torch-cuda-version': '11.7', 'python-version': '3.10', 'icefall-git-branch': 'zipformer_bf16', 'icefall-git-sha1': 'a288d412-clean', 'icefall-git-date': 'Mon Aug 12 11:30:21 2024', 'icefall-path': '/star-xy/softwares/icefall_development/icefall_bf16', 'k2-path': '/star-xy/softwares/pyenvs/k2_cuda11/k2_cuda11/lib/python3.10/site-packages/k2/__init__.py', 'lhotse-path': '/star-xy/softwares/pyenvs/k2_cuda11/k2_cuda11/lib/python3.10/site-packages/lhotse/__init__.py', 'hostname': 'de-74279-k2-train-2-1207150844-f49d8c4f4-c49d5', 'IP address': '10.177.22.19'}, 'epoch': 40, 'iter': 0, 'avg': 17, 'use_averaged_model': True, 'exp_dir': PosixPath('zipformer/exp-full-libri-full-bf16-md1000'), 'bpe_model': 'data/lang_bpe_500/bpe.model', 'lang_dir': PosixPath('data/lang_bpe_500'), 'decoding_method': 'greedy_search', 'beam_size': 4, 'beam': 20.0, 'ngram_lm_scale': 0.01, 'max_contexts': 8, 'max_states': 64, 'context_size': 2, 'max_sym_per_frame': 1, 'num_paths': 200, 'nbest_scale': 0.5, 'use_shallow_fusion': False, 'lm_type': 'rnn', 'lm_scale': 0.3, 'tokens_ngram': 2, 'backoff_id': 500, 'context_score': 2, 'context_file': '', 'num_encoder_layers': '2,2,3,4,3,2', 'downsampling_factor': '1,2,4,8,4,2', 'feedforward_dim': '512,768,1024,1536,1024,768', 'num_heads': '4,4,4,8,4,4', 'encoder_dim': '192,256,384,512,384,256', 'query_head_dim': '32', 'value_head_dim': '12', 'pos_head_dim': '4', 'pos_dim': 48, 'encoder_unmasked_dim': '192,192,256,256,256,192', 'cnn_module_kernel': '31,31,15,15,15,31', 'decoder_dim': 512, 'joiner_dim': 512, 'attention_decoder_dim': 512, 'attention_decoder_num_layers': 6, 'attention_decoder_attention_dim': 512, 'attention_decoder_num_heads': 8, 'attention_decoder_feedforward_dim': 2048, 'causal': False, 'chunk_size': '16,32,64,-1', 'left_context_frames': '64,128,256,-1', 'use_transducer': True, 'use_ctc': False, 'use_attention_decoder': False, 'full_libri': True, 'mini_libri': False, 'manifest_dir': PosixPath('data/fbank'), 'max_duration': 600, 'bucketing_sampler': True, 'num_buckets': 30, 'concatenate_cuts': False, 'duration_factor': 1.0, 'gap': 1.0, 'on_the_fly_feats': False, 'shuffle': True, 'drop_last': True, 'return_cuts': True, 'num_workers': 2, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'input_strategy': 'PrecomputedFeatures', 'lm_vocab_size': 500, 'lm_epoch': 7, 'lm_avg': 1, 'lm_exp_dir': None, 'rnn_lm_embedding_dim': 2048, 'rnn_lm_hidden_dim': 2048, 'rnn_lm_num_layers': 3, 'rnn_lm_tie_weights': True, 'transformer_lm_exp_dir': None, 'transformer_lm_dim_feedforward': 2048, 'transformer_lm_encoder_dim': 768, 'transformer_lm_embedding_dim': 768, 'transformer_lm_nhead': 8, 'transformer_lm_num_layers': 16, 'transformer_lm_tie_weights': True, 'res_dir': PosixPath('zipformer/exp-full-libri-full-bf16-md1000/greedy_search'), 'has_contexts': False, 'suffix': 'epoch-40-avg-17-context-2-max-sym-per-frame-1-use-averaged-model', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
4
+ 2024-08-13 10:42:13,092 INFO [decode.py:851] About to create model
5
+ 2024-08-13 10:42:13,809 INFO [decode.py:918] Calculating the averaged model over epoch range from 23 (excluded) to 40
6
+ 2024-08-13 10:42:18,591 INFO [decode.py:1012] Number of model parameters: 65549011
7
+ 2024-08-13 10:42:18,591 INFO [asr_datamodule.py:467] About to get test-clean cuts
8
+ 2024-08-13 10:42:18,600 INFO [asr_datamodule.py:474] About to get test-other cuts
9
+ 2024-08-13 10:42:21,742 INFO [decode.py:706] batch 0/?, cuts processed until now is 36
10
+ 2024-08-13 10:42:47,055 INFO [decode.py:722] The transcripts are stored in zipformer/exp-full-libri-full-bf16-md1000/greedy_search/recogs-test-clean-greedy_search-epoch-40-avg-17-context-2-max-sym-per-frame-1-use-averaged-model.txt
11
+ 2024-08-13 10:42:47,193 INFO [utils.py:657] [test-clean-greedy_search] %WER 2.39% [1257 / 52576, 138 ins, 107 del, 1012 sub ]
12
+ 2024-08-13 10:42:47,570 INFO [decode.py:735] Wrote detailed error stats to zipformer/exp-full-libri-full-bf16-md1000/greedy_search/errs-test-clean-greedy_search-epoch-40-avg-17-context-2-max-sym-per-frame-1-use-averaged-model.txt
13
+ 2024-08-13 10:42:47,574 INFO [decode.py:751]
14
+ For test-clean, WER of different settings are:
15
+ greedy_search 2.39 best for test-clean
16
+
17
+ 2024-08-13 10:42:48,866 INFO [decode.py:706] batch 0/?, cuts processed until now is 43
18
+ 2024-08-13 10:43:00,718 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([3.5879, 4.5445, 4.5348, 2.5455], device='cuda:0')
19
+ 2024-08-13 10:43:12,267 INFO [decode.py:722] The transcripts are stored in zipformer/exp-full-libri-full-bf16-md1000/greedy_search/recogs-test-other-greedy_search-epoch-40-avg-17-context-2-max-sym-per-frame-1-use-averaged-model.txt
20
+ 2024-08-13 10:43:12,384 INFO [utils.py:657] [test-other-greedy_search] %WER 5.44% [2848 / 52343, 286 ins, 264 del, 2298 sub ]
21
+ 2024-08-13 10:43:12,665 INFO [decode.py:735] Wrote detailed error stats to zipformer/exp-full-libri-full-bf16-md1000/greedy_search/errs-test-other-greedy_search-epoch-40-avg-17-context-2-max-sym-per-frame-1-use-averaged-model.txt
22
+ 2024-08-13 10:43:12,668 INFO [decode.py:751]
23
+ For test-other, WER of different settings are:
24
+ greedy_search 5.44 best for test-other
25
+
26
+ 2024-08-13 10:43:12,669 INFO [decode.py:1047] Done!
decoding_results/greedy_search/log-decode-epoch-40-avg-19-context-2-max-sym-per-frame-1-use-averaged-model-2024-08-13-10-40-58 ADDED
@@ -0,0 +1,25 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2024-08-13 10:40:58,910 INFO [decode.py:833] Decoding started
2
+ 2024-08-13 10:40:58,910 INFO [decode.py:839] Device: cuda:0
3
+ 2024-08-13 10:40:58,917 INFO [decode.py:849] {'best_train_loss': inf, 'best_valid_loss': inf, 'best_train_epoch': -1, 'best_valid_epoch': -1, 'batch_idx_train': 0, 'log_interval': 50, 'reset_interval': 200, 'valid_interval': 3000, 'feature_dim': 80, 'subsampling_factor': 4, 'ignore_id': -1, 'label_smoothing': 0.1, 'warm_step': 2000, 'env_info': {'k2-version': '1.24.4', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': 'ff1d435a8d3c4eaa15828a84a7240678a70539a7', 'k2-git-date': 'Fri Feb 23 01:48:38 2024', 'lhotse-version': '1.26.0', 'torch-version': '2.0.1+cu117', 'torch-cuda-available': True, 'torch-cuda-version': '11.7', 'python-version': '3.10', 'icefall-git-branch': 'zipformer_bf16', 'icefall-git-sha1': 'a288d412-clean', 'icefall-git-date': 'Mon Aug 12 11:30:21 2024', 'icefall-path': '/star-xy/softwares/icefall_development/icefall_bf16', 'k2-path': '/star-xy/softwares/pyenvs/k2_cuda11/k2_cuda11/lib/python3.10/site-packages/k2/__init__.py', 'lhotse-path': '/star-xy/softwares/pyenvs/k2_cuda11/k2_cuda11/lib/python3.10/site-packages/lhotse/__init__.py', 'hostname': 'de-74279-k2-train-2-1207150844-f49d8c4f4-c49d5', 'IP address': '10.177.22.19'}, 'epoch': 40, 'iter': 0, 'avg': 19, 'use_averaged_model': True, 'exp_dir': PosixPath('zipformer/exp-full-libri-full-bf16-md1000'), 'bpe_model': 'data/lang_bpe_500/bpe.model', 'lang_dir': PosixPath('data/lang_bpe_500'), 'decoding_method': 'greedy_search', 'beam_size': 4, 'beam': 20.0, 'ngram_lm_scale': 0.01, 'max_contexts': 8, 'max_states': 64, 'context_size': 2, 'max_sym_per_frame': 1, 'num_paths': 200, 'nbest_scale': 0.5, 'use_shallow_fusion': False, 'lm_type': 'rnn', 'lm_scale': 0.3, 'tokens_ngram': 2, 'backoff_id': 500, 'context_score': 2, 'context_file': '', 'num_encoder_layers': '2,2,3,4,3,2', 'downsampling_factor': '1,2,4,8,4,2', 'feedforward_dim': '512,768,1024,1536,1024,768', 'num_heads': '4,4,4,8,4,4', 'encoder_dim': '192,256,384,512,384,256', 'query_head_dim': '32', 'value_head_dim': '12', 'pos_head_dim': '4', 'pos_dim': 48, 'encoder_unmasked_dim': '192,192,256,256,256,192', 'cnn_module_kernel': '31,31,15,15,15,31', 'decoder_dim': 512, 'joiner_dim': 512, 'attention_decoder_dim': 512, 'attention_decoder_num_layers': 6, 'attention_decoder_attention_dim': 512, 'attention_decoder_num_heads': 8, 'attention_decoder_feedforward_dim': 2048, 'causal': False, 'chunk_size': '16,32,64,-1', 'left_context_frames': '64,128,256,-1', 'use_transducer': True, 'use_ctc': False, 'use_attention_decoder': False, 'full_libri': True, 'mini_libri': False, 'manifest_dir': PosixPath('data/fbank'), 'max_duration': 600, 'bucketing_sampler': True, 'num_buckets': 30, 'concatenate_cuts': False, 'duration_factor': 1.0, 'gap': 1.0, 'on_the_fly_feats': False, 'shuffle': True, 'drop_last': True, 'return_cuts': True, 'num_workers': 2, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'input_strategy': 'PrecomputedFeatures', 'lm_vocab_size': 500, 'lm_epoch': 7, 'lm_avg': 1, 'lm_exp_dir': None, 'rnn_lm_embedding_dim': 2048, 'rnn_lm_hidden_dim': 2048, 'rnn_lm_num_layers': 3, 'rnn_lm_tie_weights': True, 'transformer_lm_exp_dir': None, 'transformer_lm_dim_feedforward': 2048, 'transformer_lm_encoder_dim': 768, 'transformer_lm_embedding_dim': 768, 'transformer_lm_nhead': 8, 'transformer_lm_num_layers': 16, 'transformer_lm_tie_weights': True, 'res_dir': PosixPath('zipformer/exp-full-libri-full-bf16-md1000/greedy_search'), 'has_contexts': False, 'suffix': 'epoch-40-avg-19-context-2-max-sym-per-frame-1-use-averaged-model', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
4
+ 2024-08-13 10:40:58,918 INFO [decode.py:851] About to create model
5
+ 2024-08-13 10:40:59,703 INFO [decode.py:918] Calculating the averaged model over epoch range from 21 (excluded) to 40
6
+ 2024-08-13 10:41:06,624 INFO [decode.py:1012] Number of model parameters: 65549011
7
+ 2024-08-13 10:41:06,625 INFO [asr_datamodule.py:467] About to get test-clean cuts
8
+ 2024-08-13 10:41:06,644 INFO [asr_datamodule.py:474] About to get test-other cuts
9
+ 2024-08-13 10:41:11,474 INFO [decode.py:706] batch 0/?, cuts processed until now is 36
10
+ 2024-08-13 10:41:38,833 INFO [decode.py:722] The transcripts are stored in zipformer/exp-full-libri-full-bf16-md1000/greedy_search/recogs-test-clean-greedy_search-epoch-40-avg-19-context-2-max-sym-per-frame-1-use-averaged-model.txt
11
+ 2024-08-13 10:41:38,953 INFO [utils.py:657] [test-clean-greedy_search] %WER 2.39% [1254 / 52576, 136 ins, 105 del, 1013 sub ]
12
+ 2024-08-13 10:41:39,217 INFO [decode.py:735] Wrote detailed error stats to zipformer/exp-full-libri-full-bf16-md1000/greedy_search/errs-test-clean-greedy_search-epoch-40-avg-19-context-2-max-sym-per-frame-1-use-averaged-model.txt
13
+ 2024-08-13 10:41:39,221 INFO [decode.py:751]
14
+ For test-clean, WER of different settings are:
15
+ greedy_search 2.39 best for test-clean
16
+
17
+ 2024-08-13 10:41:40,479 INFO [decode.py:706] batch 0/?, cuts processed until now is 43
18
+ 2024-08-13 10:42:05,991 INFO [decode.py:722] The transcripts are stored in zipformer/exp-full-libri-full-bf16-md1000/greedy_search/recogs-test-other-greedy_search-epoch-40-avg-19-context-2-max-sym-per-frame-1-use-averaged-model.txt
19
+ 2024-08-13 10:42:06,107 INFO [utils.py:657] [test-other-greedy_search] %WER 5.38% [2814 / 52343, 281 ins, 254 del, 2279 sub ]
20
+ 2024-08-13 10:42:06,374 INFO [decode.py:735] Wrote detailed error stats to zipformer/exp-full-libri-full-bf16-md1000/greedy_search/errs-test-other-greedy_search-epoch-40-avg-19-context-2-max-sym-per-frame-1-use-averaged-model.txt
21
+ 2024-08-13 10:42:06,378 INFO [decode.py:751]
22
+ For test-other, WER of different settings are:
23
+ greedy_search 5.38 best for test-other
24
+
25
+ 2024-08-13 10:42:06,378 INFO [decode.py:1047] Done!
decoding_results/greedy_search/log-decode-epoch-40-avg-7-context-2-max-sym-per-frame-1-use-averaged-model-2024-08-13-10-47-51 ADDED
@@ -0,0 +1,27 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2024-08-13 10:47:51,660 INFO [decode.py:833] Decoding started
2
+ 2024-08-13 10:47:51,661 INFO [decode.py:839] Device: cuda:0
3
+ 2024-08-13 10:47:51,665 INFO [decode.py:849] {'best_train_loss': inf, 'best_valid_loss': inf, 'best_train_epoch': -1, 'best_valid_epoch': -1, 'batch_idx_train': 0, 'log_interval': 50, 'reset_interval': 200, 'valid_interval': 3000, 'feature_dim': 80, 'subsampling_factor': 4, 'ignore_id': -1, 'label_smoothing': 0.1, 'warm_step': 2000, 'env_info': {'k2-version': '1.24.4', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': 'ff1d435a8d3c4eaa15828a84a7240678a70539a7', 'k2-git-date': 'Fri Feb 23 01:48:38 2024', 'lhotse-version': '1.26.0', 'torch-version': '2.0.1+cu117', 'torch-cuda-available': True, 'torch-cuda-version': '11.7', 'python-version': '3.10', 'icefall-git-branch': 'zipformer_bf16', 'icefall-git-sha1': 'a288d412-clean', 'icefall-git-date': 'Mon Aug 12 11:30:21 2024', 'icefall-path': '/star-xy/softwares/icefall_development/icefall_bf16', 'k2-path': '/star-xy/softwares/pyenvs/k2_cuda11/k2_cuda11/lib/python3.10/site-packages/k2/__init__.py', 'lhotse-path': '/star-xy/softwares/pyenvs/k2_cuda11/k2_cuda11/lib/python3.10/site-packages/lhotse/__init__.py', 'hostname': 'de-74279-k2-train-2-1207150844-f49d8c4f4-c49d5', 'IP address': '10.177.22.19'}, 'epoch': 40, 'iter': 0, 'avg': 7, 'use_averaged_model': True, 'exp_dir': PosixPath('zipformer/exp-full-libri-full-bf16-md1000'), 'bpe_model': 'data/lang_bpe_500/bpe.model', 'lang_dir': PosixPath('data/lang_bpe_500'), 'decoding_method': 'greedy_search', 'beam_size': 4, 'beam': 20.0, 'ngram_lm_scale': 0.01, 'max_contexts': 8, 'max_states': 64, 'context_size': 2, 'max_sym_per_frame': 1, 'num_paths': 200, 'nbest_scale': 0.5, 'use_shallow_fusion': False, 'lm_type': 'rnn', 'lm_scale': 0.3, 'tokens_ngram': 2, 'backoff_id': 500, 'context_score': 2, 'context_file': '', 'num_encoder_layers': '2,2,3,4,3,2', 'downsampling_factor': '1,2,4,8,4,2', 'feedforward_dim': '512,768,1024,1536,1024,768', 'num_heads': '4,4,4,8,4,4', 'encoder_dim': '192,256,384,512,384,256', 'query_head_dim': '32', 'value_head_dim': '12', 'pos_head_dim': '4', 'pos_dim': 48, 'encoder_unmasked_dim': '192,192,256,256,256,192', 'cnn_module_kernel': '31,31,15,15,15,31', 'decoder_dim': 512, 'joiner_dim': 512, 'attention_decoder_dim': 512, 'attention_decoder_num_layers': 6, 'attention_decoder_attention_dim': 512, 'attention_decoder_num_heads': 8, 'attention_decoder_feedforward_dim': 2048, 'causal': False, 'chunk_size': '16,32,64,-1', 'left_context_frames': '64,128,256,-1', 'use_transducer': True, 'use_ctc': False, 'use_attention_decoder': False, 'full_libri': True, 'mini_libri': False, 'manifest_dir': PosixPath('data/fbank'), 'max_duration': 600, 'bucketing_sampler': True, 'num_buckets': 30, 'concatenate_cuts': False, 'duration_factor': 1.0, 'gap': 1.0, 'on_the_fly_feats': False, 'shuffle': True, 'drop_last': True, 'return_cuts': True, 'num_workers': 2, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'input_strategy': 'PrecomputedFeatures', 'lm_vocab_size': 500, 'lm_epoch': 7, 'lm_avg': 1, 'lm_exp_dir': None, 'rnn_lm_embedding_dim': 2048, 'rnn_lm_hidden_dim': 2048, 'rnn_lm_num_layers': 3, 'rnn_lm_tie_weights': True, 'transformer_lm_exp_dir': None, 'transformer_lm_dim_feedforward': 2048, 'transformer_lm_encoder_dim': 768, 'transformer_lm_embedding_dim': 768, 'transformer_lm_nhead': 8, 'transformer_lm_num_layers': 16, 'transformer_lm_tie_weights': True, 'res_dir': PosixPath('zipformer/exp-full-libri-full-bf16-md1000/greedy_search'), 'has_contexts': False, 'suffix': 'epoch-40-avg-7-context-2-max-sym-per-frame-1-use-averaged-model', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
4
+ 2024-08-13 10:47:51,665 INFO [decode.py:851] About to create model
5
+ 2024-08-13 10:47:52,448 INFO [decode.py:918] Calculating the averaged model over epoch range from 33 (excluded) to 40
6
+ 2024-08-13 10:47:58,181 INFO [decode.py:1012] Number of model parameters: 65549011
7
+ 2024-08-13 10:47:58,182 INFO [asr_datamodule.py:467] About to get test-clean cuts
8
+ 2024-08-13 10:47:58,191 INFO [asr_datamodule.py:474] About to get test-other cuts
9
+ 2024-08-13 10:48:01,303 INFO [decode.py:706] batch 0/?, cuts processed until now is 36
10
+ 2024-08-13 10:48:20,058 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([4.9036, 4.0826, 4.4653, 4.1816], device='cuda:0')
11
+ 2024-08-13 10:48:27,787 INFO [decode.py:722] The transcripts are stored in zipformer/exp-full-libri-full-bf16-md1000/greedy_search/recogs-test-clean-greedy_search-epoch-40-avg-7-context-2-max-sym-per-frame-1-use-averaged-model.txt
12
+ 2024-08-13 10:48:27,897 INFO [utils.py:657] [test-clean-greedy_search] %WER 2.40% [1263 / 52576, 140 ins, 110 del, 1013 sub ]
13
+ 2024-08-13 10:48:28,161 INFO [decode.py:735] Wrote detailed error stats to zipformer/exp-full-libri-full-bf16-md1000/greedy_search/errs-test-clean-greedy_search-epoch-40-avg-7-context-2-max-sym-per-frame-1-use-averaged-model.txt
14
+ 2024-08-13 10:48:28,164 INFO [decode.py:751]
15
+ For test-clean, WER of different settings are:
16
+ greedy_search 2.4 best for test-clean
17
+
18
+ 2024-08-13 10:48:29,411 INFO [decode.py:706] batch 0/?, cuts processed until now is 43
19
+ 2024-08-13 10:48:34,450 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([3.3629, 4.1523, 4.1857, 2.4414], device='cuda:0')
20
+ 2024-08-13 10:48:53,579 INFO [decode.py:722] The transcripts are stored in zipformer/exp-full-libri-full-bf16-md1000/greedy_search/recogs-test-other-greedy_search-epoch-40-avg-7-context-2-max-sym-per-frame-1-use-averaged-model.txt
21
+ 2024-08-13 10:48:53,698 INFO [utils.py:657] [test-other-greedy_search] %WER 5.46% [2860 / 52343, 283 ins, 267 del, 2310 sub ]
22
+ 2024-08-13 10:48:53,981 INFO [decode.py:735] Wrote detailed error stats to zipformer/exp-full-libri-full-bf16-md1000/greedy_search/errs-test-other-greedy_search-epoch-40-avg-7-context-2-max-sym-per-frame-1-use-averaged-model.txt
23
+ 2024-08-13 10:48:53,984 INFO [decode.py:751]
24
+ For test-other, WER of different settings are:
25
+ greedy_search 5.46 best for test-other
26
+
27
+ 2024-08-13 10:48:53,984 INFO [decode.py:1047] Done!
decoding_results/greedy_search/log-decode-epoch-40-avg-9-context-2-max-sym-per-frame-1-use-averaged-model-2024-08-13-10-46-41 ADDED
@@ -0,0 +1,27 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2024-08-13 10:46:41,885 INFO [decode.py:833] Decoding started
2
+ 2024-08-13 10:46:41,886 INFO [decode.py:839] Device: cuda:0
3
+ 2024-08-13 10:46:41,889 INFO [decode.py:849] {'best_train_loss': inf, 'best_valid_loss': inf, 'best_train_epoch': -1, 'best_valid_epoch': -1, 'batch_idx_train': 0, 'log_interval': 50, 'reset_interval': 200, 'valid_interval': 3000, 'feature_dim': 80, 'subsampling_factor': 4, 'ignore_id': -1, 'label_smoothing': 0.1, 'warm_step': 2000, 'env_info': {'k2-version': '1.24.4', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': 'ff1d435a8d3c4eaa15828a84a7240678a70539a7', 'k2-git-date': 'Fri Feb 23 01:48:38 2024', 'lhotse-version': '1.26.0', 'torch-version': '2.0.1+cu117', 'torch-cuda-available': True, 'torch-cuda-version': '11.7', 'python-version': '3.10', 'icefall-git-branch': 'zipformer_bf16', 'icefall-git-sha1': 'a288d412-clean', 'icefall-git-date': 'Mon Aug 12 11:30:21 2024', 'icefall-path': '/star-xy/softwares/icefall_development/icefall_bf16', 'k2-path': '/star-xy/softwares/pyenvs/k2_cuda11/k2_cuda11/lib/python3.10/site-packages/k2/__init__.py', 'lhotse-path': '/star-xy/softwares/pyenvs/k2_cuda11/k2_cuda11/lib/python3.10/site-packages/lhotse/__init__.py', 'hostname': 'de-74279-k2-train-2-1207150844-f49d8c4f4-c49d5', 'IP address': '10.177.22.19'}, 'epoch': 40, 'iter': 0, 'avg': 9, 'use_averaged_model': True, 'exp_dir': PosixPath('zipformer/exp-full-libri-full-bf16-md1000'), 'bpe_model': 'data/lang_bpe_500/bpe.model', 'lang_dir': PosixPath('data/lang_bpe_500'), 'decoding_method': 'greedy_search', 'beam_size': 4, 'beam': 20.0, 'ngram_lm_scale': 0.01, 'max_contexts': 8, 'max_states': 64, 'context_size': 2, 'max_sym_per_frame': 1, 'num_paths': 200, 'nbest_scale': 0.5, 'use_shallow_fusion': False, 'lm_type': 'rnn', 'lm_scale': 0.3, 'tokens_ngram': 2, 'backoff_id': 500, 'context_score': 2, 'context_file': '', 'num_encoder_layers': '2,2,3,4,3,2', 'downsampling_factor': '1,2,4,8,4,2', 'feedforward_dim': '512,768,1024,1536,1024,768', 'num_heads': '4,4,4,8,4,4', 'encoder_dim': '192,256,384,512,384,256', 'query_head_dim': '32', 'value_head_dim': '12', 'pos_head_dim': '4', 'pos_dim': 48, 'encoder_unmasked_dim': '192,192,256,256,256,192', 'cnn_module_kernel': '31,31,15,15,15,31', 'decoder_dim': 512, 'joiner_dim': 512, 'attention_decoder_dim': 512, 'attention_decoder_num_layers': 6, 'attention_decoder_attention_dim': 512, 'attention_decoder_num_heads': 8, 'attention_decoder_feedforward_dim': 2048, 'causal': False, 'chunk_size': '16,32,64,-1', 'left_context_frames': '64,128,256,-1', 'use_transducer': True, 'use_ctc': False, 'use_attention_decoder': False, 'full_libri': True, 'mini_libri': False, 'manifest_dir': PosixPath('data/fbank'), 'max_duration': 600, 'bucketing_sampler': True, 'num_buckets': 30, 'concatenate_cuts': False, 'duration_factor': 1.0, 'gap': 1.0, 'on_the_fly_feats': False, 'shuffle': True, 'drop_last': True, 'return_cuts': True, 'num_workers': 2, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'input_strategy': 'PrecomputedFeatures', 'lm_vocab_size': 500, 'lm_epoch': 7, 'lm_avg': 1, 'lm_exp_dir': None, 'rnn_lm_embedding_dim': 2048, 'rnn_lm_hidden_dim': 2048, 'rnn_lm_num_layers': 3, 'rnn_lm_tie_weights': True, 'transformer_lm_exp_dir': None, 'transformer_lm_dim_feedforward': 2048, 'transformer_lm_encoder_dim': 768, 'transformer_lm_embedding_dim': 768, 'transformer_lm_nhead': 8, 'transformer_lm_num_layers': 16, 'transformer_lm_tie_weights': True, 'res_dir': PosixPath('zipformer/exp-full-libri-full-bf16-md1000/greedy_search'), 'has_contexts': False, 'suffix': 'epoch-40-avg-9-context-2-max-sym-per-frame-1-use-averaged-model', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
4
+ 2024-08-13 10:46:41,890 INFO [decode.py:851] About to create model
5
+ 2024-08-13 10:46:42,660 INFO [decode.py:918] Calculating the averaged model over epoch range from 31 (excluded) to 40
6
+ 2024-08-13 10:46:49,190 INFO [decode.py:1012] Number of model parameters: 65549011
7
+ 2024-08-13 10:46:49,191 INFO [asr_datamodule.py:467] About to get test-clean cuts
8
+ 2024-08-13 10:46:49,207 INFO [asr_datamodule.py:474] About to get test-other cuts
9
+ 2024-08-13 10:46:51,782 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([6.2635, 5.6943, 6.0289, 5.6558], device='cuda:0')
10
+ 2024-08-13 10:46:52,522 INFO [decode.py:706] batch 0/?, cuts processed until now is 36
11
+ 2024-08-13 10:47:18,640 INFO [decode.py:722] The transcripts are stored in zipformer/exp-full-libri-full-bf16-md1000/greedy_search/recogs-test-clean-greedy_search-epoch-40-avg-9-context-2-max-sym-per-frame-1-use-averaged-model.txt
12
+ 2024-08-13 10:47:18,758 INFO [utils.py:657] [test-clean-greedy_search] %WER 2.42% [1271 / 52576, 141 ins, 108 del, 1022 sub ]
13
+ 2024-08-13 10:47:19,017 INFO [decode.py:735] Wrote detailed error stats to zipformer/exp-full-libri-full-bf16-md1000/greedy_search/errs-test-clean-greedy_search-epoch-40-avg-9-context-2-max-sym-per-frame-1-use-averaged-model.txt
14
+ 2024-08-13 10:47:19,022 INFO [decode.py:751]
15
+ For test-clean, WER of different settings are:
16
+ greedy_search 2.42 best for test-clean
17
+
18
+ 2024-08-13 10:47:20,272 INFO [decode.py:706] batch 0/?, cuts processed until now is 43
19
+ 2024-08-13 10:47:32,932 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([5.1170, 4.5429, 4.4889, 4.5820], device='cuda:0')
20
+ 2024-08-13 10:47:44,411 INFO [decode.py:722] The transcripts are stored in zipformer/exp-full-libri-full-bf16-md1000/greedy_search/recogs-test-other-greedy_search-epoch-40-avg-9-context-2-max-sym-per-frame-1-use-averaged-model.txt
21
+ 2024-08-13 10:47:44,528 INFO [utils.py:657] [test-other-greedy_search] %WER 5.49% [2873 / 52343, 281 ins, 275 del, 2317 sub ]
22
+ 2024-08-13 10:47:44,799 INFO [decode.py:735] Wrote detailed error stats to zipformer/exp-full-libri-full-bf16-md1000/greedy_search/errs-test-other-greedy_search-epoch-40-avg-9-context-2-max-sym-per-frame-1-use-averaged-model.txt
23
+ 2024-08-13 10:47:44,803 INFO [decode.py:751]
24
+ For test-other, WER of different settings are:
25
+ greedy_search 5.49 best for test-other
26
+
27
+ 2024-08-13 10:47:44,803 INFO [decode.py:1047] Done!
decoding_results/greedy_search/recogs-test-clean-greedy_search-epoch-30-avg-11-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/recogs-test-clean-greedy_search-epoch-30-avg-13-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/recogs-test-clean-greedy_search-epoch-30-avg-15-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/recogs-test-clean-greedy_search-epoch-30-avg-7-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/recogs-test-clean-greedy_search-epoch-30-avg-9-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/recogs-test-clean-greedy_search-epoch-40-avg-11-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/recogs-test-clean-greedy_search-epoch-40-avg-13-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/recogs-test-clean-greedy_search-epoch-40-avg-15-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/recogs-test-clean-greedy_search-epoch-40-avg-17-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/recogs-test-clean-greedy_search-epoch-40-avg-19-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
decoding_results/greedy_search/recogs-test-clean-greedy_search-epoch-40-avg-7-context-2-max-sym-per-frame-1-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff