add files

README.md

@@ -1,3 +1,7 @@
 ---
 license: apache-2.0
 ---
+Introduction
+This repo contains pre-trained models, checkpoints, training logs and decoding results of the following pull-request:
+
+https://github.com/k2-fsa/icefall/pull/730

data/lang_bpe_500/HLG.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8ec0554f89485063a4f6fb314843416ee8f7c2c5c71dcb5e375987bc8d50f00a
+size 845007559

data/lang_bpe_500/L.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f513bf1f243d9a9cf40eaeaf871d2cf3c76bcfcf678a1c8297bab1ce00b9e240
+size 19025679

data/lang_bpe_500/LG.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6f9b5668d3a11f4bc78044ca8e62a3186b3de5a5c06df56fd582759770515259
+size 249852170

data/lang_bpe_500/Linv.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:806b26522f4558068d45a5ba67cc782da293422c095e5c464957bb2e932f5ed1
+size 19025691

data/lang_bpe_500/bpe.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c53433de083c4a6ad12d034550ef22de68cec62c4f58932a7b6b8b2f1e743fa5
+size 244865

data/lang_bpe_500/tokens.txt

@@ -0,0 +1,502 @@
+<blk> 0
+<sos/eos> 1
+<unk> 2
+S 3
+▁THE 4
+▁A 5
+T 6
+▁AND 7
+ED 8
+▁OF 9
+▁TO 10
+E 11
+D 12
+N 13
+ING 14
+▁IN 15
+Y 16
+M 17
+C 18
+▁I 19
+A 20
+P 21
+▁HE 22
+R 23
+O 24
+L 25
+RE 26
+I 27
+U 28
+ER 29
+▁IT 30
+LY 31
+▁THAT 32
+▁WAS 33
+▁ 34
+▁S 35
+AR 36
+▁BE 37
+F 38
+▁C 39
+IN 40
+B 41
+▁FOR 42
+OR 43
+LE 44
+' 45
+▁HIS 46
+▁YOU 47
+AL 48
+▁RE 49
+V 50
+▁B 51
+G 52
+RI 53
+▁E 54
+▁WITH 55
+▁T 56
+▁AS 57
+LL 58
+▁P 59
+▁HER 60
+ST 61
+▁HAD 62
+▁SO 63
+▁F 64
+W 65
+CE 66
+▁IS 67
+ND 68
+▁NOT 69
+TH 70
+▁BUT 71
+EN 72
+▁SHE 73
+▁ON 74
+VE 75
+ON 76
+SE 77
+▁DE 78
+UR 79
+▁G 80
+CH 81
+K 82
+TER 83
+▁AT 84
+IT 85
+▁ME 86
+RO 87
+NE 88
+RA 89
+ES 90
+IL 91
+NG 92
+IC 93
+▁NO 94
+▁HIM 95
+ENT 96
+IR 97
+▁WE 98
+H 99
+▁DO 100
+▁ALL 101
+▁HAVE 102
+LO 103
+▁BY 104
+▁MY 105
+▁MO 106
+▁THIS 107
+LA 108
+▁ST 109
+▁WHICH 110
+▁CON 111
+▁THEY 112
+CK 113
+TE 114
+▁SAID 115
+▁FROM 116
+▁GO 117
+▁WHO 118
+▁TH 119
+▁OR 120
+▁D 121
+▁W 122
+VER 123
+LI 124
+▁SE 125
+▁ONE 126
+▁CA 127
+▁AN 128
+▁LA 129
+▁WERE 130
+EL 131
+▁HA 132
+▁MAN 133
+▁FA 134
+▁EX 135
+AD 136
+▁SU 137
+RY 138
+▁MI 139
+AT 140
+▁BO 141
+▁WHEN 142
+AN 143
+THER 144
+PP 145
+ATION 146
+▁FI 147
+▁WOULD 148
+▁PRO 149
+OW 150
+ET 151
+▁O 152
+▁THERE 153
+▁HO 154
+ION 155
+▁WHAT 156
+▁FE 157
+▁PA 158
+US 159
+MENT 160
+▁MA 161
+UT 162
+▁OUT 163
+▁THEIR 164
+▁IF 165
+▁LI 166
+▁K 167
+▁WILL 168
+▁ARE 169
+ID 170
+▁RO 171
+DE 172
+TION 173
+▁WA 174
+PE 175
+▁UP 176
+▁SP 177
+▁PO 178
+IGHT 179
+▁UN 180
+RU 181
+▁LO 182
+AS 183
+OL 184
+▁LE 185
+▁BEEN 186
+▁SH 187
+▁RA 188
+▁SEE 189
+KE 190
+UL 191
+TED 192
+▁SA 193
+UN 194
+UND 195
+ANT 196
+▁NE 197
+IS 198
+▁THEM 199
+CI 200
+GE 201
+▁COULD 202
+▁DIS 203
+OM 204
+ISH 205
+HE 206
+EST 207
+▁SOME 208
+ENCE 209
+ITY 210
+IVE 211
+▁US 212
+▁MORE 213
+▁EN 214
+ARD 215
+ATE 216
+▁YOUR 217
+▁INTO 218
+▁KNOW 219
+▁CO 220
+ANCE 221
+▁TIME 222
+▁WI 223
+▁YE 224
+AGE 225
+▁NOW 226
+TI 227
+FF 228
+ABLE 229
+▁VERY 230
+▁LIKE 231
+AM 232
+HI 233
+Z 234
+▁OTHER 235
+▁THAN 236
+▁LITTLE 237
+▁DID 238
+▁LOOK 239
+TY 240
+ERS 241
+▁CAN 242
+▁CHA 243
+▁AR 244
+X 245
+FUL 246
+UGH 247
+▁BA 248
+▁DAY 249
+▁ABOUT 250
+TEN 251
+IM 252
+▁ANY 253
+▁PRE 254
+▁OVER 255
+IES 256
+NESS 257
+ME 258
+BLE 259
+▁M 260
+ROW 261
+▁HAS 262
+▁GREAT 263
+▁VI 264
+TA 265
+▁AFTER 266
+PER 267
+▁AGAIN 268
+HO 269
+SH 270
+▁UPON 271
+▁DI 272
+▁HAND 273
+▁COM 274
+IST 275
+TURE 276
+▁STA 277
+▁THEN 278
+▁SHOULD 279
+▁GA 280
+OUS 281
+OUR 282
+▁WELL 283
+▁ONLY 284
+MAN 285
+▁GOOD 286
+▁TWO 287
+▁MAR 288
+▁SAY 289
+▁HU 290
+TING 291
+▁OUR 292
+RESS 293
+▁DOWN 294
+IOUS 295
+▁BEFORE 296
+▁DA 297
+▁NA 298
+QUI 299
+▁MADE 300
+▁EVERY 301
+▁OLD 302
+▁EVEN 303
+IG 304
+▁COME 305
+▁GRA 306
+▁RI 307
+▁LONG 308
+OT 309
+SIDE 310
+WARD 311
+▁FO 312
+▁WHERE 313
+MO 314
+LESS 315
+▁SC 316
+▁MUST 317
+▁NEVER 318
+▁HOW 319
+▁CAME 320
+▁SUCH 321
+▁RU 322
+▁TAKE 323
+▁WO 324
+▁CAR 325
+UM 326
+AK 327
+▁THINK 328
+▁MUCH 329
+▁MISTER 330
+▁MAY 331
+▁JO 332
+▁WAY 333
+▁COMP 334
+▁THOUGHT 335
+▁STO 336
+▁MEN 337
+▁BACK 338
+▁DON 339
+J 340
+▁LET 341
+▁TRA 342
+▁FIRST 343
+▁JUST 344
+▁VA 345
+▁OWN 346
+▁PLA 347
+▁MAKE 348
+ATED 349
+▁HIMSELF 350
+▁WENT 351
+▁PI 352
+GG 353
+RING 354
+▁DU 355
+▁MIGHT 356
+▁PART 357
+▁GIVE 358
+▁IMP 359
+▁BU 360
+▁PER 361
+▁PLACE 362
+▁HOUSE 363
+▁THROUGH 364
+IAN 365
+▁SW 366
+▁UNDER 367
+QUE 368
+▁AWAY 369
+▁LOVE 370
+QUA 371
+▁LIFE 372
+▁GET 373
+▁WITHOUT 374
+▁PASS 375
+▁TURN 376
+IGN 377
+▁HEAD 378
+▁MOST 379
+▁THOSE 380
+▁SHALL 381
+▁EYES 382
+▁COL 383
+▁STILL 384
+▁NIGHT 385
+▁NOTHING 386
+ITION 387
+HA 388
+▁TELL 389
+▁WORK 390
+▁LAST 391
+▁NEW 392
+▁FACE 393
+▁HI 394
+▁WORD 395
+▁FOUND 396
+▁COUNT 397
+▁OB 398
+▁WHILE 399
+▁SHA 400
+▁MEAN 401
+▁SAW 402
+▁PEOPLE 403
+▁FRIEND 404
+▁THREE 405
+▁ROOM 406
+▁SAME 407
+▁THOUGH 408
+▁RIGHT 409
+▁CHILD 410
+▁FATHER 411
+▁ANOTHER 412
+▁HEART 413
+▁WANT 414
+▁TOOK 415
+OOK 416
+▁LIGHT 417
+▁MISSUS 418
+▁OPEN 419
+▁JU 420
+▁ASKED 421
+PORT 422
+▁LEFT 423
+▁JA 424
+▁WORLD 425
+▁HOME 426
+▁WHY 427
+▁ALWAYS 428
+▁ANSWER 429
+▁SEEMED 430
+▁SOMETHING 431
+▁GIRL 432
+▁BECAUSE 433
+▁NAME 434
+▁TOLD 435
+▁NI 436
+▁HIGH 437
+IZE 438
+▁WOMAN 439
+▁FOLLOW 440
+▁RETURN 441
+▁KNEW 442
+▁EACH 443
+▁KIND 444
+▁JE 445
+▁ACT 446
+▁LU 447
+▁CERTAIN 448
+▁YEARS 449
+▁QUITE 450
+▁APPEAR 451
+▁BETTER 452
+▁HALF 453
+▁PRESENT 454
+▁PRINCE 455
+SHIP 456
+▁ALSO 457
+▁BEGAN 458
+▁HAVING 459
+▁ENOUGH 460
+▁PERSON 461
+▁LADY 462
+▁WHITE 463
+▁COURSE 464
+▁VOICE 465
+▁SPEAK 466
+▁POWER 467
+▁MORNING 468
+▁BETWEEN 469
+▁AMONG 470
+▁KEEP 471
+▁WALK 472
+▁MATTER 473
+▁TEA 474
+▁BELIEVE 475
+▁SMALL 476
+▁TALK 477
+▁FELT 478
+▁HORSE 479
+▁MYSELF 480
+▁SIX 481
+▁HOWEVER 482
+▁FULL 483
+▁HERSELF 484
+▁POINT 485
+▁STOOD 486
+▁HUNDRED 487
+▁ALMOST 488
+▁SINCE 489
+▁LARGE 490
+▁LEAVE 491
+▁PERHAPS 492
+▁DARK 493
+▁SUDDEN 494
+▁REPLIED 495
+▁ANYTHING 496
+▁WONDER 497
+▁UNTIL 498
+Q 499
+#0 500
+#1 501

data/lm/G_4_gram.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ff7d17fc8bf0482b9f0cb9113ea83612ab88df85118aba19c0b81238df80d8ca
+size 3700956590

decoding-results/ctc-decoding/log-decode-epoch-30-avg-10-use-averaged-model-2022-12-14-15-17-22

@@ -0,0 +1,27 @@
+2022-12-14 15:17:22,943 INFO [ctc_decode.py:608] Decoding started
+2022-12-14 15:17:22,944 INFO [ctc_decode.py:614] Device: cuda:0
+2022-12-14 15:17:22,944 INFO [ctc_decode.py:615] {'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, 'beam_size': 10, 'use_double_scores': True, 'warm_step': 2000, 'env_info': {'k2-version': '1.22', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': '6df2d56bd9097bba8d8af12d6c1ef8cb66bf9c17', 'k2-git-date': 'Thu Nov 17 19:06:54 2022', 'lhotse-version': '1.10.0', 'torch-version': '1.13.0', 'torch-cuda-available': True, 'torch-cuda-version': '11.6', 'python-version': '3.1', 'icefall-git-branch': 'blankskip', 'icefall-git-sha1': 'cf69804-dirty', 'icefall-git-date': 'Sat Dec 3 16:30:31 2022', 'icefall-path': '/home/yfy62/icefall', 'k2-path': '/home/yfy62/anaconda3/envs/icefall/lib/python3.10/site-packages/k2-1.22.dev20221122+cuda11.6.torch1.13.0-py3.10-linux-x86_64.egg/k2/__init__.py', 'lhotse-path': '/home/yfy62/anaconda3/envs/icefall/lib/python3.10/site-packages/lhotse/__init__.py', 'hostname': 'd3-hpc-sjtu-test-004', 'IP address': '10.11.11.11'}, 'frame_shift_ms': 10, 'search_beam': 20, 'output_beam': 8, 'min_active_states': 30, 'max_active_states': 10000, 'epoch': 30, 'iter': 0, 'avg': 10, 'use_averaged_model': True, 'exp_dir': PosixPath('pruned_transducer_stateless7_ctc_bk/exp_lconv_scaling'), 'bpe_model': 'data/lang_bpe_500/bpe.model', 'lang_dir': PosixPath('data/lang_bpe_500'), 'context_size': 2, 'decoding_method': 'ctc-decoding', 'num_paths': 100, 'nbest_scale': 0.5, 'hlg_scale': 0.8, 'lm_dir': PosixPath('data/lm'), 'num_encoder_layers': '2,4,3,2,4', 'feedforward_dims': '1024,1024,2048,2048,1024', 'nhead': '8,8,8,8,8', 'encoder_dims': '384,384,384,384,384', 'attention_dims': '192,192,192,192,192', 'encoder_unmasked_dims': '256,256,256,256,256', 'zipformer_downsampling_factors': '1,2,4,8,2', 'cnn_module_kernels': '31,31,31,31,31', 'decoder_dim': 512, 'joiner_dim': 512, 'full_libri': True, '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', 'res_dir': PosixPath('pruned_transducer_stateless7_ctc_bk/exp_lconv_scaling/ctc-decoding'), 'suffix': 'epoch-30-avg-10-use-averaged-model'}
+2022-12-14 15:17:23,300 INFO [lexicon.py:168] Loading pre-compiled data/lang_bpe_500/Linv.pt
+2022-12-14 15:17:24,967 INFO [ctc_decode.py:693] About to create model
+2022-12-14 15:17:25,311 INFO [zipformer.py:179] At encoder stack 4, which has downsampling_factor=2, we will combine the outputs of layers 1 and 3, with downsampling_factors=2 and 8.
+2022-12-14 15:17:25,325 INFO [ctc_decode.py:760] Calculating the averaged model over epoch range from 20 (excluded) to 30
+2022-12-14 15:17:28,896 INFO [ctc_decode.py:777] Number of model parameters: 71164387
+2022-12-14 15:17:28,896 INFO [asr_datamodule.py:443] About to get test-clean cuts
+2022-12-14 15:17:28,897 INFO [asr_datamodule.py:450] About to get test-other cuts
+2022-12-14 15:17:32,819 INFO [ctc_decode.py:526] batch 0/?, cuts processed until now is 43
+2022-12-14 15:17:53,968 INFO [ctc_decode.py:544] The transcripts are stored in pruned_transducer_stateless7_ctc_bk/exp_lconv_scaling/ctc-decoding/recogs-test-clean-ctc-decoding-epoch-30-avg-10-use-averaged-model.txt
+2022-12-14 15:17:54,111 INFO [utils.py:536] [test-clean-ctc-decoding] %WER 6.24% [3280 / 52576, 327 ins, 240 del, 2713 sub ]
+2022-12-14 15:17:54,339 INFO [ctc_decode.py:555] Wrote detailed error stats to pruned_transducer_stateless7_ctc_bk/exp_lconv_scaling/ctc-decoding/errs-test-clean-ctc-decoding-epoch-30-avg-10-use-averaged-model.txt
+2022-12-14 15:17:54,340 INFO [ctc_decode.py:572] 
+For test-clean, WER of different settings are:
+ctc-decoding	6.24	best for test-clean
+
+2022-12-14 15:17:55,518 INFO [ctc_decode.py:526] batch 0/?, cuts processed until now is 52
+2022-12-14 15:18:17,880 INFO [ctc_decode.py:544] The transcripts are stored in pruned_transducer_stateless7_ctc_bk/exp_lconv_scaling/ctc-decoding/recogs-test-other-ctc-decoding-epoch-30-avg-10-use-averaged-model.txt
+2022-12-14 15:18:17,966 INFO [utils.py:536] [test-other-ctc-decoding] %WER 16.97% [8883 / 52343, 842 ins, 805 del, 7236 sub ]
+2022-12-14 15:18:18,137 INFO [ctc_decode.py:555] Wrote detailed error stats to pruned_transducer_stateless7_ctc_bk/exp_lconv_scaling/ctc-decoding/errs-test-other-ctc-decoding-epoch-30-avg-10-use-averaged-model.txt
+2022-12-14 15:18:18,138 INFO [ctc_decode.py:572] 
+For test-other, WER of different settings are:
+ctc-decoding	16.97	best for test-other
+
+2022-12-14 15:18:18,138 INFO [ctc_decode.py:810] Done!

decoding-results/ctc-decoding/log-decode-epoch-30-avg-10-use-averaged-model-2022-12-14-15-24-23

@@ -0,0 +1,27 @@
+2022-12-14 15:24:23,674 INFO [ctc_decode.py:608] Decoding started
+2022-12-14 15:24:23,675 INFO [ctc_decode.py:614] Device: cuda:0
+2022-12-14 15:24:23,675 INFO [ctc_decode.py:615] {'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, 'beam_size': 10, 'use_double_scores': True, 'warm_step': 2000, 'env_info': {'k2-version': '1.22', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': '6df2d56bd9097bba8d8af12d6c1ef8cb66bf9c17', 'k2-git-date': 'Thu Nov 17 19:06:54 2022', 'lhotse-version': '1.10.0', 'torch-version': '1.13.0', 'torch-cuda-available': True, 'torch-cuda-version': '11.6', 'python-version': '3.1', 'icefall-git-branch': 'blankskip', 'icefall-git-sha1': 'cf69804-dirty', 'icefall-git-date': 'Sat Dec 3 16:30:31 2022', 'icefall-path': '/home/yfy62/icefall', 'k2-path': '/home/yfy62/anaconda3/envs/icefall/lib/python3.10/site-packages/k2-1.22.dev20221122+cuda11.6.torch1.13.0-py3.10-linux-x86_64.egg/k2/__init__.py', 'lhotse-path': '/home/yfy62/anaconda3/envs/icefall/lib/python3.10/site-packages/lhotse/__init__.py', 'hostname': 'd3-hpc-sjtu-test-004', 'IP address': '10.11.11.11'}, 'frame_shift_ms': 10, 'search_beam': 20, 'output_beam': 8, 'min_active_states': 30, 'max_active_states': 10000, 'epoch': 30, 'iter': 0, 'avg': 10, 'use_averaged_model': True, 'exp_dir': PosixPath('pruned_transducer_stateless7_ctc_bk/exp_lconv_scaling'), 'bpe_model': 'data/lang_bpe_500/bpe.model', 'lang_dir': PosixPath('data/lang_bpe_500'), 'context_size': 2, 'decoding_method': 'ctc-decoding', 'num_paths': 100, 'nbest_scale': 0.5, 'hlg_scale': 0.8, 'lm_dir': PosixPath('data/lm'), 'num_encoder_layers': '2,4,3,2,4', 'feedforward_dims': '1024,1024,2048,2048,1024', 'nhead': '8,8,8,8,8', 'encoder_dims': '384,384,384,384,384', 'attention_dims': '192,192,192,192,192', 'encoder_unmasked_dims': '256,256,256,256,256', 'zipformer_downsampling_factors': '1,2,4,8,2', 'cnn_module_kernels': '31,31,31,31,31', 'decoder_dim': 512, 'joiner_dim': 512, 'full_libri': True, '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', 'res_dir': PosixPath('pruned_transducer_stateless7_ctc_bk/exp_lconv_scaling/ctc-decoding'), 'suffix': 'epoch-30-avg-10-use-averaged-model'}
+2022-12-14 15:24:24,047 INFO [lexicon.py:168] Loading pre-compiled data/lang_bpe_500/Linv.pt
+2022-12-14 15:24:25,648 INFO [ctc_decode.py:693] About to create model
+2022-12-14 15:24:25,996 INFO [zipformer.py:179] At encoder stack 4, which has downsampling_factor=2, we will combine the outputs of layers 1 and 3, with downsampling_factors=2 and 8.
+2022-12-14 15:24:26,011 INFO [ctc_decode.py:760] Calculating the averaged model over epoch range from 20 (excluded) to 30
+2022-12-14 15:24:28,399 INFO [ctc_decode.py:777] Number of model parameters: 71164387
+2022-12-14 15:24:28,399 INFO [asr_datamodule.py:443] About to get test-clean cuts
+2022-12-14 15:24:28,400 INFO [asr_datamodule.py:450] About to get test-other cuts
+2022-12-14 15:24:32,204 INFO [ctc_decode.py:526] batch 0/?, cuts processed until now is 43
+2022-12-14 15:24:53,572 INFO [ctc_decode.py:544] The transcripts are stored in pruned_transducer_stateless7_ctc_bk/exp_lconv_scaling/ctc-decoding/recogs-test-clean-ctc-decoding-epoch-30-avg-10-use-averaged-model.txt
+2022-12-14 15:24:53,690 INFO [utils.py:536] [test-clean-ctc-decoding] %WER 6.24% [3280 / 52576, 327 ins, 240 del, 2713 sub ]
+2022-12-14 15:24:53,920 INFO [ctc_decode.py:555] Wrote detailed error stats to pruned_transducer_stateless7_ctc_bk/exp_lconv_scaling/ctc-decoding/errs-test-clean-ctc-decoding-epoch-30-avg-10-use-averaged-model.txt
+2022-12-14 15:24:53,920 INFO [ctc_decode.py:572] 
+For test-clean, WER of different settings are:
+ctc-decoding	6.24	best for test-clean
+
+2022-12-14 15:24:55,073 INFO [ctc_decode.py:526] batch 0/?, cuts processed until now is 52
+2022-12-14 15:25:17,528 INFO [ctc_decode.py:544] The transcripts are stored in pruned_transducer_stateless7_ctc_bk/exp_lconv_scaling/ctc-decoding/recogs-test-other-ctc-decoding-epoch-30-avg-10-use-averaged-model.txt
+2022-12-14 15:25:17,661 INFO [utils.py:536] [test-other-ctc-decoding] %WER 16.97% [8883 / 52343, 842 ins, 805 del, 7236 sub ]
+2022-12-14 15:25:17,856 INFO [ctc_decode.py:555] Wrote detailed error stats to pruned_transducer_stateless7_ctc_bk/exp_lconv_scaling/ctc-decoding/errs-test-other-ctc-decoding-epoch-30-avg-10-use-averaged-model.txt
+2022-12-14 15:25:17,856 INFO [ctc_decode.py:572] 
+For test-other, WER of different settings are:
+ctc-decoding	16.97	best for test-other
+
+2022-12-14 15:25:17,856 INFO [ctc_decode.py:810] Done!

decoding-results/ctc-decoding/wer-summary-test-clean-ctc-decoding-epoch-30-avg-10-use-averaged-model.txt

@@ -0,0 +1,2 @@
+settings	WER
+ctc-decoding	6.24

decoding-results/ctc-decoding/wer-summary-test-other-ctc-decoding-epoch-30-avg-10-use-averaged-model.txt

@@ -0,0 +1,2 @@
+settings	WER
+ctc-decoding	16.97

decoding-results/fast_beam_search/log-decode-epoch-30-avg-13-beam-20.0-max-contexts-8-max-states-64-use-averaged-model-2022-12-13-15-17-20

@@ -0,0 +1,40 @@
+2022-12-13 15:17:20,132 INFO [ctc_guild_decode_bk.py:710] Decoding started
+2022-12-13 15:17:20,132 INFO [ctc_guild_decode_bk.py:716] Device: cuda:0
+2022-12-13 15:17:20,134 INFO [ctc_guild_decode_bk.py:731] {'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, 'beam_size': 4, 'use_double_scores': True, 'warm_step': 2000, 'env_info': {'k2-version': '1.22', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': '6df2d56bd9097bba8d8af12d6c1ef8cb66bf9c17', 'k2-git-date': 'Thu Nov 17 19:06:54 2022', 'lhotse-version': '1.10.0', 'torch-version': '1.13.0', 'torch-cuda-available': True, 'torch-cuda-version': '11.6', 'python-version': '3.1', 'icefall-git-branch': 'blankskip', 'icefall-git-sha1': 'cf69804-dirty', 'icefall-git-date': 'Sat Dec 3 16:30:31 2022', 'icefall-path': '/home/yfy62/icefall', 'k2-path': '/home/yfy62/anaconda3/envs/icefall/lib/python3.10/site-packages/k2-1.22.dev20221122+cuda11.6.torch1.13.0-py3.10-linux-x86_64.egg/k2/__init__.py', 'lhotse-path': '/home/yfy62/anaconda3/envs/icefall/lib/python3.10/site-packages/lhotse/__init__.py', 'hostname': 'd3-hpc-sjtu-test-004', 'IP address': '10.11.11.11'}, 'epoch': 30, 'iter': 0, 'avg': 13, 'use_averaged_model': True, 'exp_dir': PosixPath('pruned_transducer_stateless7_ctc_bk/exp_lconv_scaling'), 'bpe_model': 'data/lang_bpe_500/bpe.model', 'lang_dir': PosixPath('data/lang_bpe_500'), 'decoding_method': 'fast_beam_search', '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, 'simulate_streaming': False, 'decode_chunk_size': 16, 'left_context': 64, 'num_encoder_layers': '2,4,3,2,4', 'feedforward_dims': '1024,1024,2048,2048,1024', 'nhead': '8,8,8,8,8', 'encoder_dims': '384,384,384,384,384', 'attention_dims': '192,192,192,192,192', 'encoder_unmasked_dims': '256,256,256,256,256', 'zipformer_downsampling_factors': '1,2,4,8,2', 'cnn_module_kernels': '31,31,31,31,31', 'decoder_dim': 512, 'joiner_dim': 512, 'full_libri': True, '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', 'res_dir': PosixPath('pruned_transducer_stateless7_ctc_bk/exp_lconv_scaling/fast_beam_search'), 'suffix': 'epoch-30-avg-13-beam-20.0-max-contexts-8-max-states-64-use-averaged-model', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
+2022-12-13 15:17:20,134 INFO [ctc_guild_decode_bk.py:733] About to create model
+2022-12-13 15:17:20,561 INFO [zipformer.py:179] At encoder stack 4, which has downsampling_factor=2, we will combine the outputs of layers 1 and 3, with downsampling_factors=2 and 8.
+2022-12-13 15:17:20,576 INFO [ctc_guild_decode_bk.py:800] Calculating the averaged model over epoch range from 17 (excluded) to 30
+2022-12-13 15:17:25,971 INFO [ctc_guild_decode_bk.py:836] Number of model parameters: 71164387
+2022-12-13 15:17:25,971 INFO [asr_datamodule.py:443] About to get test-clean cuts
+2022-12-13 15:17:25,972 INFO [asr_datamodule.py:450] About to get test-other cuts
+2022-12-13 15:17:31,374 INFO [ctc_guild_decode_bk.py:608] batch 0/?, cuts processed until now is 43
+2022-12-13 15:17:44,998 INFO [ctc_guild_decode_bk.py:608] batch 20/?, cuts processed until now is 1434
+2022-12-13 15:17:57,210 INFO [ctc_guild_decode_bk.py:608] batch 40/?, cuts processed until now is 2574
+2022-12-13 15:17:57,816 INFO [ctc_guild_decode_bk.py:626] The transcripts are stored in pruned_transducer_stateless7_ctc_bk/exp_lconv_scaling/fast_beam_search/recogs-test-clean-beam_20.0_max_contexts_8_max_states_64-epoch-30-avg-13-beam-20.0-max-contexts-8-max-states-64-use-averaged-model.txt
+2022-12-13 15:17:57,952 INFO [utils.py:536] [test-clean-beam_20.0_max_contexts_8_max_states_64] %WER 7.72% [4061 / 52576, 331 ins, 585 del, 3145 sub ]
+2022-12-13 15:17:58,178 INFO [ctc_guild_decode_bk.py:639] Wrote detailed error stats to pruned_transducer_stateless7_ctc_bk/exp_lconv_scaling/fast_beam_search/errs-test-clean-beam_20.0_max_contexts_8_max_states_64-epoch-30-avg-13-beam-20.0-max-contexts-8-max-states-64-use-averaged-model.txt
+2022-12-13 15:17:58,179 INFO [ctc_guild_decode_bk.py:656] 
+For test-clean, WER of different settings are:
+beam_20.0_max_contexts_8_max_states_64	7.72	best for test-clean
+
+2022-12-13 15:17:59,469 INFO [ctc_guild_decode_bk.py:608] batch 0/?, cuts processed until now is 52
+2022-12-13 15:18:08,253 INFO [zipformer.py:1414] attn_weights_entropy = tensor([2.4268, 2.5728, 2.1033, 2.5628, 1.5892, 2.2766, 2.0166, 2.5159],
+       device='cuda:0'), covar=tensor([0.0206, 0.0204, 0.0389, 0.0143, 0.0417, 0.0170, 0.0340, 0.0311],
+       device='cuda:0'), in_proj_covar=tensor([0.0058, 0.0043, 0.0064, 0.0047, 0.0061, 0.0046, 0.0059, 0.0047],
+       device='cuda:0'), out_proj_covar=tensor([7.3030e-05, 5.2734e-05, 8.0003e-05, 5.6515e-05, 7.6844e-05, 5.6425e-05,
+        7.4780e-05, 6.0481e-05], device='cuda:0')
+2022-12-13 15:18:12,258 INFO [ctc_guild_decode_bk.py:608] batch 20/?, cuts processed until now is 1646
+2022-12-13 15:18:23,247 INFO [zipformer.py:1414] attn_weights_entropy = tensor([2.6448, 3.1406, 2.3828, 2.9393, 1.9249, 2.7527, 2.2939, 2.9757],
+       device='cuda:0'), covar=tensor([0.0206, 0.0196, 0.0335, 0.0147, 0.0424, 0.0154, 0.0334, 0.0352],
+       device='cuda:0'), in_proj_covar=tensor([0.0058, 0.0043, 0.0064, 0.0047, 0.0061, 0.0046, 0.0059, 0.0047],
+       device='cuda:0'), out_proj_covar=tensor([7.3030e-05, 5.2734e-05, 8.0003e-05, 5.6515e-05, 7.6844e-05, 5.6425e-05,
+        7.4780e-05, 6.0481e-05], device='cuda:0')
+2022-12-13 15:18:23,820 INFO [ctc_guild_decode_bk.py:608] batch 40/?, cuts processed until now is 2870
+2022-12-13 15:18:24,826 INFO [ctc_guild_decode_bk.py:626] The transcripts are stored in pruned_transducer_stateless7_ctc_bk/exp_lconv_scaling/fast_beam_search/recogs-test-other-beam_20.0_max_contexts_8_max_states_64-epoch-30-avg-13-beam-20.0-max-contexts-8-max-states-64-use-averaged-model.txt
+2022-12-13 15:18:24,985 INFO [utils.py:536] [test-other-beam_20.0_max_contexts_8_max_states_64] %WER 17.51% [9163 / 52343, 737 ins, 1362 del, 7064 sub ]
+2022-12-13 15:18:25,235 INFO [ctc_guild_decode_bk.py:639] Wrote detailed error stats to pruned_transducer_stateless7_ctc_bk/exp_lconv_scaling/fast_beam_search/errs-test-other-beam_20.0_max_contexts_8_max_states_64-epoch-30-avg-13-beam-20.0-max-contexts-8-max-states-64-use-averaged-model.txt
+2022-12-13 15:18:25,235 INFO [ctc_guild_decode_bk.py:656] 
+For test-other, WER of different settings are:
+beam_20.0_max_contexts_8_max_states_64	17.51	best for test-other
+
+2022-12-13 15:18:25,235 INFO [ctc_guild_decode_bk.py:867] Done!

decoding-results/fast_beam_search/wer-summary-test-clean-beam_20.0_max_contexts_8_max_states_64-epoch-30-avg-13-beam-20.0-max-contexts-8-max-states-64-use-averaged-model.txt

@@ -0,0 +1,2 @@
+settings	WER
+beam_20.0_max_contexts_8_max_states_64	7.72

decoding-results/fast_beam_search/wer-summary-test-other-beam_20.0_max_contexts_8_max_states_64-epoch-30-avg-13-beam-20.0-max-contexts-8-max-states-64-use-averaged-model.txt

@@ -0,0 +1,2 @@
+settings	WER
+beam_20.0_max_contexts_8_max_states_64	17.51

decoding-results/greedy_search/log-decode-epoch-30-avg-13-context-2-max-sym-per-frame-1-use-averaged-model-2022-12-14-16-40-31

@@ -0,0 +1,46 @@
+2022-12-14 16:40:31,275 INFO [ctc_guild_decode_bk.py:710] Decoding started
+2022-12-14 16:40:31,276 INFO [ctc_guild_decode_bk.py:716] Device: cuda:0
+2022-12-14 16:40:31,278 INFO [ctc_guild_decode_bk.py:731] {'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, 'beam_size': 4, 'use_double_scores': True, 'warm_step': 2000, 'env_info': {'k2-version': '1.22', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': '6df2d56bd9097bba8d8af12d6c1ef8cb66bf9c17', 'k2-git-date': 'Thu Nov 17 19:06:54 2022', 'lhotse-version': '1.10.0', 'torch-version': '1.13.0', 'torch-cuda-available': True, 'torch-cuda-version': '11.6', 'python-version': '3.1', 'icefall-git-branch': 'blankskip', 'icefall-git-sha1': 'cf69804-dirty', 'icefall-git-date': 'Sat Dec 3 16:30:31 2022', 'icefall-path': '/home/yfy62/icefall', 'k2-path': '/home/yfy62/anaconda3/envs/icefall/lib/python3.10/site-packages/k2-1.22.dev20221122+cuda11.6.torch1.13.0-py3.10-linux-x86_64.egg/k2/__init__.py', 'lhotse-path': '/home/yfy62/anaconda3/envs/icefall/lib/python3.10/site-packages/lhotse/__init__.py', 'hostname': 'd3-hpc-sjtu-test-004', 'IP address': '10.11.11.11'}, 'epoch': 30, 'iter': 0, 'avg': 13, 'use_averaged_model': True, 'exp_dir': PosixPath('pruned_transducer_stateless7_ctc_bk/exp_lconv_scaling'), 'bpe_model': 'data/lang_bpe_500/bpe.model', 'lang_dir': PosixPath('data/lang_bpe_500'), 'decoding_method': 'greedy_search', '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, 'simulate_streaming': False, 'decode_chunk_size': 16, 'left_context': 64, 'num_encoder_layers': '2,4,3,2,4', 'feedforward_dims': '1024,1024,2048,2048,1024', 'nhead': '8,8,8,8,8', 'encoder_dims': '384,384,384,384,384', 'attention_dims': '192,192,192,192,192', 'encoder_unmasked_dims': '256,256,256,256,256', 'zipformer_downsampling_factors': '1,2,4,8,2', 'cnn_module_kernels': '31,31,31,31,31', 'decoder_dim': 512, 'joiner_dim': 512, 'full_libri': True, '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', 'res_dir': PosixPath('pruned_transducer_stateless7_ctc_bk/exp_lconv_scaling/greedy_search'), 'suffix': 'epoch-30-avg-13-context-2-max-sym-per-frame-1-use-averaged-model', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
+2022-12-14 16:40:31,278 INFO [ctc_guild_decode_bk.py:733] About to create model
+2022-12-14 16:40:31,774 INFO [zipformer.py:179] At encoder stack 4, which has downsampling_factor=2, we will combine the outputs of layers 1 and 3, with downsampling_factors=2 and 8.
+2022-12-14 16:40:31,794 INFO [ctc_guild_decode_bk.py:800] Calculating the averaged model over epoch range from 17 (excluded) to 30
+2022-12-14 16:40:36,459 INFO [ctc_guild_decode_bk.py:836] Number of model parameters: 71164387
+2022-12-14 16:40:36,459 INFO [asr_datamodule.py:443] About to get test-clean cuts
+2022-12-14 16:40:36,460 INFO [asr_datamodule.py:450] About to get test-other cuts
+2022-12-14 16:40:41,112 INFO [ctc_guild_decode_bk.py:608] batch 0/?, cuts processed until now is 43
+2022-12-14 16:40:42,842 INFO [zipformer.py:1414] attn_weights_entropy = tensor([5.0845, 5.2382, 5.3612, 5.0017, 5.2092, 4.7563, 4.6855, 4.7861],
+       device='cuda:0'), covar=tensor([0.0388, 0.0189, 0.0125, 0.0220, 0.0208, 0.0217, 0.0272, 0.0309],
+       device='cuda:0'), in_proj_covar=tensor([0.0182, 0.0146, 0.0126, 0.0149, 0.0135, 0.0156, 0.0173, 0.0173],
+       device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002],
+       device='cuda:0')
+2022-12-14 16:40:45,678 INFO [zipformer.py:1414] attn_weights_entropy = tensor([3.1129, 3.3022, 3.6867, 2.1906, 3.2836, 3.0820, 3.2884, 2.9239],
+       device='cuda:0'), covar=tensor([0.0854, 0.0538, 0.0214, 0.2138, 0.0361, 0.0956, 0.0574, 0.1603],
+       device='cuda:0'), in_proj_covar=tensor([0.0183, 0.0129, 0.0122, 0.0206, 0.0134, 0.0180, 0.0172, 0.0206],
+       device='cuda:0'), out_proj_covar=tensor([1.2879e-04, 9.0075e-05, 8.3072e-05, 1.4434e-04, 9.0510e-05, 1.2637e-04,
+        1.1752e-04, 1.4144e-04], device='cuda:0')
+2022-12-14 16:40:53,272 INFO [ctc_guild_decode_bk.py:626] The transcripts are stored in pruned_transducer_stateless7_ctc_bk/exp_lconv_scaling/greedy_search/recogs-test-clean-greedy_search-epoch-30-avg-13-context-2-max-sym-per-frame-1-use-averaged-model.txt
+2022-12-14 16:40:53,345 INFO [utils.py:536] [test-clean-greedy_search] %WER 5.74% [3016 / 52576, 326 ins, 267 del, 2423 sub ]
+2022-12-14 16:40:53,491 INFO [ctc_guild_decode_bk.py:639] Wrote detailed error stats to pruned_transducer_stateless7_ctc_bk/exp_lconv_scaling/greedy_search/errs-test-clean-greedy_search-epoch-30-avg-13-context-2-max-sym-per-frame-1-use-averaged-model.txt
+2022-12-14 16:40:53,491 INFO [ctc_guild_decode_bk.py:656] 
+For test-clean, WER of different settings are:
+greedy_search	5.74	best for test-clean
+
+2022-12-14 16:40:54,313 INFO [ctc_guild_decode_bk.py:608] batch 0/?, cuts processed until now is 52
+2022-12-14 16:40:54,451 INFO [zipformer.py:1414] attn_weights_entropy = tensor([4.4679, 4.4290, 4.5958, 5.0097, 4.3190, 4.8124, 4.5159, 4.5238],
+       device='cuda:0'), covar=tensor([0.0281, 0.0263, 0.0248, 0.0141, 0.0244, 0.0115, 0.0241, 0.0221],
+       device='cuda:0'), in_proj_covar=tensor([0.0103, 0.0086, 0.0093, 0.0085, 0.0073, 0.0079, 0.0079, 0.0085],
+       device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001],
+       device='cuda:0')
+2022-12-14 16:40:55,386 INFO [zipformer.py:1414] attn_weights_entropy = tensor([2.8303, 2.5061, 2.4607, 3.5364, 3.2174, 2.9463, 3.1359, 3.0660],
+       device='cuda:0'), covar=tensor([0.0377, 0.0609, 0.0640, 0.0145, 0.0443, 0.0493, 0.0152, 0.0411],
+       device='cuda:0'), in_proj_covar=tensor([0.0073, 0.0065, 0.0084, 0.0054, 0.0053, 0.0054, 0.0063, 0.0053],
+       device='cuda:0'), out_proj_covar=tensor([5.8793e-05, 5.3987e-05, 8.1245e-05, 3.9913e-05, 4.5528e-05, 4.6460e-05,
+        4.7543e-05, 4.2467e-05], device='cuda:0')
+2022-12-14 16:41:06,368 INFO [ctc_guild_decode_bk.py:626] The transcripts are stored in pruned_transducer_stateless7_ctc_bk/exp_lconv_scaling/greedy_search/recogs-test-other-greedy_search-epoch-30-avg-13-context-2-max-sym-per-frame-1-use-averaged-model.txt
+2022-12-14 16:41:06,453 INFO [utils.py:536] [test-other-greedy_search] %WER 15.46% [8093 / 52343, 758 ins, 887 del, 6448 sub ]
+2022-12-14 16:41:06,620 INFO [ctc_guild_decode_bk.py:639] Wrote detailed error stats to pruned_transducer_stateless7_ctc_bk/exp_lconv_scaling/greedy_search/errs-test-other-greedy_search-epoch-30-avg-13-context-2-max-sym-per-frame-1-use-averaged-model.txt
+2022-12-14 16:41:06,620 INFO [ctc_guild_decode_bk.py:656] 
+For test-other, WER of different settings are:
+greedy_search	15.46	best for test-other
+
+2022-12-14 16:41:06,620 INFO [ctc_guild_decode_bk.py:867] Done!

decoding-results/greedy_search/wer-summary-test-clean-greedy_search-epoch-30-avg-13-context-2-max-sym-per-frame-1-use-averaged-model.txt

@@ -0,0 +1,2 @@
+settings	WER
+greedy_search	5.74

decoding-results/greedy_search/wer-summary-test-other-greedy_search-epoch-30-avg-13-context-2-max-sym-per-frame-1-use-averaged-model.txt

@@ -0,0 +1,2 @@
+settings	WER
+greedy_search	15.46

decoding-results/modified_beam_search/log-decode-epoch-30-avg-13-modified_beam_search-beam-size-4-use-averaged-model-2022-12-14-16-41-57

@@ -0,0 +1,35 @@
+2022-12-14 16:41:57,346 INFO [ctc_guild_decode_bk.py:710] Decoding started
+2022-12-14 16:41:57,346 INFO [ctc_guild_decode_bk.py:716] Device: cuda:0
+2022-12-14 16:41:57,348 INFO [ctc_guild_decode_bk.py:731] {'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, 'beam_size': 4, 'use_double_scores': True, 'warm_step': 2000, 'env_info': {'k2-version': '1.22', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': '6df2d56bd9097bba8d8af12d6c1ef8cb66bf9c17', 'k2-git-date': 'Thu Nov 17 19:06:54 2022', 'lhotse-version': '1.10.0', 'torch-version': '1.13.0', 'torch-cuda-available': True, 'torch-cuda-version': '11.6', 'python-version': '3.1', 'icefall-git-branch': 'blankskip', 'icefall-git-sha1': 'cf69804-dirty', 'icefall-git-date': 'Sat Dec 3 16:30:31 2022', 'icefall-path': '/home/yfy62/icefall', 'k2-path': '/home/yfy62/anaconda3/envs/icefall/lib/python3.10/site-packages/k2-1.22.dev20221122+cuda11.6.torch1.13.0-py3.10-linux-x86_64.egg/k2/__init__.py', 'lhotse-path': '/home/yfy62/anaconda3/envs/icefall/lib/python3.10/site-packages/lhotse/__init__.py', 'hostname': 'd3-hpc-sjtu-test-004', 'IP address': '10.11.11.11'}, 'epoch': 30, 'iter': 0, 'avg': 13, 'use_averaged_model': True, 'exp_dir': PosixPath('pruned_transducer_stateless7_ctc_bk/exp_lconv_scaling'), 'bpe_model': 'data/lang_bpe_500/bpe.model', 'lang_dir': PosixPath('data/lang_bpe_500'), 'decoding_method': 'modified_beam_search', '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, 'simulate_streaming': False, 'decode_chunk_size': 16, 'left_context': 64, 'num_encoder_layers': '2,4,3,2,4', 'feedforward_dims': '1024,1024,2048,2048,1024', 'nhead': '8,8,8,8,8', 'encoder_dims': '384,384,384,384,384', 'attention_dims': '192,192,192,192,192', 'encoder_unmasked_dims': '256,256,256,256,256', 'zipformer_downsampling_factors': '1,2,4,8,2', 'cnn_module_kernels': '31,31,31,31,31', 'decoder_dim': 512, 'joiner_dim': 512, 'full_libri': True, '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', 'res_dir': PosixPath('pruned_transducer_stateless7_ctc_bk/exp_lconv_scaling/modified_beam_search'), 'suffix': 'epoch-30-avg-13-modified_beam_search-beam-size-4-use-averaged-model', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
+2022-12-14 16:41:57,348 INFO [ctc_guild_decode_bk.py:733] About to create model
+2022-12-14 16:41:57,749 INFO [zipformer.py:179] At encoder stack 4, which has downsampling_factor=2, we will combine the outputs of layers 1 and 3, with downsampling_factors=2 and 8.
+2022-12-14 16:41:57,766 INFO [ctc_guild_decode_bk.py:800] Calculating the averaged model over epoch range from 17 (excluded) to 30
+2022-12-14 16:42:02,099 INFO [ctc_guild_decode_bk.py:836] Number of model parameters: 71164387
+2022-12-14 16:42:02,099 INFO [asr_datamodule.py:443] About to get test-clean cuts
+2022-12-14 16:42:02,100 INFO [asr_datamodule.py:450] About to get test-other cuts
+2022-12-14 16:42:07,791 INFO [ctc_guild_decode_bk.py:608] batch 0/?, cuts processed until now is 43
+2022-12-14 16:42:28,479 INFO [ctc_guild_decode_bk.py:608] batch 20/?, cuts processed until now is 1434
+2022-12-14 16:42:44,750 INFO [ctc_guild_decode_bk.py:608] batch 40/?, cuts processed until now is 2574
+2022-12-14 16:42:45,603 INFO [ctc_guild_decode_bk.py:626] The transcripts are stored in pruned_transducer_stateless7_ctc_bk/exp_lconv_scaling/modified_beam_search/recogs-test-clean-beam_size_4-epoch-30-avg-13-modified_beam_search-beam-size-4-use-averaged-model.txt
+2022-12-14 16:42:45,722 INFO [utils.py:536] [test-clean-beam_size_4] %WER 5.66% [2978 / 52576, 335 ins, 241 del, 2402 sub ]
+2022-12-14 16:42:45,870 INFO [ctc_guild_decode_bk.py:639] Wrote detailed error stats to pruned_transducer_stateless7_ctc_bk/exp_lconv_scaling/modified_beam_search/errs-test-clean-beam_size_4-epoch-30-avg-13-modified_beam_search-beam-size-4-use-averaged-model.txt
+2022-12-14 16:42:45,870 INFO [ctc_guild_decode_bk.py:656] 
+For test-clean, WER of different settings are:
+beam_size_4	5.66	best for test-clean
+
+2022-12-14 16:42:46,534 INFO [zipformer.py:1414] attn_weights_entropy = tensor([2.6509, 2.3458, 2.3456, 3.1348, 2.9894, 2.8388, 2.8162, 2.8047],
+       device='cuda:0'), covar=tensor([0.0352, 0.0657, 0.0608, 0.0179, 0.0396, 0.0467, 0.0155, 0.0386],
+       device='cuda:0'), in_proj_covar=tensor([0.0073, 0.0065, 0.0084, 0.0054, 0.0053, 0.0054, 0.0063, 0.0053],
+       device='cuda:0'), out_proj_covar=tensor([5.8793e-05, 5.3987e-05, 8.1245e-05, 3.9913e-05, 4.5528e-05, 4.6460e-05,
+        4.7543e-05, 4.2467e-05], device='cuda:0')
+2022-12-14 16:42:47,542 INFO [ctc_guild_decode_bk.py:608] batch 0/?, cuts processed until now is 52
+2022-12-14 16:43:06,573 INFO [ctc_guild_decode_bk.py:608] batch 20/?, cuts processed until now is 1646
+2022-12-14 16:43:22,122 INFO [ctc_guild_decode_bk.py:608] batch 40/?, cuts processed until now is 2870
+2022-12-14 16:43:22,829 INFO [ctc_guild_decode_bk.py:626] The transcripts are stored in pruned_transducer_stateless7_ctc_bk/exp_lconv_scaling/modified_beam_search/recogs-test-other-beam_size_4-epoch-30-avg-13-modified_beam_search-beam-size-4-use-averaged-model.txt
+2022-12-14 16:43:22,906 INFO [utils.py:536] [test-other-beam_size_4] %WER 15.24% [7975 / 52343, 761 ins, 830 del, 6384 sub ]
+2022-12-14 16:43:23,086 INFO [ctc_guild_decode_bk.py:639] Wrote detailed error stats to pruned_transducer_stateless7_ctc_bk/exp_lconv_scaling/modified_beam_search/errs-test-other-beam_size_4-epoch-30-avg-13-modified_beam_search-beam-size-4-use-averaged-model.txt
+2022-12-14 16:43:23,086 INFO [ctc_guild_decode_bk.py:656] 
+For test-other, WER of different settings are:
+beam_size_4	15.24	best for test-other
+
+2022-12-14 16:43:23,087 INFO [ctc_guild_decode_bk.py:867] Done!

decoding-results/modified_beam_search/wer-summary-test-clean-beam_size_4-epoch-30-avg-13-modified_beam_search-beam-size-4-use-averaged-model.txt

@@ -0,0 +1,2 @@
+settings	WER
+beam_size_4	5.66

decoding-results/modified_beam_search/wer-summary-test-other-beam_size_4-epoch-30-avg-13-modified_beam_search-beam-size-4-use-averaged-model.txt

@@ -0,0 +1,2 @@
+settings	WER
+beam_size_4	15.24

exp/cpu_jit.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:da57e855d2bcd3d3230e3cfb0905d6e0230b616a76c5290db9fad37999ebac45
+size 284936924

exp/decode.sh

@@ -0,0 +1,31 @@
+#!/usr/bin/env bash
+
+export CUDA_VISIBLE_DEVICES=
+
+# greedy search
+./pruned_transducer_stateless7_ctc_bs/ctc_guild_decode_bs.py \
+    --epoch 30 \
+    --avg 13 \
+    --exp-dir ./pruned_transducer_stateless7_ctc_bs/exp \
+    --max-duration 600 \
+    --decoding-method greedy_search
+
+# modified beam search
+./pruned_transducer_stateless7_ctc_bs/ctc_guild_decode_bs.py \
+    --epoch 30 \
+    --avg 13 \
+    --exp-dir ./pruned_transducer_stateless7_ctc_bs/exp \
+    --max-duration 600 \
+    --decoding-method modified_beam_search \
+    --beam-size 4
+
+# fast beam search (one best)
+./pruned_transducer_stateless7_ctc_bs/ctc_guild_decode_bs.py \
+    --epoch 30 \
+    --avg 13 \
+    --exp-dir ./pruned_transducer_stateless7_ctc_bs/exp \
+    --max-duration 600 \
+    --decoding-method fast_beam_search \
+    --beam 20.0 \
+    --max-contexts 8 \
+    --max-states 64

exp/epoch-30.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a51723afbc76d9a3ff9e522f3e3eb385ace7916c2f977ae6fcedd5b3b142d36f
+size 1139299611

exp/export.sh

@@ -0,0 +1,10 @@
+#!/usr/bin/env bash
+
+export CUDA_VISIBLE_DEVICES=
+
+./pruned_transducer_stateless7_ctc_bs/export.py \
+  --exp-dir ./pruned_transducer_stateless7_ctc_bs/exp \
+  --bpe-model data/lang_bpe_500/bpe.model \
+  --epoch 30 \
+  --avg 13 \
+  --jit 1

exp/pretrained.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:63d958507e38bd4a627755ae46fd27bd1bbdfc9d4ab10c0903e4d51bae23593a
+size 284952187

exp/run.sh

@@ -0,0 +1,11 @@
+#!/usr/bin/env bash
+
+export CUDA_VISIBLE_DEVICES="0,1,2,3"
+
+./pruned_transducer_stateless7_ctc_bs/train.py \
+  --world-size 4 \
+  --num-epochs 30 \
+  --start-epoch 1 \
+  --exp-dir pruned_transducer_stateless7_ctc_bs/exp \
+  --full-libri 0 \
+  --max-duration 300

exp/tensorboard/events.out.tfevents.1670841928.d3-hpc-sjtu-test-004.143030.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0db07f460a0ded8858fc69f7d13034073c826f216b137a25ddc8a28504961b57
+size 331437