initial commit

data/lang_bpe_500/HLG.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b5dbbe8b485c0cb37d11e07e8e734990f1e40a2d00fe9689d8da2e7b6fe72883
+size 845007583

data/lang_bpe_500/L.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b1b88996f918737fba67fbd29152018b51a537c16ce0718a2b43d5140583224e
+size 19025703

data/lang_bpe_500/LG.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3bb9f021c7aad79d45dc275ba8154a430c4f660a319dcb872cd52500f25553d6
+size 249852195

data/lang_bpe_500/Linv.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cbc8b3687a1b8f0811a84106b3b310642566c7b1bc282a929878f9269507a2c6
+size 19025703

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

decoding-results/fast_beam_search/log-decode-epoch-20-avg-4-streaming-chunk-size-32-beam-20.0-max-contexts-8-max-states-64-right-padding-64-use-averaged-model-2023-04-06-13-11-21

@@ -0,0 +1,35 @@
+2023-04-06 13:11:21,982 INFO [decode.py:659] Decoding started
+2023-04-06 13:11:21,982 INFO [decode.py:665] Device: cuda:0
+2023-04-06 13:11:21,985 INFO [decode.py:675] {'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, 'warm_step': 2000, 'env_info': {'k2-version': '1.22', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': '96c9a2aece2a3a7633da07740e24fa3d96f5498c', 'k2-git-date': 'Thu Nov 10 08:14:02 2022', 'lhotse-version': '1.13.0.dev+git.527d964.clean', 'torch-version': '1.12.1', 'torch-cuda-available': True, 'torch-cuda-version': '11.6', 'python-version': '3.8', 'icefall-git-branch': 'zipformer_libri_small_models', 'icefall-git-sha1': 'b52e7ae-dirty', 'icefall-git-date': 'Tue Apr 4 14:07:45 2023', 'icefall-path': '/ceph-data4/yangxiaoyu/softwares/icefall_development/icefall_small_models', 'k2-path': '/ceph-data4/yangxiaoyu/softwares/anaconda3/envs/k2_latest/lib/python3.8/site-packages/k2/__init__.py', 'lhotse-path': '/ceph-data4/yangxiaoyu/softwares/lhotse_development/lhotse_random_padding_left/lhotse/__init__.py', 'hostname': 'de-74279-k2-train-6-1219221738-65dd59bbf8-2ghmr', 'IP address': '10.177.28.85'}, 'epoch': 20, 'iter': 0, 'avg': 4, 'use_averaged_model': True, 'exp_dir': PosixPath('pruned_transducer_stateless7_streaming_multi/exp'), 'bpe_model': 'data/lang_bpe_500/bpe.model', 'lang_dir': PosixPath('data/lang_bpe_500'), 'decoding_method': 'fast_beam_search', 'beam_size': 4, 'beam': 20.0, 'ngram_lm_scale': 0.01, 'max_contexts': 8, 'max_states': 64, 'context_size': 2, 'right_padding': 64, 'max_sym_per_frame': 1, 'num_paths': 200, 'nbest_scale': 0.5, '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, 'short_chunk_size': 50, 'num_left_chunks': 4, 'decode_chunk_len': 32, 'max_duration': 600, 'bucketing_sampler': True, 'num_buckets': 30, 'shuffle': True, 'return_cuts': True, 'num_workers': 2, 'on_the_fly_num_workers': 0, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'manifest_dir': PosixPath('data/fbank'), 'on_the_fly_feats': False, 'res_dir': PosixPath('pruned_transducer_stateless7_streaming_multi/exp/fast_beam_search'), 'suffix': 'epoch-20-avg-4-streaming-chunk-size-32-beam-20.0-max-contexts-8-max-states-64-right-padding-64-use-averaged-model', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
+2023-04-06 13:11:21,986 INFO [decode.py:677] About to create model
+2023-04-06 13:11:22,808 INFO [zipformer.py:405] 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.
+2023-04-06 13:11:22,828 INFO [decode.py:748] Calculating the averaged model over epoch range from 16 (excluded) to 20
+2023-04-06 13:11:28,037 INFO [decode.py:782] Number of model parameters: 70369391
+2023-04-06 13:11:28,038 INFO [librispeech.py:58] About to get test-clean cuts from data/fbank/librispeech_cuts_test-clean.jsonl.gz
+2023-04-06 13:11:28,040 INFO [librispeech.py:63] About to get test-other cuts from data/fbank/librispeech_cuts_test-other.jsonl.gz
+2023-04-06 13:11:35,436 INFO [decode.py:569] batch 0/?, cuts processed until now is 26
+2023-04-06 13:12:19,715 INFO [decode.py:569] batch 20/?, cuts processed until now is 1545
+2023-04-06 13:12:56,621 INFO [decode.py:569] batch 40/?, cuts processed until now is 2375
+2023-04-06 13:13:27,169 INFO [decode.py:583] The transcripts are stored in pruned_transducer_stateless7_streaming_multi/exp/fast_beam_search/recogs-test-clean-epoch-20-avg-4-streaming-chunk-size-32-beam-20.0-max-contexts-8-max-states-64-right-padding-64-use-averaged-model.txt
+2023-04-06 13:13:27,321 INFO [utils.py:558] [test-clean-beam_20.0_max_contexts_8_max_states_64] %WER 2.43% [1277 / 52576, 147 ins, 97 del, 1033 sub ]
+2023-04-06 13:13:27,669 INFO [decode.py:594] Wrote detailed error stats to pruned_transducer_stateless7_streaming_multi/exp/fast_beam_search/errs-test-clean-epoch-20-avg-4-streaming-chunk-size-32-beam-20.0-max-contexts-8-max-states-64-right-padding-64-use-averaged-model.txt
+2023-04-06 13:13:27,670 INFO [decode.py:608] 
+For test-clean, WER of different settings are:
+beam_20.0_max_contexts_8_max_states_64	2.43	best for test-clean
+
+2023-04-06 13:13:31,899 INFO [decode.py:569] batch 0/?, cuts processed until now is 30
+2023-04-06 13:14:12,501 INFO [decode.py:569] batch 20/?, cuts processed until now is 1771
+2023-04-06 13:14:45,255 INFO [decode.py:569] batch 40/?, cuts processed until now is 2696
+2023-04-06 13:15:02,874 INFO [zipformer.py:2441] attn_weights_entropy = tensor([1.7310, 1.5854, 0.8361, 1.3923, 1.6120, 1.5549, 1.4788, 1.4736],
+       device='cuda:0'), covar=tensor([0.0463, 0.0283, 0.0402, 0.0525, 0.0249, 0.0464, 0.0430, 0.0527],
+       device='cuda:0'), in_proj_covar=tensor([0.0035, 0.0028, 0.0026, 0.0035, 0.0023, 0.0035, 0.0035, 0.0037],
+       device='cuda:0'), out_proj_covar=tensor([0.0053, 0.0045, 0.0039, 0.0056, 0.0039, 0.0054, 0.0054, 0.0057],
+       device='cuda:0')
+2023-04-06 13:15:12,927 INFO [decode.py:583] The transcripts are stored in pruned_transducer_stateless7_streaming_multi/exp/fast_beam_search/recogs-test-other-epoch-20-avg-4-streaming-chunk-size-32-beam-20.0-max-contexts-8-max-states-64-right-padding-64-use-averaged-model.txt
+2023-04-06 13:15:13,083 INFO [utils.py:558] [test-other-beam_20.0_max_contexts_8_max_states_64] %WER 5.99% [3136 / 52343, 326 ins, 311 del, 2499 sub ]
+2023-04-06 13:15:13,436 INFO [decode.py:594] Wrote detailed error stats to pruned_transducer_stateless7_streaming_multi/exp/fast_beam_search/errs-test-other-epoch-20-avg-4-streaming-chunk-size-32-beam-20.0-max-contexts-8-max-states-64-right-padding-64-use-averaged-model.txt
+2023-04-06 13:15:13,437 INFO [decode.py:608] 
+For test-other, WER of different settings are:
+beam_20.0_max_contexts_8_max_states_64	5.99	best for test-other
+
+2023-04-06 13:15:13,437 INFO [decode.py:814] Done!

decoding-results/fast_beam_search/log-decode-epoch-20-avg-4-streaming-chunk-size-64-beam-20.0-max-contexts-8-max-states-64-right-padding-64-use-averaged-model-2023-04-06-13-07-27

@@ -0,0 +1,35 @@
+2023-04-06 13:07:27,764 INFO [decode.py:659] Decoding started
+2023-04-06 13:07:27,764 INFO [decode.py:665] Device: cuda:0
+2023-04-06 13:07:27,768 INFO [decode.py:675] {'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, 'warm_step': 2000, 'env_info': {'k2-version': '1.22', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': '96c9a2aece2a3a7633da07740e24fa3d96f5498c', 'k2-git-date': 'Thu Nov 10 08:14:02 2022', 'lhotse-version': '1.13.0.dev+git.527d964.clean', 'torch-version': '1.12.1', 'torch-cuda-available': True, 'torch-cuda-version': '11.6', 'python-version': '3.8', 'icefall-git-branch': 'zipformer_libri_small_models', 'icefall-git-sha1': 'b52e7ae-dirty', 'icefall-git-date': 'Tue Apr 4 14:07:45 2023', 'icefall-path': '/ceph-data4/yangxiaoyu/softwares/icefall_development/icefall_small_models', 'k2-path': '/ceph-data4/yangxiaoyu/softwares/anaconda3/envs/k2_latest/lib/python3.8/site-packages/k2/__init__.py', 'lhotse-path': '/ceph-data4/yangxiaoyu/softwares/lhotse_development/lhotse_random_padding_left/lhotse/__init__.py', 'hostname': 'de-74279-k2-train-6-1219221738-65dd59bbf8-2ghmr', 'IP address': '10.177.28.85'}, 'epoch': 20, 'iter': 0, 'avg': 4, 'use_averaged_model': True, 'exp_dir': PosixPath('pruned_transducer_stateless7_streaming_multi/exp'), 'bpe_model': 'data/lang_bpe_500/bpe.model', 'lang_dir': PosixPath('data/lang_bpe_500'), 'decoding_method': 'fast_beam_search', 'beam_size': 4, 'beam': 20.0, 'ngram_lm_scale': 0.01, 'max_contexts': 8, 'max_states': 64, 'context_size': 2, 'right_padding': 64, 'max_sym_per_frame': 1, 'num_paths': 200, 'nbest_scale': 0.5, '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, 'short_chunk_size': 50, 'num_left_chunks': 4, 'decode_chunk_len': 64, 'max_duration': 600, 'bucketing_sampler': True, 'num_buckets': 30, 'shuffle': True, 'return_cuts': True, 'num_workers': 2, 'on_the_fly_num_workers': 0, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'manifest_dir': PosixPath('data/fbank'), 'on_the_fly_feats': False, 'res_dir': PosixPath('pruned_transducer_stateless7_streaming_multi/exp/fast_beam_search'), 'suffix': 'epoch-20-avg-4-streaming-chunk-size-64-beam-20.0-max-contexts-8-max-states-64-right-padding-64-use-averaged-model', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
+2023-04-06 13:07:27,768 INFO [decode.py:677] About to create model
+2023-04-06 13:07:28,588 INFO [zipformer.py:405] 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.
+2023-04-06 13:07:28,608 INFO [decode.py:748] Calculating the averaged model over epoch range from 16 (excluded) to 20
+2023-04-06 13:07:33,838 INFO [decode.py:782] Number of model parameters: 70369391
+2023-04-06 13:07:33,838 INFO [librispeech.py:58] About to get test-clean cuts from data/fbank/librispeech_cuts_test-clean.jsonl.gz
+2023-04-06 13:07:33,841 INFO [librispeech.py:63] About to get test-other cuts from data/fbank/librispeech_cuts_test-other.jsonl.gz
+2023-04-06 13:07:40,955 INFO [decode.py:569] batch 0/?, cuts processed until now is 26
+2023-04-06 13:08:05,123 INFO [zipformer.py:2441] attn_weights_entropy = tensor([3.8971, 3.9030, 2.9892, 4.4202, 3.9925, 3.8669, 1.8646, 3.8496],
+       device='cuda:0'), covar=tensor([0.1528, 0.0781, 0.3562, 0.0964, 0.2049, 0.2446, 0.6687, 0.2063],
+       device='cuda:0'), in_proj_covar=tensor([0.0265, 0.0238, 0.0290, 0.0333, 0.0331, 0.0273, 0.0293, 0.0287],
+       device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002],
+       device='cuda:0')
+2023-04-06 13:08:24,743 INFO [decode.py:569] batch 20/?, cuts processed until now is 1545
+2023-04-06 13:09:01,013 INFO [decode.py:569] batch 40/?, cuts processed until now is 2375
+2023-04-06 13:09:31,347 INFO [decode.py:583] The transcripts are stored in pruned_transducer_stateless7_streaming_multi/exp/fast_beam_search/recogs-test-clean-epoch-20-avg-4-streaming-chunk-size-64-beam-20.0-max-contexts-8-max-states-64-right-padding-64-use-averaged-model.txt
+2023-04-06 13:09:31,497 INFO [utils.py:558] [test-clean-beam_20.0_max_contexts_8_max_states_64] %WER 2.27% [1196 / 52576, 129 ins, 100 del, 967 sub ]
+2023-04-06 13:09:31,841 INFO [decode.py:594] Wrote detailed error stats to pruned_transducer_stateless7_streaming_multi/exp/fast_beam_search/errs-test-clean-epoch-20-avg-4-streaming-chunk-size-64-beam-20.0-max-contexts-8-max-states-64-right-padding-64-use-averaged-model.txt
+2023-04-06 13:09:31,842 INFO [decode.py:608] 
+For test-clean, WER of different settings are:
+beam_20.0_max_contexts_8_max_states_64	2.27	best for test-clean
+
+2023-04-06 13:09:36,034 INFO [decode.py:569] batch 0/?, cuts processed until now is 30
+2023-04-06 13:10:16,221 INFO [decode.py:569] batch 20/?, cuts processed until now is 1771
+2023-04-06 13:10:49,392 INFO [decode.py:569] batch 40/?, cuts processed until now is 2696
+2023-04-06 13:11:16,867 INFO [decode.py:583] The transcripts are stored in pruned_transducer_stateless7_streaming_multi/exp/fast_beam_search/recogs-test-other-epoch-20-avg-4-streaming-chunk-size-64-beam-20.0-max-contexts-8-max-states-64-right-padding-64-use-averaged-model.txt
+2023-04-06 13:11:17,025 INFO [utils.py:558] [test-other-beam_20.0_max_contexts_8_max_states_64] %WER 5.54% [2900 / 52343, 305 ins, 250 del, 2345 sub ]
+2023-04-06 13:11:17,379 INFO [decode.py:594] Wrote detailed error stats to pruned_transducer_stateless7_streaming_multi/exp/fast_beam_search/errs-test-other-epoch-20-avg-4-streaming-chunk-size-64-beam-20.0-max-contexts-8-max-states-64-right-padding-64-use-averaged-model.txt
+2023-04-06 13:11:17,380 INFO [decode.py:608] 
+For test-other, WER of different settings are:
+beam_20.0_max_contexts_8_max_states_64	5.54	best for test-other
+
+2023-04-06 13:11:17,380 INFO [decode.py:814] Done!

decoding-results/fast_beam_search/log-decode-epoch-99-avg-1-streaming-chunk-size-32-beam-20.0-max-contexts-8-max-states-64-2023-04-04-11-02-26

@@ -0,0 +1,30 @@
+2023-04-04 11:02:26,737 INFO [decode.py:651] Decoding started
+2023-04-04 11:02:26,737 INFO [decode.py:657] Device: cuda:0
+2023-04-04 11:02:26,740 INFO [decode.py:667] {'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, 'warm_step': 2000, 'env_info': {'k2-version': '1.22', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': '96c9a2aece2a3a7633da07740e24fa3d96f5498c', 'k2-git-date': 'Thu Nov 10 08:14:02 2022', 'lhotse-version': '1.13.0.dev+git.527d964.clean', 'torch-version': '1.12.1', 'torch-cuda-available': True, 'torch-cuda-version': '11.6', 'python-version': '3.8', 'icefall-git-branch': 'zipformer_libri_small_models', 'icefall-git-sha1': '0994afb-dirty', 'icefall-git-date': 'Tue Apr 4 10:59:02 2023', 'icefall-path': '/ceph-data4/yangxiaoyu/softwares/icefall_development/icefall_small_models', 'k2-path': '/ceph-data4/yangxiaoyu/softwares/anaconda3/envs/k2_latest/lib/python3.8/site-packages/k2/__init__.py', 'lhotse-path': '/ceph-data4/yangxiaoyu/softwares/lhotse_development/lhotse_random_padding_left/lhotse/__init__.py', 'hostname': 'de-74279-k2-train-1-1220091118-57c4d55446-mlpzc', 'IP address': '10.177.22.19'}, 'epoch': 99, 'iter': 0, 'avg': 1, 'use_averaged_model': False, 'exp_dir': PosixPath('pruned_transducer_stateless7_streaming_multi/exp'), 'bpe_model': 'data/lang_bpe_500/bpe.model', 'lang_dir': PosixPath('data/lang_bpe_500'), 'decoding_method': 'fast_beam_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, '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, 'short_chunk_size': 50, 'num_left_chunks': 4, 'decode_chunk_len': 32, 'max_duration': 600, 'bucketing_sampler': True, 'num_buckets': 30, 'shuffle': True, 'return_cuts': True, 'num_workers': 2, 'on_the_fly_num_workers': 0, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'manifest_dir': PosixPath('data/fbank'), 'on_the_fly_feats': False, 'res_dir': PosixPath('pruned_transducer_stateless7_streaming_multi/exp/fast_beam_search'), 'suffix': 'epoch-99-avg-1-streaming-chunk-size-32-beam-20.0-max-contexts-8-max-states-64', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
+2023-04-04 11:02:26,740 INFO [decode.py:669] About to create model
+2023-04-04 11:02:27,365 INFO [zipformer.py:405] 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.
+2023-04-04 11:02:27,378 INFO [checkpoint.py:112] Loading checkpoint from pruned_transducer_stateless7_streaming_multi/exp/epoch-99.pt
+2023-04-04 11:02:29,736 INFO [decode.py:774] Number of model parameters: 70369391
+2023-04-04 11:02:29,736 INFO [librispeech.py:58] About to get test-clean cuts from data/fbank/librispeech_cuts_test-clean.jsonl.gz
+2023-04-04 11:02:29,738 INFO [librispeech.py:63] About to get test-other cuts from data/fbank/librispeech_cuts_test-other.jsonl.gz
+2023-04-04 11:02:37,465 INFO [decode.py:562] batch 0/?, cuts processed until now is 26
+2023-04-04 11:03:33,946 INFO [decode.py:562] batch 20/?, cuts processed until now is 1545
+2023-04-04 11:04:22,204 INFO [decode.py:562] batch 40/?, cuts processed until now is 2375
+2023-04-04 11:05:08,814 INFO [decode.py:576] The transcripts are stored in pruned_transducer_stateless7_streaming_multi/exp/fast_beam_search/recogs-test-clean-epoch-99-avg-1-streaming-chunk-size-32-beam-20.0-max-contexts-8-max-states-64.txt
+2023-04-04 11:05:08,911 INFO [utils.py:558] [test-clean-beam_20.0_max_contexts_8_max_states_64] %WER 2.47% [1298 / 52576, 160 ins, 96 del, 1042 sub ]
+2023-04-04 11:05:09,110 INFO [decode.py:587] Wrote detailed error stats to pruned_transducer_stateless7_streaming_multi/exp/fast_beam_search/errs-test-clean-epoch-99-avg-1-streaming-chunk-size-32-beam-20.0-max-contexts-8-max-states-64.txt
+2023-04-04 11:05:09,111 INFO [decode.py:601] 
+For test-clean, WER of different settings are:
+beam_20.0_max_contexts_8_max_states_64	2.47	best for test-clean
+
+2023-04-04 11:05:15,419 INFO [decode.py:562] batch 0/?, cuts processed until now is 30
+2023-04-04 11:06:16,018 INFO [decode.py:562] batch 20/?, cuts processed until now is 1771
+2023-04-04 11:06:47,363 INFO [decode.py:562] batch 40/?, cuts processed until now is 2696
+2023-04-04 11:07:09,305 INFO [decode.py:576] The transcripts are stored in pruned_transducer_stateless7_streaming_multi/exp/fast_beam_search/recogs-test-other-epoch-99-avg-1-streaming-chunk-size-32-beam-20.0-max-contexts-8-max-states-64.txt
+2023-04-04 11:07:09,398 INFO [utils.py:558] [test-other-beam_20.0_max_contexts_8_max_states_64] %WER 6.11% [3200 / 52343, 352 ins, 290 del, 2558 sub ]
+2023-04-04 11:07:09,598 INFO [decode.py:587] Wrote detailed error stats to pruned_transducer_stateless7_streaming_multi/exp/fast_beam_search/errs-test-other-epoch-99-avg-1-streaming-chunk-size-32-beam-20.0-max-contexts-8-max-states-64.txt
+2023-04-04 11:07:09,599 INFO [decode.py:601] 
+For test-other, WER of different settings are:
+beam_20.0_max_contexts_8_max_states_64	6.11	best for test-other
+
+2023-04-04 11:07:09,599 INFO [decode.py:806] Done!

decoding-results/fast_beam_search/log-decode-epoch-99-avg-1-streaming-chunk-size-64-beam-20.0-max-contexts-8-max-states-64-right-padding-64-2023-04-04-12-26-13

@@ -0,0 +1,50 @@
+2023-04-04 12:26:13,984 INFO [decode.py:659] Decoding started
+2023-04-04 12:26:13,984 INFO [decode.py:665] Device: cuda:0
+2023-04-04 12:26:13,992 INFO [decode.py:675] {'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, 'warm_step': 2000, 'env_info': {'k2-version': '1.22', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': '96c9a2aece2a3a7633da07740e24fa3d96f5498c', 'k2-git-date': 'Thu Nov 10 08:14:02 2022', 'lhotse-version': '1.13.0.dev+git.527d964.clean', 'torch-version': '1.12.1', 'torch-cuda-available': True, 'torch-cuda-version': '11.6', 'python-version': '3.8', 'icefall-git-branch': 'zipformer_libri_small_models', 'icefall-git-sha1': '475430b-dirty', 'icefall-git-date': 'Tue Apr 4 11:28:58 2023', 'icefall-path': '/ceph-data4/yangxiaoyu/softwares/icefall_development/icefall_small_models', 'k2-path': '/ceph-data4/yangxiaoyu/softwares/anaconda3/envs/k2_latest/lib/python3.8/site-packages/k2/__init__.py', 'lhotse-path': '/ceph-data4/yangxiaoyu/softwares/lhotse_development/lhotse_random_padding_left/lhotse/__init__.py', 'hostname': 'de-74279-k2-train-1-1220091118-57c4d55446-mlpzc', 'IP address': '10.177.22.19'}, 'epoch': 99, 'iter': 0, 'avg': 1, 'use_averaged_model': False, 'exp_dir': PosixPath('pruned_transducer_stateless7_streaming_multi/exp'), 'bpe_model': 'data/lang_bpe_500/bpe.model', 'lang_dir': PosixPath('data/lang_bpe_500'), 'decoding_method': 'fast_beam_search', 'beam_size': 4, 'beam': 20.0, 'ngram_lm_scale': 0.01, 'max_contexts': 8, 'max_states': 64, 'context_size': 2, 'right_padding': 64, 'max_sym_per_frame': 1, 'num_paths': 200, 'nbest_scale': 0.5, '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, 'short_chunk_size': 50, 'num_left_chunks': 4, 'decode_chunk_len': 64, 'max_duration': 600, 'bucketing_sampler': True, 'num_buckets': 30, 'shuffle': True, 'return_cuts': True, 'num_workers': 2, 'on_the_fly_num_workers': 0, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'manifest_dir': PosixPath('data/fbank'), 'on_the_fly_feats': False, 'res_dir': PosixPath('pruned_transducer_stateless7_streaming_multi/exp/fast_beam_search'), 'suffix': 'epoch-99-avg-1-streaming-chunk-size-64-beam-20.0-max-contexts-8-max-states-64-right-padding-64', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
+2023-04-04 12:26:13,992 INFO [decode.py:677] About to create model
+2023-04-04 12:26:14,594 INFO [zipformer.py:405] 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.
+2023-04-04 12:26:14,607 INFO [checkpoint.py:112] Loading checkpoint from pruned_transducer_stateless7_streaming_multi/exp/epoch-99.pt
+2023-04-04 12:26:16,838 INFO [decode.py:782] Number of model parameters: 70369391
+2023-04-04 12:26:16,838 INFO [librispeech.py:58] About to get test-clean cuts from data/fbank/librispeech_cuts_test-clean.jsonl.gz
+2023-04-04 12:26:16,841 INFO [librispeech.py:63] About to get test-other cuts from data/fbank/librispeech_cuts_test-other.jsonl.gz
+2023-04-04 12:26:22,248 INFO [decode.py:569] batch 0/?, cuts processed until now is 26
+2023-04-04 12:26:37,082 INFO [zipformer.py:2401] attn_weights_entropy = tensor([3.9816, 3.9442, 2.9777, 4.5003, 4.1074, 3.9360, 1.7938, 3.8771],
+       device='cuda:0'), covar=tensor([0.1328, 0.0678, 0.3053, 0.0896, 0.1707, 0.1995, 0.6684, 0.2012],
+       device='cuda:0'), in_proj_covar=tensor([0.0258, 0.0230, 0.0281, 0.0324, 0.0320, 0.0267, 0.0288, 0.0280],
+       device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002],
+       device='cuda:0')
+2023-04-04 12:26:59,982 INFO [decode.py:569] batch 20/?, cuts processed until now is 1545
+2023-04-04 12:27:01,565 INFO [zipformer.py:2401] attn_weights_entropy = tensor([1.6970, 1.9492, 0.9371, 1.3477, 2.0484, 1.5397, 1.4495, 1.4722],
+       device='cuda:0'), covar=tensor([0.0490, 0.0291, 0.0386, 0.0581, 0.0219, 0.0526, 0.0508, 0.0593],
+       device='cuda:0'), in_proj_covar=tensor([0.0035, 0.0028, 0.0026, 0.0035, 0.0023, 0.0035, 0.0034, 0.0036],
+       device='cuda:0'), out_proj_covar=tensor([0.0051, 0.0043, 0.0038, 0.0053, 0.0037, 0.0052, 0.0052, 0.0054],
+       device='cuda:0')
+2023-04-04 12:27:19,645 INFO [zipformer.py:2401] attn_weights_entropy = tensor([4.0814, 4.0133, 2.9767, 4.5842, 4.2312, 4.0406, 1.7885, 3.9094],
+       device='cuda:0'), covar=tensor([0.1133, 0.0571, 0.2740, 0.0753, 0.1763, 0.1667, 0.6636, 0.1986],
+       device='cuda:0'), in_proj_covar=tensor([0.0258, 0.0230, 0.0281, 0.0324, 0.0320, 0.0267, 0.0288, 0.0280],
+       device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002],
+       device='cuda:0')
+2023-04-04 12:27:30,069 INFO [decode.py:569] batch 40/?, cuts processed until now is 2375
+2023-04-04 12:27:54,759 INFO [decode.py:583] The transcripts are stored in pruned_transducer_stateless7_streaming_multi/exp/fast_beam_search/recogs-test-clean-epoch-99-avg-1-streaming-chunk-size-64-beam-20.0-max-contexts-8-max-states-64-right-padding-64.txt
+2023-04-04 12:27:54,855 INFO [utils.py:558] [test-clean-beam_20.0_max_contexts_8_max_states_64] %WER 2.34% [1231 / 52576, 142 ins, 99 del, 990 sub ]
+2023-04-04 12:27:55,057 INFO [decode.py:594] Wrote detailed error stats to pruned_transducer_stateless7_streaming_multi/exp/fast_beam_search/errs-test-clean-epoch-99-avg-1-streaming-chunk-size-64-beam-20.0-max-contexts-8-max-states-64-right-padding-64.txt
+2023-04-04 12:27:55,057 INFO [decode.py:608] 
+For test-clean, WER of different settings are:
+beam_20.0_max_contexts_8_max_states_64	2.34	best for test-clean
+
+2023-04-04 12:27:58,391 INFO [decode.py:569] batch 0/?, cuts processed until now is 30
+2023-04-04 12:28:33,992 INFO [decode.py:569] batch 20/?, cuts processed until now is 1771
+2023-04-04 12:28:51,985 INFO [zipformer.py:2401] attn_weights_entropy = tensor([2.2778, 2.1054, 2.4532, 2.7704, 2.6064, 2.0049, 1.4088, 2.3488],
+       device='cuda:0'), covar=tensor([0.0651, 0.0710, 0.0442, 0.0255, 0.0299, 0.0649, 0.0782, 0.0321],
+       device='cuda:0'), in_proj_covar=tensor([0.0204, 0.0213, 0.0191, 0.0172, 0.0174, 0.0193, 0.0167, 0.0184],
+       device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001],
+       device='cuda:0')
+2023-04-04 12:29:01,114 INFO [decode.py:569] batch 40/?, cuts processed until now is 2696
+2023-04-04 12:29:23,342 INFO [decode.py:583] The transcripts are stored in pruned_transducer_stateless7_streaming_multi/exp/fast_beam_search/recogs-test-other-epoch-99-avg-1-streaming-chunk-size-64-beam-20.0-max-contexts-8-max-states-64-right-padding-64.txt
+2023-04-04 12:29:23,443 INFO [utils.py:558] [test-other-beam_20.0_max_contexts_8_max_states_64] %WER 5.67% [2970 / 52343, 310 ins, 273 del, 2387 sub ]
+2023-04-04 12:29:23,648 INFO [decode.py:594] Wrote detailed error stats to pruned_transducer_stateless7_streaming_multi/exp/fast_beam_search/errs-test-other-epoch-99-avg-1-streaming-chunk-size-64-beam-20.0-max-contexts-8-max-states-64-right-padding-64.txt
+2023-04-04 12:29:23,649 INFO [decode.py:608] 
+For test-other, WER of different settings are:
+beam_20.0_max_contexts_8_max_states_64	5.67	best for test-other
+
+2023-04-04 12:29:23,649 INFO [decode.py:814] Done!

decoding-results/fast_beam_search/wer-summary-test-clean-epoch-20-avg-4-streaming-chunk-size-32-beam-20.0-max-contexts-8-max-states-64-right-padding-64-use-averaged-model.txt

@@ -0,0 +1,2 @@
+settings	WER
+beam_20.0_max_contexts_8_max_states_64	2.43

decoding-results/fast_beam_search/wer-summary-test-clean-epoch-20-avg-4-streaming-chunk-size-64-beam-20.0-max-contexts-8-max-states-64-right-padding-64-use-averaged-model.txt

@@ -0,0 +1,2 @@
+settings	WER
+beam_20.0_max_contexts_8_max_states_64	2.27

decoding-results/fast_beam_search/wer-summary-test-clean-epoch-99-avg-1-streaming-chunk-size-32-beam-20.0-max-contexts-8-max-states-64.txt

@@ -0,0 +1,2 @@
+settings	WER
+beam_20.0_max_contexts_8_max_states_64	2.47

decoding-results/fast_beam_search/wer-summary-test-clean-epoch-99-avg-1-streaming-chunk-size-64-beam-20.0-max-contexts-8-max-states-64-right-padding-64.txt

@@ -0,0 +1,2 @@
+settings	WER
+beam_20.0_max_contexts_8_max_states_64	2.34

decoding-results/fast_beam_search/wer-summary-test-other-epoch-20-avg-4-streaming-chunk-size-32-beam-20.0-max-contexts-8-max-states-64-right-padding-64-use-averaged-model.txt

@@ -0,0 +1,2 @@
+settings	WER
+beam_20.0_max_contexts_8_max_states_64	5.99

decoding-results/fast_beam_search/wer-summary-test-other-epoch-20-avg-4-streaming-chunk-size-64-beam-20.0-max-contexts-8-max-states-64-right-padding-64-use-averaged-model.txt

@@ -0,0 +1,2 @@
+settings	WER
+beam_20.0_max_contexts_8_max_states_64	5.54

decoding-results/fast_beam_search/wer-summary-test-other-epoch-99-avg-1-streaming-chunk-size-32-beam-20.0-max-contexts-8-max-states-64.txt

@@ -0,0 +1,2 @@
+settings	WER
+beam_20.0_max_contexts_8_max_states_64	6.11

decoding-results/fast_beam_search/wer-summary-test-other-epoch-99-avg-1-streaming-chunk-size-64-beam-20.0-max-contexts-8-max-states-64-right-padding-64.txt

@@ -0,0 +1,2 @@
+settings	WER
+beam_20.0_max_contexts_8_max_states_64	5.67

decoding-results/greedy_search/log-decode-epoch-20-avg-4-streaming-chunk-size-32-context-2-max-sym-per-frame-1-use-averaged-model-2023-04-06-10-45-22

@@ -0,0 +1,43 @@
+2023-04-06 10:45:22,213 INFO [decode.py:659] Decoding started
+2023-04-06 10:45:22,213 INFO [decode.py:665] Device: cuda:0
+2023-04-06 10:45:22,217 INFO [decode.py:675] {'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, 'warm_step': 2000, 'env_info': {'k2-version': '1.22', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': '96c9a2aece2a3a7633da07740e24fa3d96f5498c', 'k2-git-date': 'Thu Nov 10 08:14:02 2022', 'lhotse-version': '1.13.0.dev+git.527d964.clean', 'torch-version': '1.12.1', 'torch-cuda-available': True, 'torch-cuda-version': '11.6', 'python-version': '3.8', 'icefall-git-branch': 'zipformer_libri_small_models', 'icefall-git-sha1': 'b52e7ae-dirty', 'icefall-git-date': 'Tue Apr 4 14:07:45 2023', 'icefall-path': '/ceph-data4/yangxiaoyu/softwares/icefall_development/icefall_small_models', 'k2-path': '/ceph-data4/yangxiaoyu/softwares/anaconda3/envs/k2_latest/lib/python3.8/site-packages/k2/__init__.py', 'lhotse-path': '/ceph-data4/yangxiaoyu/softwares/lhotse_development/lhotse_random_padding_left/lhotse/__init__.py', 'hostname': 'de-74279-k2-train-6-1219221738-65dd59bbf8-2ghmr', 'IP address': '10.177.28.85'}, 'epoch': 20, 'iter': 0, 'avg': 4, 'use_averaged_model': True, 'exp_dir': PosixPath('pruned_transducer_stateless7_streaming_multi/exp'), '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, 'right_padding': 64, 'max_sym_per_frame': 1, 'num_paths': 200, 'nbest_scale': 0.5, '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, 'short_chunk_size': 50, 'num_left_chunks': 4, 'decode_chunk_len': 32, 'max_duration': 600, 'bucketing_sampler': True, 'num_buckets': 30, 'shuffle': True, 'return_cuts': True, 'num_workers': 2, 'on_the_fly_num_workers': 0, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'manifest_dir': PosixPath('data/fbank'), 'on_the_fly_feats': False, 'res_dir': PosixPath('pruned_transducer_stateless7_streaming_multi/exp/greedy_search'), 'suffix': 'epoch-20-avg-4-streaming-chunk-size-32-context-2-max-sym-per-frame-1-use-averaged-model', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
+2023-04-06 10:45:22,217 INFO [decode.py:677] About to create model
+2023-04-06 10:45:23,051 INFO [zipformer.py:405] 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.
+2023-04-06 10:45:23,071 INFO [decode.py:748] Calculating the averaged model over epoch range from 16 (excluded) to 20
+2023-04-06 10:45:28,421 INFO [decode.py:782] Number of model parameters: 70369391
+2023-04-06 10:45:28,421 INFO [librispeech.py:58] About to get test-clean cuts from data/fbank/librispeech_cuts_test-clean.jsonl.gz
+2023-04-06 10:45:28,424 INFO [librispeech.py:63] About to get test-other cuts from data/fbank/librispeech_cuts_test-other.jsonl.gz
+2023-04-06 10:45:34,938 INFO [decode.py:569] batch 0/?, cuts processed until now is 26
+2023-04-06 10:46:11,317 INFO [zipformer.py:2441] attn_weights_entropy = tensor([0.6097, 1.6395, 1.8269, 1.2424, 1.8390, 1.4426, 2.2329, 1.6805],
+       device='cuda:0'), covar=tensor([0.4370, 0.1894, 0.5571, 0.3727, 0.1697, 0.2827, 0.1812, 0.4848],
+       device='cuda:0'), in_proj_covar=tensor([0.0381, 0.0394, 0.0474, 0.0408, 0.0456, 0.0428, 0.0445, 0.0457],
+       device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001],
+       device='cuda:0')
+2023-04-06 10:46:38,790 INFO [decode.py:569] batch 50/?, cuts processed until now is 2526
+2023-04-06 10:46:44,017 INFO [decode.py:583] The transcripts are stored in pruned_transducer_stateless7_streaming_multi/exp/greedy_search/recogs-test-clean-epoch-20-avg-4-streaming-chunk-size-32-context-2-max-sym-per-frame-1-use-averaged-model.txt
+2023-04-06 10:46:44,167 INFO [utils.py:558] [test-clean-greedy_search] %WER 2.43% [1280 / 52576, 146 ins, 105 del, 1029 sub ]
+2023-04-06 10:46:44,511 INFO [decode.py:594] Wrote detailed error stats to pruned_transducer_stateless7_streaming_multi/exp/greedy_search/errs-test-clean-epoch-20-avg-4-streaming-chunk-size-32-context-2-max-sym-per-frame-1-use-averaged-model.txt
+2023-04-06 10:46:44,511 INFO [decode.py:608] 
+For test-clean, WER of different settings are:
+greedy_search	2.43	best for test-clean
+
+2023-04-06 10:46:47,347 INFO [decode.py:569] batch 0/?, cuts processed until now is 30
+2023-04-06 10:47:20,989 INFO [zipformer.py:2441] attn_weights_entropy = tensor([1.4628, 2.1425, 1.6960, 2.0355, 1.7270, 1.8142, 1.7535, 1.5162],
+       device='cuda:0'), covar=tensor([0.1946, 0.0790, 0.0849, 0.0845, 0.2228, 0.0746, 0.1337, 0.1364],
+       device='cuda:0'), in_proj_covar=tensor([0.0321, 0.0340, 0.0246, 0.0312, 0.0328, 0.0285, 0.0290, 0.0303],
+       device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001],
+       device='cuda:0')
+2023-04-06 10:47:32,540 INFO [zipformer.py:2441] attn_weights_entropy = tensor([1.5169, 1.4365, 0.6674, 1.2004, 1.4270, 1.3728, 1.2824, 1.2656],
+       device='cuda:0'), covar=tensor([0.0450, 0.0286, 0.0379, 0.0536, 0.0204, 0.0468, 0.0452, 0.0543],
+       device='cuda:0'), in_proj_covar=tensor([0.0035, 0.0028, 0.0026, 0.0035, 0.0023, 0.0035, 0.0035, 0.0037],
+       device='cuda:0'), out_proj_covar=tensor([0.0053, 0.0045, 0.0039, 0.0056, 0.0039, 0.0054, 0.0054, 0.0057],
+       device='cuda:0')
+2023-04-06 10:47:39,863 INFO [decode.py:569] batch 50/?, cuts processed until now is 2840
+2023-04-06 10:47:44,252 INFO [decode.py:583] The transcripts are stored in pruned_transducer_stateless7_streaming_multi/exp/greedy_search/recogs-test-other-epoch-20-avg-4-streaming-chunk-size-32-context-2-max-sym-per-frame-1-use-averaged-model.txt
+2023-04-06 10:47:44,408 INFO [utils.py:558] [test-other-greedy_search] %WER 6.00% [3138 / 52343, 313 ins, 301 del, 2524 sub ]
+2023-04-06 10:47:44,755 INFO [decode.py:594] Wrote detailed error stats to pruned_transducer_stateless7_streaming_multi/exp/greedy_search/errs-test-other-epoch-20-avg-4-streaming-chunk-size-32-context-2-max-sym-per-frame-1-use-averaged-model.txt
+2023-04-06 10:47:44,756 INFO [decode.py:608] 
+For test-other, WER of different settings are:
+greedy_search	6.0	best for test-other
+
+2023-04-06 10:47:44,756 INFO [decode.py:814] Done!

decoding-results/greedy_search/log-decode-epoch-20-avg-4-streaming-chunk-size-64-context-2-max-sym-per-frame-1-use-averaged-model-2023-04-06-10-42-45

@@ -0,0 +1,28 @@
+2023-04-06 10:42:45,288 INFO [decode.py:659] Decoding started
+2023-04-06 10:42:45,288 INFO [decode.py:665] Device: cuda:0
+2023-04-06 10:42:45,291 INFO [decode.py:675] {'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, 'warm_step': 2000, 'env_info': {'k2-version': '1.22', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': '96c9a2aece2a3a7633da07740e24fa3d96f5498c', 'k2-git-date': 'Thu Nov 10 08:14:02 2022', 'lhotse-version': '1.13.0.dev+git.527d964.clean', 'torch-version': '1.12.1', 'torch-cuda-available': True, 'torch-cuda-version': '11.6', 'python-version': '3.8', 'icefall-git-branch': 'zipformer_libri_small_models', 'icefall-git-sha1': 'b52e7ae-dirty', 'icefall-git-date': 'Tue Apr 4 14:07:45 2023', 'icefall-path': '/ceph-data4/yangxiaoyu/softwares/icefall_development/icefall_small_models', 'k2-path': '/ceph-data4/yangxiaoyu/softwares/anaconda3/envs/k2_latest/lib/python3.8/site-packages/k2/__init__.py', 'lhotse-path': '/ceph-data4/yangxiaoyu/softwares/lhotse_development/lhotse_random_padding_left/lhotse/__init__.py', 'hostname': 'de-74279-k2-train-6-1219221738-65dd59bbf8-2ghmr', 'IP address': '10.177.28.85'}, 'epoch': 20, 'iter': 0, 'avg': 4, 'use_averaged_model': True, 'exp_dir': PosixPath('pruned_transducer_stateless7_streaming_multi/exp'), '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, 'right_padding': 64, 'max_sym_per_frame': 1, 'num_paths': 200, 'nbest_scale': 0.5, '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, 'short_chunk_size': 50, 'num_left_chunks': 4, 'decode_chunk_len': 64, 'max_duration': 600, 'bucketing_sampler': True, 'num_buckets': 30, 'shuffle': True, 'return_cuts': True, 'num_workers': 2, 'on_the_fly_num_workers': 0, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'manifest_dir': PosixPath('data/fbank'), 'on_the_fly_feats': False, 'res_dir': PosixPath('pruned_transducer_stateless7_streaming_multi/exp/greedy_search'), 'suffix': 'epoch-20-avg-4-streaming-chunk-size-64-context-2-max-sym-per-frame-1-use-averaged-model', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
+2023-04-06 10:42:45,292 INFO [decode.py:677] About to create model
+2023-04-06 10:42:46,106 INFO [zipformer.py:405] 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.
+2023-04-06 10:42:46,126 INFO [decode.py:748] Calculating the averaged model over epoch range from 16 (excluded) to 20
+2023-04-06 10:42:56,583 INFO [decode.py:782] Number of model parameters: 70369391
+2023-04-06 10:42:56,583 INFO [librispeech.py:58] About to get test-clean cuts from data/fbank/librispeech_cuts_test-clean.jsonl.gz
+2023-04-06 10:42:56,586 INFO [librispeech.py:63] About to get test-other cuts from data/fbank/librispeech_cuts_test-other.jsonl.gz
+2023-04-06 10:43:02,305 INFO [decode.py:569] batch 0/?, cuts processed until now is 26
+2023-04-06 10:44:00,603 INFO [decode.py:569] batch 50/?, cuts processed until now is 2526
+2023-04-06 10:44:05,834 INFO [decode.py:583] The transcripts are stored in pruned_transducer_stateless7_streaming_multi/exp/greedy_search/recogs-test-clean-epoch-20-avg-4-streaming-chunk-size-64-context-2-max-sym-per-frame-1-use-averaged-model.txt
+2023-04-06 10:44:05,981 INFO [utils.py:558] [test-clean-greedy_search] %WER 2.26% [1190 / 52576, 122 ins, 97 del, 971 sub ]
+2023-04-06 10:44:06,327 INFO [decode.py:594] Wrote detailed error stats to pruned_transducer_stateless7_streaming_multi/exp/greedy_search/errs-test-clean-epoch-20-avg-4-streaming-chunk-size-64-context-2-max-sym-per-frame-1-use-averaged-model.txt
+2023-04-06 10:44:06,328 INFO [decode.py:608] 
+For test-clean, WER of different settings are:
+greedy_search	2.26	best for test-clean
+
+2023-04-06 10:44:09,181 INFO [decode.py:569] batch 0/?, cuts processed until now is 30
+2023-04-06 10:45:10,920 INFO [decode.py:569] batch 50/?, cuts processed until now is 2840
+2023-04-06 10:45:17,011 INFO [decode.py:583] The transcripts are stored in pruned_transducer_stateless7_streaming_multi/exp/greedy_search/recogs-test-other-epoch-20-avg-4-streaming-chunk-size-64-context-2-max-sym-per-frame-1-use-averaged-model.txt
+2023-04-06 10:45:17,175 INFO [utils.py:558] [test-other-greedy_search] %WER 5.58% [2920 / 52343, 285 ins, 260 del, 2375 sub ]
+2023-04-06 10:45:17,534 INFO [decode.py:594] Wrote detailed error stats to pruned_transducer_stateless7_streaming_multi/exp/greedy_search/errs-test-other-epoch-20-avg-4-streaming-chunk-size-64-context-2-max-sym-per-frame-1-use-averaged-model.txt
+2023-04-06 10:45:17,535 INFO [decode.py:608] 
+For test-other, WER of different settings are:
+greedy_search	5.58	best for test-other
+
+2023-04-06 10:45:17,535 INFO [decode.py:814] Done!

decoding-results/greedy_search/log-decode-epoch-99-avg-1-context-2-max-sym-per-frame-1-2023-04-04-10-36-37

@@ -0,0 +1,48 @@
+2023-04-04 10:36:37,574 INFO [decode.py:683] Decoding started
+2023-04-04 10:36:37,574 INFO [decode.py:689] Device: cuda:0
+2023-04-04 10:36:37,579 INFO [decode.py:699] {'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, 'warm_step': 2000, 'env_info': {'k2-version': '1.22', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': '96c9a2aece2a3a7633da07740e24fa3d96f5498c', 'k2-git-date': 'Thu Nov 10 08:14:02 2022', 'lhotse-version': '1.13.0.dev+git.527d964.clean', 'torch-version': '1.12.1', 'torch-cuda-available': True, 'torch-cuda-version': '11.6', 'python-version': '3.8', 'icefall-git-branch': 'zipformer_libri_small_models', 'icefall-git-sha1': '1a059bd-dirty', 'icefall-git-date': 'Mon Apr 3 23:17:14 2023', 'icefall-path': '/ceph-data4/yangxiaoyu/softwares/icefall_development/icefall_small_models', 'k2-path': '/ceph-data4/yangxiaoyu/softwares/anaconda3/envs/k2_latest/lib/python3.8/site-packages/k2/__init__.py', 'lhotse-path': '/ceph-data4/yangxiaoyu/softwares/lhotse_development/lhotse_random_padding_left/lhotse/__init__.py', 'hostname': 'de-74279-k2-train-1-1220091118-57c4d55446-mlpzc', 'IP address': '10.177.22.19'}, 'epoch': 99, 'iter': 0, 'avg': 1, 'use_averaged_model': False, 'exp_dir': PosixPath('pruned_transducer_stateless7_streaming_multi/exp'), '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, '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, 'short_chunk_size': 50, 'num_left_chunks': 4, 'decode_chunk_len': 32, 'max_duration': 600, 'bucketing_sampler': True, 'num_buckets': 30, 'shuffle': True, 'return_cuts': True, 'num_workers': 2, 'on_the_fly_num_workers': 0, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'manifest_dir': PosixPath('data/fbank'), 'on_the_fly_feats': False, 'res_dir': PosixPath('pruned_transducer_stateless7_streaming_multi/exp/greedy_search'), 'suffix': 'epoch-99-avg-1-context-2-max-sym-per-frame-1', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
+2023-04-04 10:36:37,580 INFO [decode.py:701] About to create model
+2023-04-04 10:36:38,169 INFO [zipformer.py:405] 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.
+2023-04-04 10:36:38,183 INFO [checkpoint.py:112] Loading checkpoint from pruned_transducer_stateless7_streaming_multi/exp/epoch-99.pt
+2023-04-04 10:36:40,938 INFO [decode.py:806] Number of model parameters: 70369391
+2023-04-04 10:36:40,938 INFO [librispeech.py:58] About to get test-clean cuts from data/fbank/librispeech_cuts_test-clean.jsonl.gz
+2023-04-04 10:36:40,940 INFO [librispeech.py:63] About to get test-other cuts from data/fbank/librispeech_cuts_test-other.jsonl.gz
+2023-04-04 10:36:45,198 INFO [decode.py:586] batch 0/?, cuts processed until now is 26
+2023-04-04 10:37:25,210 INFO [zipformer.py:2401] attn_weights_entropy = tensor([2.5399, 1.3419, 1.7835, 1.8547, 1.8171, 1.8067, 1.4053, 1.6264],
+       device='cuda:0'), covar=tensor([0.1138, 0.6186, 0.2528, 0.2486, 0.5720, 0.5618, 0.6822, 0.3789],
+       device='cuda:0'), in_proj_covar=tensor([0.0340, 0.0501, 0.0399, 0.0388, 0.0431, 0.0436, 0.0483, 0.0419],
+       device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002],
+       device='cuda:0')
+2023-04-04 10:37:34,207 INFO [decode.py:586] batch 50/?, cuts processed until now is 2526
+2023-04-04 10:37:38,421 INFO [decode.py:602] The transcripts are stored in pruned_transducer_stateless7_streaming_multi/exp/greedy_search/recogs-test-clean-greedy_search-epoch-99-avg-1-context-2-max-sym-per-frame-1.txt
+2023-04-04 10:37:38,518 INFO [utils.py:558] [test-clean-greedy_search] %WER 2.69% [1414 / 52576, 143 ins, 107 del, 1164 sub ]
+2023-04-04 10:37:38,718 INFO [decode.py:615] Wrote detailed error stats to pruned_transducer_stateless7_streaming_multi/exp/greedy_search/errs-test-clean-greedy_search-epoch-99-avg-1-context-2-max-sym-per-frame-1.txt
+2023-04-04 10:37:38,719 INFO [decode.py:631] 
+For test-clean, WER of different settings are:
+greedy_search	2.69	best for test-clean
+
+2023-04-04 10:37:40,990 INFO [decode.py:586] batch 0/?, cuts processed until now is 30
+2023-04-04 10:38:09,117 INFO [zipformer.py:2401] attn_weights_entropy = tensor([1.6377, 1.5390, 0.6662, 1.2690, 1.5933, 1.4750, 1.3637, 1.3537],
+       device='cuda:0'), covar=tensor([0.0499, 0.0341, 0.0432, 0.0619, 0.0252, 0.0532, 0.0529, 0.0637],
+       device='cuda:0'), in_proj_covar=tensor([0.0035, 0.0028, 0.0026, 0.0035, 0.0023, 0.0035, 0.0034, 0.0036],
+       device='cuda:0'), out_proj_covar=tensor([0.0051, 0.0043, 0.0038, 0.0053, 0.0037, 0.0052, 0.0052, 0.0054],
+       device='cuda:0')
+2023-04-04 10:38:18,838 INFO [zipformer.py:2401] attn_weights_entropy = tensor([1.9599, 2.2519, 0.9427, 1.1409, 1.5647, 1.2337, 2.8050, 1.6180],
+       device='cuda:0'), covar=tensor([0.0496, 0.0360, 0.0586, 0.1042, 0.0491, 0.0775, 0.0378, 0.0566],
+       device='cuda:0'), in_proj_covar=tensor([0.0058, 0.0075, 0.0053, 0.0051, 0.0056, 0.0056, 0.0094, 0.0054],
+       device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0010, 0.0007, 0.0008, 0.0008, 0.0008, 0.0012, 0.0007],
+       device='cuda:0')
+2023-04-04 10:38:24,826 INFO [decode.py:586] batch 50/?, cuts processed until now is 2840
+2023-04-04 10:38:25,708 INFO [zipformer.py:2401] attn_weights_entropy = tensor([2.0014, 1.1741, 1.8564, 2.1889, 1.8678, 1.7254, 1.6541, 1.7504],
+       device='cuda:0'), covar=tensor([0.2613, 0.7024, 0.3231, 0.5580, 0.4379, 0.6271, 0.6111, 0.6292],
+       device='cuda:0'), in_proj_covar=tensor([0.0507, 0.0581, 0.0659, 0.0635, 0.0561, 0.0612, 0.0649, 0.0635],
+       device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002],
+       device='cuda:0')
+2023-04-04 10:38:28,421 INFO [decode.py:602] The transcripts are stored in pruned_transducer_stateless7_streaming_multi/exp/greedy_search/recogs-test-other-greedy_search-epoch-99-avg-1-context-2-max-sym-per-frame-1.txt
+2023-04-04 10:38:28,524 INFO [utils.py:558] [test-other-greedy_search] %WER 6.35% [3323 / 52343, 322 ins, 322 del, 2679 sub ]
+2023-04-04 10:38:28,727 INFO [decode.py:615] Wrote detailed error stats to pruned_transducer_stateless7_streaming_multi/exp/greedy_search/errs-test-other-greedy_search-epoch-99-avg-1-context-2-max-sym-per-frame-1.txt
+2023-04-04 10:38:28,727 INFO [decode.py:631] 
+For test-other, WER of different settings are:
+greedy_search	6.35	best for test-other
+
+2023-04-04 10:38:28,727 INFO [decode.py:836] Done!

decoding-results/greedy_search/log-decode-epoch-99-avg-1-streaming-chunk-size-32-context-2-max-sym-per-frame-1-2023-04-04-10-58-09

@@ -0,0 +1,38 @@
+2023-04-04 10:58:09,872 INFO [decode.py:650] Decoding started
+2023-04-04 10:58:09,872 INFO [decode.py:656] Device: cuda:0
+2023-04-04 10:58:09,874 INFO [decode.py:666] {'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, 'warm_step': 2000, 'env_info': {'k2-version': '1.22', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': '96c9a2aece2a3a7633da07740e24fa3d96f5498c', 'k2-git-date': 'Thu Nov 10 08:14:02 2022', 'lhotse-version': '1.13.0.dev+git.527d964.clean', 'torch-version': '1.12.1', 'torch-cuda-available': True, 'torch-cuda-version': '11.6', 'python-version': '3.8', 'icefall-git-branch': 'zipformer_libri_small_models', 'icefall-git-sha1': '1a059bd-dirty', 'icefall-git-date': 'Mon Apr 3 23:17:14 2023', 'icefall-path': '/ceph-data4/yangxiaoyu/softwares/icefall_development/icefall_small_models', 'k2-path': '/ceph-data4/yangxiaoyu/softwares/anaconda3/envs/k2_latest/lib/python3.8/site-packages/k2/__init__.py', 'lhotse-path': '/ceph-data4/yangxiaoyu/softwares/lhotse_development/lhotse_random_padding_left/lhotse/__init__.py', 'hostname': 'de-74279-k2-train-1-1220091118-57c4d55446-mlpzc', 'IP address': '10.177.22.19'}, 'epoch': 99, 'iter': 0, 'avg': 1, 'use_averaged_model': False, 'exp_dir': PosixPath('pruned_transducer_stateless7_streaming_multi/exp'), '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, '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, 'short_chunk_size': 50, 'num_left_chunks': 4, 'decode_chunk_len': 32, 'max_duration': 600, 'bucketing_sampler': True, 'num_buckets': 30, 'shuffle': True, 'return_cuts': True, 'num_workers': 2, 'on_the_fly_num_workers': 0, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'manifest_dir': PosixPath('data/fbank'), 'on_the_fly_feats': False, 'res_dir': PosixPath('pruned_transducer_stateless7_streaming_multi/exp/greedy_search'), 'suffix': 'epoch-99-avg-1-streaming-chunk-size-32-context-2-max-sym-per-frame-1', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
+2023-04-04 10:58:09,874 INFO [decode.py:668] About to create model
+2023-04-04 10:58:10,475 INFO [zipformer.py:405] 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.
+2023-04-04 10:58:10,488 INFO [checkpoint.py:112] Loading checkpoint from pruned_transducer_stateless7_streaming_multi/exp/epoch-99.pt
+2023-04-04 10:58:12,820 INFO [decode.py:773] Number of model parameters: 70369391
+2023-04-04 10:58:12,820 INFO [librispeech.py:58] About to get test-clean cuts from data/fbank/librispeech_cuts_test-clean.jsonl.gz
+2023-04-04 10:58:12,822 INFO [librispeech.py:63] About to get test-other cuts from data/fbank/librispeech_cuts_test-other.jsonl.gz
+2023-04-04 10:58:17,163 INFO [decode.py:561] batch 0/?, cuts processed until now is 26
+2023-04-04 10:59:04,873 INFO [decode.py:561] batch 50/?, cuts processed until now is 2526
+2023-04-04 10:59:08,082 INFO [zipformer.py:2401] attn_weights_entropy = tensor([1.3428, 1.1291, 3.8454, 3.5745, 3.3987, 3.8075, 3.9097, 3.3450],
+       device='cuda:0'), covar=tensor([0.7416, 0.6198, 0.1054, 0.1629, 0.1179, 0.1146, 0.0398, 0.1268],
+       device='cuda:0'), in_proj_covar=tensor([0.0336, 0.0311, 0.0437, 0.0450, 0.0361, 0.0415, 0.0346, 0.0390],
+       device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002],
+       device='cuda:0')
+2023-04-04 10:59:08,817 INFO [decode.py:575] The transcripts are stored in pruned_transducer_stateless7_streaming_multi/exp/greedy_search/recogs-test-clean-epoch-99-avg-1-streaming-chunk-size-32-context-2-max-sym-per-frame-1.txt
+2023-04-04 10:59:08,914 INFO [utils.py:558] [test-clean-greedy_search] %WER 2.46% [1292 / 52576, 143 ins, 101 del, 1048 sub ]
+2023-04-04 10:59:09,114 INFO [decode.py:586] Wrote detailed error stats to pruned_transducer_stateless7_streaming_multi/exp/greedy_search/errs-test-clean-epoch-99-avg-1-streaming-chunk-size-32-context-2-max-sym-per-frame-1.txt
+2023-04-04 10:59:09,114 INFO [decode.py:600] 
+For test-clean, WER of different settings are:
+greedy_search	2.46	best for test-clean
+
+2023-04-04 10:59:11,273 INFO [decode.py:561] batch 0/?, cuts processed until now is 30
+2023-04-04 10:59:38,454 INFO [zipformer.py:2401] attn_weights_entropy = tensor([1.3609, 1.1987, 3.8710, 3.5310, 3.4165, 3.6922, 3.8854, 3.3341],
+       device='cuda:0'), covar=tensor([0.6862, 0.5960, 0.0934, 0.1721, 0.1150, 0.1188, 0.0519, 0.1388],
+       device='cuda:0'), in_proj_covar=tensor([0.0336, 0.0311, 0.0437, 0.0450, 0.0361, 0.0415, 0.0346, 0.0390],
+       device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002],
+       device='cuda:0')
+2023-04-04 10:59:53,538 INFO [decode.py:561] batch 50/?, cuts processed until now is 2840
+2023-04-04 10:59:57,073 INFO [decode.py:575] The transcripts are stored in pruned_transducer_stateless7_streaming_multi/exp/greedy_search/recogs-test-other-epoch-99-avg-1-streaming-chunk-size-32-context-2-max-sym-per-frame-1.txt
+2023-04-04 10:59:57,175 INFO [utils.py:558] [test-other-greedy_search] %WER 6.16% [3224 / 52343, 322 ins, 303 del, 2599 sub ]
+2023-04-04 10:59:57,384 INFO [decode.py:586] Wrote detailed error stats to pruned_transducer_stateless7_streaming_multi/exp/greedy_search/errs-test-other-epoch-99-avg-1-streaming-chunk-size-32-context-2-max-sym-per-frame-1.txt
+2023-04-04 10:59:57,385 INFO [decode.py:600] 
+For test-other, WER of different settings are:
+greedy_search	6.16	best for test-other
+
+2023-04-04 10:59:57,385 INFO [decode.py:805] Done!

decoding-results/greedy_search/log-decode-epoch-99-avg-1-streaming-chunk-size-32-context-2-max-sym-per-frame-1-2023-04-04-11-22-52

@@ -0,0 +1,38 @@
+2023-04-04 11:22:52,235 INFO [decode.py:659] Decoding started
+2023-04-04 11:22:52,235 INFO [decode.py:665] Device: cuda:0
+2023-04-04 11:22:52,239 INFO [decode.py:675] {'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, 'warm_step': 2000, 'env_info': {'k2-version': '1.22', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': '96c9a2aece2a3a7633da07740e24fa3d96f5498c', 'k2-git-date': 'Thu Nov 10 08:14:02 2022', 'lhotse-version': '1.13.0.dev+git.527d964.clean', 'torch-version': '1.12.1', 'torch-cuda-available': True, 'torch-cuda-version': '11.6', 'python-version': '3.8', 'icefall-git-branch': 'zipformer_libri_small_models', 'icefall-git-sha1': '0994afb-dirty', 'icefall-git-date': 'Tue Apr 4 10:59:02 2023', 'icefall-path': '/ceph-data4/yangxiaoyu/softwares/icefall_development/icefall_small_models', 'k2-path': '/ceph-data4/yangxiaoyu/softwares/anaconda3/envs/k2_latest/lib/python3.8/site-packages/k2/__init__.py', 'lhotse-path': '/ceph-data4/yangxiaoyu/softwares/lhotse_development/lhotse_random_padding_left/lhotse/__init__.py', 'hostname': 'de-74279-k2-train-1-1220091118-57c4d55446-mlpzc', 'IP address': '10.177.22.19'}, 'epoch': 99, 'iter': 0, 'avg': 1, 'use_averaged_model': False, 'exp_dir': PosixPath('pruned_transducer_stateless7_streaming_multi/exp'), '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, 'right_padding': 64, 'max_sym_per_frame': 1, 'num_paths': 200, 'nbest_scale': 0.5, '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, 'short_chunk_size': 50, 'num_left_chunks': 4, 'decode_chunk_len': 32, 'max_duration': 600, 'bucketing_sampler': True, 'num_buckets': 30, 'shuffle': True, 'return_cuts': True, 'num_workers': 2, 'on_the_fly_num_workers': 0, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'manifest_dir': PosixPath('data/fbank'), 'on_the_fly_feats': False, 'res_dir': PosixPath('pruned_transducer_stateless7_streaming_multi/exp/greedy_search'), 'suffix': 'epoch-99-avg-1-streaming-chunk-size-32-context-2-max-sym-per-frame-1', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
+2023-04-04 11:22:52,239 INFO [decode.py:677] About to create model
+2023-04-04 11:22:52,830 INFO [zipformer.py:405] 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.
+2023-04-04 11:22:52,846 INFO [checkpoint.py:112] Loading checkpoint from pruned_transducer_stateless7_streaming_multi/exp/epoch-99.pt
+2023-04-04 11:22:55,051 INFO [decode.py:782] Number of model parameters: 70369391
+2023-04-04 11:22:55,052 INFO [librispeech.py:58] About to get test-clean cuts from data/fbank/librispeech_cuts_test-clean.jsonl.gz
+2023-04-04 11:22:55,054 INFO [librispeech.py:63] About to get test-other cuts from data/fbank/librispeech_cuts_test-other.jsonl.gz
+2023-04-04 11:22:59,267 INFO [decode.py:569] batch 0/?, cuts processed until now is 26
+2023-04-04 11:23:47,552 INFO [decode.py:569] batch 50/?, cuts processed until now is 2526
+2023-04-04 11:23:51,606 INFO [decode.py:583] The transcripts are stored in pruned_transducer_stateless7_streaming_multi/exp/greedy_search/recogs-test-clean-epoch-99-avg-1-streaming-chunk-size-32-context-2-max-sym-per-frame-1.txt
+2023-04-04 11:23:51,699 INFO [utils.py:558] [test-clean-greedy_search] %WER 2.46% [1292 / 52576, 148 ins, 99 del, 1045 sub ]
+2023-04-04 11:23:51,895 INFO [decode.py:594] Wrote detailed error stats to pruned_transducer_stateless7_streaming_multi/exp/greedy_search/errs-test-clean-epoch-99-avg-1-streaming-chunk-size-32-context-2-max-sym-per-frame-1.txt
+2023-04-04 11:23:51,896 INFO [decode.py:608] 
+For test-clean, WER of different settings are:
+greedy_search	2.46	best for test-clean
+
+2023-04-04 11:23:54,113 INFO [decode.py:569] batch 0/?, cuts processed until now is 30
+2023-04-04 11:24:01,474 INFO [zipformer.py:2401] attn_weights_entropy = tensor([2.1750, 2.5418, 1.0587, 1.2736, 1.7698, 1.3217, 3.1138, 1.7876],
+       device='cuda:0'), covar=tensor([0.0603, 0.0441, 0.0655, 0.1150, 0.0538, 0.0888, 0.0428, 0.0611],
+       device='cuda:0'), in_proj_covar=tensor([0.0058, 0.0075, 0.0053, 0.0051, 0.0056, 0.0056, 0.0094, 0.0054],
+       device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0010, 0.0007, 0.0008, 0.0008, 0.0008, 0.0012, 0.0007],
+       device='cuda:0')
+2023-04-04 11:24:12,789 INFO [zipformer.py:2401] attn_weights_entropy = tensor([1.8796, 1.6856, 1.9582, 2.2762, 2.1299, 1.6889, 1.1918, 1.9793],
+       device='cuda:0'), covar=tensor([0.0562, 0.0823, 0.0437, 0.0278, 0.0337, 0.0568, 0.0736, 0.0329],
+       device='cuda:0'), in_proj_covar=tensor([0.0204, 0.0213, 0.0191, 0.0172, 0.0174, 0.0193, 0.0167, 0.0184],
+       device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001],
+       device='cuda:0')
+2023-04-04 11:24:38,235 INFO [decode.py:569] batch 50/?, cuts processed until now is 2840
+2023-04-04 11:24:41,805 INFO [decode.py:583] The transcripts are stored in pruned_transducer_stateless7_streaming_multi/exp/greedy_search/recogs-test-other-epoch-99-avg-1-streaming-chunk-size-32-context-2-max-sym-per-frame-1.txt
+2023-04-04 11:24:41,911 INFO [utils.py:558] [test-other-greedy_search] %WER 6.18% [3235 / 52343, 335 ins, 303 del, 2597 sub ]
+2023-04-04 11:24:42,118 INFO [decode.py:594] Wrote detailed error stats to pruned_transducer_stateless7_streaming_multi/exp/greedy_search/errs-test-other-epoch-99-avg-1-streaming-chunk-size-32-context-2-max-sym-per-frame-1.txt
+2023-04-04 11:24:42,119 INFO [decode.py:608] 
+For test-other, WER of different settings are:
+greedy_search	6.18	best for test-other
+
+2023-04-04 11:24:42,119 INFO [decode.py:814] Done!