Duplicate from danielrosehill/Fritter

.gitattributes

@@ -0,0 +1,42 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tar filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text
+Fritter.nemo filter=lfs diff=lfs merge=lfs -text
+onnx/fp32/encoder-model.onnx.data filter=lfs diff=lfs merge=lfs -text
+gguf/Fritter-Q8_0.gguf filter=lfs diff=lfs merge=lfs -text
+gguf/Fritter-Q4_K_M.gguf filter=lfs diff=lfs merge=lfs -text
+gguf/Fritter-F16.gguf filter=lfs diff=lfs merge=lfs -text
+figures/banner.png filter=lfs diff=lfs merge=lfs -text
+figures/objective.png filter=lfs diff=lfs merge=lfs -text

Fritter.nemo

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8ad23df599ef57186b07842d79bc9be83d35a91c20d818600a2a73edde481f7f
+size 2472212480

README.md

@@ -0,0 +1,99 @@
+---
+license: cc-by-nc-4.0
+language:
+- en
+base_model: nvidia/parakeet-tdt-0.6b-v2
+base_model_relation: finetune
+pipeline_tag: automatic-speech-recognition
+library_name: nemo
+tags:
+- parakeet
+- tdt
+- speaker-adaptation
+- english
+- tech-vocabulary
+- irish-english
+- dictation
+- fritter
+datasets:
+- danielrosehill/Read-Aloud-Voice-Sentences
+- openslr/librispeech_asr
+---
+
+# Fritter v1.3: Parakeet TDT 0.6B v2 for English dictation, no code-switching
+
+![Fritter: English dictation for technical vocabulary](figures/banner.png)
+
+## What it's for
+
+Stock speech models stumble on the words technical people say all day. Fritter is tuned for higher accuracy on technical vocabulary, including AI: model and product names, CLI and cloud terms, brands, acronyms and numbers in digits.
+
+![What Fritter is for: AI model names, CLI and cloud terms, brands and acronyms, numbers in digits](figures/objective.png)
+
+**App developers:** you're welcome to add Fritter to your app or model catalogue, with no strings attached, for non-commercial use with attribution. The [Fritter collection](https://huggingface.co/collections/danielrosehill/fritter) has every release; a streaming version is coming soon and will appear there.
+
+A personal speech-recognition fine-tune for **English dictation without code-switching**: tech vocabulary (model and product names, CLI and cloud terms), brand names, acronyms, numbers and currency in digits, Irish-English phrasing, fast and slow speech, whispering, and everyday email and message dictation, all in one speaker's voice.
+
+**Name.** Fritter is the code name of this model line; later retrains keep it and bump the version. The first release's run id is `r11-general`.
+
+A full fine-tune of [`nvidia/parakeet-tdt-0.6b-v2`](https://huggingface.co/nvidia/parakeet-tdt-0.6b-v2) on one
+speaker's read-aloud phone recordings (an Irish English speaker living in Israel), run `r11-general`, trained
+2026-10-27. Output keeps v2's casing and punctuation; numbers are written in digits.
+
+**Current version: v1.3** (run `r35-general`, trained 2026-11-08): v1.2 continued on 184.7 min of new own takes plus a replay sample (clips over 25 s left out). The weights and every build in this repo are v1.3. v1.1 to v1.3 have not been benchmarked, so the results below are v1.0's (run `r11-general`).
+
+**Try it in your browser:** [danielrosehill/Fritter-Demo](https://huggingface.co/spaces/danielrosehill/Fritter-Demo) runs the int8 ONNX build on your own device.
+
+**Licence and access.** Public and ungated. CC-BY-NC-4.0: free for non-commercial use with attribution. A personal project by Daniel Rosehill; not affiliated with NVIDIA.
+
+## Results
+
+All models scored on the same test clips (dataset revision `520e7506e379`, blend `general` test sets;
+never trained on, never used for checkpoint selection). WER is normalised (lowercase, no punctuation).
+
+| Model | Hebrew-free English WER (319 clips) | tech WER | own English WER (364) | LibriSpeech WER (200) | code-switching WER (614) | held-out Hebrew term recall | long-form WER |
+|---|---|---|---|---|---|---|---|
+| stock `nvidia/parakeet-tdt-0.6b-v2` | 7.46% | 7.53% | 7.23% | 1.23% | 12.64% | 24.90% | 11.05% |
+| **this model** (`r11-general`) | 4.73% | 4.01% | 4.85% | 1.47% | 11.69% | 24.27% | 12.16% |
+
+On its target, Fritter cuts tech-sentence WER from 7.53% to 4.01% and Hebrew-free English WER from 7.46% to 4.73%, and recalls 79.6% of the tech and English target terms in the test takes against stock's 65.1% (447 occurrences of 262 terms; exact-case 74.1% vs 58.2%). Numbers recall rises from 33.9% to 50.0% (62 tokens). LibriSpeech is 0.24 points worse than stock (1.47% vs 1.23%, 200 utterances) and long-form dictation 1.1 points worse (12.16% vs 11.05%, 23 files), so general English outside the speaker's own voice is about where stock is, not better. It is not a code-switching model: its code-switching WER is only slightly better than stock's. Test takes are one speaker's, and their target terms are disjoint from training. Validation WER at the best epoch (17 of 20) was 4.24%.
+
+## Training data
+
+Source: [`danielrosehill/Read-Aloud-Voice-Sentences`](https://huggingface.co/datasets/danielrosehill/Read-Aloud-Voice-Sentences) at revision `520e7506e379`; the mix is rebuilt from it by `prepare` (blend `general`).
+
+- The speaker's takes that are not code-switching (`is_code_switching()` false). Every task subset is in except `name-cipher` and the held-back tags (weird voices); 10% LibriSpeech-PC replay.
+- Mix: 319.9 min of the speaker's takes
+  plus 35.6 min of LibriSpeech-PC replay (cased, punctuated; CC-BY-4.0) = 391.0 min, 3,848 clips.
+- Validation: 1 in 16 of the speaker's train takes by id hash, used only to pick the checkpoint.
+
+## Training
+
+NeMo 2.7.3 on one L40S (Modal), bf16. AdamW, lr 3e-05 (cosine, 10% warmup), 20 epochs,
+batch 16, seed 7. Best checkpoint `best-epoch=17-val_wer=0.0424.ckpt`,
+38.2 min of training.
+
+## Usage
+
+```python
+import nemo.collections.asr as nemo_asr
+from huggingface_hub import hf_hub_download
+path = hf_hub_download("danielrosehill/Fritter", "Fritter.nemo")
+model = nemo_asr.models.ASRModel.restore_from(path)
+print(model.transcribe(["clip.wav"])[0].text)   # 16 kHz mono
+```
+
+GGUF builds for transcribe.cpp (Offline Voice Input on Android, Handy on Linux) are in `gguf/`: Q8_0 (recommended),
+Q4_K_M and F16.
+
+## Limitations
+
+- **One speaker.** It is adapted to one voice, accent and phone; others will see less benefit.
+- The test sets are small (Hebrew-free English is 319 clips), so differences under about
+  one WER point are within noise.
+- Not trained on English-Hebrew code-switching: romanized Hebrew words come out as stock v2 writes them.
+- Disfluent takes were trained on with clean targets, so the model may drop some fillers and repeated words.
+
+## Licence
+
+CC-BY-NC-4.0. The base model, `nvidia/parakeet-tdt-0.6b-v2`, is CC-BY-4.0, by NVIDIA.

gguf/Fritter-F16.gguf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:344224ec2987ecfbd2a8c0ea1a0164987c06970548c52f810a74848705c52a7e
+size 1237334624

gguf/Fritter-Q4_K_M.gguf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2f93e1782cb768ef36daed4d86d1e8224087f309e583ad35f23abe075caa9ddf
+size 475491872

gguf/Fritter-Q8_0.gguf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:64bd0e84de4d540e7b4ac1c76770ca639eef48736d4e9b68a4f76b654fa91a3c
+size 729574944

onnx/fp32/config.json

@@ -0,0 +1,5 @@
+{
+    "model_type": "nemo-conformer-tdt",
+    "features_size": 128,
+    "subsampling_factor": 8
+}

onnx/fp32/decoder_joint-model.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1f7d3277a97f21c15d4ecbb85d104d56984c3c3f07314e8181f9fd5d51997e44
+size 35792060

onnx/fp32/encoder-model.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2eee8df543d6223d0da376c136387262418d1573763d0e25d489dd58b142be44
+size 41842955

onnx/fp32/encoder-model.onnx.data

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ffa8504d8e8d46f8679edb590ebbd05ee2bddf8f52b9c8e8ab944693dd9296f6
+size 2435420160

onnx/fp32/vocab.txt

@@ -0,0 +1,1025 @@
+<unk> 0
+▁t 1
+▁th 2
+▁a 3
+in 4
+▁the 5
+re 6
+▁w 7
+▁o 8
+▁s 9
+at 10
+ou 11
+er 12
+nd 13
+▁i 14
+▁b 15
+▁c 16
+on 17
+▁h 18
+ing 19
+▁to 20
+▁m 21
+en 22
+▁f 23
+▁p 24
+an 25
+▁d 26
+es 27
+or 28
+ll 29
+▁of 30
+▁and 31
+▁y 32
+▁l 33
+▁I 34
+it 35
+▁in 36
+is 37
+ed 38
+▁g 39
+▁you 40
+ar 41
+▁that 42
+om 43
+as 44
+▁n 45
+ve 46
+us 47
+ic 48
+ow 49
+al 50
+▁it 51
+▁be 52
+▁wh 53
+le 54
+ion 55
+ut 56
+ot 57
+▁we 58
+▁is 59
+▁e 60
+et 61
+ay 62
+▁re 63
+▁on 64
+▁T 65
+▁A 66
+▁ha 67
+ent 68
+ke 69
+ct 70
+▁S 71
+ig 72
+ver 73
+▁Th 74
+all 75
+id 76
+▁for 77
+ro 78
+▁he 79
+se 80
+▁this 81
+ld 82
+ly 83
+▁go 84
+▁k 85
+▁st 86
+st 87
+ch 88
+▁li 89
+▁u 90
+am 91
+ur 92
+ce 93
+ith 94
+im 95
+▁so 96
+▁have 97
+▁do 98
+ht 99
+th 100
+▁an 101
+▁with 102
+ad 103
+▁r 104
+ir 105
+▁was 106
+▁as 107
+▁W 108
+▁are 109
+ust 110
+ally 111
+▁j 112
+▁se 113
+ation 114
+od 115
+ere 116
+▁like 117
+▁not 118
+▁kn 119
+ight 120
+▁B 121
+▁they 122
+▁And 123
+▁know 124
+ome 125
+op 126
+▁can 127
+▁or 128
+▁sh 129
+▁me 130
+ill 131
+ant 132
+ck 133
+▁what 134
+▁at 135
+▁ab 136
+ould 137
+ol 138
+▁So 139
+▁C 140
+use 141
+ter 142
+il 143
+▁but 144
+▁just 145
+▁ne 146
+▁de 147
+ra 148
+ore 149
+▁there 150
+ul 151
+out 152
+▁con 153
+▁all 154
+▁The 155
+ers 156
+▁H 157
+▁fr 158
+▁pro 159
+ge 160
+ea 161
+▁Y 162
+▁O 163
+▁M 164
+pp 165
+▁com 166
+ess 167
+▁ch 168
+▁al 169
+est 170
+ate 171
+qu 172
+▁lo 173
+▁ex 174
+very 175
+▁su 176
+ain 177
+▁one 178
+ca 179
+art 180
+ist 181
+if 182
+ive 183
+▁if 184
+ink 185
+nt 186
+ab 187
+▁about 188
+▁going 189
+▁v 190
+▁wor 191
+um 192
+ok 193
+▁your 194
+▁my 195
+ind 196
+▁get 197
+cause 198
+▁from 199
+▁don 200
+ri 201
+pe 202
+un 203
+ity 204
+▁up 205
+▁P 206
+▁out 207
+ort 208
+▁L 209
+ment 210
+el 211
+▁N 212
+▁some 213
+ich 214
+and 215
+▁think 216
+em 217
+oug 218
+▁G 219
+os 220
+▁D 221
+res 222
+▁because 223
+▁by 224
+ake 225
+▁int 226
+ie 227
+▁us 228
+▁tr 229
+▁then 230
+ack 231
+▁pl 232
+▁here 233
+▁pe 234
+her 235
+▁will 236
+▁F 237
+▁which 238
+ard 239
+▁right 240
+▁thing 241
+▁want 242
+ies 243
+ople 244
+▁It 245
+▁them 246
+ame 247
+▁We 248
+our 249
+▁say 250
+▁R 251
+▁people 252
+▁see 253
+▁who 254
+ast 255
+ure 256
+ect 257
+ear 258
+▁tim 259
+▁E 260
+▁You 261
+▁would 262
+▁when 263
+ven 264
+▁our 265
+ci 266
+▁really 267
+▁more 268
+ound 269
+ose 270
+ak 271
+▁co 272
+ide 273
+ough 274
+▁had 275
+so 276
+▁qu 277
+eah 278
+▁were 279
+ine 280
+▁act 281
+ther 282
+▁these 283
+▁how 284
+▁now 285
+▁sa 286
+ud 287
+▁Wh 288
+▁man 289
+ous 290
+one 291
+pt 292
+ff 293
+ong 294
+▁has 295
+▁any 296
+▁very 297
+▁But 298
+▁look 299
+iv 300
+itt 301
+▁time 302
+▁mo 303
+▁ar 304
+hing 305
+▁le 306
+▁work 307
+▁their 308
+are 309
+▁his 310
+per 311
+ions 312
+▁im 313
+▁ag 314
+▁J 315
+▁no 316
+▁en 317
+▁got 318
+ag 319
+▁sp 320
+ans 321
+act 322
+▁te 323
+▁also 324
+iz 325
+ice 326
+▁That 327
+▁cl 328
+▁been 329
+▁way 330
+▁fe 331
+▁did 332
+ple 333
+ually 334
+▁other 335
+▁U 336
+ite 337
+age 338
+omet 339
+ber 340
+reat 341
+ree 342
+▁into 343
+own 344
+▁tw 345
+▁part 346
+alk 347
+▁where 348
+▁need 349
+▁every 350
+pl 351
+▁ad 352
+ry 353
+▁over 354
+ble 355
+ap 356
+ue 357
+▁kind 358
+▁po 359
+▁back 360
+▁cont 361
+iff 362
+▁somet 363
+▁pr 364
+nder 365
+ire 366
+▁good 367
+▁than 368
+ace 369
+▁gu 370
+ep 371
+og 372
+ick 373
+way 374
+▁lot 375
+▁un 376
+▁things 377
+▁In 378
+ish 379
+kay 380
+▁well 381
+▁could 382
+▁pre 383
+▁two 384
+irst 385
+▁diff 386
+ach 387
+cc 388
+ittle 389
+int 390
+▁He 391
+▁those 392
+ence 393
+ip 394
+ase 395
+▁him 396
+▁make 397
+▁little 398
+ical 399
+▁gr 400
+▁year 401
+ass 402
+▁thr 403
+uch 404
+ated 405
+▁This 406
+▁off 407
+▁res 408
+ac 409
+ance 410
+▁actually 411
+▁talk 412
+ult 413
+able 414
+orm 415
+▁dis 416
+▁first 417
+ations 418
+▁something 419
+▁she 420
+sel 421
+▁let 422
+ord 423
+▁may 424
+ia 425
+▁am 426
+▁her 427
+▁said 428
+▁bo 429
+be 430
+ount 431
+▁much 432
+▁per 433
+▁even 434
+▁differe 435
+vel 436
+ary 437
+▁app 438
+ving 439
+▁comm 440
+▁imp 441
+ys 442
+▁again 443
+ress 444
+▁yeah 445
+▁down 446
+ang 447
+▁mean 448
+na 449
+ens 450
+▁does 451
+▁fo 452
+▁comp 453
+▁ro 454
+▁bl 455
+ody 456
+▁K 457
+▁through 458
+▁start 459
+uct 460
+▁only 461
+▁bet 462
+▁under 463
+▁br 464
+▁take 465
+ning 466
+▁bu 467
+▁use 468
+▁Ch 469
+xt 470
+co 471
+ory 472
+ild 473
+▁put 474
+▁call 475
+▁new 476
+other 477
+ting 478
+▁happ 479
+ater 480
+▁inc 481
+ition 482
+▁different 483
+▁should 484
+ade 485
+ign 486
+thing 487
+▁day 488
+fore 489
+▁Yeah 490
+ark 491
+ile 492
+ial 493
+▁come 494
+▁They 495
+��being 496
+▁try 497
+ious 498
+▁sc 499
+▁bit 500
+▁spe 501
+ub 502
+fe 503
+▁doing 504
+▁St 505
+vers 506
+av 507
+ty 508
+ian 509
+onna 510
+red 511
+wn 512
+▁ke 513
+form 514
+ors 515
+▁fl 516
+fter 517
+ail 518
+ents 519
+▁gonna 520
+▁point 521
+ces 522
+▁There 523
+self 524
+▁many 525
+▁If 526
+▁same 527
+▁sy 528
+▁quest 529
+▁most 530
+▁great 531
+▁What 532
+▁fu 533
+ug 534
+▁show 535
+we 536
+ual 537
+ons 538
+▁Be 539
+ically 540
+▁ser 541
+▁rem 542
+▁ind 543
+▁pers 544
+▁V 545
+he 546
+▁str 547
+ved 548
+▁still 549
+ank 550
+▁rec 551
+▁wr 552
+ought 553
+day 554
+ath 555
+▁end 556
+▁bas 557
+ft 558
+erm 559
+body 560
+ph 561
+ject 562
+ict 563
+▁play 564
+▁Is 565
+ates 566
+▁ph 567
+oth 568
+▁acc 569
+get 570
+▁years 571
+▁em 572
+▁id 573
+▁Oh 574
+ves 575
+ever 576
+▁inter 577
+▁rel 578
+▁before 579
+▁feel 580
+igh 581
+▁three 582
+iss 583
+▁des 584
+ne 585
+▁why 586
+▁uh 587
+▁To 588
+▁cons 589
+▁hel 590
+▁after 591
+ower 592
+urn 593
+▁okay 594
+▁long 595
+▁bel 596
+▁around 597
+ful 598
+te 599
+ise 600
+▁ob 601
+▁supp 602
+ady 603
+ange 604
+aking 605
+▁pos 606
+atch 607
+▁tra 608
+gr 609
+▁might 610
+ert 611
+▁help 612
+ost 613
+▁too 614
+cial 615
+▁world 616
+▁give 617
+ike 618
+▁Okay 619
+ways 620
+▁min 621
+ward 622
+ily 623
+▁gen 624
+▁find 625
+▁dec 626
+ular 627
+ob 628
+▁tell 629
+▁Now 630
+▁sm 631
+▁cour 632
+▁real 633
+cess 634
+nds 635
+▁big 636
+▁num 637
+ction 638
+▁add 639
+▁set 640
+▁um 641
+ood 642
+ible 643
+▁own 644
+▁life 645
+ities 646
+▁its 647
+▁God 648
+pect 649
+▁didn 650
+stem 651
+les 652
+uc 653
+ib 654
+ating 655
+olog 656
+▁person 657
+▁inv 658
+ably 659
+▁sure 660
+▁reg 661
+lic 662
+▁stu 663
+▁cr 664
+▁ev 665
+ments 666
+▁another 667
+▁la 668
+▁last 669
+▁sub 670
+▁att 671
+▁op 672
+▁inst 673
+▁sl 674
+▁happen 675
+▁rep 676
+▁import 677
+ific 678
+ix 679
+▁made 680
+▁ear 681
+▁ac 682
+▁def 683
+ute 684
+▁next 685
+ative 686
+▁form 687
+▁guys 688
+▁system 689
+ew 690
+▁able 691
+ied 692
+▁always 693
+ren 694
+erest 695
+▁As 696
+▁mod 697
+▁done 698
+ings 699
+▁love 700
+ism 701
+▁ask 702
+old 703
+ered 704
+▁trans 705
+▁count 706
+ility 707
+▁high 708
+▁fin 709
+▁prob 710
+▁pol 711
+▁exam 712
+▁pres 713
+▁maybe 714
+ell 715
+▁stud 716
+▁prod 717
+▁car 718
+ock 719
+▁used 720
+oy 721
+stand 722
+▁No 723
+▁mon 724
+ks 725
+▁interest 726
+▁ent 727
+ited 728
+▁sort 729
+▁For 730
+▁today 731
+ics 732
+▁vide 733
+▁bec 734
+▁Well 735
+▁Al 736
+▁important 737
+▁such 738
+▁run 739
+▁keep 740
+▁fact 741
+ata 742
+ss 743
+▁never 744
+ween 745
+▁stuff 746
+ract 747
+▁question 748
+als 749
+▁sim 750
+vern 751
+ather 752
+▁course 753
+▁Of 754
+oc 755
+ness 756
+arch 757
+ize 758
+▁All 759
+ense 760
+blem 761
+▁probably 762
+hip 763
+▁number 764
+ention 765
+▁saying 766
+▁commun 767
+▁An 768
+akes 769
+▁belie 770
+▁between 771
+▁better 772
+cus 773
+▁place 774
+▁gener 775
+▁ca 776
+▁ins 777
+▁ass 778
+cond 779
+cept 780
+ull 781
+▁understand 782
+▁fun 783
+▁thought 784
+gan 785
+iew 786
+cy 787
+ution 788
+ope 789
+ason 790
+▁problem 791
+▁doesn 792
+ational 793
+▁read 794
+▁trying 795
+▁sch 796
+▁el 797
+ah 798
+atter 799
+▁exper 800
+▁four 801
+▁ele 802
+▁cou 803
+ont 804
+▁called 805
+▁partic 806
+▁open 807
+▁gl 808
+▁everything 809
+▁eff 810
+▁getting 811
+▁ty 812
+▁Am 813
+▁Because 814
+ave 815
+▁met 816
+▁Like 817
+oney 818
+▁ 819
+e 820
+t 821
+o 822
+a 823
+n 824
+i 825
+s 826
+h 827
+r 828
+l 829
+d 830
+u 831
+c 832
+y 833
+m 834
+g 835
+w 836
+f 837
+p 838
+, 839
+b 840
+. 841
+k 842
+v 843
+' 844
+I 845
+T 846
+A 847
+S 848
+j 849
+x 850
+W 851
+B 852
+C 853
+? 854
+0 855
+O 856
+- 857
+M 858
+H 859
+Y 860
+q 861
+1 862
+P 863
+z 864
+L 865
+D 866
+N 867
+G 868
+F 869
+R 870
+E 871
+2 872
+J 873
+U 874
+: 875
+5 876
+9 877
+3 878
+K 879
+4 880
+V 881
+8 882
+6 883
+7 884
+! 885
+% 886
+Q 887
+$ 888
+Z 889
+X 890
+é 891
+/ 892
+í 893
+á 894
+£ 895
+ó 896
+ā 897
+ü 898
+ñ 899
+ö 900
+è 901
+ç 902
+à 903
+¿ 904
+μ 905
+π 906
+ä 907
+ú 908
+θ 909
+ã 910
+φ 911
+ī 912
+σ 913
+ê 914
+ρ 915
+â 916
+ô 917
+^ 918
+€ 919
+É 920
+ū 921
+Δ 922
+λ 923
+α 924
+τ 925
+æ 926
+а 927
+о 928
+ν 929
+î 930
+γ 931
+ψ 932
+ē 933
+т 934
+ß 935
+ω 936
+ï 937
+ć 938
+č 939
+ε 940
+е 941
+и 942
+ò 943
+р 944
+β 945
+ø 946
+ł 947
+δ 948
+η 949
+п 950
+ë 951
+н 952
+с 953
+š 954
+Ü 955
+å 956
+ń 957
+ś 958
+я 959
+đ 960
+л 961
+м 962
+Ö 963
+û 964
+ș 965
+в 966
+Á 967
+Ø 968
+ù 969
+ο 970
+ч 971
+ь 972
+ž 973
+Φ 974
+у 975
+ę 976
+ι 977
+б 978
+г 979
+к 980
+ő 981
+Ś 982
+Ω 983
+κ 984
+υ 985
+ì 986
+Č 987
+έ 988
+х 989
+ы 990
+Å 991
+Ç 992
+ż 993
+ί 994
+ζ 995
+χ 996
+э 997
+Æ 998
+Í 999
+õ 1000
+ě 1001
+ħ 1002
+Ł 1003
+œ 1004
+Ž 1005
+ț 1006
+Γ 1007
+П 1008
+д 1009
+з 1010
+ф 1011
+¡ 1012
+À 1013
+Î 1014
+Ā 1015
+ė 1016
+Š 1017
+ź 1018
+Κ 1019
+Ψ 1020
+ά 1021
+ξ 1022
+ό 1023
+<blk> 1024

onnx/int8/config.json

@@ -0,0 +1,5 @@
+{
+    "model_type": "nemo-conformer-tdt",
+    "features_size": 128,
+    "subsampling_factor": 8
+}

onnx/int8/decoder_joint-model.int8.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ce6346af81e1fa27a66d896071469c577bc8591a7020682e19cb63bc9617a884
+size 8998284

onnx/int8/encoder-model.int8.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fcd1ec6b8a92b857907090be457aa725035268a6778077d5e55c4f35dfad379f
+size 652282023

onnx/int8/vocab.txt

@@ -0,0 +1,1025 @@
+<unk> 0
+▁t 1
+▁th 2
+▁a 3
+in 4
+▁the 5
+re 6
+▁w 7
+▁o 8
+▁s 9
+at 10
+ou 11
+er 12
+nd 13
+▁i 14
+▁b 15
+▁c 16
+on 17
+▁h 18
+ing 19
+▁to 20
+▁m 21
+en 22
+▁f 23
+▁p 24
+an 25
+▁d 26
+es 27
+or 28
+ll 29
+▁of 30
+▁and 31
+▁y 32
+▁l 33
+▁I 34
+it 35
+▁in 36
+is 37
+ed 38
+▁g 39
+▁you 40
+ar 41
+▁that 42
+om 43
+as 44
+▁n 45
+ve 46
+us 47
+ic 48
+ow 49
+al 50
+▁it 51
+▁be 52
+▁wh 53
+le 54
+ion 55
+ut 56
+ot 57
+▁we 58
+▁is 59
+▁e 60
+et 61
+ay 62
+▁re 63
+▁on 64
+▁T 65
+▁A 66
+▁ha 67
+ent 68
+ke 69
+ct 70
+▁S 71
+ig 72
+ver 73
+▁Th 74
+all 75
+id 76
+▁for 77
+ro 78
+▁he 79
+se 80
+▁this 81
+ld 82
+ly 83
+▁go 84
+▁k 85
+▁st 86
+st 87
+ch 88
+▁li 89
+▁u 90
+am 91
+ur 92
+ce 93
+ith 94
+im 95
+▁so 96
+▁have 97
+▁do 98
+ht 99
+th 100
+▁an 101
+▁with 102
+ad 103
+▁r 104
+ir 105
+▁was 106
+▁as 107
+▁W 108
+▁are 109
+ust 110
+ally 111
+▁j 112
+▁se 113
+ation 114
+od 115
+ere 116
+▁like 117
+▁not 118
+▁kn 119
+ight 120
+▁B 121
+▁they 122
+▁And 123
+▁know 124
+ome 125
+op 126
+▁can 127
+▁or 128
+▁sh 129
+▁me 130
+ill 131
+ant 132
+ck 133
+▁what 134
+▁at 135
+▁ab 136
+ould 137
+ol 138
+▁So 139
+▁C 140
+use 141
+ter 142
+il 143
+▁but 144
+▁just 145
+▁ne 146
+▁de 147
+ra 148
+ore 149
+▁there 150
+ul 151
+out 152
+▁con 153
+▁all 154
+▁The 155
+ers 156
+▁H 157
+▁fr 158
+▁pro 159
+ge 160
+ea 161
+▁Y 162
+▁O 163
+▁M 164
+pp 165
+▁com 166
+ess 167
+▁ch 168
+▁al 169
+est 170
+ate 171
+qu 172
+▁lo 173
+▁ex 174
+very 175
+▁su 176
+ain 177
+▁one 178
+ca 179
+art 180
+ist 181
+if 182
+ive 183
+▁if 184
+ink 185
+nt 186
+ab 187
+▁about 188
+▁going 189
+▁v 190
+▁wor 191
+um 192
+ok 193
+▁your 194
+▁my 195
+ind 196
+▁get 197
+cause 198
+▁from 199
+▁don 200
+ri 201
+pe 202
+un 203
+ity 204
+▁up 205
+▁P 206
+▁out 207
+ort 208
+▁L 209
+ment 210
+el 211
+▁N 212
+▁some 213
+ich 214
+and 215
+▁think 216
+em 217
+oug 218
+▁G 219
+os 220
+▁D 221
+res 222
+▁because 223
+▁by 224
+ake 225
+▁int 226
+ie 227
+▁us 228
+▁tr 229
+▁then 230
+ack 231
+▁pl 232
+▁here 233
+▁pe 234
+her 235
+▁will 236
+▁F 237
+▁which 238
+ard 239
+▁right 240
+▁thing 241
+▁want 242
+ies 243
+ople 244
+▁It 245
+▁them 246
+ame 247
+▁We 248
+our 249
+▁say 250
+▁R 251
+▁people 252
+▁see 253
+▁who 254
+ast 255
+ure 256
+ect 257
+ear 258
+▁tim 259
+▁E 260
+▁You 261
+▁would 262
+▁when 263
+ven 264
+▁our 265
+ci 266
+▁really 267
+▁more 268
+ound 269
+ose 270
+ak 271
+▁co 272
+ide 273
+ough 274
+▁had 275
+so 276
+▁qu 277
+eah 278
+▁were 279
+ine 280
+▁act 281
+ther 282
+▁these 283
+▁how 284
+▁now 285
+▁sa 286
+ud 287
+▁Wh 288
+▁man 289
+ous 290
+one 291
+pt 292
+ff 293
+ong 294
+▁has 295
+▁any 296
+▁very 297
+▁But 298
+▁look 299
+iv 300
+itt 301
+▁time 302
+▁mo 303
+▁ar 304
+hing 305
+▁le 306
+▁work 307
+▁their 308
+are 309
+▁his 310
+per 311
+ions 312
+▁im 313
+▁ag 314
+▁J 315
+▁no 316
+▁en 317
+▁got 318
+ag 319
+▁sp 320
+ans 321
+act 322
+▁te 323
+▁also 324
+iz 325
+ice 326
+▁That 327
+▁cl 328
+▁been 329
+▁way 330
+▁fe 331
+▁did 332
+ple 333
+ually 334
+▁other 335
+▁U 336
+ite 337
+age 338
+omet 339
+ber 340
+reat 341
+ree 342
+▁into 343
+own 344
+▁tw 345
+▁part 346
+alk 347
+▁where 348
+▁need 349
+▁every 350
+pl 351
+▁ad 352
+ry 353
+▁over 354
+ble 355
+ap 356
+ue 357
+▁kind 358
+▁po 359
+▁back 360
+▁cont 361
+iff 362
+▁somet 363
+▁pr 364
+nder 365
+ire 366
+▁good 367
+▁than 368
+ace 369
+▁gu 370
+ep 371
+og 372
+ick 373
+way 374
+▁lot 375
+▁un 376
+▁things 377
+▁In 378
+ish 379
+kay 380
+▁well 381
+▁could 382
+▁pre 383
+▁two 384
+irst 385
+▁diff 386
+ach 387
+cc 388
+ittle 389
+int 390
+▁He 391
+▁those 392
+ence 393
+ip 394
+ase 395
+▁him 396
+▁make 397
+▁little 398
+ical 399
+▁gr 400
+▁year 401
+ass 402
+▁thr 403
+uch 404
+ated 405
+▁This 406
+▁off 407
+▁res 408
+ac 409
+ance 410
+▁actually 411
+▁talk 412
+ult 413
+able 414
+orm 415
+▁dis 416
+▁first 417
+ations 418
+▁something 419
+▁she 420
+sel 421
+▁let 422
+ord 423
+▁may 424
+ia 425
+▁am 426
+▁her 427
+▁said 428
+▁bo 429
+be 430
+ount 431
+▁much 432
+▁per 433
+▁even 434
+▁differe 435
+vel 436
+ary 437
+▁app 438
+ving 439
+▁comm 440
+▁imp 441
+ys 442
+▁again 443
+ress 444
+▁yeah 445
+▁down 446
+ang 447
+▁mean 448
+na 449
+ens 450
+▁does 451
+▁fo 452
+▁comp 453
+▁ro 454
+▁bl 455
+ody 456
+▁K 457
+▁through 458
+▁start 459
+uct 460
+▁only 461
+▁bet 462
+▁under 463
+▁br 464
+▁take 465
+ning 466
+▁bu 467
+▁use 468
+▁Ch 469
+xt 470
+co 471
+ory 472
+ild 473
+▁put 474
+▁call 475
+▁new 476
+other 477
+ting 478
+▁happ 479
+ater 480
+▁inc 481
+ition 482
+▁different 483
+▁should 484
+ade 485
+ign 486
+thing 487
+▁day 488
+fore 489
+▁Yeah 490
+ark 491
+ile 492
+ial 493
+▁come 494
+▁They 495
+��being 496
+▁try 497
+ious 498
+▁sc 499
+▁bit 500
+▁spe 501
+ub 502
+fe 503
+▁doing 504
+▁St 505
+vers 506
+av 507
+ty 508
+ian 509
+onna 510
+red 511
+wn 512
+▁ke 513
+form 514
+ors 515
+▁fl 516
+fter 517
+ail 518
+ents 519
+▁gonna 520
+▁point 521
+ces 522
+▁There 523
+self 524
+▁many 525
+▁If 526
+▁same 527
+▁sy 528
+▁quest 529
+▁most 530
+▁great 531
+▁What 532
+▁fu 533
+ug 534
+▁show 535
+we 536
+ual 537
+ons 538
+▁Be 539
+ically 540
+▁ser 541
+▁rem 542
+▁ind 543
+▁pers 544
+▁V 545
+he 546
+▁str 547
+ved 548
+▁still 549
+ank 550
+▁rec 551
+▁wr 552
+ought 553
+day 554
+ath 555
+▁end 556
+▁bas 557
+ft 558
+erm 559
+body 560
+ph 561
+ject 562
+ict 563
+▁play 564
+▁Is 565
+ates 566
+▁ph 567
+oth 568
+▁acc 569
+get 570
+▁years 571
+▁em 572
+▁id 573
+▁Oh 574
+ves 575
+ever 576
+▁inter 577
+▁rel 578
+▁before 579
+▁feel 580
+igh 581
+▁three 582
+iss 583
+▁des 584
+ne 585
+▁why 586
+▁uh 587
+▁To 588
+▁cons 589
+▁hel 590
+▁after 591
+ower 592
+urn 593
+▁okay 594
+▁long 595
+▁bel 596
+▁around 597
+ful 598
+te 599
+ise 600
+▁ob 601
+▁supp 602
+ady 603
+ange 604
+aking 605
+▁pos 606
+atch 607
+▁tra 608
+gr 609
+▁might 610
+ert 611
+▁help 612
+ost 613
+▁too 614
+cial 615
+▁world 616
+▁give 617
+ike 618
+▁Okay 619
+ways 620
+▁min 621
+ward 622
+ily 623
+▁gen 624
+▁find 625
+▁dec 626
+ular 627
+ob 628
+▁tell 629
+▁Now 630
+▁sm 631
+▁cour 632
+▁real 633
+cess 634
+nds 635
+▁big 636
+▁num 637
+ction 638
+▁add 639
+▁set 640
+▁um 641
+ood 642
+ible 643
+▁own 644
+▁life 645
+ities 646
+▁its 647
+▁God 648
+pect 649
+▁didn 650
+stem 651
+les 652
+uc 653
+ib 654
+ating 655
+olog 656
+▁person 657
+▁inv 658
+ably 659
+▁sure 660
+▁reg 661
+lic 662
+▁stu 663
+▁cr 664
+▁ev 665
+ments 666
+▁another 667
+▁la 668
+▁last 669
+▁sub 670
+▁att 671
+▁op 672
+▁inst 673
+▁sl 674
+▁happen 675
+▁rep 676
+▁import 677
+ific 678
+ix 679
+▁made 680
+▁ear 681
+▁ac 682
+▁def 683
+ute 684
+▁next 685
+ative 686
+▁form 687
+▁guys 688
+▁system 689
+ew 690
+▁able 691
+ied 692
+▁always 693
+ren 694
+erest 695
+▁As 696
+▁mod 697
+▁done 698
+ings 699
+▁love 700
+ism 701
+▁ask 702
+old 703
+ered 704
+▁trans 705
+▁count 706
+ility 707
+▁high 708
+▁fin 709
+▁prob 710
+▁pol 711
+▁exam 712
+▁pres 713
+▁maybe 714
+ell 715
+▁stud 716
+▁prod 717
+▁car 718
+ock 719
+▁used 720
+oy 721
+stand 722
+▁No 723
+▁mon 724
+ks 725
+▁interest 726
+▁ent 727
+ited 728
+▁sort 729
+▁For 730
+▁today 731
+ics 732
+▁vide 733
+▁bec 734
+▁Well 735
+▁Al 736
+▁important 737
+▁such 738
+▁run 739
+▁keep 740
+▁fact 741
+ata 742
+ss 743
+▁never 744
+ween 745
+▁stuff 746
+ract 747
+▁question 748
+als 749
+▁sim 750
+vern 751
+ather 752
+▁course 753
+▁Of 754
+oc 755
+ness 756
+arch 757
+ize 758
+▁All 759
+ense 760
+blem 761
+▁probably 762
+hip 763
+▁number 764
+ention 765
+▁saying 766
+▁commun 767
+▁An 768
+akes 769
+▁belie 770
+▁between 771
+▁better 772
+cus 773
+▁place 774
+▁gener 775
+▁ca 776
+▁ins 777
+▁ass 778
+cond 779
+cept 780
+ull 781
+▁understand 782
+▁fun 783
+▁thought 784
+gan 785
+iew 786
+cy 787
+ution 788
+ope 789
+ason 790
+▁problem 791
+▁doesn 792
+ational 793
+▁read 794
+▁trying 795
+▁sch 796
+▁el 797
+ah 798
+atter 799
+▁exper 800
+▁four 801
+▁ele 802
+▁cou 803
+ont 804
+▁called 805
+▁partic 806
+▁open 807
+▁gl 808
+▁everything 809
+▁eff 810
+▁getting 811
+▁ty 812
+▁Am 813
+▁Because 814
+ave 815
+▁met 816
+▁Like 817
+oney 818
+▁ 819
+e 820
+t 821
+o 822
+a 823
+n 824
+i 825
+s 826
+h 827
+r 828
+l 829
+d 830
+u 831
+c 832
+y 833
+m 834
+g 835
+w 836
+f 837
+p 838
+, 839
+b 840
+. 841
+k 842
+v 843
+' 844
+I 845
+T 846
+A 847
+S 848
+j 849
+x 850
+W 851
+B 852
+C 853
+? 854
+0 855
+O 856
+- 857
+M 858
+H 859
+Y 860
+q 861
+1 862
+P 863
+z 864
+L 865
+D 866
+N 867
+G 868
+F 869
+R 870
+E 871
+2 872
+J 873
+U 874
+: 875
+5 876
+9 877
+3 878
+K 879
+4 880
+V 881
+8 882
+6 883
+7 884
+! 885
+% 886
+Q 887
+$ 888
+Z 889
+X 890
+é 891
+/ 892
+í 893
+á 894
+£ 895
+ó 896
+ā 897
+ü 898
+ñ 899
+ö 900
+è 901
+ç 902
+à 903
+¿ 904
+μ 905
+π 906
+ä 907
+ú 908
+θ 909
+ã 910
+φ 911
+ī 912
+σ 913
+ê 914
+ρ 915
+â 916
+ô 917
+^ 918
+€ 919
+É 920
+ū 921
+Δ 922
+λ 923
+α 924
+τ 925
+æ 926
+а 927
+о 928
+ν 929
+î 930
+γ 931
+ψ 932
+ē 933
+т 934
+ß 935
+ω 936
+ï 937
+ć 938
+č 939
+ε 940
+е 941
+и 942
+ò 943
+р 944
+β 945
+ø 946
+ł 947
+δ 948
+η 949
+п 950
+ë 951
+н 952
+с 953
+š 954
+Ü 955
+å 956
+ń 957
+ś 958
+я 959
+đ 960
+л 961
+м 962
+Ö 963
+û 964
+ș 965
+в 966
+Á 967
+Ø 968
+ù 969
+ο 970
+ч 971
+ь 972
+ž 973
+Φ 974
+у 975
+ę 976
+ι 977
+б 978
+г 979
+к 980
+ő 981
+Ś 982
+Ω 983
+κ 984
+υ 985
+ì 986
+Č 987
+έ 988
+х 989
+ы 990
+Å 991
+Ç 992
+ż 993
+ί 994
+ζ 995
+χ 996
+э 997
+Æ 998
+Í 999
+õ 1000
+ě 1001
+ħ 1002
+Ł 1003
+œ 1004
+Ž 1005
+ț 1006
+Γ 1007
+П 1008
+д 1009
+з 1010
+ф 1011
+¡ 1012
+À 1013
+Î 1014
+Ā 1015
+ė 1016
+Š 1017
+ź 1018
+Κ 1019
+Ψ 1020
+ά 1021
+ξ 1022
+ό 1023
+<blk> 1024

sherpa-onnx/int8/bpe.vocab

@@ -0,0 +1,1024 @@
+<unk>	0.0
+▁t	-0.0
+▁th	-1.0
+▁a	-2.0
+in	-3.0
+▁the	-4.0
+re	-5.0
+▁w	-6.0
+▁o	-7.0
+▁s	-8.0
+at	-9.0
+ou	-10.0
+er	-11.0
+nd	-12.0
+▁i	-13.0
+▁b	-14.0
+▁c	-15.0
+on	-16.0
+▁h	-17.0
+ing	-18.0
+▁to	-19.0
+▁m	-20.0
+en	-21.0
+▁f	-22.0
+▁p	-23.0
+an	-24.0
+▁d	-25.0
+es	-26.0
+or	-27.0
+ll	-28.0
+▁of	-29.0
+▁and	-30.0
+▁y	-31.0
+▁l	-32.0
+▁I	-33.0
+it	-34.0
+▁in	-35.0
+is	-36.0
+ed	-37.0
+▁g	-38.0
+▁you	-39.0
+ar	-40.0
+▁that	-41.0
+om	-42.0
+as	-43.0
+▁n	-44.0
+ve	-45.0
+us	-46.0
+ic	-47.0
+ow	-48.0
+al	-49.0
+▁it	-50.0
+▁be	-51.0
+▁wh	-52.0
+le	-53.0
+ion	-54.0
+ut	-55.0
+ot	-56.0
+▁we	-57.0
+▁is	-58.0
+▁e	-59.0
+et	-60.0
+ay	-61.0
+▁re	-62.0
+▁on	-63.0
+▁T	-64.0
+▁A	-65.0
+▁ha	-66.0
+ent	-67.0
+ke	-68.0
+ct	-69.0
+▁S	-70.0
+ig	-71.0
+ver	-72.0
+▁Th	-73.0
+all	-74.0
+id	-75.0
+▁for	-76.0
+ro	-77.0
+▁he	-78.0
+se	-79.0
+▁this	-80.0
+ld	-81.0
+ly	-82.0
+▁go	-83.0
+▁k	-84.0
+▁st	-85.0
+st	-86.0
+ch	-87.0
+▁li	-88.0
+▁u	-89.0
+am	-90.0
+ur	-91.0
+ce	-92.0
+ith	-93.0
+im	-94.0
+▁so	-95.0
+▁have	-96.0
+▁do	-97.0
+ht	-98.0
+th	-99.0
+▁an	-100.0
+▁with	-101.0
+ad	-102.0
+▁r	-103.0
+ir	-104.0
+▁was	-105.0
+▁as	-106.0
+▁W	-107.0
+▁are	-108.0
+ust	-109.0
+ally	-110.0
+▁j	-111.0
+▁se	-112.0
+ation	-113.0
+od	-114.0
+ere	-115.0
+▁like	-116.0
+▁not	-117.0
+▁kn	-118.0
+ight	-119.0
+▁B	-120.0
+▁they	-121.0
+▁And	-122.0
+▁know	-123.0
+ome	-124.0
+op	-125.0
+▁can	-126.0
+▁or	-127.0
+▁sh	-128.0
+▁me	-129.0
+ill	-130.0
+ant	-131.0
+ck	-132.0
+▁what	-133.0
+▁at	-134.0
+▁ab	-135.0
+ould	-136.0
+ol	-137.0
+▁So	-138.0
+▁C	-139.0
+use	-140.0
+ter	-141.0
+il	-142.0
+▁but	-143.0
+▁just	-144.0
+▁ne	-145.0
+▁de	-146.0
+ra	-147.0
+ore	-148.0
+▁there	-149.0
+ul	-150.0
+out	-151.0
+▁con	-152.0
+▁all	-153.0
+▁The	-154.0
+ers	-155.0
+▁H	-156.0
+▁fr	-157.0
+▁pro	-158.0
+ge	-159.0
+ea	-160.0
+▁Y	-161.0
+▁O	-162.0
+▁M	-163.0
+pp	-164.0
+▁com	-165.0
+ess	-166.0
+▁ch	-167.0
+▁al	-168.0
+est	-169.0
+ate	-170.0
+qu	-171.0
+▁lo	-172.0
+▁ex	-173.0
+very	-174.0
+▁su	-175.0
+ain	-176.0
+▁one	-177.0
+ca	-178.0
+art	-179.0
+ist	-180.0
+if	-181.0
+ive	-182.0
+▁if	-183.0
+ink	-184.0
+nt	-185.0
+ab	-186.0
+▁about	-187.0
+▁going	-188.0
+▁v	-189.0
+▁wor	-190.0
+um	-191.0
+ok	-192.0
+▁your	-193.0
+▁my	-194.0
+ind	-195.0
+▁get	-196.0
+cause	-197.0
+▁from	-198.0
+▁don	-199.0
+ri	-200.0
+pe	-201.0
+un	-202.0
+ity	-203.0
+▁up	-204.0
+▁P	-205.0
+▁out	-206.0
+ort	-207.0
+▁L	-208.0
+ment	-209.0
+el	-210.0
+▁N	-211.0
+▁some	-212.0
+ich	-213.0
+and	-214.0
+▁think	-215.0
+em	-216.0
+oug	-217.0
+▁G	-218.0
+os	-219.0
+▁D	-220.0
+res	-221.0
+▁because	-222.0
+▁by	-223.0
+ake	-224.0
+▁int	-225.0
+ie	-226.0
+▁us	-227.0
+▁tr	-228.0
+▁then	-229.0
+ack	-230.0
+▁pl	-231.0
+▁here	-232.0
+▁pe	-233.0
+her	-234.0
+▁will	-235.0
+▁F	-236.0
+▁which	-237.0
+ard	-238.0
+▁right	-239.0
+▁thing	-240.0
+▁want	-241.0
+ies	-242.0
+ople	-243.0
+▁It	-244.0
+▁them	-245.0
+ame	-246.0
+▁We	-247.0
+our	-248.0
+▁say	-249.0
+▁R	-250.0
+▁people	-251.0
+▁see	-252.0
+▁who	-253.0
+ast	-254.0
+ure	-255.0
+ect	-256.0
+ear	-257.0
+▁tim	-258.0
+▁E	-259.0
+▁You	-260.0
+▁would	-261.0
+▁when	-262.0
+ven	-263.0
+▁our	-264.0
+ci	-265.0
+▁really	-266.0
+▁more	-267.0
+ound	-268.0
+ose	-269.0
+ak	-270.0
+▁co	-271.0
+ide	-272.0
+ough	-273.0
+▁had	-274.0
+so	-275.0
+▁qu	-276.0
+eah	-277.0
+▁were	-278.0
+ine	-279.0
+▁act	-280.0
+ther	-281.0
+▁these	-282.0
+▁how	-283.0
+▁now	-284.0
+▁sa	-285.0
+ud	-286.0
+▁Wh	-287.0
+▁man	-288.0
+ous	-289.0
+one	-290.0
+pt	-291.0
+ff	-292.0
+ong	-293.0
+▁has	-294.0
+▁any	-295.0
+▁very	-296.0
+▁But	-297.0
+▁look	-298.0
+iv	-299.0
+itt	-300.0
+▁time	-301.0
+▁mo	-302.0
+▁ar	-303.0
+hing	-304.0
+▁le	-305.0
+▁work	-306.0
+▁their	-307.0
+are	-308.0
+▁his	-309.0
+per	-310.0
+ions	-311.0
+▁im	-312.0
+▁ag	-313.0
+▁J	-314.0
+▁no	-315.0
+▁en	-316.0
+▁got	-317.0
+ag	-318.0
+▁sp	-319.0
+ans	-320.0
+act	-321.0
+▁te	-322.0
+▁also	-323.0
+iz	-324.0
+ice	-325.0
+▁That	-326.0
+▁cl	-327.0
+▁been	-328.0
+▁way	-329.0
+▁fe	-330.0
+▁did	-331.0
+ple	-332.0
+ually	-333.0
+▁other	-334.0
+▁U	-335.0
+ite	-336.0
+age	-337.0
+omet	-338.0
+ber	-339.0
+reat	-340.0
+ree	-341.0
+▁into	-342.0
+own	-343.0
+▁tw	-344.0
+▁part	-345.0
+alk	-346.0
+▁where	-347.0
+▁need	-348.0
+▁every	-349.0
+pl	-350.0
+▁ad	-351.0
+ry	-352.0
+▁over	-353.0
+ble	-354.0
+ap	-355.0
+ue	-356.0
+▁kind	-357.0
+▁po	-358.0
+▁back	-359.0
+▁cont	-360.0
+iff	-361.0
+▁somet	-362.0
+▁pr	-363.0
+nder	-364.0
+ire	-365.0
+▁good	-366.0
+▁than	-367.0
+ace	-368.0
+▁gu	-369.0
+ep	-370.0
+og	-371.0
+ick	-372.0
+way	-373.0
+▁lot	-374.0
+▁un	-375.0
+▁things	-376.0
+▁In	-377.0
+ish	-378.0
+kay	-379.0
+▁well	-380.0
+▁could	-381.0
+▁pre	-382.0
+▁two	-383.0
+irst	-384.0
+▁diff	-385.0
+ach	-386.0
+cc	-387.0
+ittle	-388.0
+int	-389.0
+▁He	-390.0
+▁those	-391.0
+ence	-392.0
+ip	-393.0
+ase	-394.0
+▁him	-395.0
+▁make	-396.0
+▁little	-397.0
+ical	-398.0
+▁gr	-399.0
+▁year	-400.0
+ass	-401.0
+▁thr	-402.0
+uch	-403.0
+ated	-404.0
+▁This	-405.0
+▁off	-406.0
+▁res	-407.0
+ac	-408.0
+ance	-409.0
+▁actually	-410.0
+▁talk	-411.0
+ult	-412.0
+able	-413.0
+orm	-414.0
+▁dis	-415.0
+▁first	-416.0
+ations	-417.0
+▁something	-418.0
+▁she	-419.0
+sel	-420.0
+▁let	-421.0
+ord	-422.0
+▁may	-423.0
+ia	-424.0
+▁am	-425.0
+▁her	-426.0
+▁said	-427.0
+▁bo	-428.0
+be	-429.0
+ount	-430.0
+▁much	-431.0
+▁per	-432.0
+▁even	-433.0
+▁differe	-434.0
+vel	-435.0
+ary	-436.0
+▁app	-437.0
+ving	-438.0
+▁comm	-439.0
+▁imp	-440.0
+ys	-441.0
+▁again	-442.0
+ress	-443.0
+▁yeah	-444.0
+▁down	-445.0
+ang	-446.0
+▁mean	-447.0
+na	-448.0
+ens	-449.0
+▁does	-450.0
+▁fo	-451.0
+▁comp	-452.0
+▁ro	-453.0
+▁bl	-454.0
+ody	-455.0
+▁K	-456.0
+▁through	-457.0
+▁start	-458.0
+uct	-459.0
+▁only	-460.0
+▁bet	-461.0
+▁under	-462.0
+▁br	-463.0
+▁take	-464.0
+ning	-465.0
+▁bu	-466.0
+▁use	-467.0
+▁Ch	-468.0
+xt	-469.0
+co	-470.0
+ory	-471.0
+ild	-472.0
+▁put	-473.0
+▁call	-474.0
+▁new	-475.0
+other	-476.0
+ting	-477.0
+▁happ	-478.0
+ater	-479.0
+▁inc	-480.0
+ition	-481.0
+▁different	-482.0
+▁should	-483.0
+ade	-484.0
+ign	-485.0
+thing	-486.0
+▁day	-487.0
+fore	-488.0
+▁Yeah	-489.0
+ark	-490.0
+ile	-491.0
+ial	-492.0
+▁come	-493.0
+▁They	-494.0
+▁being	-495.0
+▁try	-496.0
+ious	-497.0
+▁sc	-498.0
+▁bit	-499.0
+▁spe	-500.0
+ub	-501.0
+fe	-502.0
+▁doing	-503.0
+▁St	-504.0
+vers	-505.0
+av	-506.0
+ty	-507.0
+ian	-508.0
+onna	-509.0
+red	-510.0
+wn	-511.0
+▁ke	-512.0
+form	-513.0
+ors	-514.0
+▁fl	-515.0
+fter	-516.0
+ail	-517.0
+ents	-518.0
+▁gonna	-519.0
+▁point	-520.0
+ces	-521.0
+▁There	-522.0
+self	-523.0
+▁many	-524.0
+▁If	-525.0
+▁same	-526.0
+▁sy	-527.0
+▁quest	-528.0
+▁most	-529.0
+▁great	-530.0
+▁What	-531.0
+▁fu	-532.0
+ug	-533.0
+▁show	-534.0
+we	-535.0
+ual	-536.0
+ons	-537.0
+▁Be	-538.0
+ically	-539.0
+▁ser	-540.0
+▁rem	-541.0
+▁ind	-542.0
+▁pers	-543.0
+▁V	-544.0
+he	-545.0
+▁str	-546.0
+ved	-547.0
+▁still	-548.0
+ank	-549.0
+▁rec	-550.0
+▁wr	-551.0
+ought	-552.0
+day	-553.0
+ath	-554.0
+▁end	-555.0
+▁bas	-556.0
+ft	-557.0
+erm	-558.0
+body	-559.0
+ph	-560.0
+ject	-561.0
+ict	-562.0
+▁play	-563.0
+▁Is	-564.0
+ates	-565.0
+▁ph	-566.0
+oth	-567.0
+▁acc	-568.0
+get	-569.0
+▁years	-570.0
+▁em	-571.0
+▁id	-572.0
+▁Oh	-573.0
+ves	-574.0
+ever	-575.0
+▁inter	-576.0
+▁rel	-577.0
+▁before	-578.0
+▁feel	-579.0
+igh	-580.0
+▁three	-581.0
+iss	-582.0
+▁des	-583.0
+ne	-584.0
+▁why	-585.0
+▁uh	-586.0
+▁To	-587.0
+▁cons	-588.0
+▁hel	-589.0
+▁after	-590.0
+ower	-591.0
+urn	-592.0
+▁okay	-593.0
+▁long	-594.0
+▁bel	-595.0
+▁around	-596.0
+ful	-597.0
+te	-598.0
+ise	-599.0
+▁ob	-600.0
+▁supp	-601.0
+ady	-602.0
+ange	-603.0
+aking	-604.0
+▁pos	-605.0
+atch	-606.0
+▁tra	-607.0
+gr	-608.0
+▁might	-609.0
+ert	-610.0
+▁help	-611.0
+ost	-612.0
+▁too	-613.0
+cial	-614.0
+▁world	-615.0
+▁give	-616.0
+ike	-617.0
+▁Okay	-618.0
+ways	-619.0
+▁min	-620.0
+ward	-621.0
+ily	-622.0
+▁gen	-623.0
+▁find	-624.0
+▁dec	-625.0
+ular	-626.0
+ob	-627.0
+▁tell	-628.0
+▁Now	-629.0
+▁sm	-630.0
+▁cour	-631.0
+▁real	-632.0
+cess	-633.0
+nds	-634.0
+▁big	-635.0
+▁num	-636.0
+ction	-637.0
+▁add	-638.0
+▁set	-639.0
+▁um	-640.0
+ood	-641.0
+ible	-642.0
+▁own	-643.0
+▁life	-644.0
+ities	-645.0
+▁its	-646.0
+▁God	-647.0
+pect	-648.0
+▁didn	-649.0
+stem	-650.0
+les	-651.0
+uc	-652.0
+ib	-653.0
+ating	-654.0
+olog	-655.0
+▁person	-656.0
+▁inv	-657.0
+ably	-658.0
+▁sure	-659.0
+▁reg	-660.0
+lic	-661.0
+▁stu	-662.0
+▁cr	-663.0
+▁ev	-664.0
+ments	-665.0
+▁another	-666.0
+▁la	-667.0
+▁last	-668.0
+▁sub	-669.0
+▁att	-670.0
+▁op	-671.0
+▁inst	-672.0
+▁sl	-673.0
+▁happen	-674.0
+▁rep	-675.0
+▁import	-676.0
+ific	-677.0
+ix	-678.0
+▁made	-679.0
+▁ear	-680.0
+▁ac	-681.0
+▁def	-682.0
+ute	-683.0
+▁next	-684.0
+ative	-685.0
+▁form	-686.0
+▁guys	-687.0
+▁system	-688.0
+ew	-689.0
+▁able	-690.0
+ied	-691.0
+▁always	-692.0
+ren	-693.0
+erest	-694.0
+▁As	-695.0
+▁mod	-696.0
+▁done	-697.0
+ings	-698.0
+▁love	-699.0
+ism	-700.0
+▁ask	-701.0
+old	-702.0
+ered	-703.0
+▁trans	-704.0
+▁count	-705.0
+ility	-706.0
+▁high	-707.0
+▁fin	-708.0
+▁prob	-709.0
+▁pol	-710.0
+▁exam	-711.0
+▁pres	-712.0
+▁maybe	-713.0
+ell	-714.0
+▁stud	-715.0
+▁prod	-716.0
+▁car	-717.0
+ock	-718.0
+▁used	-719.0
+oy	-720.0
+stand	-721.0
+▁No	-722.0
+▁mon	-723.0
+ks	-724.0
+▁interest	-725.0
+▁ent	-726.0
+ited	-727.0
+▁sort	-728.0
+▁For	-729.0
+▁today	-730.0
+ics	-731.0
+▁vide	-732.0
+▁bec	-733.0
+▁Well	-734.0
+▁Al	-735.0
+▁important	-736.0
+▁such	-737.0
+▁run	-738.0
+▁keep	-739.0
+▁fact	-740.0
+ata	-741.0
+ss	-742.0
+▁never	-743.0
+ween	-744.0
+▁stuff	-745.0
+ract	-746.0
+▁question	-747.0
+als	-748.0
+▁sim	-749.0
+vern	-750.0
+ather	-751.0
+▁course	-752.0
+▁Of	-753.0
+oc	-754.0
+ness	-755.0
+arch	-756.0
+ize	-757.0
+▁All	-758.0
+ense	-759.0
+blem	-760.0
+▁probably	-761.0
+hip	-762.0
+▁number	-763.0
+ention	-764.0
+▁saying	-765.0
+▁commun	-766.0
+▁An	-767.0
+akes	-768.0
+▁belie	-769.0
+▁between	-770.0
+▁better	-771.0
+cus	-772.0
+▁place	-773.0
+▁gener	-774.0
+▁ca	-775.0
+▁ins	-776.0
+▁ass	-777.0
+cond	-778.0
+cept	-779.0
+ull	-780.0
+▁understand	-781.0
+▁fun	-782.0
+▁thought	-783.0
+gan	-784.0
+iew	-785.0
+cy	-786.0
+ution	-787.0
+ope	-788.0
+ason	-789.0
+▁problem	-790.0
+▁doesn	-791.0
+ational	-792.0
+▁read	-793.0
+▁trying	-794.0
+▁sch	-795.0
+▁el	-796.0
+ah	-797.0
+atter	-798.0
+▁exper	-799.0
+▁four	-800.0
+▁ele	-801.0
+▁cou	-802.0
+ont	-803.0
+▁called	-804.0
+▁partic	-805.0
+▁open	-806.0
+▁gl	-807.0
+▁everything	-808.0
+▁eff	-809.0
+▁getting	-810.0
+▁ty	-811.0
+▁Am	-812.0
+▁Because	-813.0
+ave	-814.0
+▁met	-815.0
+▁Like	-816.0
+oney	-817.0
+▁	-818.0
+e	-819.0
+t	-820.0
+o	-821.0
+a	-822.0
+n	-823.0
+i	-824.0
+s	-825.0
+h	-826.0
+r	-827.0
+l	-828.0
+d	-829.0
+u	-830.0
+c	-831.0
+y	-832.0
+m	-833.0
+g	-834.0
+w	-835.0
+f	-836.0
+p	-837.0
+,	-838.0
+b	-839.0
+.	-840.0
+k	-841.0
+v	-842.0
+'	-843.0
+I	-844.0
+T	-845.0
+A	-846.0
+S	-847.0
+j	-848.0
+x	-849.0
+W	-850.0
+B	-851.0
+C	-852.0
+?	-853.0
+0	-854.0
+O	-855.0
+-	-856.0
+M	-857.0
+H	-858.0
+Y	-859.0
+q	-860.0
+1	-861.0
+P	-862.0
+z	-863.0
+L	-864.0
+D	-865.0
+N	-866.0
+G	-867.0
+F	-868.0
+R	-869.0
+E	-870.0
+2	-871.0
+J	-872.0
+U	-873.0
+:	-874.0
+5	-875.0
+9	-876.0
+3	-877.0
+K	-878.0
+4	-879.0
+V	-880.0
+8	-881.0
+6	-882.0
+7	-883.0
+!	-884.0
+%	-885.0
+Q	-886.0
+$	-887.0
+Z	-888.0
+X	-889.0
+é	-890.0
+/	-891.0
+í	-892.0
+á	-893.0
+£	-894.0
+ó	-895.0
+ā	-896.0
+ü	-897.0
+ñ	-898.0
+ö	-899.0
+è	-900.0
+ç	-901.0
+à	-902.0
+¿	-903.0
+μ	-904.0
+π	-905.0
+ä	-906.0
+ú	-907.0
+θ	-908.0
+ã	-909.0
+φ	-910.0
+ī	-911.0
+σ	-912.0
+ê	-913.0
+ρ	-914.0
+â	-915.0
+ô	-916.0
+^	-917.0
+€	-918.0
+É	-919.0
+ū	-920.0
+Δ	-921.0
+λ	-922.0
+α	-923.0
+τ	-924.0
+æ	-925.0
+а	-926.0
+о	-927.0
+ν	-928.0
+î	-929.0
+γ	-930.0
+ψ	-931.0
+ē	-932.0
+т	-933.0
+ß	-934.0
+ω	-935.0
+ï	-936.0
+ć	-937.0
+č	-938.0
+ε	-939.0
+е	-940.0
+и	-941.0
+ò	-942.0
+р	-943.0
+β	-944.0
+ø	-945.0
+ł	-946.0
+δ	-947.0
+η	-948.0
+п	-949.0
+ë	-950.0
+н	-951.0
+с	-952.0
+š	-953.0
+Ü	-954.0
+å	-955.0
+ń	-956.0
+ś	-957.0
+я	-958.0
+đ	-959.0
+л	-960.0
+м	-961.0
+Ö	-962.0
+û	-963.0
+ș	-964.0
+в	-965.0
+Á	-966.0
+Ø	-967.0
+ù	-968.0
+ο	-969.0
+ч	-970.0
+ь	-971.0
+ž	-972.0
+Φ	-973.0
+у	-974.0
+ę	-975.0
+ι	-976.0
+б	-977.0
+г	-978.0
+к	-979.0
+ő	-980.0
+Ś	-981.0
+Ω	-982.0
+κ	-983.0
+υ	-984.0
+ì	-985.0
+Č	-986.0
+έ	-987.0
+х	-988.0
+ы	-989.0
+Å	-990.0
+Ç	-991.0
+ż	-992.0
+ί	-993.0
+ζ	-994.0
+χ	-995.0
+э	-996.0
+Æ	-997.0
+Í	-998.0
+õ	-999.0
+ě	-1000.0
+ħ	-1001.0
+Ł	-1002.0
+œ	-1003.0
+Ž	-1004.0
+ț	-1005.0
+Γ	-1006.0
+П	-1007.0
+д	-1008.0
+з	-1009.0
+ф	-1010.0
+¡	-1011.0
+À	-1012.0
+Î	-1013.0
+Ā	-1014.0
+ė	-1015.0
+Š	-1016.0
+ź	-1017.0
+Κ	-1018.0
+Ψ	-1019.0
+ά	-1020.0
+ξ	-1021.0
+ό	-1022.0

sherpa-onnx/int8/decoder.int8.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:24cf4df5ade1d2a81b25f879c255d7e656ca7055d83cd6a7641c6e64ac638d9c
+size 7257753

sherpa-onnx/int8/encoder.int8.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0a411fa880171c5576832246146b1966e7ed97169f1243a0781e869c09c7a1f3
+size 652282305

sherpa-onnx/int8/joiner.int8.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3050c3167b933269c6aea3f9bee4dea30860199285169e0f6b48fed7855eb5d9
+size 1739080

sherpa-onnx/int8/tokens.txt

@@ -0,0 +1,1025 @@
+<unk> 0
+▁t 1
+▁th 2
+▁a 3
+in 4
+▁the 5
+re 6
+▁w 7
+▁o 8
+▁s 9
+at 10
+ou 11
+er 12
+nd 13
+▁i 14
+▁b 15
+▁c 16
+on 17
+▁h 18
+ing 19
+▁to 20
+▁m 21
+en 22
+▁f 23
+▁p 24
+an 25
+▁d 26
+es 27
+or 28
+ll 29
+▁of 30
+▁and 31
+▁y 32
+▁l 33
+▁I 34
+it 35
+▁in 36
+is 37
+ed 38
+▁g 39
+▁you 40
+ar 41
+▁that 42
+om 43
+as 44
+▁n 45
+ve 46
+us 47
+ic 48
+ow 49
+al 50
+▁it 51
+▁be 52
+▁wh 53
+le 54
+ion 55
+ut 56
+ot 57
+▁we 58
+▁is 59
+▁e 60
+et 61
+ay 62
+▁re 63
+▁on 64
+▁T 65
+▁A 66
+▁ha 67
+ent 68
+ke 69
+ct 70
+▁S 71
+ig 72
+ver 73
+▁Th 74
+all 75
+id 76
+▁for 77
+ro 78
+▁he 79
+se 80
+▁this 81
+ld 82
+ly 83
+▁go 84
+▁k 85
+▁st 86
+st 87
+ch 88
+▁li 89
+▁u 90
+am 91
+ur 92
+ce 93
+ith 94
+im 95
+▁so 96
+▁have 97
+▁do 98
+ht 99
+th 100
+▁an 101
+▁with 102
+ad 103
+▁r 104
+ir 105
+▁was 106
+▁as 107
+▁W 108
+▁are 109
+ust 110
+ally 111
+▁j 112
+▁se 113
+ation 114
+od 115
+ere 116
+▁like 117
+▁not 118
+▁kn 119
+ight 120
+▁B 121
+▁they 122
+▁And 123
+▁know 124
+ome 125
+op 126
+▁can 127
+▁or 128
+▁sh 129
+▁me 130
+ill 131
+ant 132
+ck 133
+▁what 134
+▁at 135
+▁ab 136
+ould 137
+ol 138
+▁So 139
+▁C 140
+use 141
+ter 142
+il 143
+▁but 144
+▁just 145
+▁ne 146
+▁de 147
+ra 148
+ore 149
+▁there 150
+ul 151
+out 152
+▁con 153
+▁all 154
+▁The 155
+ers 156
+▁H 157
+▁fr 158
+▁pro 159
+ge 160
+ea 161
+▁Y 162
+▁O 163
+▁M 164
+pp 165
+▁com 166
+ess 167
+▁ch 168
+▁al 169
+est 170
+ate 171
+qu 172
+▁lo 173
+▁ex 174
+very 175
+▁su 176
+ain 177
+▁one 178
+ca 179
+art 180
+ist 181
+if 182
+ive 183
+▁if 184
+ink 185
+nt 186
+ab 187
+▁about 188
+▁going 189
+▁v 190
+▁wor 191
+um 192
+ok 193
+▁your 194
+▁my 195
+ind 196
+▁get 197
+cause 198
+▁from 199
+▁don 200
+ri 201
+pe 202
+un 203
+ity 204
+▁up 205
+▁P 206
+▁out 207
+ort 208
+▁L 209
+ment 210
+el 211
+▁N 212
+▁some 213
+ich 214
+and 215
+▁think 216
+em 217
+oug 218
+▁G 219
+os 220
+▁D 221
+res 222
+▁because 223
+▁by 224
+ake 225
+▁int 226
+ie 227
+▁us 228
+▁tr 229
+▁then 230
+ack 231
+▁pl 232
+▁here 233
+▁pe 234
+her 235
+▁will 236
+▁F 237
+▁which 238
+ard 239
+▁right 240
+▁thing 241
+▁want 242
+ies 243
+ople 244
+▁It 245
+▁them 246
+ame 247
+▁We 248
+our 249
+▁say 250
+▁R 251
+▁people 252
+▁see 253
+▁who 254
+ast 255
+ure 256
+ect 257
+ear 258
+▁tim 259
+▁E 260
+▁You 261
+▁would 262
+▁when 263
+ven 264
+▁our 265
+ci 266
+▁really 267
+▁more 268
+ound 269
+ose 270
+ak 271
+▁co 272
+ide 273
+ough 274
+▁had 275
+so 276
+▁qu 277
+eah 278
+▁were 279
+ine 280
+▁act 281
+ther 282
+▁these 283
+▁how 284
+▁now 285
+▁sa 286
+ud 287
+▁Wh 288
+▁man 289
+ous 290
+one 291
+pt 292
+ff 293
+ong 294
+▁has 295
+▁any 296
+▁very 297
+▁But 298
+▁look 299
+iv 300
+itt 301
+▁time 302
+▁mo 303
+▁ar 304
+hing 305
+▁le 306
+▁work 307
+▁their 308
+are 309
+▁his 310
+per 311
+ions 312
+▁im 313
+▁ag 314
+▁J 315
+▁no 316
+▁en 317
+▁got 318
+ag 319
+▁sp 320
+ans 321
+act 322
+▁te 323
+▁also 324
+iz 325
+ice 326
+▁That 327
+▁cl 328
+▁been 329
+▁way 330
+▁fe 331
+▁did 332
+ple 333
+ually 334
+▁other 335
+▁U 336
+ite 337
+age 338
+omet 339
+ber 340
+reat 341
+ree 342
+▁into 343
+own 344
+▁tw 345
+▁part 346
+alk 347
+▁where 348
+▁need 349
+▁every 350
+pl 351
+▁ad 352
+ry 353
+▁over 354
+ble 355
+ap 356
+ue 357
+▁kind 358
+▁po 359
+▁back 360
+▁cont 361
+iff 362
+▁somet 363
+▁pr 364
+nder 365
+ire 366
+▁good 367
+▁than 368
+ace 369
+▁gu 370
+ep 371
+og 372
+ick 373
+way 374
+▁lot 375
+▁un 376
+▁things 377
+▁In 378
+ish 379
+kay 380
+▁well 381
+▁could 382
+▁pre 383
+▁two 384
+irst 385
+▁diff 386
+ach 387
+cc 388
+ittle 389
+int 390
+▁He 391
+▁those 392
+ence 393
+ip 394
+ase 395
+▁him 396
+▁make 397
+▁little 398
+ical 399
+▁gr 400
+▁year 401
+ass 402
+▁thr 403
+uch 404
+ated 405
+▁This 406
+▁off 407
+▁res 408
+ac 409
+ance 410
+▁actually 411
+▁talk 412
+ult 413
+able 414
+orm 415
+▁dis 416
+▁first 417
+ations 418
+▁something 419
+▁she 420
+sel 421
+▁let 422
+ord 423
+▁may 424
+ia 425
+▁am 426
+▁her 427
+▁said 428
+▁bo 429
+be 430
+ount 431
+▁much 432
+▁per 433
+▁even 434
+▁differe 435
+vel 436
+ary 437
+▁app 438
+ving 439
+▁comm 440
+▁imp 441
+ys 442
+▁again 443
+ress 444
+▁yeah 445
+▁down 446
+ang 447
+▁mean 448
+na 449
+ens 450
+▁does 451
+▁fo 452
+▁comp 453
+▁ro 454
+▁bl 455
+ody 456
+▁K 457
+▁through 458
+▁start 459
+uct 460
+▁only 461
+▁bet 462
+▁under 463
+▁br 464
+▁take 465
+ning 466
+▁bu 467
+▁use 468
+▁Ch 469
+xt 470
+co 471
+ory 472
+ild 473
+▁put 474
+▁call 475
+▁new 476
+other 477
+ting 478
+▁happ 479
+ater 480
+▁inc 481
+ition 482
+▁different 483
+▁should 484
+ade 485
+ign 486
+thing 487
+▁day 488
+fore 489
+▁Yeah 490
+ark 491
+ile 492
+ial 493
+▁come 494
+▁They 495
+��being 496
+▁try 497
+ious 498
+▁sc 499
+▁bit 500
+▁spe 501
+ub 502
+fe 503
+▁doing 504
+▁St 505
+vers 506
+av 507
+ty 508
+ian 509
+onna 510
+red 511
+wn 512
+▁ke 513
+form 514
+ors 515
+▁fl 516
+fter 517
+ail 518
+ents 519
+▁gonna 520
+▁point 521
+ces 522
+▁There 523
+self 524
+▁many 525
+▁If 526
+▁same 527
+▁sy 528
+▁quest 529
+▁most 530
+▁great 531
+▁What 532
+▁fu 533
+ug 534
+▁show 535
+we 536
+ual 537
+ons 538
+▁Be 539
+ically 540
+▁ser 541
+▁rem 542
+▁ind 543
+▁pers 544
+▁V 545
+he 546
+▁str 547
+ved 548
+▁still 549
+ank 550
+▁rec 551
+▁wr 552
+ought 553
+day 554
+ath 555
+▁end 556
+▁bas 557
+ft 558
+erm 559
+body 560
+ph 561
+ject 562
+ict 563
+▁play 564
+▁Is 565
+ates 566
+▁ph 567
+oth 568
+▁acc 569
+get 570
+▁years 571
+▁em 572
+▁id 573
+▁Oh 574
+ves 575
+ever 576
+▁inter 577
+▁rel 578
+▁before 579
+▁feel 580
+igh 581
+▁three 582
+iss 583
+▁des 584
+ne 585
+▁why 586
+▁uh 587
+▁To 588
+▁cons 589
+▁hel 590
+▁after 591
+ower 592
+urn 593
+▁okay 594
+▁long 595
+▁bel 596
+▁around 597
+ful 598
+te 599
+ise 600
+▁ob 601
+▁supp 602
+ady 603
+ange 604
+aking 605
+▁pos 606
+atch 607
+▁tra 608
+gr 609
+▁might 610
+ert 611
+▁help 612
+ost 613
+▁too 614
+cial 615
+▁world 616
+▁give 617
+ike 618
+▁Okay 619
+ways 620
+▁min 621
+ward 622
+ily 623
+▁gen 624
+▁find 625
+▁dec 626
+ular 627
+ob 628
+▁tell 629
+▁Now 630
+▁sm 631
+▁cour 632
+▁real 633
+cess 634
+nds 635
+▁big 636
+▁num 637
+ction 638
+▁add 639
+▁set 640
+▁um 641
+ood 642
+ible 643
+▁own 644
+▁life 645
+ities 646
+▁its 647
+▁God 648
+pect 649
+▁didn 650
+stem 651
+les 652
+uc 653
+ib 654
+ating 655
+olog 656
+▁person 657
+▁inv 658
+ably 659
+▁sure 660
+▁reg 661
+lic 662
+▁stu 663
+▁cr 664
+▁ev 665
+ments 666
+▁another 667
+▁la 668
+▁last 669
+▁sub 670
+▁att 671
+▁op 672
+▁inst 673
+▁sl 674
+▁happen 675
+▁rep 676
+▁import 677
+ific 678
+ix 679
+▁made 680
+▁ear 681
+▁ac 682
+▁def 683
+ute 684
+▁next 685
+ative 686
+▁form 687
+▁guys 688
+▁system 689
+ew 690
+▁able 691
+ied 692
+▁always 693
+ren 694
+erest 695
+▁As 696
+▁mod 697
+▁done 698
+ings 699
+▁love 700
+ism 701
+▁ask 702
+old 703
+ered 704
+▁trans 705
+▁count 706
+ility 707
+▁high 708
+▁fin 709
+▁prob 710
+▁pol 711
+▁exam 712
+▁pres 713
+▁maybe 714
+ell 715
+▁stud 716
+▁prod 717
+▁car 718
+ock 719
+▁used 720
+oy 721
+stand 722
+▁No 723
+▁mon 724
+ks 725
+▁interest 726
+▁ent 727
+ited 728
+▁sort 729
+▁For 730
+▁today 731
+ics 732
+▁vide 733
+▁bec 734
+▁Well 735
+▁Al 736
+▁important 737
+▁such 738
+▁run 739
+▁keep 740
+▁fact 741
+ata 742
+ss 743
+▁never 744
+ween 745
+▁stuff 746
+ract 747
+▁question 748
+als 749
+▁sim 750
+vern 751
+ather 752
+▁course 753
+▁Of 754
+oc 755
+ness 756
+arch 757
+ize 758
+▁All 759
+ense 760
+blem 761
+▁probably 762
+hip 763
+▁number 764
+ention 765
+▁saying 766
+▁commun 767
+▁An 768
+akes 769
+▁belie 770
+▁between 771
+▁better 772
+cus 773
+▁place 774
+▁gener 775
+▁ca 776
+▁ins 777
+▁ass 778
+cond 779
+cept 780
+ull 781
+▁understand 782
+▁fun 783
+▁thought 784
+gan 785
+iew 786
+cy 787
+ution 788
+ope 789
+ason 790
+▁problem 791
+▁doesn 792
+ational 793
+▁read 794
+▁trying 795
+▁sch 796
+▁el 797
+ah 798
+atter 799
+▁exper 800
+▁four 801
+▁ele 802
+▁cou 803
+ont 804
+▁called 805
+▁partic 806
+▁open 807
+▁gl 808
+▁everything 809
+▁eff 810
+▁getting 811
+▁ty 812
+▁Am 813
+▁Because 814
+ave 815
+▁met 816
+▁Like 817
+oney 818
+▁ 819
+e 820
+t 821
+o 822
+a 823
+n 824
+i 825
+s 826
+h 827
+r 828
+l 829
+d 830
+u 831
+c 832
+y 833
+m 834
+g 835
+w 836
+f 837
+p 838
+, 839
+b 840
+. 841
+k 842
+v 843
+' 844
+I 845
+T 846
+A 847
+S 848
+j 849
+x 850
+W 851
+B 852
+C 853
+? 854
+0 855
+O 856
+- 857
+M 858
+H 859
+Y 860
+q 861
+1 862
+P 863
+z 864
+L 865
+D 866
+N 867
+G 868
+F 869
+R 870
+E 871
+2 872
+J 873
+U 874
+: 875
+5 876
+9 877
+3 878
+K 879
+4 880
+V 881
+8 882
+6 883
+7 884
+! 885
+% 886
+Q 887
+$ 888
+Z 889
+X 890
+é 891
+/ 892
+í 893
+á 894
+£ 895
+ó 896
+ā 897
+ü 898
+ñ 899
+ö 900
+è 901
+ç 902
+à 903
+¿ 904
+μ 905
+π 906
+ä 907
+ú 908
+θ 909
+ã 910
+φ 911
+ī 912
+σ 913
+ê 914
+ρ 915
+â 916
+ô 917
+^ 918
+€ 919
+É 920
+ū 921
+Δ 922
+λ 923
+α 924
+τ 925
+æ 926
+а 927
+о 928
+ν 929
+î 930
+γ 931
+ψ 932
+ē 933
+т 934
+ß 935
+ω 936
+ï 937
+ć 938
+č 939
+ε 940
+е 941
+и 942
+ò 943
+р 944
+β 945
+ø 946
+ł 947
+δ 948
+η 949
+п 950
+ë 951
+н 952
+с 953
+š 954
+Ü 955
+å 956
+ń 957
+ś 958
+я 959
+đ 960
+л 961
+м 962
+Ö 963
+û 964
+ș 965
+в 966
+Á 967
+Ø 968
+ù 969
+ο 970
+ч 971
+ь 972
+ž 973
+Φ 974
+у 975
+ę 976
+ι 977
+б 978
+г 979
+к 980
+ő 981
+Ś 982
+Ω 983
+κ 984
+υ 985
+ì 986
+Č 987
+έ 988
+х 989
+ы 990
+Å 991
+Ç 992
+ż 993
+ί 994
+ζ 995
+χ 996
+э 997
+Æ 998
+Í 999
+õ 1000
+ě 1001
+ħ 1002
+Ł 1003
+œ 1004
+Ž 1005
+ț 1006
+Γ 1007
+П 1008
+д 1009
+з 1010
+ф 1011
+¡ 1012
+À 1013
+Î 1014
+Ā 1015
+ė 1016
+Š 1017
+ź 1018
+Κ 1019
+Ψ 1020
+ά 1021
+ξ 1022
+ό 1023
+<blk> 1024

sherpa-onnx/int8/version.json

@@ -0,0 +1 @@
+{"version": "Fritter v1.3", "run": "r35-general"}

training/r11-general/config.json

@@ -0,0 +1,26 @@
+{
+ "run": "r11-general",
+ "base_model": "nvidia/parakeet-tdt-0.6b-v2",
+ "rav_sha": "520e7506e37920b911981714a1b4616c6fa0faf7",
+ "data": "general",
+ "english_share": null,
+ "replay_share": 0.1,
+ "epochs": 20,
+ "lr": 3e-05,
+ "batch_size": 16,
+ "seed": 7,
+ "mix": {
+  "blend": "general",
+  "cs_min": 319.9,
+  "own_en_min": 0.0,
+  "replay_min": 35.6,
+  "total_min": 391.0,
+  "n": 3848,
+  "priority_rows": 381,
+  "priority_extra_min": 35.5,
+  "priority_repeat": 3,
+  "primary_min": 319.9,
+  "primary_cs_min": 0.0,
+  "replay_share_actual": 0.091
+ }
+}

training/r11-general/result.json

@@ -0,0 +1,30 @@
+{
+ "run": "r11-general",
+ "base_model": "nvidia/parakeet-tdt-0.6b-v2",
+ "rav_sha": "520e7506e37920b911981714a1b4616c6fa0faf7",
+ "data": "general",
+ "english_share": null,
+ "replay_share": 0.1,
+ "epochs": 20,
+ "lr": 3e-05,
+ "batch_size": 16,
+ "seed": 7,
+ "mix": {
+  "blend": "general",
+  "cs_min": 319.9,
+  "own_en_min": 0.0,
+  "replay_min": 35.6,
+  "total_min": 391.0,
+  "n": 3848,
+  "priority_rows": 381,
+  "priority_extra_min": 35.5,
+  "priority_repeat": 3,
+  "primary_min": 319.9,
+  "primary_cs_min": 0.0,
+  "replay_share_actual": 0.091
+ },
+ "best_checkpoint": "best-epoch=17-val_wer=0.0424.ckpt",
+ "best_val_wer": 0.04243668541312218,
+ "train_minutes": 38.2,
+ "model": "runs/r11-general/model.nemo"
+}

training/r25-general/config.json

@@ -0,0 +1,53 @@
+{
+ "run": "r25-general",
+ "base_model": "nvidia/parakeet-tdt-0.6b-v2",
+ "init_from": null,
+ "parent": null,
+ "rav_sha": "5ce184058554ddf5b22b63ad5f062c6de6d96f04",
+ "data": "general",
+ "english_share": null,
+ "replay_share": 0.1,
+ "speed_perturb": {
+  "prob": 1.0,
+  "sr": 16000,
+  "resample_type": "kaiser_fast",
+  "min_speed_rate": 0.9,
+  "max_speed_rate": 1.1,
+  "num_rates": 3
+ },
+ "avg_last": 5,
+ "max_rows": null,
+ "epochs": 20,
+ "lr": 3e-05,
+ "batch_size": 16,
+ "seed": 7,
+ "mix": {
+  "blend": "general",
+  "cs_min": 694.8,
+  "own_en_min": 0.0,
+  "replay_min": 77.3,
+  "total_min": 849.2,
+  "n": 8871,
+  "priority_rows": 1204,
+  "priority_extra_min": 77.2,
+  "priority_repeat": 3,
+  "primary_min": 694.8,
+  "primary_cs_min": 0.0,
+  "replay_share_actual": 0.091
+ },
+ "targets": {
+  "train": {
+   "plain": 8869,
+   "full": 0,
+   "fallback": 2,
+   "dropped": 0
+  },
+  "val": {
+   "plain": 468,
+   "full": 0,
+   "fallback": 1,
+   "dropped": 0
+  },
+  "charset": "parakeet-tdt-0.6b-v2"
+ }
+}

training/r25-general/result.json

@@ -0,0 +1,68 @@
+{
+ "run": "r25-general",
+ "base_model": "nvidia/parakeet-tdt-0.6b-v2",
+ "init_from": null,
+ "parent": null,
+ "rav_sha": "5ce184058554ddf5b22b63ad5f062c6de6d96f04",
+ "data": "general",
+ "english_share": null,
+ "replay_share": 0.1,
+ "speed_perturb": {
+  "prob": 1.0,
+  "sr": 16000,
+  "resample_type": "kaiser_fast",
+  "min_speed_rate": 0.9,
+  "max_speed_rate": 1.1,
+  "num_rates": 3
+ },
+ "avg_last": 5,
+ "max_rows": null,
+ "epochs": 20,
+ "lr": 3e-05,
+ "batch_size": 16,
+ "seed": 7,
+ "mix": {
+  "blend": "general",
+  "cs_min": 694.8,
+  "own_en_min": 0.0,
+  "replay_min": 77.3,
+  "total_min": 849.2,
+  "n": 8871,
+  "priority_rows": 1204,
+  "priority_extra_min": 77.2,
+  "priority_repeat": 3,
+  "primary_min": 694.8,
+  "primary_cs_min": 0.0,
+  "replay_share_actual": 0.091
+ },
+ "targets": {
+  "train": {
+   "plain": 8869,
+   "full": 0,
+   "fallback": 2,
+   "dropped": 0
+  },
+  "val": {
+   "plain": 468,
+   "full": 0,
+   "fallback": 1,
+   "dropped": 0
+  },
+  "charset": "parakeet-tdt-0.6b-v2"
+ },
+ "best_checkpoint": "best-epoch=18-val_wer=0.0402.ckpt",
+ "best_val_wer": 0.04022550582885742,
+ "averaged": {
+  "checkpoints": [
+   "w-epoch=15.ckpt",
+   "w-epoch=16.ckpt",
+   "w-epoch=17.ckpt",
+   "w-epoch=18.ckpt",
+   "w-epoch=19.ckpt"
+  ],
+  "val_wer": 0.04144446179270744,
+  "kept": false
+ },
+ "train_minutes": 74.1,
+ "model": "runs/r25-general/model.nemo"
+}

training/r30-general/config.json

@@ -0,0 +1,65 @@
+{
+ "run": "r30-general",
+ "base_model": "nvidia/parakeet-tdt-0.6b-v2",
+ "init_from": "runs/r25-general/model.nemo",
+ "parent": "r25-general",
+ "rav_sha": "306a7450433d99c5892ac62acdb54f54f94a717c",
+ "data": "general",
+ "english_share": null,
+ "replay_share": 0.1,
+ "speed_perturb": {
+  "prob": 1.0,
+  "sr": 16000,
+  "resample_type": "kaiser_fast",
+  "min_speed_rate": 0.9,
+  "max_speed_rate": 1.1,
+  "num_rates": 3
+ },
+ "avg_last": 0,
+ "max_rows": null,
+ "epochs": 3,
+ "lr": 1e-05,
+ "batch_size": 16,
+ "seed": 7,
+ "mix": {
+  "blend": "general",
+  "cs_min": 858.8,
+  "own_en_min": 0.0,
+  "replay_min": 95.6,
+  "total_min": 1049.8,
+  "n": 11745,
+  "priority_rows": 2776,
+  "priority_extra_min": 95.4,
+  "priority_repeat": 3,
+  "primary_min": 858.8,
+  "primary_cs_min": 0.0,
+  "replay_share_actual": 0.091,
+  "delta": {
+   "parent": "r25-general",
+   "new_rows": 3199,
+   "new_unique": 2619,
+   "new_min": 210.0,
+   "new_cs_min": 0.0,
+   "old_rows": 7937,
+   "old_min": 789.6,
+   "full_mix_rows": 11745,
+   "rows": 11136,
+   "total_min": 999.7
+  }
+ },
+ "targets": {
+  "train": {
+   "plain": 11127,
+   "full": 0,
+   "fallback": 9,
+   "dropped": 0
+  },
+  "val": {
+   "plain": 653,
+   "full": 0,
+   "fallback": 1,
+   "dropped": 0
+  },
+  "charset": "parakeet-tdt-0.6b-v2"
+ }
+}

training/r30-general/result.json

@@ -0,0 +1,70 @@
+{
+ "run": "r30-general",
+ "base_model": "nvidia/parakeet-tdt-0.6b-v2",
+ "init_from": "runs/r25-general/model.nemo",
+ "parent": "r25-general",
+ "rav_sha": "306a7450433d99c5892ac62acdb54f54f94a717c",
+ "data": "general",
+ "english_share": null,
+ "replay_share": 0.1,
+ "speed_perturb": {
+  "prob": 1.0,
+  "sr": 16000,
+  "resample_type": "kaiser_fast",
+  "min_speed_rate": 0.9,
+  "max_speed_rate": 1.1,
+  "num_rates": 3
+ },
+ "avg_last": 0,
+ "max_rows": null,
+ "epochs": 3,
+ "lr": 1e-05,
+ "batch_size": 16,
+ "seed": 7,
+ "mix": {
+  "blend": "general",
+  "cs_min": 858.8,
+  "own_en_min": 0.0,
+  "replay_min": 95.6,
+  "total_min": 1049.8,
+  "n": 11745,
+  "priority_rows": 2776,
+  "priority_extra_min": 95.4,
+  "priority_repeat": 3,
+  "primary_min": 858.8,
+  "primary_cs_min": 0.0,
+  "replay_share_actual": 0.091,
+  "delta": {
+   "parent": "r25-general",
+   "new_rows": 3199,
+   "new_unique": 2619,
+   "new_min": 210.0,
+   "new_cs_min": 0.0,
+   "old_rows": 7937,
+   "old_min": 789.6,
+   "full_mix_rows": 11745,
+   "rows": 11136,
+   "total_min": 999.7
+  }
+ },
+ "targets": {
+  "train": {
+   "plain": 11127,
+   "full": 0,
+   "fallback": 9,
+   "dropped": 0
+  },
+  "val": {
+   "plain": 653,
+   "full": 0,
+   "fallback": 1,
+   "dropped": 0
+  },
+  "charset": "parakeet-tdt-0.6b-v2"
+ },
+ "best_checkpoint": "best-epoch=1-val_wer=0.0463.ckpt",
+ "best_val_wer": 0.04633016511797905,
+ "averaged": null,
+ "train_minutes": 14.4,
+ "model": "runs/r30-general/model.nemo"
+}

training/r35-general/config.json

@@ -0,0 +1,65 @@
+{
+ "run": "r35-general",
+ "base_model": "nvidia/parakeet-tdt-0.6b-v2",
+ "init_from": "runs/r30-general/model.nemo",
+ "parent": "r30-general",
+ "rav_sha": "39f21d71410306cd9d7a5b2b35b668f1284d5b6f",
+ "data": "general",
+ "english_share": null,
+ "replay_share": 0.1,
+ "speed_perturb": {
+  "prob": 1.0,
+  "sr": 16000,
+  "resample_type": "kaiser_fast",
+  "min_speed_rate": 0.9,
+  "max_speed_rate": 1.1,
+  "num_rates": 3
+ },
+ "avg_last": 0,
+ "max_rows": null,
+ "epochs": 3,
+ "lr": 1e-05,
+ "batch_size": 16,
+ "seed": 7,
+ "mix": {
+  "blend": "general",
+  "cs_min": 1033.0,
+  "own_en_min": 0.0,
+  "replay_min": 114.8,
+  "total_min": 1262.6,
+  "n": 13781,
+  "priority_rows": 3042,
+  "priority_extra_min": 114.8,
+  "priority_repeat": 3,
+  "primary_min": 1033.0,
+  "primary_cs_min": 0.0,
+  "replay_share_actual": 0.091,
+  "delta": {
+   "parent": "r30-general",
+   "new_rows": 1838,
+   "new_unique": 1722,
+   "new_min": 184.7,
+   "new_cs_min": 0.0,
+   "old_rows": 5514,
+   "old_min": 500.3,
+   "full_mix_rows": 13781,
+   "rows": 7352,
+   "total_min": 685.0
+  }
+ },
+ "targets": {
+  "train": {
+   "plain": 7347,
+   "full": 0,
+   "fallback": 5,
+   "dropped": 0
+  },
+  "val": {
+   "plain": 765,
+   "full": 0,
+   "fallback": 1,
+   "dropped": 0
+  },
+  "charset": "parakeet-tdt-0.6b-v2"
+ }
+}

training/r35-general/result.json

@@ -0,0 +1,70 @@
+{
+ "run": "r35-general",
+ "base_model": "nvidia/parakeet-tdt-0.6b-v2",
+ "init_from": "runs/r30-general/model.nemo",
+ "parent": "r30-general",
+ "rav_sha": "39f21d71410306cd9d7a5b2b35b668f1284d5b6f",
+ "data": "general",
+ "english_share": null,
+ "replay_share": 0.1,
+ "speed_perturb": {
+  "prob": 1.0,
+  "sr": 16000,
+  "resample_type": "kaiser_fast",
+  "min_speed_rate": 0.9,
+  "max_speed_rate": 1.1,
+  "num_rates": 3
+ },
+ "avg_last": 0,
+ "max_rows": null,
+ "epochs": 3,
+ "lr": 1e-05,
+ "batch_size": 16,
+ "seed": 7,
+ "mix": {
+  "blend": "general",
+  "cs_min": 1033.0,
+  "own_en_min": 0.0,
+  "replay_min": 114.8,
+  "total_min": 1262.6,
+  "n": 13781,
+  "priority_rows": 3042,
+  "priority_extra_min": 114.8,
+  "priority_repeat": 3,
+  "primary_min": 1033.0,
+  "primary_cs_min": 0.0,
+  "replay_share_actual": 0.091,
+  "delta": {
+   "parent": "r30-general",
+   "new_rows": 1838,
+   "new_unique": 1722,
+   "new_min": 184.7,
+   "new_cs_min": 0.0,
+   "old_rows": 5514,
+   "old_min": 500.3,
+   "full_mix_rows": 13781,
+   "rows": 7352,
+   "total_min": 685.0
+  }
+ },
+ "targets": {
+  "train": {
+   "plain": 7347,
+   "full": 0,
+   "fallback": 5,
+   "dropped": 0
+  },
+  "val": {
+   "plain": 765,
+   "full": 0,
+   "fallback": 1,
+   "dropped": 0
+  },
+  "charset": "parakeet-tdt-0.6b-v2"
+ },
+ "best_checkpoint": "best-epoch=1-val_wer=0.0534.ckpt",
+ "best_val_wer": 0.05344599485397339,
+ "averaged": null,
+ "train_minutes": 10.9,
+ "model": "runs/r35-general/model.nemo"
+}