First checkpoint version

.gitattributes

@@ -32,3 +32,7 @@ saved_model/**/* 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
+checkpoints/avg_top5_25-29.ckpt.data-00000-of-00001 filter=lfs diff=lfs merge=lfs -text
+checkpoints/avg_top5_25-29.ckpt.index filter=lfs diff=lfs merge=lfs -text
+checkpoints/avg_top5_27-32.ckpt.data-00000-of-00001 filter=lfs diff=lfs merge=lfs -text
+checkpoints/avg_top5_27-32.ckpt.index filter=lfs diff=lfs merge=lfs -text

README.md

@@ -1,3 +1,11 @@
 ---
 license: mit
 ---
+
+### How to clone this repo
+```
+sudo apt-get install git-lfs
+git clone https://huggingface.co/thanhtvt/uetasr-conformer_30.3m
+cd uetasr-conformer_30.3m
+git lfs pull
+```

checkpoints/avg_top5_25-29.ckpt.data-00000-of-00001

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:49fc029829a19daaa12a9a0803a953b65151b242201c4fafa8935898079efb75
+size 246353751

checkpoints/avg_top5_25-29.ckpt.index

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5e767271b8c9e5d7254a5da8e23c659adf3fa080909076d0dedd0eb61e8e28be
+size 105822

checkpoints/avg_top5_27-32.ckpt.data-00000-of-00001

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5806d3e06fac6c2fde5a70f5c5d29bf31dcf81557ef744fd67339804ef736050
+size 246353751

checkpoints/avg_top5_27-32.ckpt.index

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:620561277420e8326531c0036fe013b9f57aa6e577fb7390236679ae15c70ca7
+size 105822

config.yaml

@@ -0,0 +1,79 @@
+root_dir: ..
+ckpt_path: !ref <root_dir>/checkpoints/avg_top5_27-32.ckpt
+
+text_encoder: !new:uetasr.featurizers.text.Subword
+    model_prefix: vocabs/subword_vietnamese_500_v3
+    data_path: data/transcript_v3.1.txt
+    character_coverage: 1.0
+    model_type: bpe # word bpe unigram char
+    num_threads: 16
+    unk_id: 1
+    pad_id: 0
+    eos_id: -1
+    unk_piece: <unk>
+    pad_piece: <blank>
+    eos_piece: </s>
+    vocab_size: 500
+
+audio_encoder: !new:uetasr.featurizers.audio.LogMelSpectrogram
+    fs: 16000
+    n_fft: 512
+    win_length: 400
+    hop_length: 160
+    n_mels: 80
+    fmin: 0
+    fmax: 8000
+    htk: False
+
+d_model: 256
+
+encoder_model: !new:uetasr.models.encoders.Conformer
+    num_features: 80
+    window_size: 1
+    d_model: !ref <d_model>
+    input_layer: vgg2l
+    pos_enc_layer_type: rel_pos
+    dropout_rate_pos_enc: 0.2
+    selfattention_layer_type: rel_selfattn
+    attention_heads: 4
+    dropout_rate_att: 0.1
+    dropout_rate_pos_wise: 0.1
+    dropout_rate: 0.1
+    positionwise_layer_type: linear
+    linear_units: 1024
+    conv_mod_kernel: 31
+    num_blocks: 18
+    use_macaron: True
+    use_cnn_module: True
+    eps_layer_norm: 0.000000000001
+
+decoder_model: !new:uetasr.models.decoders.RNNDecoder
+    vocab_size: !ref <text_encoder.vocab_size>
+    embedding_dim: 256
+    num_layers: 1
+    hidden_dim: !ref <d_model>
+    dropout_embed: 0.2
+    dropout_rnn: 0.1
+    rnn_type: LSTM
+
+jointer_model: !new:uetasr.layers.jointer.RNNTJointer
+    encoder_dim: !ref <d_model>
+    decoder_dim: !ref <d_model>
+    hidden_dim: 512
+    output_dim: !ref <text_encoder.vocab_size>
+
+ctc_lin: null
+
+model: !new:uetasr.models.rnnt.RNNT
+    encoder: !ref <encoder_model>
+    decoder: !ref <decoder_model>
+    jointer: !ref <jointer_model>
+    ctc_lin: !ref <ctc_lin>
+    ctc_dropout: 0.1
+    use_cmvn: True
+
+decoder: !new:uetasr.searchers.GreedyRNNTV2
+    decoder: !ref <decoder_model>
+    jointer: !ref <jointer_model>
+    text_decoder: !ref <text_encoder>
+    max_symbols_per_step: 10

vocabs/subword_vietnamese_500_v3.vocab

@@ -0,0 +1,500 @@
+<blank>	0
+<unk>	0
+▁t	-0
+ng	-1
+▁c	-2
+▁đ	-3
+▁n	-4
+▁l	-5
+▁th	-6
+▁v	-7
+▁m	-8
+▁h	-9
+▁ch	-10
+▁b	-11
+nh	-12
+▁k	-13
+▁nh	-14
+▁tr	-15
+▁s	-16
+▁g	-17
+▁kh	-18
+▁ng	-19
+▁là	-20
+▁p	-21
+▁cá	-22
+▁ph	-23
+▁d	-24
+ông	-25
+▁gi	-26
+iệ	-27
+▁r	-28
+▁q	-29
+▁qu	-30
+ên	-31
+▁và	-32
+ườ	-33
+▁có	-34
+▁x	-35
+ột	-36
+▁củ	-37
+▁của	-38
+iế	-39
+▁cái	-40
+▁một	-41
+▁các	-42
+ới	-43
+▁không	-44
+ượ	-45
+ình	-46
+ất	-47
+▁nà	-48
+ai	-49
+ại	-50
+ười	-51
+ân	-52
+uy	-53
+ong	-54
+▁thì	-55
+ươ	-56
+ôi	-57
+ược	-58
+▁như	-59
+ăm	-60
+iề	-61
+▁ngh	-62
+ch	-63
+▁nó	-64
+▁được	-65
+anh	-66
+▁cho	-67
+ây	-68
+▁người	-69
+úng	-70
+▁này	-71
+▁ta	-72
+▁trong	-73
+an	-74
+ấy	-75
+ải	-76
+ướ	-77
+▁mà	-78
+ạn	-79
+iể	-80
+ững	-81
+ần	-82
+▁đó	-83
+em	-84
+▁với	-85
+▁những	-86
+iên	-87
+inh	-88
+▁hai	-89
+ay	-90
+iều	-91
+ao	-92
+▁đã	-93
+▁tôi	-94
+ước	-95
+ương	-96
+▁việ	-97
+▁chúng	-98
+▁chí	-99
+▁họ	-100
+▁em	-101
+▁ở	-102
+ành	-103
+▁sẽ	-104
+▁để	-105
+ến	-106
+ản	-107
+ường	-108
+ức	-109
+ác	-110
+iện	-111
+ũng	-112
+▁bạn	-113
+ang	-114
+▁ho	-115
+ài	-116
+ực	-117
+▁độ	-118
+▁cũng	-119
+▁ra	-120
+ời	-121
+▁về	-122
+▁phải	-123
+ết	-124
+▁năm	-125
+ầu	-126
+▁đến	-127
+▁thể	-128
+ống	-129
+ung	-130
+au	-131
+am	-132
+▁cả	-133
+ận	-134
+âu	-135
+ơn	-136
+▁làm	-137
+ối	-138
+▁từ	-139
+ồi	-140
+▁mình	-141
+àn	-142
+iết	-143
+ật	-144
+▁anh	-145
+▁công	-146
+àng	-147
+▁lại	-148
+ùng	-149
+▁khi	-150
+▁thế	-151
+át	-152
+ốn	-153
+▁đi	-154
+ốc	-155
+ồng	-156
+on	-157
+iến	-158
+án	-159
+ằng	-160
+▁thu	-161
+▁đây	-162
+▁gì	-163
+ăn	-164
+in	-165
+ội	-166
+▁học	-167
+▁vào	-168
+ưa	-169
+▁nhiều	-170
+áng	-171
+▁ba	-172
+▁đị	-173
+òn	-174
+▁giá	-175
+êu	-176
+ươi	-177
+▁nhà	-178
+eo	-179
+▁nước	-180
+ăng	-181
+▁rất	-182
+▁rồi	-183
+▁số	-184
+▁nào	-185
+▁quy	-186
+▁mươi	-187
+▁chính	-188
+▁nhưng	-189
+ập	-190
+▁nói	-191
+▁còn	-192
+ày	-193
+úc	-194
+▁con	-195
+▁sự	-196
+▁thành	-197
+ộc	-198
+▁trên	-199
+▁đầu	-200
+▁bị	-201
+iểm	-202
+▁tiế	-203
+ục	-204
+▁chỉ	-205
+ôn	-206
+▁trăm	-207
+ậy	-208
+▁động	-209
+▁nam	-210
+▁việc	-211
+▁tại	-212
+▁bả	-213
+ích	-214
+▁ông	-215
+ữa	-216
+▁quan	-217
+▁hiện	-218
+ền	-219
+▁việt	-220
+▁sau	-221
+▁theo	-222
+▁mười	-223
+▁lu	-224
+▁do	-225
+▁ngày	-226
+▁dân	-227
+▁nhân	-228
+▁đề	-229
+ắc	-230
+ái	-231
+▁y	-232
+▁vậy	-233
+▁mu	-234
+iệu	-235
+▁quốc	-236
+▁giờ	-237
+ách	-238
+▁cu	-239
+ổi	-240
+▁đồng	-241
+▁cô	-242
+▁hơn	-243
+▁thấy	-244
+▁định	-245
+▁to	-246
+ấn	-247
+▁xu	-248
+òng	-249
+ơi	-250
+▁vì	-251
+▁chuy	-252
+âm	-253
+ẫn	-254
+ặc	-255
+▁tư	-256
+▁trường	-257
+▁nhất	-258
+▁phần	-259
+ắt	-260
+▁kho	-261
+áp	-262
+iền	-263
+ượng	-264
+▁hàng	-265
+▁trung	-266
+▁biết	-267
+▁dụ	-268
+iểu	-269
+▁ạ	-270
+▁mới	-271
+▁thực	-272
+ìn	-273
+ếu	-274
+▁sản	-275
+▁thời	-276
+ạt	-277
+ạo	-278
+ặt	-279
+ánh	-280
+▁điều	-281
+iển	-282
+▁đấy	-283
+▁phát	-284
+▁hội	-285
+▁sinh	-286
+▁đang	-287
+▁nghiệ	-288
+▁lý	-289
+▁ý	-290
+▁thứ	-291
+ịch	-292
+▁hình	-293
+ấp	-294
+▁nguy	-295
+▁nên	-296
+▁bộ	-297
+▁gia	-298
+▁cùng	-299
+▁cách	-300
+▁nghĩ	-301
+▁cơ	-302
+▁hợ	-303
+▁trình	-304
+▁hợp	-305
+ọi	-306
+▁chủ	-307
+ảng	-308
+ám	-309
+▁hay	-310
+ễn	-311
+ốt	-312
+▁thông	-313
+ứng	-314
+▁đúng	-315
+▁tiếp	-316
+áu	-317
+▁trước	-318
+▁tổ	-319
+▁ngo	-320
+▁khác	-321
+▁vị	-322
+▁tế	-323
+ỏi	-324
+▁lo	-325
+▁kinh	-326
+▁dự	-327
+ính	-328
+▁lên	-329
+▁điểm	-330
+▁chín	-331
+▁nay	-332
+▁tháng	-333
+▁lớ	-334
+▁viên	-335
+▁chi	-336
+ởng	-337
+▁qua	-338
+▁bằng	-339
+áo	-340
+▁đối	-341
+▁tuy	-342
+êm	-343
+iêu	-344
+▁rằng	-345
+ện	-346
+▁bốn	-347
+▁nhận	-348
+▁tự	-349
+▁tu	-350
+ọng	-351
+▁mạ	-352
+ấu	-353
+▁nếu	-354
+▁tin	-355
+▁à	-356
+▁tới	-357
+iệt	-358
+▁nghiệp	-359
+▁muốn	-360
+▁cuộc	-361
+▁chưa	-362
+▁cứ	-363
+▁thị	-364
+▁vụ	-365
+▁a	-366
+▁cao	-367
+▁sao	-368
+ừa	-369
+▁hành	-370
+▁câu	-371
+ào	-372
+▁tài	-373
+▁dụng	-374
+▁bản	-375
+ầy	-376
+▁lúc	-377
+▁	-378
+n	-379
+h	-380
+t	-381
+c	-382
+i	-383
+g	-384
+a	-385
+m	-386
+à	-387
+đ	-388
+u	-389
+l	-390
+o	-391
+á	-392
+v	-393
+ư	-394
+r	-395
+y	-396
+b	-397
+p	-398
+k	-399
+ô	-400
+s	-401
+ó	-402
+ế	-403
+ạ	-404
+ì	-405
+ộ	-406
+ả	-407
+ờ	-408
+d	-409
+ê	-410
+ấ	-411
+â	-412
+ệ	-413
+ố	-414
+ớ	-415
+ề	-416
+ơ	-417
+ể	-418
+ủ	-419
+q	-420
+ợ	-421
+e	-422
+ậ	-423
+í	-424
+ă	-425
+ị	-426
+ú	-427
+x	-428
+ầ	-429
+ứ	-430
+ữ	-431
+ự	-432
+ọ	-433
+ở	-434
+ồ	-435
+ã	-436
+ụ	-437
+ắ	-438
+ừ	-439
+ò	-440
+ổ	-441
+ũ	-442
+ù	-443
+ẽ	-444
+ặ	-445
+ý	-446
+ỉ	-447
+ỏ	-448
+ằ	-449
+ử	-450
+é	-451
+ĩ	-452
+ẩ	-453
+ễ	-454
+ẫ	-455
+ỗ	-456
+ẹ	-457
+ẻ	-458
+ỹ	-459
+è	-460
+ỳ	-461
+õ	-462
+ẳ	-463
+ỡ	-464
+ỷ	-465
+f	-466
+w	-467
+ẵ	-468
+z	-469
+j	-470
+-	-471
+ỵ	-472
+2	-473
+1	-474
+'	-475
+3	-476
+0	-477
+ε	-478
+λ	-479
+4	-480
+9	-481
+P	-482
+"	-483
+6	-484
+7	-485
+H	-486
+5	-487
+N	-488
+O	-489
+[	-490
+ń	-491
+ǀ	-492
+ǹ	-493
+α	-494
+β	-495
+γ	-496
+δ	-497