Configuration Parsing Warning: In adapter_config.json: "peft.task_type" must be a string

Whisper Small ko

This model is a fine-tuned version of openai/whisper-large-v3-turbo on the custom dataset. It achieves the following results on the evaluation set:

  • Loss: 0.3680

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 0.0001
  • train_batch_size: 64
  • eval_batch_size: 256
  • seed: 42
  • optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 200
  • training_steps: 2000
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss
0.597 0.0153 10 1.5896
0.6039 0.0306 20 1.5751
0.5982 0.0459 30 1.5476
0.5547 0.0612 40 1.4984
0.5346 0.0765 50 1.4090
0.4456 0.0917 60 1.2126
0.3208 0.1070 70 0.9587
0.2289 0.1223 80 0.9004
0.1958 0.1376 90 0.8680
0.1832 0.1529 100 0.8425
0.1618 0.1682 110 0.8191
0.1649 0.1835 120 0.7813
0.1416 0.1988 130 0.7523
0.142 0.2141 140 0.7328
0.1386 0.2294 150 0.7280
0.1362 0.2446 160 0.7054
0.1293 0.2599 170 0.6869
0.1313 0.2752 180 0.6788
0.1356 0.2905 190 0.6736
0.1377 0.3058 200 0.6692
0.1332 0.3211 210 0.6521
0.1237 0.3364 220 0.6593
0.1153 0.3517 230 0.6551
0.1236 0.3670 240 0.6427
0.1084 0.3823 250 0.6343
0.1292 0.3976 260 0.6168
0.1182 0.4128 270 0.6030
0.1142 0.4281 280 0.6027
0.1198 0.4434 290 0.6147
0.1111 0.4587 300 0.6141
0.1014 0.4740 310 0.6062
0.1115 0.4893 320 0.5983
0.1082 0.5046 330 0.6016
0.1024 0.5199 340 0.5988
0.0968 0.5352 350 0.5907
0.1052 0.5505 360 0.5890
0.1038 0.5657 370 0.5807
0.1116 0.5810 380 0.5696
0.1005 0.5963 390 0.5662
0.1215 0.6116 400 0.5562
0.0969 0.6269 410 0.5628
0.106 0.6422 420 0.5664
0.0949 0.6575 430 0.5553
0.0952 0.6728 440 0.5636
0.1101 0.6881 450 0.5532
0.1125 0.7034 460 0.5539
0.1052 0.7187 470 0.5584
0.1041 0.7339 480 0.5624
0.1024 0.7492 490 0.5480
0.1006 0.7645 500 0.5385
0.0957 0.7798 510 0.5456
0.1042 0.7951 520 0.5415
0.1067 0.8104 530 0.5426
0.1047 0.8257 540 0.5346
0.1072 0.8410 550 0.5315
0.098 0.8563 560 0.5237
0.1091 0.8716 570 0.5197
0.096 0.8869 580 0.5240
0.0895 0.9021 590 0.5254
0.0975 0.9174 600 0.5307
0.0998 0.9327 610 0.5235
0.0972 0.9480 620 0.5197
0.0967 0.9633 630 0.5041
0.0968 0.9786 640 0.5032
0.0981 0.9939 650 0.4923
0.083 1.0092 660 0.4843
0.0885 1.0245 670 0.4822
0.1085 1.0398 680 0.4807
0.0889 1.0550 690 0.4807
0.0897 1.0703 700 0.4757
0.0941 1.0856 710 0.4670
0.092 1.1009 720 0.4679
0.0901 1.1162 730 0.4637
0.1065 1.1315 740 0.4724
0.0807 1.1468 750 0.4709
0.0893 1.1621 760 0.4728
0.104 1.1774 770 0.4781
0.0847 1.1927 780 0.4764
0.0981 1.2080 790 0.4723
0.0871 1.2232 800 0.4650
0.0933 1.2385 810 0.4613
0.0855 1.2538 820 0.4582
0.0877 1.2691 830 0.4634
0.0887 1.2844 840 0.4703
0.091 1.2997 850 0.4668
0.0902 1.3150 860 0.4621
0.0876 1.3303 870 0.4485
0.0808 1.3456 880 0.4487
0.0811 1.3609 890 0.4614
0.0828 1.3761 900 0.4684
0.0895 1.3914 910 0.4599
0.0852 1.4067 920 0.4549
0.0927 1.4220 930 0.4458
0.0879 1.4373 940 0.4458
0.1015 1.4526 950 0.4390
0.0919 1.4679 960 0.4417
0.0817 1.4832 970 0.4380
0.0833 1.4985 980 0.4360
0.093 1.5138 990 0.4375
0.0897 1.5291 1000 0.4353
0.0818 1.5443 1010 0.4348
0.0891 1.5596 1020 0.4243
0.0942 1.5749 1030 0.4263
0.0852 1.5902 1040 0.4286
0.0815 1.6055 1050 0.4272
0.081 1.6208 1060 0.4189
0.0848 1.6361 1070 0.4159
0.0897 1.6514 1080 0.4139
0.0845 1.6667 1090 0.4101
0.0815 1.6820 1100 0.4091
0.0789 1.6972 1110 0.4083
0.0995 1.7125 1120 0.4109
0.0827 1.7278 1130 0.4105
0.085 1.7431 1140 0.4105
0.0905 1.7584 1150 0.4116
0.0844 1.7737 1160 0.4087
0.0898 1.7890 1170 0.4080
0.0789 1.8043 1180 0.4106
0.0864 1.8196 1190 0.4126
0.0715 1.8349 1200 0.4133
0.0934 1.8502 1210 0.4086
0.081 1.8654 1220 0.4057
0.0829 1.8807 1230 0.4040
0.0795 1.8960 1240 0.4050
0.082 1.9113 1250 0.4068
0.0831 1.9266 1260 0.4021
0.094 1.9419 1270 0.4007
0.0786 1.9572 1280 0.3993
0.0837 1.9725 1290 0.3973
0.0773 1.9878 1300 0.3987
0.0993 2.0031 1310 0.4047
0.0851 2.0183 1320 0.4009
0.0826 2.0336 1330 0.3985
0.0766 2.0489 1340 0.3947
0.0778 2.0642 1350 0.3930
0.0729 2.0795 1360 0.3951
0.0769 2.0948 1370 0.3961
0.0802 2.1101 1380 0.3949
0.0787 2.1254 1390 0.3959
0.0773 2.1407 1400 0.3940
0.0729 2.1560 1410 0.3945
0.0795 2.1713 1420 0.3962
0.083 2.1865 1430 0.3942
0.0796 2.2018 1440 0.3894
0.0715 2.2171 1450 0.3886
0.0771 2.2324 1460 0.3878
0.0846 2.2477 1470 0.3856
0.0723 2.2630 1480 0.3849
0.0931 2.2783 1490 0.3836
0.0898 2.2936 1500 0.3799
0.0748 2.3089 1510 0.3800
0.0852 2.3242 1520 0.3811
0.0701 2.3394 1530 0.3814
0.085 2.3547 1540 0.3814
0.076 2.3700 1550 0.3784
0.0758 2.3853 1560 0.3785
0.0886 2.4006 1570 0.3792
0.0733 2.4159 1580 0.3823
0.0809 2.4312 1590 0.3809
0.0788 2.4465 1600 0.3768
0.0761 2.4618 1610 0.3762
0.0757 2.4771 1620 0.3768
0.0774 2.4924 1630 0.3764
0.0721 2.5076 1640 0.3759
0.066 2.5229 1650 0.3763
0.075 2.5382 1660 0.3765
0.0833 2.5535 1670 0.3765
0.0779 2.5688 1680 0.3761
0.0802 2.5841 1690 0.3761
0.0772 2.5994 1700 0.3755
0.0739 2.6147 1710 0.3743
0.0773 2.6300 1720 0.3736
0.0793 2.6453 1730 0.3719
0.0775 2.6606 1740 0.3718
0.08 2.6758 1750 0.3724
0.0751 2.6911 1760 0.3731
0.0696 2.7064 1770 0.3726
0.0792 2.7217 1780 0.3706
0.069 2.7370 1790 0.3705
0.076 2.7523 1800 0.3713
0.0724 2.7676 1810 0.3719
0.0767 2.7829 1820 0.3719
0.079 2.7982 1830 0.3713
0.0761 2.8135 1840 0.3714
0.0722 2.8287 1850 0.3711
0.0904 2.8440 1860 0.3703
0.0808 2.8593 1870 0.3695
0.0718 2.8746 1880 0.3691
0.0789 2.8899 1890 0.3690
0.089 2.9052 1900 0.3687
0.0751 2.9205 1910 0.3689
0.0856 2.9358 1920 0.3690
0.0812 2.9511 1930 0.3689
0.077 2.9664 1940 0.3687
0.0764 2.9817 1950 0.3685
0.0747 2.9969 1960 0.3682
0.065 3.0122 1970 0.3682
0.0796 3.0275 1980 0.3681
0.0767 3.0428 1990 0.3681
0.0792 3.0581 2000 0.3680

Framework versions

  • PEFT 0.14.0
  • Transformers 4.47.1
  • Pytorch 2.5.1+cu124
  • Datasets 3.2.0
  • Tokenizers 0.21.0
Downloads last month
0
Inference Providers NEW
This model is not currently available via any of the supported third-party Inference Providers, and HF Inference API was unable to determine this model’s pipeline type.

Model tree for nomnoos37/stt-turbo-1227-v1.1-peft-eng-2k-rank64-full-train

Adapter
(47)
this model