wav2vec2-common_voice-tr-demo

This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the MOZILLA-FOUNDATION/COMMON_VOICE_13_0 - TR dataset. It achieves the following results on the evaluation set:

  • Loss: 0.3097
  • Wer: 0.4725

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: 0.0003
  • train_batch_size: 16
  • eval_batch_size: 8
  • seed: 42
  • gradient_accumulation_steps: 2
  • total_train_batch_size: 32
  • optimizer: Use 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: 500
  • num_epochs: 15.0
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
No log 0.0763 100 3.9145 1.0
No log 0.1526 200 3.1491 1.0
No log 0.2289 300 2.4589 1.0151
No log 0.3052 400 0.8488 0.9319
4.6713 0.3815 500 0.7336 0.9263
4.6713 0.4578 600 0.6043 0.8163
4.6713 0.5341 700 0.5496 0.7753
4.6713 0.6105 800 0.5273 0.7605
4.6713 0.6868 900 0.5153 0.7516
0.509 0.7631 1000 0.4648 0.7192
0.509 0.8394 1100 0.4232 0.6961
0.509 0.9157 1200 0.4227 0.6983
0.509 0.9920 1300 0.4336 0.6945
0.509 1.0679 1400 0.4217 0.6709
0.37 1.1442 1500 0.4050 0.6767
0.37 1.2205 1600 0.4036 0.6879
0.37 1.2968 1700 0.3946 0.6651
0.37 1.3731 1800 0.3887 0.6583
0.37 1.4494 1900 0.3976 0.6819
0.3179 1.5258 2000 0.3707 0.6320
0.3179 1.6021 2100 0.3758 0.6387
0.3179 1.6784 2200 0.3636 0.6324
0.3179 1.7547 2300 0.3880 0.6593
0.3179 1.8310 2400 0.3533 0.6304
0.3065 1.9073 2500 0.3608 0.6312
0.3065 1.9836 2600 0.3474 0.6154
0.3065 2.0595 2700 0.3638 0.6165
0.3065 2.1358 2800 0.3606 0.6085
0.3065 2.2121 2900 0.3374 0.6053
0.2597 2.2884 3000 0.3315 0.6051
0.2597 2.3647 3100 0.3499 0.6164
0.2597 2.4411 3200 0.3410 0.6064
0.2597 2.5174 3300 0.3465 0.6099
0.2597 2.5937 3400 0.3382 0.6009
0.2536 2.6700 3500 0.3367 0.6025
0.2536 2.7463 3600 0.3397 0.6056
0.2536 2.8226 3700 0.3341 0.5914
0.2536 2.8989 3800 0.3267 0.5887
0.2536 2.9752 3900 0.3312 0.5836
0.2372 3.0511 4000 0.3317 0.5821
0.2372 3.1274 4100 0.3258 0.5747
0.2372 3.2037 4200 0.3350 0.5808
0.2372 3.2800 4300 0.3392 0.5798
0.2372 3.3564 4400 0.3209 0.5779
0.2128 3.4327 4500 0.3227 0.5719
0.2128 3.5090 4600 0.3217 0.5740
0.2128 3.5853 4700 0.3210 0.5713
0.2128 3.6616 4800 0.3093 0.5662
0.2128 3.7379 4900 0.3242 0.5701
0.2114 3.8142 5000 0.3142 0.5646
0.2114 3.8905 5100 0.3135 0.5680
0.2114 3.9668 5200 0.3219 0.5646
0.2114 4.0427 5300 0.3347 0.5610
0.2114 4.1190 5400 0.3303 0.5682
0.1987 4.1953 5500 0.3368 0.5690
0.1987 4.2717 5600 0.3214 0.5629
0.1987 4.3480 5700 0.3072 0.5621
0.1987 4.4243 5800 0.3213 0.5606
0.1987 4.5006 5900 0.3184 0.5711
0.1889 4.5769 6000 0.3135 0.5601
0.1889 4.6532 6100 0.3201 0.5596
0.1889 4.7295 6200 0.3049 0.5567
0.1889 4.8058 6300 0.3081 0.5545
0.1889 4.8821 6400 0.3112 0.5569
0.183 4.9584 6500 0.3060 0.5504
0.183 5.0343 6600 0.3161 0.5569
0.183 5.1106 6700 0.3288 0.5550
0.183 5.1870 6800 0.3199 0.5502
0.183 5.2633 6900 0.3248 0.5558
0.1648 5.3396 7000 0.3216 0.5566
0.1648 5.4159 7100 0.3026 0.5543
0.1648 5.4922 7200 0.3135 0.5611
0.1648 5.5685 7300 0.3137 0.5582
0.1648 5.6448 7400 0.3179 0.5566
0.161 5.7211 7500 0.3256 0.5522
0.161 5.7974 7600 0.3109 0.5553
0.161 5.8737 7700 0.3047 0.5430
0.161 5.9500 7800 0.3128 0.5436
0.161 6.0259 7900 0.3150 0.5427
0.1564 6.1023 8000 0.3025 0.5405
0.1564 6.1786 8100 0.3155 0.5533
0.1564 6.2549 8200 0.3101 0.5377
0.1564 6.3312 8300 0.3302 0.5407
0.1564 6.4075 8400 0.3180 0.5361
0.1454 6.4838 8500 0.3060 0.5397
0.1454 6.5601 8600 0.3084 0.5347
0.1454 6.6364 8700 0.3218 0.5443
0.1454 6.7127 8800 0.3086 0.5433
0.1454 6.7890 8900 0.3248 0.5337
0.1425 6.8653 9000 0.2960 0.5370
0.1425 6.9416 9100 0.2962 0.5367
0.1425 7.0176 9200 0.3222 0.5313
0.1425 7.0939 9300 0.3115 0.5329
0.1425 7.1702 9400 0.3048 0.5261
0.1352 7.2465 9500 0.3170 0.5269
0.1352 7.3228 9600 0.2940 0.5186
0.1352 7.3991 9700 0.3054 0.5236
0.1352 7.4754 9800 0.3175 0.5262
0.1352 7.5517 9900 0.2949 0.5267
0.1297 7.6280 10000 0.3060 0.5255
0.1297 7.7043 10100 0.2849 0.5254
0.1297 7.7806 10200 0.2876 0.5240
0.1297 7.8569 10300 0.2971 0.5172
0.1297 7.9332 10400 0.3026 0.5179
0.1277 8.0092 10500 0.2984 0.5161
0.1277 8.0855 10600 0.3032 0.5241
0.1277 8.1618 10700 0.3111 0.5218
0.1277 8.2381 10800 0.2936 0.5216
0.1277 8.3144 10900 0.2961 0.5129
0.1168 8.3907 11000 0.3053 0.5126
0.1168 8.4670 11100 0.2934 0.5134
0.1168 8.5433 11200 0.3029 0.5145
0.1168 8.6196 11300 0.3097 0.5153
0.1168 8.6959 11400 0.2979 0.5128
0.1149 8.7722 11500 0.2925 0.5146
0.1149 8.8485 11600 0.3051 0.5169
0.1149 8.9248 11700 0.3001 0.5126
0.1149 9.0008 11800 0.2969 0.5063
0.1149 9.0771 11900 0.3101 0.5083
0.1076 9.1534 12000 0.3153 0.5171
0.1076 9.2297 12100 0.3266 0.5200
0.1076 9.3060 12200 0.3139 0.5173
0.1076 9.3823 12300 0.3133 0.5172
0.1076 9.4586 12400 0.3073 0.5225
0.0988 9.5349 12500 0.3108 0.5160
0.0988 9.6112 12600 0.3038 0.5265
0.0988 9.6875 12700 0.2941 0.5110
0.0988 9.7638 12800 0.2985 0.5013
0.0988 9.8401 12900 0.2984 0.5099
0.103 9.9164 13000 0.3067 0.5174
0.103 9.9928 13100 0.2915 0.5039
0.103 10.0687 13200 0.2970 0.5034
0.103 10.1450 13300 0.3185 0.5169
0.103 10.2213 13400 0.3166 0.5052
0.0925 10.2976 13500 0.3081 0.5068
0.0925 10.3739 13600 0.3048 0.5103
0.0925 10.4502 13700 0.3028 0.5000
0.0925 10.5265 13800 0.3043 0.5043
0.0925 10.6028 13900 0.3030 0.4949
0.0909 10.6791 14000 0.2920 0.4975
0.0909 10.7554 14100 0.3046 0.4992
0.0909 10.8317 14200 0.3047 0.5025
0.0909 10.9081 14300 0.2946 0.4972
0.0909 10.9844 14400 0.2917 0.4996
0.0884 11.0603 14500 0.3081 0.5003
0.0884 11.1366 14600 0.3083 0.4955
0.0884 11.2129 14700 0.3057 0.4937
0.0884 11.2892 14800 0.3083 0.4987
0.0884 11.3655 14900 0.3074 0.4962
0.0798 11.4418 15000 0.2997 0.4956
0.0798 11.5181 15100 0.2995 0.4990
0.0798 11.5944 15200 0.2942 0.4926
0.0798 11.6707 15300 0.2961 0.4880
0.0798 11.7470 15400 0.3124 0.4902
0.0803 11.8233 15500 0.2986 0.4894
0.0803 11.8997 15600 0.2985 0.4902
0.0803 11.9760 15700 0.3027 0.4963
0.0803 12.0519 15800 0.3083 0.4878
0.0803 12.1282 15900 0.2997 0.4869
0.0781 12.2045 16000 0.3017 0.4911
0.0781 12.2808 16100 0.3071 0.4881
0.0781 12.3571 16200 0.3135 0.4893
0.0781 12.4334 16300 0.3038 0.4859
0.0781 12.5097 16400 0.3072 0.4845
0.0721 12.5860 16500 0.3146 0.4840
0.0721 12.6623 16600 0.3058 0.4839
0.0721 12.7386 16700 0.3020 0.4816
0.0721 12.8150 16800 0.3081 0.4811
0.0721 12.8913 16900 0.3005 0.4798
0.0701 12.9676 17000 0.3087 0.4836
0.0701 13.0435 17100 0.3191 0.4888
0.0701 13.1198 17200 0.3174 0.4839
0.0701 13.1961 17300 0.3073 0.4846
0.0701 13.2724 17400 0.3044 0.4812
0.0625 13.3487 17500 0.3117 0.4788
0.0625 13.4250 17600 0.3130 0.4782
0.0625 13.5013 17700 0.3029 0.4764
0.0625 13.5776 17800 0.3101 0.4812
0.0625 13.6539 17900 0.2972 0.4814
0.0627 13.7303 18000 0.3097 0.4764
0.0627 13.8066 18100 0.3035 0.4761
0.0627 13.8829 18200 0.3074 0.4775
0.0627 13.9592 18300 0.3059 0.4755
0.0627 14.0351 18400 0.3084 0.4755
0.0623 14.1114 18500 0.3097 0.4778
0.0623 14.1877 18600 0.3160 0.4770
0.0623 14.2640 18700 0.3148 0.4762
0.0623 14.3403 18800 0.3184 0.4737
0.0623 14.4166 18900 0.3177 0.4750
0.0554 14.4929 19000 0.3141 0.4744
0.0554 14.5692 19100 0.3114 0.4746
0.0554 14.6456 19200 0.3114 0.4736
0.0554 14.7219 19300 0.3094 0.4736
0.0554 14.7982 19400 0.3086 0.4727
0.0553 14.8745 19500 0.3094 0.4722
0.0553 14.9508 19600 0.3097 0.4720

Framework versions

  • Transformers 4.47.0.dev0
  • Pytorch 2.5.1+cu124
  • Datasets 3.1.0
  • Tokenizers 0.20.3
Downloads last month
46
Safetensors
Model size
315M params
Tensor type
F32
·
Inference Providers NEW
This model is not currently available via any of the supported Inference Providers.

Model tree for utakumi/wav2vec2-common_voice-tr-demo

Finetuned
(246)
this model

Evaluation results