Edit model card

output2

This model is a fine-tuned version of facebook/wav2vec2-base on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.7711
  • Wer: 0.3693

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: 8
  • eval_batch_size: 8
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 1000
  • num_epochs: 30
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
0.9858 0.5 500 0.8322 0.6842
0.7867 1.0 1000 0.6777 0.6137
0.6252 1.5 1500 0.6082 0.5503
0.5833 2.0 2000 0.5441 0.5066
0.4611 2.5 2500 0.5498 0.4922
0.4511 3.0 3000 0.5262 0.4654
0.37 3.5 3500 0.5422 0.4554
0.375 4.0 4000 0.6414 0.4659
0.3149 4.5 4500 0.5149 0.4353
0.3103 5.0 5000 0.5329 0.4423
0.2735 5.5 5500 0.9014 0.4359
0.2711 6.0 6000 3.1838 0.4374
0.26 6.5 6500 0.5987 0.4288
0.2451 7.0 7000 0.5245 0.4206
0.2184 7.5 7500 0.5627 0.4138
0.2115 8.0 8000 0.6408 0.4245
0.187 8.5 8500 0.5788 0.4093
0.1955 9.0 9000 0.5591 0.4214
0.1725 9.5 9500 0.5812 0.4135
0.1758 10.0 10000 0.5863 0.4051
0.1592 10.5 10500 0.6263 0.4116
0.1576 11.0 11000 0.5829 0.4028
0.1427 11.5 11500 0.6378 0.4016
0.1476 12.0 12000 0.5706 0.3988
0.1289 12.5 12500 0.6381 0.4104
0.1366 13.0 13000 0.6326 0.3975
0.1183 13.5 13500 0.6256 0.3916
0.1225 14.0 14000 0.6376 0.3971
0.1083 14.5 14500 0.6493 0.3905
0.1134 15.0 15000 0.6686 0.3951
0.1003 15.5 15500 0.6983 0.3967
0.104 16.0 16000 0.6324 0.3927
0.0928 16.5 16500 0.6482 0.3907
0.0944 17.0 17000 0.6790 0.3912
0.0925 17.5 17500 0.6877 0.3902
0.0847 18.0 18000 0.6572 0.3845
0.0808 18.5 18500 0.6551 0.3910
0.0836 19.0 19000 0.6832 0.3859
0.0757 19.5 19500 0.7594 0.3905
0.0751 20.0 20000 0.6960 0.3880
0.0715 20.5 20500 0.7244 0.3840
0.07 21.0 21000 0.7233 0.3848
0.0654 21.5 21500 0.7428 0.3833
0.0657 22.0 22000 0.7014 0.3842
0.0641 22.5 22500 0.7357 0.3796
0.0624 23.0 23000 0.7338 0.3796
0.0575 23.5 23500 0.7375 0.3804
0.0578 24.0 24000 0.7386 0.3782
0.0542 24.5 24500 0.7405 0.3758
0.0509 25.0 25000 0.7719 0.3774
0.0495 25.5 25500 0.7505 0.3763
0.0521 26.0 26000 0.7345 0.3742
0.0477 26.5 26500 0.7776 0.3740
0.0442 27.0 27000 0.7742 0.3738
0.0473 27.5 27500 0.7695 0.3719
0.0452 28.0 28000 0.7737 0.3705
0.0425 28.5 28500 0.7937 0.3702
0.0415 29.0 29000 0.7970 0.3713
0.0432 29.5 29500 0.7714 0.3700
0.041 30.0 30000 0.7711 0.3693

Framework versions

  • Transformers 4.35.2
  • Pytorch 2.1.0+cu118
  • Datasets 2.15.0
  • Tokenizers 0.15.0
Downloads last month
6
Safetensors
Model size
94.4M params
Tensor type
F32
·
Inference API
or
This model can be loaded on Inference API (serverless).

Finetuned from