Edit model card

w2v-bert-2.0-yogera-fleurs-cv-lg-50hrs-v2

This model is a fine-tuned version of facebook/w2v-bert-2.0 on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.6495
  • Wer: 0.2689
  • Cer: 0.0609

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: 5e-05
  • train_batch_size: 8
  • eval_batch_size: 4
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 100
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer Cer
0.4233 1.0 2164 0.3504 0.3618 0.0788
0.2409 2.0 4328 0.3031 0.3348 0.0701
0.1932 3.0 6492 0.3135 0.3000 0.0650
0.1644 4.0 8656 0.3037 0.3156 0.0682
0.1433 5.0 10820 0.3281 0.2822 0.0631
0.1265 6.0 12984 0.3193 0.3039 0.0655
0.1108 7.0 15148 0.3772 0.2879 0.0637
0.0978 8.0 17312 0.3621 0.2834 0.0629
0.0859 9.0 19476 0.3588 0.2789 0.0605
0.0745 10.0 21640 0.3374 0.2751 0.0609
0.065 11.0 23804 0.4067 0.2785 0.0610
0.0559 12.0 25968 0.3573 0.2781 0.0613
0.0465 13.0 28132 0.4279 0.2778 0.0615
0.0412 14.0 30296 0.4062 0.2776 0.0617
0.0363 15.0 32460 0.4455 0.2886 0.0639
0.0319 16.0 34624 0.4443 0.2765 0.0626
0.0292 17.0 36788 0.5188 0.2814 0.0633
0.0262 18.0 38952 0.4416 0.2722 0.0599
0.0229 19.0 41116 0.5748 0.2763 0.0616
0.0214 20.0 43280 0.4433 0.2795 0.0615
0.0199 21.0 45444 0.4917 0.2688 0.0615
0.0187 22.0 47608 0.4385 0.2752 0.0614
0.0166 23.0 49772 0.5818 0.2800 0.0630
0.0157 24.0 51936 0.5325 0.2699 0.0611
0.0145 25.0 54100 0.6610 0.2772 0.0621
0.0135 26.0 56264 0.5643 0.2714 0.0618
0.0124 27.0 58428 0.6077 0.2631 0.0607
0.0117 28.0 60592 0.5318 0.2614 0.0588
0.0118 29.0 62756 0.5563 0.2664 0.0603
0.0112 30.0 64920 0.6108 0.2640 0.0608
0.0101 31.0 67084 0.5950 0.2616 0.0597
0.0101 32.0 69248 0.5932 0.2651 0.0603
0.0093 33.0 71412 0.6167 0.2717 0.0602
0.009 34.0 73576 0.6527 0.2676 0.0612
0.0079 35.0 75740 0.5715 0.2696 0.0606
0.0076 36.0 77904 0.6912 0.2672 0.0611
0.0078 37.0 80068 0.6246 0.2658 0.0607
0.0071 38.0 82232 0.6495 0.2689 0.0609

Framework versions

  • Transformers 4.45.2
  • Pytorch 2.1.0+cu118
  • Datasets 3.0.1
  • Tokenizers 0.20.1
Downloads last month
2
Safetensors
Model size
606M params
Tensor type
F32
·
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Model tree for asr-africa/w2v-bert-2.0-yogera-fleurs-cv-lg-50hrs-v2

Finetuned
(185)
this model