metadata

license: apache-2.0
tags:
  - generated_from_trainer
datasets:
  - common_voice_12_0
metrics:
  - wer
model-index:
  - name: wav2vec2-large-xls-r-1b-frisian
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: common_voice_12_0
          type: common_voice_12_0
          config: fy-NL
          split: validation
          args: fy-NL
        metrics:
          - name: Wer
            type: wer
            value: 0.1685917915949865

wav2vec2-large-xls-r-1b-frisian

This model is a fine-tuned version of facebook/wav2vec2-xls-r-1b on the common_voice_12_0 dataset. It achieves the following results on the evaluation set:

Loss: 0.2748
Wer: 0.1686

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: 16
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 32
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 500
num_epochs: 50
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
4.6809	2.1	250	2.0948	0.9948
1.2928	4.2	500	0.4505	0.4003
0.7887	6.3	750	0.3410	0.3287
0.7422	8.4	1000	0.3017	0.2756
0.7277	10.5	1250	0.3014	0.2624
0.6339	12.61	1500	0.2833	0.2398
0.5284	14.71	1750	0.2970	0.2404
0.5186	16.81	2000	0.2886	0.2400
0.515	18.91	2250	0.2891	0.2335
0.5199	21.01	2500	0.2985	0.2261
0.5228	23.11	2750	0.3026	0.2187
0.5102	25.21	3000	0.2829	0.1994
0.463	27.31	3250	0.2885	0.2012
0.5072	29.41	3500	0.2936	0.1971
0.4581	31.51	3750	0.2979	0.1912
0.4103	33.61	4000	0.2935	0.1875
0.3414	35.71	4250	0.2999	0.1860
0.4484	37.82	4500	0.2917	0.1810
0.3523	39.92	4750	0.2875	0.1759
0.3763	42.02	5000	0.2901	0.1758
0.2416	44.12	5250	0.2707	0.1740
0.1878	46.22	5500	0.2707	0.1717
0.1623	48.32	5750	0.2748	0.1686

Framework versions

Transformers 4.27.3
Pytorch 2.0.0+cu117
Datasets 2.10.1
Tokenizers 0.13.2