metadata

language:
  - eu
license: apache-2.0
tags:
  - whisper-event
  - generated_from_trainer
datasets:
  - mozilla-foundation/common_voice_13_0
metrics:
  - wer
model-index:
  - name: Whisper Small Basque
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: mozilla-foundation/common_voice_13_0 eu
          type: mozilla-foundation/common_voice_13_0
          config: eu
          split: test
          args: eu
        metrics:
          - name: Wer
            type: wer
            value: 14.119648426424725

Whisper Small Basque

This model is a fine-tuned version of openai/whisper-medium on the mozilla-foundation/common_voice_13_0 eu dataset. It achieves the following results on the evaluation set:

Loss: 0.2376
Wer: 14.1196

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: 6e-06
train_batch_size: 4
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: 500
training_steps: 8000
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.443	0.06	500	0.5037	37.4296
0.4196	0.12	1000	0.4010	28.9137
0.2823	0.19	1500	0.3453	24.6851
0.2551	0.25	2000	0.3164	22.5789
0.206	0.31	2500	0.2902	19.7922
0.2327	0.38	3000	0.2707	18.9356
0.1416	1.03	3500	0.2566	17.6921
0.0998	1.09	4000	0.2551	16.8213
0.095	1.15	4500	0.2511	16.3899
0.0971	1.21	5000	0.2415	15.5393
0.0964	1.28	5500	0.2336	15.1707
0.072	1.34	6000	0.2353	14.7596
0.0658	1.4	6500	0.2340	14.6766
0.033	2.05	7000	0.2349	14.3768
0.0288	2.11	7500	0.2371	14.1865
0.0352	2.18	8000	0.2376	14.1196

Framework versions

Transformers 4.26.0.dev0
Pytorch 1.13.1+cu117
Datasets 2.8.1.dev0
Tokenizers 0.13.2