README.md

metadata

license: apache-2.0
tags:
  - generated_from_trainer
datasets:
  - common_voice_8_0
metrics:
  - wer
model-index:
  - name: wav2vec2-large-xls-r-1b-frisian-cv-8-10m
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: common_voice_8_0
          type: common_voice_8_0
          config: fy-NL
          split: validation
          args: fy-NL
        metrics:
          - name: Wer
            type: wer
            value: 0.7612841022711041

wav2vec2-large-xls-r-1b-frisian-cv-8-10m

This model is a fine-tuned version of facebook/wav2vec2-xls-r-1b on the common_voice_8_0 dataset. It achieves the following results on the evaluation set:

• Loss: 1.1618
• Wer: 0.7613

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
• optimizer: Adam with betas=(0.9,0.98) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 80
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
8.7106	6.25	50	4.0034	1.0
3.4036	12.5	100	3.1030	1.0
3.7265	18.75	150	3.0466	1.0
3.2292	25.0	200	3.0166	1.0
3.1305	31.25	250	2.9699	1.0
3.0447	37.5	300	2.9144	1.0
2.9037	43.75	350	2.2919	0.9998
2.1115	50.0	400	1.3995	0.9429
1.3456	56.25	450	1.1093	0.8435
1.3206	62.5	500	1.1573	0.8112
1.0078	68.75	550	1.1746	0.7757
1.0674	75.0	600	1.1618	0.7613

Framework versions

• Transformers 4.28.1
• Pytorch 2.0.0+cu117
• Datasets 2.11.0
• Tokenizers 0.13.3