README.md

metadata

license: apache-2.0
base_model: facebook/wav2vec2-large-xlsr-53
tags:
  - automatic-speech-recognition
  - techiaith/banc-trawsgrifiadau-bangor
  - generated_from_trainer
datasets:
  - banc-trawsgrifiadau-bangor
metrics:
  - wer
model-index:
  - name: wav2vec2-xlsr-53-ft-btb-cy
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: TECHIAITH/BANC-TRAWSGRIFIADAU-BANGOR - CY
          type: banc-trawsgrifiadau-bangor
          config: cy
          split: test
          args: 'Config: cy, Training split: train, Eval split: test'
        metrics:
          - name: Wer
            type: wer
            value: 0.44248649327805

wav2vec2-xlsr-53-ft-btb-cy

This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the TECHIAITH/BANC-TRAWSGRIFIADAU-BANGOR - CY dataset. It achieves the following results on the evaluation set:

• Loss: 0.5992
• Wer: 0.4425

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: 0.0003
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 800
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
4.6574	1.41	500	1.0889	0.7293
1.0887	2.83	1000	0.5992	0.4425
0.7223	4.24	1500	0.7267	0.4631
2.1295	5.66	2000	3.0492	0.9984

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.1+cu121
• Datasets 2.15.0
• Tokenizers 0.15.0