README.md

metadata

license: apache-2.0
base_model: openai/whisper-large-v3
tags:
  - generated_from_trainer
datasets:
  - DewiBrynJones/oscar-cy-tts
metrics:
  - wer
model-index:
  - name: whisper-large-v3-ft-tts-cy
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: DewiBrynJones/oscar-cy-tts default
          type: DewiBrynJones/oscar-cy-tts
          args: default
        metrics:
          - name: Wer
            type: wer
            value: 0.10755268881722042

whisper-large-v3-ft-tts-cy

This model is a fine-tuned version of openai/whisper-large-v3 on the DewiBrynJones/oscar-cy-tts default dataset. It achieves the following results on the evaluation set:

• Loss: 0.1365
• Wer: 0.1076

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: 1e-05
• train_batch_size: 16
• eval_batch_size: 16
• 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
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.2469	0.3232	1000	0.2313	0.1733
0.1578	0.6464	2000	0.1785	0.1373
0.1491	0.9696	3000	0.1531	0.1213
0.099	1.2928	4000	0.1434	0.1129
0.0874	1.6160	5000	0.1365	0.1076

Framework versions

• Transformers 4.44.0
• Pytorch 2.4.0+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1