whisper-small-uk

This model is a fine-tuned version of openai/whisper-small on the mozilla-foundation/common_voice_16_1 dataset. It achieves the following results on the evaluation set:

• Loss: 0.2744
• Wer: 26.3570

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: 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: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.278	0.47	1000	0.3330	31.8004
0.2662	0.94	2000	0.2961	29.4969
0.1403	1.42	3000	0.2796	27.3209
0.1105	1.89	4000	0.2702	26.2724
0.0719	2.36	5000	0.2744	26.3570

Framework versions

• Transformers 4.38.0.dev0
• Pytorch 2.2.0+cu121
• Datasets 2.17.1
• Tokenizers 0.15.2