golesheed/whisper-native-elderly-1-dutch

Whisper Large V2

This model is a fine-tuned version of openai/whisper-large-v2 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.4000
• Wer: 10.8918

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: 3e-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: 20
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.2263	0.49	30	0.3524	16.2639
0.2192	0.98	60	0.3412	16.1689
0.104	1.48	90	0.3537	12.4644
0.0967	1.97	120	0.3511	11.5567
0.0467	2.46	150	0.3665	13.7098
0.0409	2.95	180	0.3717	11.5884
0.0213	3.44	210	0.3875	11.2718
0.0156	3.93	240	0.3891	11.6834
0.0097	4.43	270	0.3946	10.7018
0.0064	4.92	300	0.4000	10.8918

Framework versions

• Transformers 4.38.0.dev0
• Pytorch 2.1.0+cu121
• Datasets 2.14.6
• Tokenizers 0.15.0