dysarthria-tiny-en

This model is a fine-tuned version of openai/whisper-tiny.en on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.2870
• Wer: 52.8131
• Cer: 36.9112

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	Cer
0.0255	13.5135	500	0.4034	199.4555	178.1629
0.0005	27.0270	1000	0.2994	17.9673	11.4738
0.0002	40.5405	1500	0.2941	17.2414	11.1373
0.0002	54.0541	2000	0.2917	16.6969	10.7672
0.0001	67.5676	2500	0.2892	15.9710	10.3634
0.0001	81.0811	3000	0.2886	53.7205	37.3486
0.0001	94.5946	3500	0.2882	53.1760	37.2813
0.0001	108.1081	4000	0.2881	53.1760	37.3486
0.0001	121.6216	4500	0.2870	52.8131	36.9112
0.0	135.1351	5000	0.2870	52.8131	36.9112

Framework versions

• Transformers 4.45.1
• Pytorch 2.4.1+cu121
• Datasets 3.0.1
• Tokenizers 0.20.0