miosipof/asr_michael_medium1

Demo Model Whisper Large

This model is a fine-tuned version of EdoAbati/whisper-medium-it on the b-brave/speech_disorders_voice dataset. It achieves the following results on the evaluation set:

• Loss: 0.4114

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: 0.0001
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 64
• training_steps: 256
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
4.1664	6.0	48	3.4000
2.46	12.0	96	1.7709
1.121	18.0	144	0.4948
0.3706	24.0	192	0.4159
0.3031	30.0	240	0.4114

Framework versions

• Transformers 4.43.3
• Pytorch 2.2.0
• Datasets 2.20.0
• Tokenizers 0.19.1