whisper-tiny-finetune

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: 0.4696
• Wer: 19.7170

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: 128
• 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: 90
• training_steps: 180

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
3.9026	0.2778	10	3.8176	40.1572
3.5226	0.5556	20	3.3623	34.1509
2.9307	0.8333	30	2.6468	32.2642
2.0034	1.1111	40	1.6171	33.7736
1.0812	1.3889	50	0.8580	29.9371
0.7061	1.6667	60	0.6809	27.0755
0.5945	1.9444	70	0.6080	25.1258
0.532	2.2222	80	0.5639	23.3962
0.5035	2.5	90	0.5357	22.2642
0.4955	2.7778	100	0.5125	21.0063
0.4208	3.0556	110	0.4960	20.6604
0.353	3.3333	120	0.4878	20.0
0.3453	3.6111	130	0.4817	19.8113
0.3945	3.8889	140	0.4768	19.8113
0.3273	4.1667	150	0.4727	19.8742
0.2941	4.4444	160	0.4713	19.7799
0.3131	4.7222	170	0.4708	19.2767
0.2948	5.0	180	0.4696	19.7170

Framework versions

• Transformers 4.41.2
• Pytorch 2.3.0+cu121
• Datasets 2.19.3.dev0
• Tokenizers 0.19.1