whisper-tiny-finetuned-gtzan

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

• Loss: 0.5357
• Accuracy: 0.865

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: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Accuracy	Validation Loss
1.8988	1.0	50	0.475	1.8064
1.2155	2.0	100	0.66	1.2221
0.9136	3.0	150	0.76	0.9259
0.7999	4.0	200	0.8	0.7412
0.4499	5.0	250	0.785	0.6758
0.2986	6.0	300	0.845	0.5601
0.2432	7.0	350	0.825	0.5678
0.1316	8.0	400	0.845	0.5153
0.1685	9.0	450	0.86	0.4840
0.1344	10.0	500	0.86	0.4803
0.0499	11.0	550	0.5167	0.855
0.0969	12.0	600	0.5370	0.85
0.0351	13.0	650	0.5022	0.86
0.0452	14.0	700	0.5289	0.855
0.0167	15.0	750	0.5357	0.865

Framework versions

• Transformers 4.34.1
• Pytorch 2.1.0+cu118
• Datasets 2.14.6
• Tokenizers 0.14.1