wav2vec2-base-finetuned-gtzan

This model is a fine-tuned version of facebook/wav2vec2-base on the GTZAN dataset. It achieves the following results on the evaluation set:

• Loss: 0.5472
• Accuracy: 0.88

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: 5e-05
• train_batch_size: 8
• 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_ratio: 0.1
• num_epochs: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.9984	1.0	113	1.9505	0.46
1.4188	2.0	226	1.5582	0.52
1.2867	3.0	339	1.1267	0.65
0.7716	4.0	452	0.9512	0.64
0.5553	5.0	565	0.9790	0.72
0.7491	6.0	678	0.7419	0.78
0.4399	7.0	791	0.5709	0.86
0.2489	8.0	904	0.6352	0.8
0.388	9.0	1017	0.5130	0.89
0.2066	10.0	1130	0.7185	0.86
0.1905	11.0	1243	0.5545	0.9
0.1312	12.0	1356	0.8126	0.85
0.0185	13.0	1469	0.4841	0.91
0.0154	14.0	1582	0.7167	0.86
0.0156	15.0	1695	0.5472	0.88

Framework versions

• Transformers 4.41.2
• Pytorch 2.3.0+cu121
• Datasets 2.19.2
• Tokenizers 0.19.1