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: 0.8879
• Accuracy: 0.84

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: 17

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.9838	1.0	113	1.8627	0.37
1.6128	2.0	226	1.5998	0.48
1.0259	3.0	339	1.3821	0.57
1.2766	4.0	452	1.1708	0.66
0.6014	5.0	565	0.7257	0.77
0.5815	6.0	678	1.0738	0.68
0.7664	7.0	791	0.7244	0.8
0.2303	8.0	904	0.5838	0.84
0.4829	9.0	1017	0.5741	0.87
0.0859	10.0	1130	0.6199	0.83
0.2983	11.0	1243	0.8117	0.84
0.0642	12.0	1356	0.5938	0.88
0.0688	13.0	1469	0.9978	0.84
0.1542	14.0	1582	0.7437	0.85
0.0117	15.0	1695	0.9100	0.84
0.039	16.0	1808	0.7757	0.85
0.0661	17.0	1921	0.8879	0.84

Framework versions

• Transformers 4.31.0
• Pytorch 2.0.1
• Datasets 2.14.0
• Tokenizers 0.13.3