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: 1.0690
• 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: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.0298	1.0	113	1.9942	0.42
1.6308	2.0	226	1.7948	0.45
1.4047	3.0	339	1.6728	0.4
1.0438	4.0	452	1.2557	0.63
1.0471	5.0	565	1.0976	0.66
0.8658	6.0	678	0.9722	0.64
0.7625	7.0	791	0.7211	0.79
0.6197	8.0	904	0.9618	0.71
0.2382	9.0	1017	0.5927	0.85
0.275	10.0	1130	0.9532	0.75
0.2681	11.0	1243	1.1366	0.76
0.2315	12.0	1356	1.1621	0.79
0.0142	13.0	1469	0.9571	0.84
0.0151	14.0	1582	0.9650	0.84
0.1348	15.0	1695	1.2902	0.8
0.0082	16.0	1808	1.0652	0.83
0.0054	17.0	1921	0.9985	0.83
0.0049	18.0	2034	1.0041	0.85
0.0052	19.0	2147	1.0800	0.85
0.0044	20.0	2260	1.0690	0.84

Framework versions

• Transformers 4.39.0.dev0
• Pytorch 2.1.0+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2