HamzaSidhu786/distilhubert-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: 0.6028
• 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: 3e-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
2.0751	1.0	113	2.0343	0.6
1.5734	2.0	226	1.6338	0.58
1.3801	3.0	339	1.2674	0.7
1.0384	4.0	452	1.1376	0.68
0.973	5.0	565	0.9849	0.73
1.0033	6.0	678	0.7686	0.76
0.6347	7.0	791	0.5909	0.87
0.6537	8.0	904	0.9489	0.75
0.359	9.0	1017	0.7478	0.81
0.2268	10.0	1130	0.6247	0.84
0.2674	11.0	1243	0.6437	0.84
0.2237	12.0	1356	0.7997	0.81
0.1418	13.0	1469	0.7738	0.84
0.1201	14.0	1582	0.5696	0.87
0.019	15.0	1695	0.8173	0.84
0.0175	16.0	1808	0.6395	0.88
0.16	17.0	1921	0.6062	0.87
0.0137	18.0	2034	0.5422	0.9
0.0127	19.0	2147	0.6421	0.88
0.0129	20.0	2260	0.6028	0.88

Framework versions

• Transformers 4.42.4
• Pytorch 2.3.1+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1