wav2vec2-base-finetuned-gtzan-bs-8

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: 0.6312
• 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

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.0125	1.0	113	1.8959	0.41
1.534	2.0	226	1.5343	0.53
1.1174	3.0	339	1.5299	0.51
1.0413	4.0	452	1.0910	0.68
0.5856	5.0	565	0.9129	0.7
0.4625	6.0	678	0.9821	0.75
0.6228	7.0	791	0.7124	0.79
0.2862	8.0	904	0.6634	0.81
0.273	9.0	1017	0.5889	0.86
0.1331	10.0	1130	0.6628	0.85
0.1616	11.0	1243	0.6544	0.86
0.0218	12.0	1356	0.6405	0.87
0.1485	13.0	1469	0.7176	0.85
0.1493	14.0	1582	0.6074	0.89
0.0163	15.0	1695	0.6312	0.88

Framework versions

• Transformers 4.32.0.dev0
• Pytorch 2.0.1+cu118
• Datasets 2.14.3
• Tokenizers 0.13.3