wav2vec2-base-finetuned-gtzan2

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: 0.6890
• 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: 7e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 25
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.2597	0.9912	56	2.0173	0.49
1.8041	2.0	113	1.5455	0.6
1.3996	2.9912	169	1.6014	0.5
1.2821	4.0	226	1.4333	0.53
1.0763	4.9912	282	1.0503	0.65
0.9496	6.0	339	1.1490	0.64
0.8102	6.9912	395	1.0606	0.7
0.5868	8.0	452	0.9500	0.72
0.3912	8.9912	508	0.5193	0.87
0.3939	10.0	565	0.8336	0.77
0.3945	10.9912	621	0.8204	0.78
0.395	12.0	678	0.5713	0.86
0.2702	12.9912	734	0.7380	0.82
0.2035	14.0	791	0.7632	0.82
0.2397	14.9912	847	0.6820	0.84
0.0944	16.0	904	0.9130	0.84
0.116	16.9912	960	0.4568	0.91
0.0989	18.0	1017	0.7468	0.87
0.1069	18.9912	1073	0.7100	0.86
0.0419	20.0	1130	0.7169	0.86
0.0344	20.9912	1186	0.7216	0.87
0.0612	22.0	1243	0.6798	0.89
0.0473	22.9912	1299	0.6920	0.88
0.0248	24.0	1356	0.6526	0.89
0.0157	24.7788	1400	0.6890	0.88

Framework versions

• Transformers 4.40.2
• Pytorch 2.8.0+cu128
• Datasets 3.6.0
• Tokenizers 0.19.1