wav2vec2-base-960h-finetuned-gtzan

This model is a fine-tuned version of facebook/wav2vec2-base-960h on the GTZAN dataset. It achieves the following results on the evaluation set:

• Loss: 0.7067
• Accuracy: 0.87

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: 0.2
• num_epochs: 25
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.3013	0.9912	56	2.2907	0.12
2.247	2.0	113	2.1508	0.31
2.025	2.9912	169	1.8583	0.44
1.8441	4.0	226	1.5589	0.49
1.6488	4.9912	282	1.5304	0.43
1.6473	6.0	339	1.4225	0.53
1.3746	6.9912	395	1.3659	0.6
1.2692	8.0	452	1.3538	0.49
1.0924	8.9912	508	1.3275	0.59
0.9971	10.0	565	1.1256	0.6
1.0647	10.9912	621	0.8908	0.71
0.8927	12.0	678	1.0522	0.71
0.764	12.9912	734	0.8975	0.72
0.7493	14.0	791	1.2048	0.68
0.606	14.9912	847	0.6913	0.81
0.5683	16.0	904	0.7876	0.75
0.5753	16.9912	960	0.6032	0.84
0.4177	18.0	1017	0.7026	0.82
0.369	18.9912	1073	0.6930	0.84
0.2902	20.0	1130	1.0089	0.81
0.2566	20.9912	1186	0.7876	0.84
0.2661	22.0	1243	0.5696	0.89
0.146	22.9912	1299	0.7424	0.85
0.1522	24.0	1356	0.6972	0.86
0.1315	24.7788	1400	0.7067	0.87

Framework versions

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