wav2vec2-base-finetuned-gtzan-plus

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: 0.6618
• Accuracy: 0.86

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: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Accuracy	Validation Loss
1.1816	1.0	29	0.41	2.1429
0.9004	2.0	58	0.61	1.7952
0.7342	3.0	87	0.62	1.6186
0.8389	4.0	116	0.67	1.4450
0.8045	5.0	145	0.72	1.2934
0.6668	6.0	174	0.81	1.1416
0.897	7.0	203	0.78	1.0876
0.5504	8.0	232	0.82	1.0412
0.4106	9.0	261	0.83	0.9954
0.4245	10.0	290	0.84	0.9752
0.2955	11.0	319	0.9407	0.8
0.2214	12.0	348	0.9383	0.8
0.5358	13.0	377	0.8892	0.81
0.3723	14.0	406	0.8386	0.82
0.3086	15.0	435	0.7704	0.86
0.2668	16.0	464	0.7496	0.81
0.2425	17.0	493	0.7069	0.85
0.2257	18.0	522	0.7138	0.85
0.1899	19.0	551	0.6733	0.86
0.1749	20.0	580	0.6618	0.86

Framework versions

• Transformers 4.53.2
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.4-dev.0