wav2vec2-base-100k-voxpopuli-finetuned-gtzan

This model is a fine-tuned version of facebook/wav2vec2-base-100k-voxpopuli on the GTZAN dataset. It achieves the following results on the evaluation set:

• Loss: 0.9034
• 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: 5e-05
• train_batch_size: 4
• eval_batch_size: 4
• 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
2.1924	1.0	225	2.1487	0.27
1.8417	2.0	450	1.8767	0.38
1.6017	3.0	675	1.5778	0.51
1.3497	4.0	900	1.4785	0.4
1.2631	5.0	1125	1.3103	0.58
0.8172	6.0	1350	1.1736	0.63
1.1657	7.0	1575	0.9690	0.74
1.1711	8.0	1800	1.3609	0.63
0.5033	9.0	2025	0.7300	0.83
0.4104	10.0	2250	0.9866	0.72
0.318	11.0	2475	0.8159	0.81
0.1074	12.0	2700	0.8024	0.85
0.093	13.0	2925	0.8285	0.85
0.7407	14.0	3150	0.8591	0.87
0.027	15.0	3375	0.9574	0.84
0.4564	16.0	3600	0.9762	0.85
0.0198	17.0	3825	0.9204	0.85
0.5467	18.0	4050	0.8703	0.87
0.2644	19.0	4275	0.8855	0.87
0.013	20.0	4500	0.9034	0.87

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.0+cu121
• Datasets 3.0.0
• Tokenizers 0.19.1