distilhubert-finetuned-gtzan

This model is a fine-tuned version of ntu-spml/distilhubert on the GTZAN dataset. It achieves the following results on the evaluation set:

• Loss: 0.7392
• Accuracy: 0.81

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: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• 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

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.3055	0.97	7	1.2863	0.73
1.2903	1.93	14	1.2504	0.7
1.2118	2.9	21	1.1450	0.77
1.1443	4.0	29	1.1224	0.74
1.006	4.97	36	1.0376	0.79
1.0174	5.93	43	0.9681	0.8
0.9155	6.9	50	0.9322	0.81
0.8781	8.0	58	0.9266	0.78
0.819	8.97	65	0.8473	0.79
0.7984	9.93	72	0.8225	0.77
0.7254	10.9	79	0.8096	0.81
0.6752	12.0	87	0.7801	0.81
0.6132	12.97	94	0.7687	0.8
0.615	13.93	101	0.7603	0.79
0.6162	14.9	108	0.7599	0.82
0.5678	16.0	116	0.7414	0.81
0.548	16.97	123	0.7423	0.81
0.5495	17.93	130	0.7378	0.81
0.5185	18.9	137	0.7396	0.81
0.5544	19.31	140	0.7392	0.81

Framework versions

• Transformers 4.32.0.dev0
• Pytorch 2.0.1
• Datasets 2.13.1
• Tokenizers 0.13.3