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: 0.5086
• Accuracy: 0.89

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: 4e-05
• train_batch_size: 6
• eval_batch_size: 6
• seed: 42
• gradient_accumulation_steps: 7
• total_train_batch_size: 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: 25
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.2912	0.98	21	2.2667	0.19
2.2263	1.96	42	2.1460	0.48
1.9552	2.99	64	1.8067	0.44
1.5982	3.97	85	1.5912	0.54
1.5182	4.99	107	1.4077	0.61
1.2855	5.97	128	1.2654	0.69
1.1649	7.0	150	1.1915	0.69
1.0742	7.98	171	1.0769	0.75
1.0495	8.96	192	1.0011	0.77
0.8827	9.99	214	0.9062	0.79
0.7886	10.97	235	0.8333	0.83
0.7019	11.99	257	0.7801	0.83
0.6642	12.97	278	0.7691	0.79
0.5982	14.0	300	0.6984	0.82
0.5002	14.98	321	0.6526	0.84
0.4789	15.96	342	0.5980	0.88
0.3908	16.99	364	0.5874	0.86
0.3892	17.97	385	0.5570	0.86
0.3675	18.99	407	0.5634	0.87
0.303	19.97	428	0.5387	0.87
0.3017	21.0	450	0.5086	0.89
0.2469	21.98	471	0.4969	0.89
0.2542	22.96	492	0.4972	0.88
0.2651	23.99	514	0.4947	0.89
0.2591	24.5	525	0.4929	0.89

Framework versions

• Transformers 4.35.0
• Pytorch 2.1.0
• Datasets 2.14.6
• Tokenizers 0.14.1