metadata

license: apache-2.0
tags:
  - generated_from_trainer
datasets:
  - marsyas/gtzan
metrics:
  - accuracy
model-index:
  - name: distilhubert-finetuned-gtzan
    results: []

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: 1.2658
Accuracy: 0.8

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
distributed_type: multi-GPU
gradient_accumulation_steps: 2
total_train_batch_size: 8
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_ratio: 0.1
num_epochs: 30
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.1071	1.0	112	2.1453	0.33
1.6165	2.0	225	1.6129	0.59
1.2842	3.0	337	1.2084	0.68
0.9805	4.0	450	0.8842	0.74
0.5216	5.0	562	0.7350	0.78
0.5017	6.0	675	0.8196	0.77
0.1998	7.0	787	0.6709	0.8
0.3662	8.0	900	0.8483	0.78
0.2711	9.0	1012	0.8567	0.81
0.0183	10.0	1125	0.8994	0.82
0.0299	11.0	1237	1.2142	0.8
0.0064	12.0	1350	1.0208	0.81
0.004	13.0	1462	1.0619	0.81
0.0031	14.0	1575	1.1454	0.79
0.0028	15.0	1687	1.1010	0.81
0.0023	16.0	1800	1.0595	0.8
0.0017	17.0	1912	1.1340	0.8
0.0015	18.0	2025	1.1760	0.81
0.0014	19.0	2137	1.1361	0.81
0.0012	20.0	2250	1.2138	0.81
0.0011	21.0	2362	1.1366	0.81
0.0012	22.0	2475	1.1662	0.8
0.0011	23.0	2587	1.1491	0.8
0.0009	24.0	2700	1.1287	0.81
0.0009	25.0	2812	1.2027	0.81
0.0009	26.0	2925	1.1740	0.81
0.0009	27.0	3037	1.2011	0.81
0.0009	28.0	3150	1.2523	0.8
0.0008	29.0	3262	1.2494	0.81
0.0007	29.87	3360	1.2658	0.8

Framework versions

Transformers 4.29.2
Pytorch 1.13.1+cu117
Datasets 2.12.0
Tokenizers 0.13.3