ntu-spml/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: 0.7428
  • 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: 3.992986714871485e-05
  • train_batch_size: 16
  • eval_batch_size: 64
  • seed: 42
  • optimizer: Adam with betas=(0.9807885777224674,0.996064720140604) and epsilon=1e-08
  • lr_scheduler_type: cosine
  • lr_scheduler_warmup_steps: 509
  • num_epochs: 100

Training results

Training Loss Epoch Step Validation Loss Accuracy
No log 1.0 57 2.2832 0.3
No log 2.0 114 2.2273 0.28
No log 3.0 171 2.0861 0.46
No log 4.0 228 1.8473 0.5
No log 5.0 285 1.5146 0.6
No log 6.0 342 1.2140 0.69
No log 7.0 399 0.9856 0.74
No log 8.0 456 0.8056 0.79
1.6591 9.0 513 0.7135 0.8
1.6591 10.0 570 0.7642 0.75
1.6591 11.0 627 0.6344 0.79
1.6591 12.0 684 0.5982 0.83
1.6591 13.0 741 0.5369 0.86
1.6591 14.0 798 0.7501 0.79
1.6591 15.0 855 0.7493 0.78
1.6591 16.0 912 0.6891 0.83
1.6591 17.0 969 0.7492 0.8
0.2402 18.0 1026 0.6663 0.88
0.2402 19.0 1083 0.5750 0.89
0.2402 20.0 1140 0.8215 0.81
0.2402 21.0 1197 0.7435 0.79
0.2402 22.0 1254 0.8305 0.86
0.2402 23.0 1311 0.7636 0.83
0.2402 24.0 1368 0.9786 0.77
0.2402 25.0 1425 0.7082 0.88
0.2402 26.0 1482 0.7698 0.85
0.0206 27.0 1539 0.7360 0.87
0.0206 28.0 1596 0.8575 0.84
0.0206 29.0 1653 0.7428 0.87

Framework versions

  • Transformers 4.32.0.dev0
  • Pytorch 2.0.1
  • Datasets 2.13.2.dev1
  • Tokenizers 0.13.3
Downloads last month
14
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Model tree for catactivationsound/distilhubert-finetuned-gtzan

Finetuned
(400)
this model

Dataset used to train catactivationsound/distilhubert-finetuned-gtzan

Evaluation results