music-genre-detector-finetuned-gtzan_dset
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: 0.3892
- Accuracy: 0.8972
- Precision: 0.8989
- Recall: 0.8972
- F1: 0.8974
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: 9e-05
- train_batch_size: 4
- eval_batch_size: 4
- seed: 42
- gradient_accumulation_steps: 16
- total_train_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 7
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | Precision | Recall | F1 |
|---|---|---|---|---|---|---|---|
| 2.2319 | 0.98 | 49 | 1.5808 | 0.5263 | 0.5682 | 0.5263 | 0.4767 |
| 1.2682 | 1.98 | 99 | 0.9750 | 0.7556 | 0.7524 | 0.7556 | 0.7510 |
| 0.9462 | 2.99 | 149 | 0.7403 | 0.7945 | 0.7964 | 0.7945 | 0.7921 |
| 0.5946 | 3.99 | 199 | 0.5921 | 0.8233 | 0.8281 | 0.8233 | 0.8214 |
| 0.4095 | 4.99 | 249 | 0.4772 | 0.8634 | 0.8663 | 0.8634 | 0.8638 |
| 0.3349 | 5.99 | 299 | 0.4167 | 0.8835 | 0.8866 | 0.8835 | 0.8841 |
| 0.2427 | 6.88 | 343 | 0.3892 | 0.8972 | 0.8989 | 0.8972 | 0.8974 |
Framework versions
- Transformers 4.33.1
- Pytorch 1.10.2+cu111
- Datasets 2.14.5
- Tokenizers 0.13.3