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: 0.8249
- Accuracy: 0.87
The initial model trained for 20 epochs and overfit, so I recovered the best epoch (10) and pushed to hub. The metrics above reflect the latest model from epoch 10/checkpoint 2250.
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
- 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 |
|---|---|---|---|---|
| 1.9486 | 1.0 | 225 | 1.8744 | 0.54 |
| 1.0616 | 2.0 | 450 | 1.2196 | 0.66 |
| 1.0193 | 3.0 | 675 | 0.7841 | 0.78 |
| 0.81 | 4.0 | 900 | 0.7212 | 0.8 |
| 0.2171 | 5.0 | 1125 | 0.7194 | 0.77 |
| 0.0458 | 6.0 | 1350 | 0.8966 | 0.81 |
| 0.3485 | 7.0 | 1575 | 0.7960 | 0.81 |
| 0.09 | 8.0 | 1800 | 1.0860 | 0.82 |
| 0.0031 | 9.0 | 2025 | 0.7744 | 0.84 |
| 0.0026 | 10.0 | 2250 | 0.8249 | 0.87 |
Framework versions
- Transformers 4.33.0.dev0
- Pytorch 2.0.1+cu118
- Datasets 2.14.4.dev0
- Tokenizers 0.13.3
