distilhubert-gtzan-exp03
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: 0.6567
- Accuracy: 0.84
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
- gradient_accumulation_steps: 2
- total_train_batch_size: 8
- optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 100
- num_epochs: 12
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 3.8615 | 1.0 | 113 | 1.8347 | 0.47 |
| 2.5022 | 2.0 | 226 | 1.2180 | 0.66 |
| 2.0256 | 3.0 | 339 | 1.0125 | 0.72 |
| 1.3321 | 4.0 | 452 | 0.7666 | 0.79 |
| 0.7712 | 5.0 | 565 | 0.6748 | 0.81 |
| 0.7126 | 6.0 | 678 | 0.5745 | 0.81 |
| 0.3654 | 7.0 | 791 | 0.5284 | 0.84 |
| 0.1403 | 8.0 | 904 | 0.6308 | 0.79 |
| 0.2026 | 9.0 | 1017 | 0.5906 | 0.86 |
| 0.0620 | 10.0 | 1130 | 0.6097 | 0.84 |
| 0.0524 | 11.0 | 1243 | 0.6419 | 0.84 |
| 0.0348 | 12.0 | 1356 | 0.6567 | 0.84 |
Framework versions
- Transformers 5.15.0
- Pytorch 2.11.0+cu128
- Datasets 2.21.0
- Tokenizers 0.22.2