hubert-large-ll60k-finetuned-gtzan
This model is a fine-tuned version of facebook/hubert-large-ll60k on the GTZAN dataset. It achieves the following results on the evaluation set:
- Loss: 0.9220
- Accuracy: 0.73
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: 7e-05
- train_batch_size: 2
- eval_batch_size: 2
- seed: 42
- gradient_accumulation_steps: 8
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 10
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 2.2456 | 1.0 | 56 | 2.2312 | 0.34 |
| 2.059 | 1.99 | 112 | 1.9662 | 0.32 |
| 1.8574 | 2.99 | 168 | 1.6258 | 0.5 |
| 1.4447 | 4.0 | 225 | 1.4547 | 0.59 |
| 1.4224 | 5.0 | 281 | 1.2372 | 0.65 |
| 1.2131 | 5.99 | 337 | 1.0879 | 0.67 |
| 1.1151 | 6.99 | 393 | 1.0599 | 0.69 |
| 0.9471 | 8.0 | 450 | 1.0339 | 0.68 |
| 1.0319 | 9.0 | 506 | 0.9568 | 0.71 |
| 0.9313 | 9.96 | 560 | 0.9220 | 0.73 |
Framework versions
- Transformers 4.30.0.dev0
- Pytorch 2.0.1+cu117
- Datasets 2.13.1
- Tokenizers 0.13.3