distilhubert-finetuned-gtzan_accuracy_93
This model is a fine-tuned version of yuval6967/distilhubert-finetuned-gtzan on the GTZAN dataset.
It achieves the following results on the evaluation set:
- Loss: 0.5121
- Accuracy: 0.93
Model description
- Fine-tuned model to demonstrate > 87% accuracy for the Huggingface Audio course
Intended uses & limitations
- Model is built to identify the genre of music based on a ~30 sec clip
Training and evaluation data
More information needed
Training procedure
- test_size = 0.20 was used for the split
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 8
- eval_batch_size: 8
- 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: 10
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 0.0316 | 1.0 | 100 | 0.4338 | 0.895 |
| 0.0031 | 2.0 | 200 | 0.7039 | 0.86 |
| 0.0069 | 3.0 | 300 | 0.4526 | 0.925 |
| 0.1799 | 4.0 | 400 | 0.7071 | 0.88 |
| 0.1783 | 5.0 | 500 | 0.5923 | 0.92 |
| 0.0011 | 6.0 | 600 | 0.5498 | 0.92 |
| 0.0005 | 7.0 | 700 | 0.4927 | 0.925 |
| 0.0005 | 8.0 | 800 | 0.6172 | 0.915 |
| 0.0004 | 9.0 | 900 | 0.4988 | 0.925 |
| 0.0004 | 10.0 | 1000 | 0.5121 | 0.93 |
Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu118
- Datasets 2.13.1
- Tokenizers 0.13.3
- Downloads last month
- 9
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.