metadata
library_name: transformers
license: bsd-3-clause
base_model: MIT/ast-finetuned-audioset-10-10-0.4593
tags:
- generated_from_trainer
datasets:
- marsyas/gtzan
metrics:
- accuracy
model-index:
- name: ast-finetuned-gtzan
results:
- task:
name: Audio Classification
type: audio-classification
dataset:
name: GTZAN
type: marsyas/gtzan
config: default
split: train
args: default
metrics:
- name: Accuracy
type: accuracy
value: 0.94
ast-finetuned-gtzan
This model is a fine-tuned version of MIT/ast-finetuned-audioset-10-10-0.4593 on the GTZAN dataset. It achieves the following results on the evaluation set:
- Loss: 0.2095
- Accuracy: 0.94
Model description
MIT/ast-finetuned-audioset-10-10-0.4593 model has been used with the head replaced for classification of the 10 music genres.
Intended uses & limitations
More information needed
Training and evaluation data
GTZAN dataset has been used for training. 20% split was used for evaluation.
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-06
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.3
- num_epochs: 10
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 2.3629 | 1.0877 | 62 | 2.0026 | 0.38 |
| 1.816 | 2.1754 | 124 | 1.0936 | 0.76 |
| 1.0625 | 3.2632 | 186 | 0.5384 | 0.87 |
| 0.5878 | 4.3509 | 248 | 0.3465 | 0.91 |
| 0.2426 | 5.4386 | 310 | 0.3506 | 0.88 |
| 0.1584 | 6.5263 | 372 | 0.2532 | 0.92 |
| 0.1067 | 7.6140 | 434 | 0.2333 | 0.9 |
| 0.0741 | 8.7018 | 496 | 0.2248 | 0.91 |
| 0.0431 | 9.7895 | 558 | 0.2095 | 0.94 |
Framework versions
- Transformers 4.53.3
- Pytorch 2.6.0+cu124
- Datasets 3.6.0
- Tokenizers 0.21.2