README.md

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