whisper-large-v3-finetuned-gtzan

This model is a fine-tuned version of openai/whisper-large-v3 on the GTZAN dataset. It achieves the following results on the evaluation set:

  • Loss: 0.2657
  • Accuracy: 0.94

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: 4e-05
  • train_batch_size: 1
  • eval_batch_size: 1
  • seed: 42
  • distributed_type: multi-GPU
  • num_devices: 4
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 16
  • total_eval_batch_size: 4
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: cosine
  • lr_scheduler_warmup_ratio: 0.1
  • num_epochs: 10
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Accuracy
2.1646 0.5 28 1.8012 0.55
1.0152 1.0 56 0.8618 0.79
1.1129 1.49 84 0.7426 0.8
0.8163 1.99 112 0.8078 0.75
0.4374 2.49 140 0.6259 0.81
0.4607 2.99 168 0.5424 0.83
0.4225 3.48 196 0.3723 0.89
0.1769 3.98 224 0.3517 0.9
0.0927 4.48 252 0.3385 0.89
0.0159 4.98 280 0.3985 0.88
0.0119 5.48 308 0.4626 0.9
0.029 5.97 336 0.4292 0.91
0.0064 6.47 364 0.2710 0.93
0.0057 6.97 392 0.2665 0.93
0.0048 7.47 420 0.2784 0.93
0.0049 7.96 448 0.2550 0.94
0.0049 8.46 476 0.3011 0.94
0.0044 8.96 504 0.2759 0.94
0.0045 9.46 532 0.2661 0.94
0.0048 9.96 560 0.2657 0.94

Framework versions

  • Transformers 4.37.2
  • Pytorch 2.2.0+cu121
  • Datasets 2.17.0
  • Tokenizers 0.15.1
Downloads last month
10
Safetensors
Model size
637M params
Tensor type
F32
·
Inference Providers NEW
This model is not currently available via any of the supported third-party Inference Providers, and the model is not deployed on the HF Inference API.

Model tree for ChuGyouk/whisper-large-v3-finetuned-gtzan

Finetuned
(388)
this model

Dataset used to train ChuGyouk/whisper-large-v3-finetuned-gtzan

Evaluation results