中文

Whisper Tiny Indonesian

This model is a fine-tuned version of openai/whisper-tiny on the mozilla-foundation/common_voice_11_0, magic_data, titml and google/fleurs dataset. It achieves the following results on the evaluation set:

  • Loss: 0.2409
  • Wer: 18.2837

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: 1e-05
  • train_batch_size: 32
  • eval_batch_size: 16
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 500
  • training_steps: 10000
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
0.4103 0.66 1000 0.3802 27.0497
0.2682 1.32 2000 0.3223 22.9365
0.2381 1.99 3000 0.2884 20.8245
0.1606 2.65 4000 0.2727 20.1928
0.1246 3.31 5000 0.2596 18.9984
0.1344 3.97 6000 0.2482 18.7540
0.0975 4.63 7000 0.2471 18.6388
0.0916 5.29 8000 0.2436 18.9615
0.0854 5.96 9000 0.2413 18.3114
0.0812 6.62 10000 0.2409 18.2837

Framework versions

  • Transformers 4.26.0.dev0
  • Pytorch 1.13.0+cu117
  • Datasets 2.7.1.dev0
  • Tokenizers 0.13.2