whisper-turbo-check / README.md
meiiny00's picture
Create README.md
0b988a4 verified
|
raw
history blame
2.23 kB
metadata
library_name: transformers
license: mit
base_model: openai/whisper-large-v3-turbo
tags:
  - generated_from_trainer
datasets:
  - fsicoli/common_voice_18_0
metrics:
  - wer
model-index:
  - name: Whisper Turbo Train
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: Common Voice 18.0
          type: fsicoli/common_voice_18_0
          split: None
        metrics:
          - name: Wer
            type: wer
            value: 15.246076710047603

Whisper Turbo Train

This model is a fine-tuned version of openai/whisper-large-v3-turbo on the Common Voice 18.0 dataset. It achieves the following results on the evaluation set:

  • Loss: 0.1156
  • Wer: 15.2461

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: 32
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 1000
  • training_steps: 8000
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
0.3715 0.4257 1000 0.3457 40.4692
0.251 0.8514 2000 0.2181 27.7065
0.1569 1.2771 3000 0.1814 24.1533
0.1436 1.7029 4000 0.1531 20.3812
0.0931 2.1286 5000 0.1374 18.4662
0.0891 2.5543 6000 0.1252 16.9349
0.0738 2.9800 7000 0.1199 15.5610
0.0544 3.4057 8000 0.1156 15.2461

Framework versions

  • Transformers 4.45.1
  • Pytorch 2.1.0
  • Datasets 3.0.1
  • Tokenizers 0.20.0