whisper-tiny-try / README.md
Zakryah's picture
End of training
f991739 verified
metadata
library_name: peft
language:
  - hu
license: apache-2.0
base_model: openai/whisper-tiny
tags:
  - generated_from_trainer
datasets:
  - mozilla-foundation/common_voice_17_0
metrics:
  - wer
model-index:
  - name: Whisper Tiny Hu Test - Zakryah
    results:
      - task:
          type: automatic-speech-recognition
          name: Automatic Speech Recognition
        dataset:
          name: Common Voice 17.0
          type: mozilla-foundation/common_voice_17_0
          config: hu
          split: test
          args: 'config: hu, split: test'
        metrics:
          - type: wer
            value: 113.13022828434707
            name: Wer

Whisper Tiny Hu Test - Zakryah

This model is a fine-tuned version of openai/whisper-tiny on the Common Voice 17.0 dataset. It achieves the following results on the evaluation set:

  • Loss: 1.2430
  • Wer: 113.1302

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: 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_steps: 500
  • training_steps: 4000
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
1.3058 0.3299 1000 1.2987 114.1613
1.3143 0.6598 2000 1.2633 112.6243
1.2969 0.9898 3000 1.2478 113.3247
1.2082 1.3197 4000 1.2430 113.1302

Framework versions

  • PEFT 0.14.1.dev0
  • Transformers 4.50.0.dev0
  • Pytorch 2.6.0+cu126
  • Datasets 3.3.2
  • Tokenizers 0.21.0