whisper-medium-da / README.md
BlueRaccoon's picture
leaderboard fix (#1)
7755c0b
metadata
language:
  - da
license: apache-2.0
tags:
  - whisper-event
  - generated_from_trainer
datasets:
  - mozilla-foundation/common_voice_11_0
metrics:
  - wer
model-index:
  - name: Whisper Medium Danish
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: mozilla-foundation/common_voice_11_0
          type: mozilla-foundation/common_voice_11_0
          config: da
          split: test
          args: da
        metrics:
          - name: Wer
            type: wer
            value: 15.36559705418201

Whisper Medium Danish

This model is a fine-tuned version of openai/whisper-medium on the mozilla-foundation/common_voice_11_0 da dataset. It achieves the following results on the evaluation set:

  • Loss: 0.5759
  • Wer: 15.3656

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: 5e-06
  • 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: 5000
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
0.016 7.58 1000 0.4492 15.7391
0.0014 15.15 2000 0.5306 15.4550
0.0004 22.73 3000 0.5759 15.3656
0.0003 30.3 4000 0.5981 15.4655
0.0002 37.88 5000 0.6072 15.5076

Framework versions

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