README.md

metadata

language:
  - ja
license: apache-2.0
tags:
  - hf-asr-leaderboard
datasets:
  - mozilla-foundation/common_voice_11_0
metrics:
  - wer
model-index:
  - name: Whisper Small Ja - kujirahand
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: Common Voice 11.0
          type: mozilla-foundation/common_voice_11_0
          config: ja
          split: test
          args: 'config: ja, split: test'
        metrics:
          - name: Wer
            type: wer
            value: 112.99945265462507

Whisper Small Ja - kujirahand

This model is a fine-tuned version of openai/whisper-small on the Common Voice 11.0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.6825
• Wer: 112.9995

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: 64
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 5
• training_steps: 40
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	0.25	10	1.7314	23.6341
No log	0.5	20	0.7887	112.5439
1.6472	0.75	30	0.7152	110.8503
1.6472	1.0	40	0.6825	112.9995

Framework versions

• Transformers 4.27.0.dev0
• Pytorch 1.13.1+cu116
• Datasets 2.9.1.dev0
• Tokenizers 0.13.2