metadata

language:
  - multilingual
license: apache-2.0
base_model: openai/whisper-large-v3
tags:
  - hf-asr-leaderboard
  - generated_from_trainer
datasets:
  - mozilla-foundation/common_voice_16_0
metrics:
  - wer
model-index:
  - name: Whisper large-v3 nan-tw
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: Common Voice 16.0
          type: mozilla-foundation/common_voice_16_0
          config: nan-tw
          split: test
          args: 'config: nan-tw, split: test'
        metrics:
          - name: Wer
            type: wer
            value: 280.9248554913295

Whisper large-v3 nan-tw

This model is a fine-tuned version of openai/whisper-large-v3 on the Common Voice 16.0 dataset. It achieves the following results on the evaluation set:

Loss: 1.0601
Wer: 280.9249

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: 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
0.2485	3.05	1000	0.9971	538.5505
0.0154	6.1	2000	1.0482	1460.5158
0.0024	9.15	3000	1.0330	261.3161
0.0014	12.2	4000	1.0554	300.3112
0.0003	15.24	5000	1.0601	280.9249

Framework versions

Transformers 4.37.1
Pytorch 2.1.0+cu121
Datasets 2.16.1
Tokenizers 0.15.1