HappyTsai/whisper-small-zh

Whisper Small Chinese

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

• Loss: 0.2141
• Wer: 40.7369

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: 8
• eval_batch_size: 2
• 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
0.2575	0.2	1000	0.2207	44.9744
0.0529	1.16	2000	0.2099	41.3101
0.0212	2.13	3000	0.2183	41.5558
0.0041	3.09	4000	0.2151	41.4944
0.003	4.06	5000	0.2141	40.7369

Framework versions

• Transformers 4.28.0.dev0
• Pytorch 1.13.1+cu117
• Datasets 2.10.2.dev0
• Tokenizers 0.13.2