haseong8012/whisper-small_child-10k

openai/whisper-small-Ko-haseong8012/korean-child-command-voice_train-0-10000_smaplingRate-16000

This model is a fine-tuned version of openai/whisper-small on the haseong8012/korean-child-command-voice_train-0-10000_smaplingRate-16000 dataset. It achieves the following results on the evaluation set:

• Loss: 0.1613
• Wer: 14.9646
• Cer: 7.4814

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: 1.25e-05
• 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	Cer
0.0098	4.0	1000	0.1831	18.0032	9.0777
0.0007	8.0	2000	0.1634	15.6271	7.7868
0.0002	12.0	3000	0.1611	15.2159	7.5300
0.0001	16.0	4000	0.1605	15.0331	7.5370
0.0001	20.0	5000	0.1613	14.9646	7.4814

Framework versions

• Transformers 4.28.0
• Pytorch 2.1.0+cu121
• Datasets 2.14.5
• Tokenizers 0.13.3