Whisper_Korean_fine-tune

This model is a fine-tuned version of openai/whisper-base on the TARARARAK/KsponSpeech_01_ALHUB_preprocess dataset. It achieves the following results on the evaluation set:

• Loss: 0.4357
• Cer: 14.3015

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: 4000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer
0.5007	0.2690	1000	0.5043	17.1656
0.4516	0.5381	2000	0.4650	15.5761
0.4293	0.8071	3000	0.4440	14.9569
0.3653	1.0761	4000	0.4357	14.3015

Framework versions

• Transformers 4.42.0.dev0
• Pytorch 2.3.0+cu121
• Datasets 2.19.2
• Tokenizers 0.19.1