whisper-korean-zeroth-v1

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

• Loss: 0.0679
• Cer: 1.2916

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: 1
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 8
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• 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	Cer
0.1323	0.3593	1000	0.1831	4.9545
0.0789	0.7187	2000	0.1235	3.6079
0.034	1.0780	3000	0.0992	2.0689
0.0403	1.4373	4000	0.0785	1.5743
0.0243	1.7967	5000	0.0679	1.2916

Framework versions

• Transformers 4.57.1
• Pytorch 2.9.0+cu126
• Datasets 4.0.0
• Tokenizers 0.22.1