yurim111/whisper_large_ko

Whisper Large Ko - yurim111

This model is a fine-tuned version of openai/whisper-large on the Common Voice 11.0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.5245
• Wer: 30.6143

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: 2e-05
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• optimizer: Use adamw_torch 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: 3000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.013	26.3158	500	0.3880	30.5136
0.0051	52.6316	1000	0.5522	36.8580
0.0001	78.9474	1500	0.5229	36.1531
0.0	105.2632	2000	0.5227	31.4199
0.0	131.5789	2500	0.5241	30.7150
0.0	157.8947	3000	0.5245	30.6143

Framework versions

• Transformers 4.51.1
• Pytorch 2.6.0+cu124
• Datasets 3.5.0
• Tokenizers 0.21.1