Whisper large-v2 Korean - ML_project_voice2text_largev2

This model is a fine-tuned version of openai/whisper-large-v2 on the Common Voice 11.0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.0754
• Cer: 1.7235

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: 0.001
• train_batch_size: 4
• 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: 50
• training_steps: 8000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer
0.2673	0.1797	1000	0.2870	8.9863
0.2145	0.3593	2000	0.2292	6.1204
0.1762	0.5390	3000	0.1896	16.8891
0.1408	0.7186	4000	0.1579	4.1889
0.1077	0.8983	5000	0.1229	4.1104
0.0763	1.0780	6000	0.0978	3.4940
0.0414	1.2576	7000	0.0832	2.2456
0.0336	1.4373	8000	0.0754	1.7235

Framework versions

• PEFT 0.11.2.dev0
• Transformers 4.41.1
• Pytorch 2.3.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1