Fine-Tuned-Whisper-Large-v2-Zeroth-STT-KO

This model was trained from scratch on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.3667
• Wer: 26.5596
• Cer: 0.3686

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: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Use adamw_bnb_8bit with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 100
• training_steps: 1500
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
0.2835	0.6944	200	0.3069	26.3590	0.3688
0.2194	1.3889	400	0.3056	26.4986	0.3670
0.1549	2.0833	600	0.3249	26.9697	0.3659
0.0791	2.7778	800	0.3215	26.2368	0.3680
0.0497	3.4722	1000	0.3388	26.9086	0.3885
0.028	4.1667	1200	0.3616	26.5684	0.3693
0.0145	4.8611	1400	0.3667	26.5596	0.3686

Framework versions

• Transformers 4.48.1
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0