deberta-korean-multilabel_base-lora

This model is a fine-tuned version of team-lucid/deberta-v3-base-korean on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1443
• F1: 0.9717
• Roc Auc: 0.8439

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: 0.0002
• train_batch_size: 64
• eval_batch_size: 128
• seed: 42
• 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
• num_epochs: 2

Training results

Training Loss	Epoch	Step	Validation Loss	F1	Roc Auc
0.1514	1.0	6259	0.1509	0.9705	0.8375
0.1396	2.0	12518	0.1443	0.9717	0.8439

Framework versions

• PEFT 0.18.0
• Transformers 4.57.3
• Pytorch 2.9.0+cu126
• Datasets 4.0.0
• Tokenizers 0.22.2