Update README.md

Replace confusion_matrix_baseline_final.png with row-normalized percentage version

Replace confusion_matrix_final_with_unknown.png with row-normalized percentage version

Replace confusion_matrix.png with row-normalized percentage version

Update README.md

Update README.md

Add LoRA CM with UNKNOWN column

Add baseline confusion matrix

Korean model card with baseline comparison + off-vocab analysis

Update README.md

Add confusion matrix visualization

Update model card: full eval metrics, confusion matrix, usage + serving guide

Upload TRAINING.md with huggingface_hub

Final LoRA adapter - eval_loss 0.023164 at step 850

initial commit