| { | |
| "base_model": "KPF/KPF-bert-ner", | |
| "dataset": "townboy/korean-pii-dataset", | |
| "dataset_config": "bio", | |
| "batch_size": 4, | |
| "validation_batch_size": 8, | |
| "gradient_accumulation_steps": 8, | |
| "effective_batch_size": 32, | |
| "max_length": 512, | |
| "early_stopping_patience": 1, | |
| "loss": "standard_cross_entropy", | |
| "class_weighted": false, | |
| "train_rows": 9227, | |
| "validation_rows": 1510, | |
| "optimizer_steps_per_epoch": 289, | |
| "final_refinement_stages": [ | |
| { | |
| "purpose": "Korean RRN/alien-number counterfactual context learning", | |
| "epochs": 1, | |
| "learning_rate": 5e-06, | |
| "seed": 20260824, | |
| "corpus": "public corpus with counterfactual Korean-ID contexts" | |
| }, | |
| { | |
| "purpose": "general-label rehearsal and stabilization", | |
| "epochs": 1, | |
| "learning_rate": 1e-06, | |
| "seed": 20260825, | |
| "corpus": "pre-counterfactual corpus snapshot" | |
| } | |
| ] | |
| } | |