pii-layout-synth-v4

This model is a fine-tuned version of answerdotai/ModernBERT-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0049
• Precision: 0.9906
• Recall: 0.9940
• F1: 0.9923
• Accuracy: 0.9986

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: 5e-05
• train_batch_size: 16
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• 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: cosine_with_restarts
• lr_scheduler_warmup_steps: 0.1
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
0.1117	0.2389	500	0.0483	0.8936	0.9383	0.9154	0.9859
0.0453	0.4778	1000	0.0176	0.9594	0.9769	0.9681	0.9947
0.0294	0.7167	1500	0.0111	0.9749	0.9877	0.9813	0.9968
0.0157	0.9556	2000	0.0089	0.9774	0.9886	0.9830	0.9970
0.0126	1.1945	2500	0.0072	0.9835	0.9896	0.9866	0.9978
0.0098	1.4333	3000	0.0064	0.9854	0.9920	0.9886	0.9980
0.0097	1.6722	3500	0.0053	0.9877	0.9936	0.9906	0.9983
0.0088	1.9111	4000	0.0056	0.9868	0.9916	0.9892	0.9982
0.0057	2.1500	4500	0.0057	0.9882	0.9921	0.9901	0.9983
0.0049	2.3889	5000	0.0055	0.9887	0.9936	0.9911	0.9984
0.0057	2.6278	5500	0.0052	0.9898	0.9946	0.9922	0.9985
0.0050	2.8667	6000	0.0052	0.9888	0.9932	0.9910	0.9984
0.0018	3.1056	6500	0.0051	0.9910	0.9946	0.9928	0.9986
0.0025	3.3445	7000	0.0051	0.9903	0.9950	0.9927	0.9986
0.0018	3.5834	7500	0.0053	0.9912	0.9939	0.9925	0.9986
0.0020	3.8223	8000	0.0049	0.9906	0.9940	0.9923	0.9986

Framework versions

• Transformers 5.11.0
• Pytorch 2.12.0+cu130
• Datasets 5.0.0
• Tokenizers 0.22.2