pii-layout-synth-v2

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: 0.0055
• Precision: 0.9870
• Recall: 0.9911
• F1: 0.9890
• Accuracy: 0.9982

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
0.0806	0.2489	500	0.0406	0.8933	0.9476	0.9196	0.9878
0.0344	0.4978	1000	0.0187	0.9575	0.9744	0.9659	0.9945
0.0197	0.7466	1500	0.0093	0.9792	0.9882	0.9837	0.9972
0.0154	0.9955	2000	0.0113	0.9750	0.9874	0.9812	0.9965
0.0125	1.2444	2500	0.0071	0.9831	0.9887	0.9859	0.9977
0.0104	1.4933	3000	0.0068	0.9841	0.9910	0.9875	0.9979
0.0108	1.7422	3500	0.0065	0.9849	0.9907	0.9878	0.9979
0.0124	1.9910	4000	0.0058	0.9863	0.9923	0.9893	0.9981
0.0050	2.2399	4500	0.0055	0.9879	0.9920	0.9900	0.9983
0.0047	2.4888	5000	0.0055	0.9872	0.9926	0.9899	0.9983
0.0050	2.7377	5500	0.0058	0.9865	0.9926	0.9895	0.9982
0.0062	2.9866	6000	0.0055	0.9870	0.9911	0.9890	0.9982

Framework versions

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