pii-layout-synth-v9

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: 0.9896
• Recall: 0.9929
• F1: 0.9912
• Accuracy: 0.9985

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: 200
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
0.0463	0.2455	500	0.0209	0.9469	0.9744	0.9605	0.9938
0.0246	0.4909	1000	0.0105	0.9749	0.9848	0.9798	0.9967
0.0191	0.7364	1500	0.0087	0.9817	0.9913	0.9865	0.9974
0.0144	0.9818	2000	0.0077	0.9785	0.9846	0.9815	0.9973
0.0093	1.2273	2500	0.0055	0.9878	0.9922	0.9900	0.9982
0.0090	1.4728	3000	0.0056	0.9875	0.9926	0.9900	0.9981
0.0080	1.7182	3500	0.0059	0.9870	0.9925	0.9897	0.9981
0.0089	1.9637	4000	0.0051	0.9890	0.9944	0.9917	0.9984
0.0045	2.2091	4500	0.0063	0.9844	0.9906	0.9875	0.9980
0.0062	2.4546	5000	0.0073	0.9870	0.9897	0.9884	0.9980
0.0035	2.7000	5500	0.0049	0.9896	0.9929	0.9912	0.9985

Framework versions

• Transformers 5.12.1
• Pytorch 2.12.1+cu130
• Datasets 5.0.0
• Tokenizers 0.22.2