modernbert-pii-ner

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: 0.1431
• Precision: 0.9096
• Recall: 0.9466
• F1: 0.9277
• Accuracy: 0.9683

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: 0.2
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
0.4505	0.4087	500	0.2343	0.6446	0.7708	0.7020	0.9337
0.2362	0.8173	1000	0.1324	0.7922	0.8659	0.8274	0.9299
0.1772	1.2256	1500	0.0934	0.8931	0.9199	0.9063	0.9640
0.1567	1.6342	2000	0.0807	0.8926	0.9343	0.9130	0.9653
0.1416	2.0425	2500	0.0772	0.8626	0.9288	0.8945	0.9570
0.1197	2.4512	3000	0.0763	0.8855	0.9370	0.9105	0.9643
0.1091	2.8598	3500	0.0731	0.9072	0.9473	0.9268	0.9688
0.0868	3.2681	4000	0.0774	0.9155	0.9486	0.9317	0.9688
0.0681	3.6767	4500	0.0880	0.9098	0.9483	0.9286	0.9684
0.0263	4.0850	5000	0.1211	0.9102	0.9460	0.9278	0.9674
0.0187	4.4937	5500	0.1431	0.9096	0.9466	0.9277	0.9683

Framework versions

• Transformers 5.7.0
• Pytorch 2.11.0+cu130
• Datasets 4.8.5
• Tokenizers 0.22.2