polygraf-ai/pii-small

pii_mini

This model is a fine-tuned version of google/bert_uncased_L-8_H-512_A-8 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.1113
• Precision: 0.9001
• Recall: 0.9290
• F1: 0.9143
• Accuracy: 0.9645

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: 3e-05
• train_batch_size: 256
• eval_batch_size: 256
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	1.0	153	0.3797	0.4943	0.5523	0.5217	0.9024
No log	2.0	306	0.1868	0.7281	0.7917	0.7586	0.9419
No log	3.0	459	0.1319	0.8339	0.8735	0.8532	0.9565
0.5069	4.0	612	0.1098	0.8690	0.8990	0.8837	0.9603
0.5069	5.0	765	0.0971	0.8725	0.9082	0.8900	0.9647
0.5069	6.0	918	0.0924	0.8887	0.9179	0.9031	0.9653
0.1032	7.0	1071	0.0920	0.8820	0.9175	0.8994	0.9632
0.1032	8.0	1224	0.0869	0.8886	0.9219	0.9050	0.9652
0.1032	9.0	1377	0.0912	0.8917	0.9235	0.9073	0.9649
0.0719	10.0	1530	0.0875	0.8995	0.9271	0.9131	0.9666
0.0719	11.0	1683	0.0964	0.8971	0.9264	0.9115	0.9649
0.0719	12.0	1836	0.1006	0.9030	0.9293	0.9159	0.9656
0.0719	13.0	1989	0.1011	0.8978	0.9291	0.9132	0.9639
0.0539	14.0	2142	0.1071	0.9007	0.9275	0.9139	0.9628
0.0539	15.0	2295	0.1113	0.9001	0.9290	0.9143	0.9645

Framework versions

• Transformers 4.37.2
• Pytorch 2.2.0+cu121
• Datasets 2.17.1
• Tokenizers 0.15.1