xXiaobuding's picture
Update README.md
76c72e1 verified
metadata
license: apache-2.0
tags:
  - generated_from_trainer
model-index:
  - name: bert-base-uncased_ai4privacy_en
    results: []

bert-base-uncased_ai4privacy_en

This model is a fine-tuned version of google-bert/bert-base-uncased on the English subset of ai4privacy/pii-masking-200k dataset. It achieves the following results on the evaluation set:

  • Loss: 0.0702
  • Overall Precision: 0.9259
  • Overall Recall: 0.9464
  • Overall F1: 0.9360
  • Overall Accuracy: 0.9706
  • Accountname F1: 0.9924
  • Accountnumber F1: 0.9905
  • Age F1: 0.9339
  • Amount F1: 0.9278
  • Bic F1: 0.9598
  • Bitcoinaddress F1: 0.9801
  • Buildingnumber F1: 0.9091
  • City F1: 0.9564
  • Companyname F1: 0.9908
  • County F1: 0.9853
  • Creditcardcvv F1: 0.9639
  • Creditcardissuer F1: 0.9868
  • Creditcardnumber F1: 0.8929
  • Currency F1: 0.7726
  • Currencycode F1: 0.8608
  • Currencyname F1: 0.3650
  • Currencysymbol F1: 0.9536
  • Date F1: 0.8590
  • Dob F1: 0.6490
  • Email F1: 0.9945
  • Ethereumaddress F1: 0.9986
  • Eyecolor F1: 0.9688
  • Firstname F1: 0.9790
  • Gender F1: 0.9832
  • Height F1: 0.9906
  • Iban F1: 1.0
  • Ip F1: 0.1025
  • Ipv4 F1: 0.8217
  • Ipv6 F1: 0.7506
  • Jobarea F1: 0.9306
  • Jobtitle F1: 0.9938
  • Jobtype F1: 0.9508
  • Lastname F1: 0.9480
  • Litecoinaddress F1: 0.9345
  • Mac F1: 1.0
  • Maskednumber F1: 0.8609
  • Middlename F1: 0.9601
  • Nearbygpscoordinate F1: 1.0
  • Ordinaldirection F1: 0.9784
  • Password F1: 0.9839
  • Phoneimei F1: 0.9986
  • Phonenumber F1: 0.9903
  • Pin F1: 0.9390
  • Prefix F1: 0.9441
  • Secondaryaddress F1: 0.9945
  • Sex F1: 0.9780
  • Ssn F1: 0.9898
  • State F1: 0.9805
  • Street F1: 0.9693
  • Time F1: 0.9843
  • Url F1: 0.9984
  • Useragent F1: 0.9918
  • Username F1: 0.9909
  • Vehiclevin F1: 0.9856
  • Vehiclevrm F1: 0.9653
  • Zipcode F1: 0.8990

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: 8
  • eval_batch_size: 16
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: cosine_with_restarts
  • lr_scheduler_warmup_ratio: 0.2
  • num_epochs: 2

Training results

Training Loss Epoch Step Validation Loss Overall Precision Overall Recall Overall F1 Overall Accuracy Accountname F1 Accountnumber F1 Age F1 Amount F1 Bic F1 Bitcoinaddress F1 Buildingnumber F1 City F1 Companyname F1 County F1 Creditcardcvv F1 Creditcardissuer F1 Creditcardnumber F1 Currency F1 Currencycode F1 Currencyname F1 Currencysymbol F1 Date F1 Dob F1 Email F1 Ethereumaddress F1 Eyecolor F1 Firstname F1 Gender F1 Height F1 Iban F1 Ip F1 Ipv4 F1 Ipv6 F1 Jobarea F1 Jobtitle F1 Jobtype F1 Lastname F1 Litecoinaddress F1 Mac F1 Maskednumber F1 Middlename F1 Nearbygpscoordinate F1 Ordinaldirection F1 Password F1 Phoneimei F1 Phonenumber F1 Pin F1 Prefix F1 Secondaryaddress F1 Sex F1 Ssn F1 State F1 Street F1 Time F1 Url F1 Useragent F1 Username F1 Vehiclevin F1 Vehiclevrm F1 Zipcode F1
0.0939 1.0 4350 0.0801 0.8951 0.9309 0.9126 0.9666 0.9840 0.9896 0.9182 0.8769 0.9127 0.9627 0.8770 0.9616 0.9847 0.9712 0.9373 0.9813 0.8406 0.3934 0.7451 0.1372 0.9266 0.8354 0.5796 0.9920 0.9877 0.9037 0.9642 0.9789 0.9906 0.9874 0.0 0.8416 0.8087 0.8854 0.9825 0.9426 0.9213 0.9015 0.9806 0.7978 0.9543 1.0 0.9828 0.9689 0.9917 0.9777 0.8764 0.9340 0.9913 0.9761 0.9949 0.9553 0.9561 0.9723 0.9921 0.9906 0.9779 0.9942 0.9684 0.8522
0.0644 2.0 8700 0.0702 0.9259 0.9464 0.9360 0.9706 0.9924 0.9905 0.9339 0.9278 0.9598 0.9801 0.9091 0.9564 0.9908 0.9853 0.9639 0.9868 0.8929 0.7726 0.8608 0.3650 0.9536 0.8590 0.6490 0.9945 0.9986 0.9688 0.9790 0.9832 0.9906 1.0 0.1025 0.8217 0.7506 0.9306 0.9938 0.9508 0.9480 0.9345 1.0 0.8609 0.9601 1.0 0.9784 0.9839 0.9986 0.9903 0.9390 0.9441 0.9945 0.9780 0.9898 0.9805 0.9693 0.9843 0.9984 0.9918 0.9909 0.9856 0.9653 0.8990

Framework versions

  • Transformers 4.26.1
  • Pytorch 2.0.0.post101
  • Datasets 2.10.1
  • Tokenizers 0.13.3