privacy-200k-masking

This model is a fine-tuned version of distilbert-base-multilingual-cased on an unknown dataset. It achieves the following results on the evaluation set:

• eval_loss: 0.0949
• eval_overall_precision: 0.9099
• eval_overall_recall: 0.9306
• eval_overall_f1: 0.9201
• eval_overall_accuracy: 0.9692
• eval_ACCOUNTNAME_f1: 0.9863
• eval_ACCOUNTNUMBER_f1: 0.9551
• eval_AGE_f1: 0.9454
• eval_AMOUNT_f1: 0.9481
• eval_BIC_f1: 0.9140
• eval_BITCOINADDRESS_f1: 0.9227
• eval_BUILDINGNUMBER_f1: 0.9056
• eval_CITY_f1: 0.9351
• eval_COMPANYNAME_f1: 0.9621
• eval_COUNTY_f1: 0.9756
• eval_CREDITCARDCVV_f1: 0.9201
• eval_CREDITCARDISSUER_f1: 0.9767
• eval_CREDITCARDNUMBER_f1: 0.8506
• eval_CURRENCY_f1: 0.7277
• eval_CURRENCYCODE_f1: 0.8398
• eval_CURRENCYNAME_f1: 0.1576
• eval_CURRENCYSYMBOL_f1: 0.9216
• eval_DATE_f1: 0.7988
• eval_DOB_f1: 0.6103
• eval_EMAIL_f1: 0.9862
• eval_ETHEREUMADDRESS_f1: 0.9624
• eval_EYECOLOR_f1: 0.9779
• eval_FIRSTNAME_f1: 0.9636
• eval_GENDER_f1: 0.9852
• eval_HEIGHT_f1: 0.9771
• eval_IBAN_f1: 0.9513
• eval_IP_f1: 0.0
• eval_IPV4_f1: 0.8240
• eval_IPV6_f1: 0.7389
• eval_JOBAREA_f1: 0.9713
• eval_JOBTITLE_f1: 0.9819
• eval_JOBTYPE_f1: 0.9743
• eval_LASTNAME_f1: 0.9439
• eval_LITECOINADDRESS_f1: 0.8069
• eval_MAC_f1: 0.9668
• eval_MASKEDNUMBER_f1: 0.8084
• eval_MIDDLENAME_f1: 0.9401
• eval_NEARBYGPSCOORDINATE_f1: 0.9963
• eval_ORDINALDIRECTION_f1: 0.9904
• eval_PASSWORD_f1: 0.9690
• eval_PHONEIMEI_f1: 0.9842
• eval_PHONENUMBER_f1: 0.9690
• eval_PIN_f1: 0.8584
• eval_PREFIX_f1: 0.9594
• eval_SECONDARYADDRESS_f1: 0.9880
• eval_SEX_f1: 0.9952
• eval_SSN_f1: 0.9813
• eval_STATE_f1: 0.9664
• eval_STREET_f1: 0.9607
• eval_TIME_f1: 0.9560
• eval_URL_f1: 0.9866
• eval_USERAGENT_f1: 0.9901
• eval_USERNAME_f1: 0.9743
• eval_VEHICLEVIN_f1: 0.9699
• eval_VEHICLEVRM_f1: 0.9725
• eval_ZIPCODE_f1: 0.9018
• eval_runtime: 3609.2787
• eval_samples_per_second: 17.394
• eval_steps_per_second: 8.697
• epoch: 1.0
• step: 73241

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: 2
• eval_batch_size: 2
• 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

Framework versions

• Transformers 4.40.0
• Pytorch 2.2.1+cu121
• Datasets 2.19.0
• Tokenizers 0.19.1