I started this project just to learn how to train and fine-tune models. I published my first version distilbert-secret-masker which currently averages 228 downloads per month. Some security company used it in a advertisement benchmark for their product, which it got the lowest score, but they called it one of the "five leading open-source scanners". So through spite I updated the model and should now be considered the #1 open-source secret detector ;) (according to my benchmarks)
Show HN: SecMask – a 66M parameter model for finding secrets in source code
github.com