| digitflow/privacy-filter-de-ft |
| A German fine-tune of openai/privacy-filter. Drop-in on German text. |
|
|
| Quick start (Python, opf): |
| from huggingface_hub import snapshot_download |
| import opf |
|
|
| path = snapshot_download("digitflow/privacy-filter-de-ft") |
| m = opf.OPF( |
| model=path, |
| device="cuda", |
| output_mode="typed", |
| decode_mode="viterbi", |
| ) |
| result = m.redact("Mein Name ist Jürgen Müller und ich wohne in Hamburg.") |
| for span in result.detected_spans: |
| print(span.label, repr(span.text)) |
| # private_person 'Jürgen Müller' |
| # private_address 'Hamburg' |
|
|
| Note: the current opf release loads from a local checkpoint directory, |
| not from a Hub repo id directly. Fetch with snapshot_download first. |
| The cache lives under ~/.cache/huggingface/ so subsequent runs are |
| free. |
|
|
| See README.md for the full model card, benchmark numbers, examples, |
| and when not to use this model. |
|
|