Rizzo PII 0.3B โ€” validated ONNX export

Runtime files used by the Local Privacy Runtime, a local gateway that pseudonymizes personal data before text reaches an external AI provider and restores it locally afterwards.

File SHA-256
config.json 6b9d5bdd4a194d40fb40c53fdc302e225514cd7940d97f657c4082454ee7e91b
tokenizer.json a0ab5772a8c5450f45a66c95985d6df2205bf09449ba7ff7bb46de034eaa118a
model.onnx 32e403d1607f05acc2e1c2582418dbeec464c50ce7362eeacc9ea9dadce1dd3c

Origin

  • Source model: rizzoaiacademy/rizzo-pii-0.3B, revision d4f989333bcda1fcb188785a12937b0f52b52580, by Simone Rizzo โ€” Rizzo AI Academy.
  • config.json, tokenizer.json: unchanged from that revision.
  • model.onnx: FP32 ONNX export (opset 18; inputs input_ids, attention_mask; output logits [batch, sequence, 45]) produced and validated by the Local Privacy Runtime project: PyTorch parity within atol=5e-5 and identical BIO labels on a synthetic corpus. It is not an official artifact of the model author.
  • Licence: MIT, inherited from the source model (see LICENSE).

Use

The runtime downloads these files with local-privacy-cli model install and accepts each one only if its SHA-256 matches the values compiled into its binaries. Do not trust the files because of where they come from: check the checksums above.

A mirror of the same files is available as the generic package rizzo-pii-0.3B/d4f9893 on git.a75.it (andrearuffini/ai-privacy-models).

Downloads last month
15
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for andrearuffini/local-privacy-runtime-pii-detector

Quantized
(2)
this model