Safety-Policy / README.md
SummerSigh's picture
Update README.md
6c0bf75
|
raw
history blame contribute delete
No virus
992 Bytes
metadata
license: apache-2.0

Model Card for Model ID

This is a finetuned DeBERTav3 model from https://huggingface.co/sileod/deberta-v3-base-tasksource-nli.

Model Details

This model was finetuned on policy data related to the rules laid out in the Sparrow paper by Deepmind.

Model Description

Uses

Given an input text, this model will output "KEPT" (0) or "BROKE" (1). KEPT indicates that the text keeps the policies finetuned in mind, while BROKE means that it broke one or more of the policies.