The promptfans reckon you can fix moderation in post

The Mistral 7B Instruct model is a quick demonstration that the base model can be easily fine-tuned to achieve compelling performance. It does not have any moderation mechanism. We’re looking forward to engaging with the community on ways to make the model finely respect guardrails, allowing for deployment in environments requiring moderated outputs.

“Whoops, it’s done now, oh well, guess we’ll have to do it later”

Go fucking directly to jail