Describe the current behavior
While whatever is returned right now works for most situtaion but there are situations where it could be a litting confusing. for example
- whenever you chose rephrase as on fail action , this is what you will get -
"safe_text": "Please rephrase the query without unsafe content. {the reason of asking for rephrase which will differ per validator}"
- profanity free validator is not specifically an output validator, yet when we run the validator with rephrase as on fail action -
"safe_text": "Please rephrase the query without unsafe content. [REDACTED] contains profanity. Please return profanity-free output."
Describe the enhancement you'd like
-
Since all our validator don't just block unsafe content and block irrelevant content as well, so "without unsafe content" should be removed.
-
rephrase the sentence "please return profanity free output" as this shows that you expect profanity free to be an output validator
Describe the current behavior
While whatever is returned right now works for most situtaion but there are situations where it could be a litting confusing. for example
Describe the enhancement you'd like
Since all our validator don't just block unsafe content and block irrelevant content as well, so "without unsafe content" should be removed.
rephrase the sentence "please return profanity free output" as this shows that you expect profanity free to be an output validator