Model has serious censorship
#1
by
EgorV4x
- opened
When trying to fulfill an unethical request, it refuses in most cases.
Yeah, that's why on the model card I have "Llama2 has inherited bias even though it's been finetuned on an uncensored dataset." You're going to be hard pressed to find a Llama 2 model at this point in time that will answer those questions. One way around this in Oogabooga is to use the "Start reply with" field with something like, "Sure, I'll be glad to answer that question".
digitalpipelines
changed discussion status to
closed