This model card mentions merging huihui-ai abliterated v2 model of Qwen 2.5, but when prompting this model with replies that may be harmful, it will reject with canned censorship reply.
· Sign up or log in to comment
