Not giving me a proper answers

[ilyaxuwu]

I've uncensored the AI, when I test it doesn't give me a proper answer, like I've wrote give me a 2000 words story and he wrote the story insdie python code do you know why?

like this

[erm14254]

Which model is it? What is the KL divergence? What are the amount of refusals? Which abliteration method did you use?

[ilyaxuwu]

It is qwen 2.5 7b parametres,
idk,
it was like the lowest one i used i remember
4 bit i guess since I had rtx 3060 with vram 12gb

[p-e-w]

There is no actual information here, and no indication of a problem in Heretic. Closing as non-actionable.