Fine tuning gguf models?

Is possible to fine tune gguf models (Q4, Q8, Q5 …) models, just like PeftModel?

have you got any idea