Possible to re-quantize Trendyol/Trendyol-Cybersecurity-LLM-Qwen3-32B-Q8_0-GGUF ?

#1127
by ykarout - opened

The model seems really useful for cybersecurity use-cases... I couldnt find the bf16/fp16 versions, is it possible to re-quantize from Q8_0 or is there a big accuracy loss that overcomes the size loss?

It's certainly possible, just not standard. I would guess that the accuracy loss will be limited, i.e. the Q4_K_M will be worse when done from Q8_0, but probably not in a noticable way, since the Q8_0 is already so good.

I will give it a try, once I find time.

Well, break a leg.

You can check for progress at http://hf.tst.eu/status.html or regularly check the model
summary page at https://hf.tst.eu/model#Trendyol-Cybersecurity-LLM-Qwen3-32B-Q8_0-GGUF-GGUF for quants to appear.

you rock buddy

Sign up or log in to comment