https://huggingface.co/Jackrong/Qwopus3.6-27B-v2-MTP-GGUF

#2544
by MK2023 - opened

https://huggingface.co/Jackrong/Qwopus3.6-27B-v2-MTP-GGUF would be really lovely.

I found your profile via https://huggingface.co/cHunter789/Qwen3.6-27B-i1-IQ4_XS-GGUF and noticed he mentions a regression that was recently fixed and now the models are smaller. e.g. his i1_IQ4_XS version is only 14.7 GB!

Do you think it is possible to squeeze Qwopus to the same size?

depends, give me a link and I try, because what you sent... it's already in gguf format

Sorry, my bad, i thought you can use the BF16 there? because the safetensors here don't have MTP if i see this correctly: https://huggingface.co/Jackrong/Qwopus3.6-27B-v2
I a matter of fact you did that repo ~3 weeks back: https://huggingface.co/mradermacher/Qwopus3.6-27B-v2-i1-GGUF but the files there are as well bigger then the cHunter ones.
But perhaps i misunderstood something, then i apologize.

MK2023 changed discussion status to closed

Sign up or log in to comment