Deepseek Unsloth's 1.58-bit gguf support?

[maximeozenne]

Hello,

First of all, thank you for your impressive work!

I'm actively following the work of Unsloth about the dynamic quantification they made on Deepseek-r1:

https://unsloth.ai/blog/deepseekr1-dynamic

But even with KTransformers, my current hardware does not allow me to run a Q2 sized locally, but I have enough RAM to run the Q1s.

I was wondering if you plan to support the Q1s Unsloth's versions of Deepseek (1.58-bit, 1.73-bit)?

Your work is a massive enhancement in the LLM world, thank you again for it!