dnhkng

/

RYS-XLarge

like 73

Thanks! Could you do the same for any other RYS models you have prepped?

I have another method that is what will be in the paper, and these models were released a bit early. The next models will be better, and I don't want to have huge amount of suboptimal models out there :)

all other RYS? sture thing

speaking of, because your Gemma introduces new layers, llama.cpp doesn't recognize it properly, so will need to either add a fix upstream or remove the extra layers

makes sense! yeah, this was prematurely released. I didn't expect it to spread so fast 😥

Some models are not optimal yet, and others are broken.