😒 disappointing
Very disappointed. Being a total fan of this model. This new version although faster is very disappointing.
I asked the same history questions with the old version and he answered exactly, but with this one he was significantly wrong. It looks like something else apart from the name
Are you using the correct format and everything?
Honestly had zero issue up to 12k context.
Also, if you're using gguf or other quants, stuff is broken there. PoSE doesn't play well with quants.
HI SAO from your fan...
i just finished my college and about to work, i will support you when i finally have my own money.
is there any plan to make qwen based model? from my experience it's crazy good. even magnum perform really good, i believe you can make even more good quality model.
and is there any plan to make MoE, or maybe MoA, i dont really catch the news about how to structure an MoA
i really like your model, i can't thank you enough knowing that you got some busy work to do.
Are you using the correct format and everything?
Honestly had zero issue up to 12k context.
Also, if you're using gguf or other quants, stuff is broken there. PoSE doesn't play well with quants.
yes of course I converted to Guff but if that's the problem then I'll give up. Hope you fix it because the Base Model is a bomb as you already know
I have no idea how to convert to GGUF. Has a button "use this model" in Backyard.ai but Backyard needs gguf.