Still 671B moe?

#1
by eric8810 - opened

any changes on the architechture?

i don't think so

any changes on the architechture?

No, its 3.1 continuing of its predecessor.

685B
图片.png

685B是包含MTP参数的,你看R1、V3也都是显示685B的

685B
图片.png

It’s the same as the previous model
Probably just an issue with how they count parameters

I believe that the management will look up to it

It’s the same as the previous model

You're right. I was hoping to squeeze a slightly larger quant into my system this time.

Sign up or log in to comment