@dehan_jl@GoodfireAI I had the same question in regards to super position. But, if the neurons are across layers it is different than super position. A column of an SAE decoder gives a distributed representation for a set of neurons within one layer
@MrSixteenth1@seanapriori DURRRR
UHHHHH
THIS SPECIFIC THING I HAVE EXPERIENCED SO IM SURE I CAN EXTRAPOLATE IT TO A GENERAL TREND
DURRRRRR LOOK AT ME IM WETARDED