X · @teortaxesTex
· X / Twitter
My only worry is that DeepSeek itself has NIH syndrome after so much of their work became foundational to this era They adopt third party tricks, but …
My only worry is that DeepSeek itself has NIH syndrome after so much of their work became foundational to this eraThey adopt third party tricks, but I am not sure they can swallow eg AttnRes being straight up better than mHC, and may waste time on minor elaborations over it.mark: @teortaxesTex Jianlin Su's word should