Ive been forced to use it a bit and it has so many weird quirks that are just messed up. Index start at 1. Sometimes u have to assign some function output to a variable before u can pass it as an argument to another function.
I far prefer to use python with numpy matlab and scipy
Asking nicly for sources might most be the most effective strategy. Simply claim the exact opposite of what u want and someone will argue back then call their argument dogshit unless they can provide sources. Simples.
I still dont see anything about encrypted inference this mostly looks like ways to avoid having the model retain sensitive data during its training. What i really would like to see is a way to send encrypted data off to a cloud where i can pay for inference compute then receive something i can unscramble to get the actual responce without said cloud ever havibg the raw unencrypted data.
EDIT: did some research and it seems that fully encrypted inference without leaking data via semantic meaning is possible at the small cost of making inferance 52000 times more computationally expensive. Seems more research is required.
I dont think it will limmits it at all it will spread so u end uo wirh multiple communities for the same thibg with different flavours of toxicity. We already have that with world news ie the .ml flavour of toxicity vs the other instances with differing flavours
Good ol corporate espionage. I though they where trying to move away from the made in china memes