TurboQuant claims 3-bit quantization with "zero accuracy loss" for LLMs. Is this the holy grail for local AI, or is an engineer's skepticism about this "magic" solution warranted?

thepixelspulse.com/posts/turboquant-3-bit-q...

#turboquant #polarquant #qjl

Oops killed him to death
#qiujianluo #qjl

A cute Qiu Jianluo in progress... The eyes upon his back are always watching.. He will rip you to shreds, if you make the wrong move.

#qiujianluo #qjl