wagesj45@kbin.runtoTechnology@lemmy.ml•OpenAI transcribed over a million hours of YouTube videos to train GPT-4
2·
7 months agoProbably, but that assumes that the transcribers went from video to video following the algorithm. I’d suspect that they would randomize the videos they chose somehow figured out some other distribution.
But that is just a guess, you could be right.
It is much easier to buy one “hefty” physical machine and run ProxMox with virtual machines for servers than it is to run multiple Raspberry Pis. After living that life for years, I’m a ProxMox shill now. Backups are important (read the other comments), and ProxMox makes backup/restore easy. Because eventually you will fuck a server up beyond repair, you will lose data, and you will feel terrible about it. Learn from my mistakes.