What is Nostr?
Alan Levine /
npub1ctj…dzdx
2024-01-24 02:44:12

Alan Levine on Nostr: Check my math. I read that the training set for ChatGPT was 570Gb. And estimates of ...

Check my math.

I read that the training set for ChatGPT was 570Gb. And estimates of the amount of data on the internet is 147 Zettabytes (my new favorite number) or 147000000000Gb

Therefore this grand wizard of statistically generated intelligence has been trained on .0000000387755102% of the Internet.

I guess they used just the best parts.
Author Public Key
npub1ctjy2x4f83gnva7kn53snzm7m46m46azpf6q9cth3gdj9wap7zuqvsdzdx