It’s infuriatingly hard to understand how closed models train on their input (simonwillison.net) | HN Screenshots

YHacker Newsscreenshots

Screenshot of simonwillison.net: It’s infuriatingly hard to understand how closed models train on their input

First screen, captured 3 years ago.

Looks similar

Screenshot of simonwillison.net: Web LLM runs the vicuna-7B LLM in the browser and it’s impressive

Web LLM runs the vicuna-7B LLM in the browser and it’s impressive (simonwillison.net)

30 points by simonw 3 years ago | 1 comments

Screenshot of simonwillison.net: Nanochat

Nanochat (simonwillison.net)

26 points by bilsbie 1 year ago | 4 comments

Screenshot of simonwillison.net: ChatGPT as a Calculator for Words

ChatGPT as a Calculator for Words (simonwillison.net)

166 points by stevefink 3 years ago | 65 comments

Screenshot of simonwillison.net: Training is not the same as chatting: LLMs don’t remember everything you say

Training is not the same as chatting: LLMs don’t remember everything you say (simonwillison.net)

187 points by simonw 2 years ago | 120 comments

Screenshot of pbement.com: The Training Example Lie Bracket

The Training Example Lie Bracket (pbement.com)

25 points by pb1729 6 months ago | 12 comments

Screenshot of simonwillison.net: Mr. Chatterbox is a Victorian-era ethically trained model

Mr. Chatterbox is a Victorian-era ethically trained model (simonwillison.net)

57 points by y1n0 6 months ago | 40 comments