logoalt Hacker News

alansaber • today at 10:05 AM • 0 replies • view on HN

Given how LLMs access compressed knowledge from their model weights, the similarities make sense