The Guardian has a neat visual explainer of the approach behind chat-GPT or bard. It’s far from complete, but it’s the best I’ve seen so far at explaining why the large language models don’t simply copy the vast amounts of text they have processed but model it.
7 Likes
That is really good. I also liked Cal Newport’s description as well.
2 Likes
Here is an explanation I found to be quite good.
Enjoy!
1 Like