Remix.run Logo
denn-gubsky 12 hours ago

I like the Claude models politeness the most. They a really trained to be patient and polite, especially the Opus line. They are also seem to adapt to the operator's conversation style, so I always start conversation from "Please". I know that this increases token load, but model seem to value it.

zeristor 11 hours ago | parent [-]

Did you mean to this for this article?

Perhaps it was for another one?

denn-gubsky 10 hours ago | parent [-]

Sorry, this is wrong page. Comment was to the Claude AI reinforcement learning models training article. Wrong click. sorry again.