Sampling Effects on Strategic Behavior in Supervised Learning Models 2024-09-24T07:44:41.677Z

Comment by Phil Bland on How LLMs are and are not myopic · 2024-02-09T10:47:08.495Z · LW · GW

Do you have any update on this? It goes strongly against my current understanding of how LLMs learn. In particular, in the supervised learning phase any output text claiming to be an LLM would be penalized unless such statements are included in the training corpus. If such behavior nevertheless arises I would be super excited to analyze this further though.