4 ms·
Claude works differently than just RLHF. "Since launching Claude, our AI assistant trained with Constitutional AI, we've heard more questions about Constitutio
by JieJie 3y ago
Claude works differently than just RLHF.
"Since launching Claude, our AI assistant trained with Constitutional AI, we've heard more questions about Constitutional AI and how it contributes to making Claude safer and more helpful. In this post, we explain what constitutional AI is, what the values in Claude’s constitution are, and how we chose them."
https://www.anthropic.com/index/claudes-constitution https://www.anthropic.com/index/claudes-constitution
- ethav1 3y agoIt works by self-generating responses to red-team prompts and self-generating safe corrections to those then using RLHF with the corrections. It isn’t a major departure from traditional RLHF so it is interesting that it performs so much better in this case.
- kmeisthax 3y agoThis sounds like reward modeling combined with RLHF.