Hacker News new | ask | show | jobs
by rahimnathwani 657 days ago
Exactly. Section 4.3.7 briefly explains how they trained the model to better follow instructions ('steerability').