Lennart Finke
Titles as the sources gave them. In 1 entry, with 0 checked quotations.
In the Base
Every entry they appear in.
Towards Robust Model Organisms for Eliciting Latent Knowledge
According to the abstract, Lennart Finke, Hardik Bhatnagar, Euodia Dodd and colleagues train model organisms that withhold solutions in a vulnerability detection task, as a testbed for extracting knowledge from a misaligned model. They report their strongest model is substantially more robust than earlier ones.