David Udell comments on David Udell’s Shortform

[David Udell]

You can think of chain-of-thought interpretability as the combination of process-based methods with adversarial training.