Teaching Models to Dream of Better Monitors through Evaluation Conditioned Training
by Alec Harris, Kasey C, Archie Chaudhury, and yix
Edit: we renamed the technique from Monitor Sensitive Training (MST) to Evaluation Conditioned Training (ECT)[1] TL;DR * We introduce Evaluation Conditioned Training (ECT), a new post-training technique where we augment training data with evaluation labels that describe how evaluation is going to be applied for each sample. We then change...
Mar 1945