3 ms·
Or just rerun that preference learning approach, but deliberately, with the filters as the reward function and the reward being for being filtered.
by gwern 5y ago
Or just rerun that preference learning approach, but deliberately, with the filters as the reward function and the reward being for being filtered.