3 ms·
It seems like you don’t understand reinforcement learning. The signal is reinforced because it correlates to behavior, hacking the signal itself is misalignment
by catigula 10mo ago
It seems like you don’t understand reinforcement learning. The signal is reinforced because it correlates to behavior, hacking the signal itself is misalignment.