3 ms·
It is a reward function it’s just a deterministic one. Reward models are often hacked preventing real reasoning from being discovered
by mountainriver 2y ago
It is a reward function it’s just a deterministic one. Reward models are often hacked preventing real reasoning from being discovered