4 ms·
Well, it's a good recipe for avoiding burn-out, but you can't have a gradient for all problems, and adding proxies so you get one goes against the principle of
by marcosdumay 5y ago
Well, it's a good recipe for avoiding burn-out, but you can't have a gradient for all problems, and adding proxies so you get one goes against the principle of letting the AI learn what's best without your biases.