2 ms·
Same thing. You need everything to be differentiable in order to run gradient descent. To first approximation, training a neural network is just gradient descen
by kccqzy 3y ago
Same thing. You need everything to be differentiable in order to run gradient descent. To first approximation, training a neural network is just gradient descent.