variance_reduced_neural_networks icon indicating copy to clipboard operation
variance_reduced_neural_networks copied to clipboard

Published 20 hours ago • verified publisher icon kilianFatras
→

Metadata

Implementation of SVRG and SAGA optimization algorithms for deep learning topics.

Results 2 variance_reduced_neural_networks issues
Sort by recently updated
recently updated
newest added

Why is the norm of the grad not used at all in the optimisation process of SAGA?

Where exactly is the equation (3) in the main paper implemented in the SAGA algorithm?

rohan1561 avatar rohan1561

FIX torch.data[0]=>torch.item()

self.partial_grad(inputs, labels, loss_function).data[0] causes error. So I changed it to self.partial_grad(inputs, labels, loss_function).item()

limaries30 avatar limaries30

← Metadata

56
Stars
19
Forks
Watchers

Owner

verified publisher icon kilianFatras

Metadata

Implementation of SVRG and SAGA optimization algorithms for deep learning topics.

Back