antsnormflows.utils.optim

Functions

clear_grad(model)	Set gradients of model parameter to None as this speeds up training,
get_requires_grad_states(module)	Snapshots the current requires_grad flag of every parameter
restore_requires_grad(module, states)	Restores requires_grad flags previously captured with get_requires_grad_states
set_requires_grad(module, flag)	Sets requires_grad flag of all parameters of a torch.nn.module
update_lipschitz(model, n_iterations)