torch.optim.functional.radam

torch.optim.functional.radam(params, grads, exp_avgs, exp_avg_sqs, state_steps, decoupled_weight_decay=False, foreach=None, differentiable=False, capturable=False, has_complex=False, maximize=False, *, beta1, beta2, lr, weight_decay, eps)

Functional API that performs RAdam algorithm computation.

This function updates the provided parameters and optimizer state in place. The caller must initialize and retain optimizer state. Unless intentionally constructing a differentiable update with a supported differentiable=True argument, call this function under torch.no_grad. See Functional optimizer API for the common functional optimizer contract and examples, and RAdam for algorithm details.