Dassl.pytorch/dassl/optim/radam.py [50:87]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
    def step(self, closure=None):

        loss = None
        if closure is not None:
            loss = closure()

        for group in self.param_groups:

            for p in group["params"]:
                if p.grad is None:
                    continue
                grad = p.grad.data.float()
                if grad.is_sparse:
                    raise RuntimeError(
                        "RAdam does not support sparse gradients"
                    )

                p_data_fp32 = p.data.float()

                state = self.state[p]

                if len(state) == 0:
                    state["step"] = 0
                    state["exp_avg"] = torch.zeros_like(p_data_fp32)
                    state["exp_avg_sq"] = torch.zeros_like(p_data_fp32)
                else:
                    state["exp_avg"] = state["exp_avg"].type_as(p_data_fp32)
                    state["exp_avg_sq"] = state["exp_avg_sq"].type_as(
                        p_data_fp32
                    )

                exp_avg, exp_avg_sq = state["exp_avg"], state["exp_avg_sq"]
                beta1, beta2 = group["betas"]

                exp_avg_sq.mul_(beta2).addcmul_(1 - beta2, grad, grad)
                exp_avg.mul_(beta1).add_(1 - beta1, grad)

                state["step"] += 1
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -



Dassl.pytorch/dassl/optim/radam.py [165:202]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
    def step(self, closure=None):

        loss = None
        if closure is not None:
            loss = closure()

        for group in self.param_groups:

            for p in group["params"]:
                if p.grad is None:
                    continue
                grad = p.grad.data.float()
                if grad.is_sparse:
                    raise RuntimeError(
                        "RAdam does not support sparse gradients"
                    )

                p_data_fp32 = p.data.float()

                state = self.state[p]

                if len(state) == 0:
                    state["step"] = 0
                    state["exp_avg"] = torch.zeros_like(p_data_fp32)
                    state["exp_avg_sq"] = torch.zeros_like(p_data_fp32)
                else:
                    state["exp_avg"] = state["exp_avg"].type_as(p_data_fp32)
                    state["exp_avg_sq"] = state["exp_avg_sq"].type_as(
                        p_data_fp32
                    )

                exp_avg, exp_avg_sq = state["exp_avg"], state["exp_avg_sq"]
                beta1, beta2 = group["betas"]

                exp_avg_sq.mul_(beta2).addcmul_(1 - beta2, grad, grad)
                exp_avg.mul_(beta1).add_(1 - beta1, grad)

                state["step"] += 1
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -