# Copyright (c) Facebook, Inc. and its affiliates.
#
# This source code is licensed under the MIT license found in the
# LICENSE file in the root directory of this source tree.


import math

import torch
import torch.optim


class FairseqAdam(torch.optim.Optimizer):
    r"""Implements Adam algorithm.

    This implementation is modified from torch.optim.Adam based on:
    `Fixed Weight Decay Regularization in Adam`
    (see https://arxiv.org/abs/1711.05101)

    It has been proposed in `Adam: A Method for Stochastic Optimization`_.

    Args:
        params (iterable): iterable of parameters to optimize or dicts defining
            parameter groups
        lr (float, optional): learning rate (default: 1e-3)
        betas (Tuple[float, float], optional): coefficients used for computing
            running averages of gradient and its square (default: (0.9, 0.999))
        eps (float, optional): term added to the denominator to improve
            numerical stability (default: 1e-8)
        weight_decay (float, optional): weight decay (L2 penalty) (default: 0)
        amsgrad (boolean, optional): whether to use the AMSGrad variant of this
            algorithm from the paper `On the Convergence of Adam and Beyond`_

    .. _Adam\: A Method for Stochastic Optimization:
        https://arxiv.org/abs/1412.6980
    .. _On the Convergence of Adam and Beyond:
        https://openreview.net/forum?id=ryQu7f-RZ
    """

    def __init__(
        self,
        params,
        lr=1e-3,
        adam_betas=(0.9, 0.999),
        adam_eps=1e-8,
        weight_decay=0,
        amsgrad=False,
    ):
        defaults = dict(
            lr=lr, betas=adam_betas, eps=adam_eps, weight_decay=weight_decay, amsgrad=amsgrad
        )
        super(FairseqAdam, self).__init__(params, defaults)
        self.optimizer_lr = lr

    @property
    def supports_memory_efficient_fp16(self):
        return True

    @property
    def supports_flat_params(self):
        return True

    def step(self, closure=None):
        """Performs a single optimization step.

        Args:
            closure (callable, optional): A closure that reevaluates the model
                and returns the loss.
        """
        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
                if grad.dtype in {torch.float16, torch.bfloat16}:
                    grad = grad.float()
                if grad.is_sparse:
                    raise RuntimeError(
                        "Adam does not support sparse gradients, please consider SparseAdam instead"
                    )
                amsgrad = group.get("amsgrad", False)

                p_data_fp32 = p.data
                if p.data.dtype in {torch.float16, torch.bfloat16}:
                    p_data_fp32 = p_data_fp32.float()

                state = self.state[p]

                # State initialization
                if len(state) == 0:
                    state["step"] = 0
                    # Exponential moving average of gradient values
                    state["exp_avg"] = torch.zeros_like(p_data_fp32)
                    # Exponential moving average of squared gradient values
                    state["exp_avg_sq"] = torch.zeros_like(p_data_fp32)
                    if amsgrad:
                        # Maintains max of all exp. moving avg. of sq. grad. values
                        state["max_exp_avg_sq"] = torch.zeros_like(p_data_fp32)
                else:
                    state["exp_avg"] = state["exp_avg"].to(p_data_fp32)
                    state["exp_avg_sq"] = state["exp_avg_sq"].to(p_data_fp32)
                    if amsgrad:
                        state["max_exp_avg_sq"] = state["max_exp_avg_sq"].to(p_data_fp32)

                exp_avg, exp_avg_sq = state["exp_avg"], state["exp_avg_sq"]
                if amsgrad:
                    max_exp_avg_sq = state["max_exp_avg_sq"]
                beta1, beta2 = group["betas"]

                state["step"] += 1

                # Decay the first and second moment running average coefficient
                exp_avg.mul_(beta1).add_(grad, alpha=1 - beta1)
                exp_avg_sq.mul_(beta2).addcmul_(grad, grad, value=1 - beta2)
                if amsgrad:
                    # Maintains the maximum of all 2nd moment running avg. till now
                    torch.max(max_exp_avg_sq, exp_avg_sq, out=max_exp_avg_sq)
                    # Use the max. for normalizing running avg. of gradient
                    denom = max_exp_avg_sq.sqrt().add_(group["eps"])
                else:
                    denom = exp_avg_sq.sqrt().add_(group["eps"])

                bias_correction1 = 1 - beta1 ** state["step"]
                bias_correction2 = 1 - beta2 ** state["step"]
                step_size = group["lr"] * math.sqrt(bias_correction2) / bias_correction1

                if group["weight_decay"] != 0:
                    p_data_fp32.add_(p_data_fp32, alpha=-group["weight_decay"] * group["lr"])

                p_data_fp32.addcdiv_(exp_avg, denom, value=-step_size)

                if p.data.dtype in {torch.float16, torch.bfloat16}:
                    p.data.copy_(p_data_fp32)

        return loss

    def set_lr(self, lr):
        """Set the learning rate."""
        for param_group in self.param_groups:
            param_group["lr"] = lr