#!/usr/bin/env python3
from __future__ import annotations

import math
from collections.abc import Callable, Iterable
from typing import Optional

import torch


class AdamW(torch.optim.Optimizer):
    def __init__(
        self,
        params: Iterable[torch.nn.parameter.Parameter],
        lr: float = 1e-3,
        betas: tuple[float, float] = (0.9, 0.999),
        eps: float = 1e-8,
        weight_decay: float = 0.01,
    ):
        if not 0.0 <= lr:
            raise ValueError("Invalid learning rate: {}".format(lr))
        if not 0.0 <= eps:
            raise ValueError("Invalid epsilon value: {}".format(eps))
        if not 0.0 <= betas[0] < 1.0:
            raise ValueError("Invalid beta parameter at index 0: {}".format(betas[0]))
        if not 0.0 <= betas[1] < 1.0:
            raise ValueError("Invalid beta parameter at index 1: {}".format(betas[1]))
        defaults = dict(lr=lr, betas=betas, eps=eps, weight_decay=weight_decay)
        super(AdamW, self).__init__(params, defaults)

    def step(self, closure: Optional[Callable] = 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

                # Can either apply weight decay here, or at the very end
                # p.data.mul_(1 - group['lr'] * group['weight_decay'])

                grad = p.grad.data
                if grad.is_sparse:
                    raise RuntimeError("Adam does not support sparse gradients")

                state = self.state[p]
                alpha = group["lr"]
                beta_1, beta_2 = group["betas"]
                eps = group["eps"]
                t = state.get("t", 1)
                prev_m_t = state.get("m", torch.zeros_like(grad))
                prev_v_t = state.get("v", torch.zeros_like(grad))

                m_t = beta_1 * prev_m_t + ((1 - beta_1) * grad)
                v_t = beta_2 * prev_v_t + ((1 - beta_2) * torch.square(grad))

                alpha_t = alpha * (math.sqrt(1 - (beta_2**t)) / (1 - (beta_1**t)))
                p.data -= alpha_t * m_t / (torch.sqrt(v_t) + eps)
                # Apply weight decay
                p.data -= alpha * group["weight_decay"] * p.data

                state["m"] = m_t
                state["v"] = v_t
                state["t"] = t + 1
        return loss