参考bilibili讲解视频

论文地址：https://arxiv.org/abs/2302.06675

代码地址：https://github.com/google/automl/blob/master/lion/lion_pytorch.py

"""PyTorch implementation of the Lion optimizer."""
import torch
from torch.optim.optimizer import Optimizerclass Lion(Optimizer):r"""Implements Lion algorithm."""def __init__(self, params, lr=1e-4, betas=(0.9, 0.99), weight_decay=0.0):"""Initialize the hyperparameters.Args:params (iterable): iterable of parameters to optimize or dicts definingparameter groupslr (float, optional): learning rate (default: 1e-4)betas (Tuple[float, float], optional): coefficients used for computingrunning averages of gradient and its square (default: (0.9, 0.99))weight_decay (float, optional): weight decay coefficient (default: 0)"""if not 0.0 <= lr:raise ValueError('Invalid learning rate: {}'.format(lr))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, weight_decay=weight_decay)super().__init__(params, defaults)@torch.no_grad()def step(self, closure=None):"""Performs a single optimization step.Args:closure (callable, optional): A closure that reevaluates the modeland returns the loss.Returns:the loss."""loss = Noneif closure is not None:with torch.enable_grad():loss = closure()for group in self.param_groups:for p in group['params']:if p.grad is None:continue# Perform stepweight decayp.data.mul_(1 - group['lr'] * group['weight_decay'])grad = p.gradstate = self.state[p]# State initializationif len(state) == 0:# Exponential moving average of gradient valuesstate['exp_avg'] = torch.zeros_like(p)exp_avg = state['exp_avg']beta1, beta2 = group['betas']# Weight updateupdate = exp_avg * beta1 + grad * (1 - beta1)p.add_(torch.sign(update), alpha=-group['lr'])# Decay the momentum running average coefficientexp_avg.mul_(beta2).add_(grad, alpha=1 - beta2)return loss