Pytorch 版本的lookahead 优化函数使用(附代码)

Lookahead 优化算法是Adam的作者继Adam之后的又一力作，论文可以参见https://arxiv.org/abs/1907.08610

这篇博客先不讲述Lookahead具体原理，先介绍如何将Lookahead集成到现有的代码中。

本人在三个项目中（涉及风格转换、物体识别）使用该优化器，最大的感受就是使用该优化器十分有利于模型收敛，原本不收敛或者收敛过慢的模型在使用lookahead后可以看到明显的收敛情况，并且最终的效果能够满足最初设计的要求。

总所周知，Adam因为其具有较好的适应性，被广泛用于各类模型的优化；其参数简单，调参方便的特点一直为大家所喜爱，尤其对于初学者较为友好。Lookahead 也继承了Adam的优点。lookahead的Pytorch版本代码如下所示：后续会针对代码进行原理讲解，该代码在Github上可以找到。

from collections import defaultdict
from torch.optim import Optimizer
import torch


class Lookahead(Optimizer):
    def __init__(self, optimizer, k=5, alpha=0.5):
        self.optimizer = optimizer

        self.k = k
        self.alpha = alpha
        self.param_groups = self.optimizer.param_groups
        self.state = defaultdict(dict)
        self.fast_state = self.optimizer.state
        for group in self.param_groups:
            group["counter"] = 0

    def update(self, group):
        for fast in group["params"]:
            param_state = self.state[fast]
            if "slow_param" not in param_state:
                param_state["slow_param"] = torch.zeros_like(fast.data)
                param_state["slow_param"].copy_(fast.data)
            slow = param_state["slow_param"]
            slow += (fast.data - slow) * self.alpha
            fast.data.copy_(slow)

    def update_lookahead(self):
        for group in self.param_groups:
            self.update(group)

    def step(self, closure=None):
        loss = self.optimizer.step(closure)
        for group in self.param_groups:
            if group["counter"] == 0:
                self.update(group)
            group["counter"] += 1
            if group["counter"] >= self.k:
                group["counter"] = 0
        return loss

    def state_dict(self):
        fast_state_dict = self.optimizer.state_dict()
        slow_state = {
            (id(k) if isinstance(k, torch.Tensor) else k): v
            for k, v in self.state.items()
        }
        fast_state = fast_state_dict["state"]
        param_groups = fast_state_dict["param_groups"]
        return {
            "fast_state": fast_state,
            "slow_state": slow_state,
            "param_groups": param_groups,
        }

    def load_state_dict(self, state_dict):
        slow_state_dict = {
            "state": state_dict["slow_state"],
            "param_groups": state_dict["param_groups"],
        }
        fast_state_dict = {
            "state": state_dict["fast_state"],
            "param_groups": state_dict["param_groups"],
        }
        super(Lookahead, self).load_state_dict(slow_state_dict)
        self.optimizer.load_state_dict(fast_state_dict)
        self.fast_state = self.optimizer.state

    def add_param_group(self, param_group):
        param_group["counter"] = 0
        self.optimizer.add_param_group(param_group)

将lookahead集成在现有代码中如下操作即可：

base_optimizer = Adam(model.parameters(), lr=args.lr, betas=(0.9, 0.999))
opt = Lookahead(base_optimizer, k=5, alpha=0.5)

此时直接将opt作为正常的优化器使用即可，就像直接使用Adam一样的步骤使用opt