Stable-Baselines 3 部分源代码解读 2 on_policy_algorithm.py

Stable-Baselines 3 部分源代码解读 ./common/on_policy_algorithm.py

前言

阅读PPO相关的源码，了解一下标准库是如何建立PPO算法以及各种tricks的，以便于自己的复现。

在Pycharm里面一直跳转，可以看到PPO类是最终继承于基类，也就是这个py文件的内容。

所以阅读源码就先从这里开始。: )

import 包

import sys
import time
from typing import Any, Dict, List, Optional, Tuple, Type, TypeVar, Union

import numpy as np
import torch as th
from gym import spaces

from stable_baselines3.common.base_class import BaseAlgorithm
from stable_baselines3.common.buffers import DictRolloutBuffer, RolloutBuffer
from stable_baselines3.common.callbacks import BaseCallback
from stable_baselines3.common.policies import ActorCriticPolicy
from stable_baselines3.common.type_aliases import GymEnv, MaybeCallback, Schedule
from stable_baselines3.common.utils import obs_as_tensor, safe_mean
from stable_baselines3.common.vec_env import VecEnv

OnPolicyAlgorithm 类

这个类是PPO算法类的中间曾，夹在底层基类和上层PPO类的之间。

主要是同策略算法，例如：A2C和PPO算法。

policy、env和learning_rate三者与基类base-class.py的一致

n_steps表示每次更新前需要经过的时间步，作者在这里给出了n_steps * n_envs的例子，可能的意思是，如果环境是重复的多个，打算做并行训练的话，那么就是每个子环境的时间步乘以环境的数量

batch_size经验回放的最小批次信息

gamma、gae_lambda、clip_range、clip_range_vf均是具有默认值的参数，分别代表“折扣因子”、“GAE奖励中平衡偏置和方差的参数”、“为网络参数而限制幅度的范围”、“为值函数网络参数而限制幅度的范围”

normalize_advantage标志是否需要归一化优势（advantage）

ent_coef、vf_coef损失计算的熵系数

max_grad_norm最大的梯度长度，梯度下降的限幅

use_sde、sde_sample_freq是状态独立性探索，只适用于连续环境，与基类base-class.py的一致

target_kl限制每次更新时KL散度不能太大，因为clipping限幅不能防止大量更新

monitor_wrapper标志是否需要Gym库提供的监视器包装器

_init_setup_model是否建立模型，也就是是否在创建这个实例过程中创建初始化模型

class OnPolicyAlgorithm(BaseAlgorithm):
    """
    The base for On-Policy algorithms (ex: A2C/PPO).

    :param policy: The policy model to use (MlpPolicy, CnnPolicy, ...)
    :param env: The environment to learn from (if registered in Gym, can be str)
    :param learning_rate: The learning rate, it can be a function
        of the current progress remaining (from 1 to 0)
    :param n_steps: The number of steps to run for each environment per update
        (i.e. batch size is n_steps * n_env where n_env is number of environment copies running in parallel)
    :param gamma: Discount factor
    :param gae_lambda: Factor for trade-off of bias vs variance for Generalized Advantage Estimator.
        Equivalent to classic advantage when set to 1.
    :param ent_coef: Entropy coefficient for the loss calculation
    :param vf_coef: Value function coefficient for the loss calculation
    :param max_grad_norm: The maximum value for the gradient clipping
    :param use_sde: Whether to use generalized State Dependent Exploration (gSDE)
        instead of action noise exploration (default: False)
    :param sde_sample_freq: Sample a new noise matrix every n steps when using gSDE
        Default: -1 (only sample at the beginning of the rollout)
    :param tensorboard_log: the log location for tensorboard (if None, no logging)
    :param monitor_wrapper: When creating an environment, whether to wrap it
        or not in a Monitor wrapper.
    :param policy_kwargs: additional arguments to be passed to the policy on creation
    :param verbose: Verbosity level: 0 for no output, 1 for info messages (such as device or wrappers used), 2 for
        debug messages
    :param seed: Seed for the pseudo random generators
    :param device: Device (cpu, cuda, ...) on which the code should be run.
        Setting it to auto, the code will be run on the GPU if possible.
    :param _init_setup_model: Whether or not to build the network at the creation of the instance
    :param supported_action_spaces: The action spaces supported by the algorithm.
    """

    def __init__(
        self,
        policy: Union[str, Type[ActorCriticPolicy]],
        env: Union[GymEnv, str],
        learning_rate: Union[float, Schedule],
        n_steps: int,
        gamma: float,
        gae_lambda: float,
        ent_coef: float,
        vf_coef: float,
        max_grad_norm: float,
        use_sde: bool,
        sde_sample_freq: int,
        tensorboard_log: Optional[str] = None,
        monitor_wrapper: bool = True,
        policy_kwargs: Optional[Dict[str, Any]] = None,
        verbose: int = 0,
        seed: Optional[int] = None,
        device: Union[th.device, str] = "auto",
        _init_setup_model: bool = True,
        supported_action_spaces: Optional[Tuple[spaces.Space, ...]] = None,
    ):
        super().__init__(
            policy=policy,
            env=env,
            learning_rate=learning_rate,
            policy_kwargs=policy_kwargs,
            verbose=verbose,
            device=device,
            use_sde=use_sde,
            sde_sample_freq=sde_sample_freq,
            support_multi_env=True,
            seed=seed,
            tensorboard_log=tensorboard_log,
            supported_action_spaces=supported_action_spaces,
        )

        self.n_steps = n_steps
        self.gamma = gamma
        self.gae_lambda = gae_lambda
        self.ent_coef = ent_coef
        self.vf_coef = vf_coef
        self.max_grad_norm = max_grad_norm
        self.rollout_buffer = None

        # 调用基类的_setup_model()模型
        if _init_setup_model:
            self._setup_model()

    def _setup_model(self) -> None:
        # 初始化学习率，让他可以调用
        self._setup_lr_schedule()
        # 设置随机数种子
        self.set_random_seed(self.seed)
		# 设置经验池子的类，如果观测空间是spaces.Dict类那么就赋值DictRolloutBuffer
        # 如果观测空间不是spaces.Dict类那么就赋值RolloutBuffer
        buffer_cls = DictRolloutBuffer if isinstance(self.observation_space, spaces.Dict) else RolloutBuffer
		# 根据类初始化实例经验池子
        # 初始化经验池子的是时候将设备信息、折扣率、GAE超参数和环境的数量也传进去了
        self.rollout_buffer = buffer_cls(
            self.n_steps,
            self.observation_space,
            self.action_space,
            device=self.device,
            gamma=self.gamma,
            gae_lambda=self.gae_lambda,
            n_envs=self.n_envs,
        )
        # 初始化策略，直接输入状态空间、动作空间、可调用的学习率、是否使用状态独立性探索，以及自己制定策略
        # 的时候自己家的模型的参数和激活函数
        self.policy = self.policy_class(  # pytype:disable=not-instantiable
            self.observation_space,
            self.action_space,
            self.lr_schedule,
            use_sde=self.use_sde,
            **self.policy_kwargs  # pytype:disable=not-instantiable
        )
        # 将策略放到GPU/CPU中
        self.policy = self.policy.to(self.device)

    def collect_rollouts(
        self,
        env: VecEnv,
        callback: BaseCallback,
        rollout_buffer: RolloutBuffer,
        n_rollout_steps: int,
    ) -> bool:
        # 收集环境交互数据
        # 这个方法使用当前的策略并将交互历史填充到RolloutBuffer经验池子中
        # rollout的意思是无模型的概念，而不是有模型的RL或规划里面的rollout的概念
        # env 用于训练的环境
        # callback 在每个时间步都会调用的回调函数
        # rollout_buffer 将收集的经验放置到rollout_buffer中
        # 在每个环境中需要收集的条数
        # 返回值是True：如果rollout_buffer收集了这么多的经验；返回值是False：如果回调函数提前终止了
        # 这个rollouts。
        """
        Collect experiences using the current policy and fill a ``RolloutBuffer``.
        The term rollout here refers to the model-free notion and should not
        be used with the concept of rollout used in model-based RL or planning.

        :param env: The training environment
        :param callback: Callback that will be called at each step
            (and at the beginning and end of the rollout)
        :param rollout_buffer: Buffer to fill with rollouts
        :param n_rollout_steps: Number of experiences to collect per environment
        :return: True if function returned with at least `n_rollout_steps`
            collected, False if callback terminated rollout prematurely.
        """
        assert self._last_obs is not None, "No previous observation was provided"
        # 将策略转变到评估模式
        # Switch to eval mode (this affects batch norm / dropout)
        self.policy.set_training_mode(False)

        # 重置经验池子，如果使用状态独立性探索，那么就重置策略的噪声
        n_steps = 0
        rollout_buffer.reset()
        # Sample new weights for the state dependent exploration
        if self.use_sde:
            self.policy.reset_noise(env.num_envs)

        # 回调函数执行on_rollout_start()命令，跳转定义时候没有看到具体定义
        callback.on_rollout_start()

        while n_steps < n_rollout_steps:
            
            # 如果使用了状态独立性探索，并且达到了探索频率的节点，那么就重置策略的噪声
            if self.use_sde and self.sde_sample_freq > 0 and n_steps % self.sde_sample_freq == 0:
                # Sample a new noise matrix
                self.policy.reset_noise(env.num_envs)

            # 在断开梯度的情况下，转换观测数据到tensor张量内，然后输入到策略中输出动作、价值和对数概率
            # 最后再将动作数据转移到numpy中
            with th.no_grad():
                # Convert to pytorch tensor or to TensorDict
                obs_tensor = obs_as_tensor(self._last_obs, self.device)
                actions, values, log_probs = self.policy(obs_tensor)
            actions = actions.cpu().numpy()

            # Rescale and perform action
            # 归一化动作信息，限制在动作空间的上下界
            clipped_actions = actions
            # Clip the actions to avoid out of bound error
            if isinstance(self.action_space, spaces.Box):
                clipped_actions = np.clip(actions, self.action_space.low, self.action_space.high)

            # 将动作信息输入到环境中，输出新的观测、奖励数值、是否完成以及其他信息。
            new_obs, rewards, dones, infos = env.step(clipped_actions)

            # 处理回调函数和更新经验池子
            self.num_timesteps += env.num_envs
            # Give access to local variables
            callback.update_locals(locals())
            if callback.on_step() is False:
                return False
            self._update_info_buffer(infos)
            n_steps += 1

            # 如果动作空间是离散空间的话，那么就转变成一个列向量
            if isinstance(self.action_space, spaces.Discrete):
                # Reshape in case of discrete action
                actions = actions.reshape(-1, 1)

            # 判断数据是否是终止的
            # 终止之后计算累计奖励
            # Handle timeout by bootstraping with value function
            # see GitHub issue #633
            for idx, done in enumerate(dones):
                if (
                    done
                    and infos[idx].get("terminal_observation") is not None
                    and infos[idx].get("TimeLimit.truncated", False)
                ):
                    terminal_obs = self.policy.obs_to_tensor(infos[idx]["terminal_observation"])[0]
                    with th.no_grad():
                        terminal_value = self.policy.predict_values(terminal_obs)[0]
                    rewards[idx] += self.gamma * terminal_value

            # 经验池子输入的是上一个状态、动作、奖励、上一个回合的开始状态、价值列表以及对数概率
            rollout_buffer.add(self._last_obs, actions, rewards, self._last_episode_starts, values, log_probs)
            self._last_obs = new_obs
            self._last_episode_starts = dones

        # 计算下一个状态的价值
        with th.no_grad():
            # Compute value for the last timestep
            values = self.policy.predict_values(obs_as_tensor(new_obs, self.device))

        # 计算回报和优势
        rollout_buffer.compute_returns_and_advantage(last_values=values, dones=dones)

        callback.on_rollout_end()

        return True

    def train(self) -> None:
        # 这个是父类的方法
        # 在子类的实际PPO类中做了重写
        """
        Consume current rollout data and update policy parameters.
        Implemented by individual algorithms.
        """
        raise NotImplementedError

    def learn(
        self: SelfOnPolicyAlgorithm,
        total_timesteps: int,
        callback: MaybeCallback = None,
        log_interval: int = 1,
        tb_log_name: str = "OnPolicyAlgorithm",
        reset_num_timesteps: bool = True,
        progress_bar: bool = False,
    ) -> SelfOnPolicyAlgorithm:
        iteration = 0

        # 初始化模型
        total_timesteps, callback = self._setup_learn(
            total_timesteps,
            callback,
            reset_num_timesteps,
            tb_log_name,
            progress_bar,
        )

        callback.on_training_start(locals(), globals())

        while self.num_timesteps < total_timesteps:
            
            # 这里开始执行上面的函数，在环境中收集数据，收集完了就继续训练
            # 如果出了故障了，就在接下来跳出循环
            continue_training = self.collect_rollouts(self.env, callback, self.rollout_buffer, n_rollout_steps=self.n_steps)

            if continue_training is False:
                break

            # 跌带次数+1,并根据当前的训练次数更新学习率
            iteration += 1
            self._update_current_progress_remaining(self.num_timesteps, total_timesteps)

            # 在控制台按照预先定义的频率输出相关信息
            # Display training infos
            if log_interval is not None and iteration % log_interval == 0:
                time_elapsed = max((time.time_ns() - self.start_time) / 1e9, sys.float_info.epsilon)
                fps = int((self.num_timesteps - self._num_timesteps_at_start) / time_elapsed)
                self.logger.record("time/iterations", iteration, exclude="tensorboard")
                if len(self.ep_info_buffer) > 0 and len(self.ep_info_buffer[0]) > 0:
                    self.logger.record("rollout/ep_rew_mean", safe_mean([ep_info["r"] for ep_info in self.ep_info_buffer]))
                    self.logger.record("rollout/ep_len_mean", safe_mean([ep_info["l"] for ep_info in self.ep_info_buffer]))
                self.logger.record("time/fps", fps)
                self.logger.record("time/time_elapsed", int(time_elapsed), exclude="tensorboard")
                self.logger.record("time/total_timesteps", self.num_timesteps, exclude="tensorboard")
                self.logger.dump(step=self.num_timesteps)

            self.train()

        callback.on_training_end()

        return self

    def _get_torch_save_params(self) -> Tuple[List[str], List[str]]:
        state_dicts = ["policy", "policy.optimizer"]

        return state_dicts, []