Tianshou/tianshou/policy/modelfree/ppo.py

from collections.abc import Callable
from typing import Any, Literal

import gymnasium as gym
import numpy as np
import torch
from torch import nn

from tianshou.data import ReplayBuffer, to_torch_as
from tianshou.data.types import LogpOldProtocol, RolloutBatchProtocol
from tianshou.policy import A2CPolicy
from tianshou.policy.base import TLearningRateScheduler
from tianshou.policy.modelfree.pg import TDistParams
from tianshou.utils.net.common import ActorCritic


class PPOPolicy(A2CPolicy):
    r"""Implementation of Proximal Policy Optimization. arXiv:1707.06347.

    :param actor: the actor network following the rules in BasePolicy. (s -> logits)
    :param critic: the critic network. (s -> V(s))
    :param optim: the optimizer for actor and critic network.
    :param dist_fn: distribution class for computing the action.
    :param action_space: env's action space
    :param eps_clip: :math:`\epsilon` in :math:`L_{CLIP}` in the original
        paper.
    :param dual_clip: a parameter c mentioned in arXiv:1912.09729 Equ. 5,
        where c > 1 is a constant indicating the lower bound. Set to None
        to disable dual-clip PPO.
    :param value_clip: a parameter mentioned in arXiv:1811.02553v3 Sec. 4.1.
    :param advantage_normalization: whether to do per mini-batch advantage
        normalization.
    :param recompute_advantage: whether to recompute advantage every update
        repeat according to https://arxiv.org/pdf/2006.05990.pdf Sec. 3.5.
    :param vf_coef: weight for value loss.
    :param ent_coef: weight for entropy loss.
    :param max_grad_norm: clipping gradients in back propagation.
    :param gae_lambda: in [0, 1], param for Generalized Advantage Estimation.
    :param max_batchsize: the maximum size of the batch when computing GAE.
    :param discount_factor: in [0, 1].
    :param reward_normalization: normalize estimated values to have std close to 1.
    :param deterministic_eval: if True, use deterministic evaluation.
    :param observation_space: the space of the observation.
    :param action_scaling: if True, scale the action from [-1, 1] to the range of
        action_space. Only used if the action_space is continuous.
    :param action_bound_method: method to bound action to range [-1, 1].
    :param lr_scheduler: if not None, will be called in `policy.update()`.

    .. seealso::

        Please refer to :class:`~tianshou.policy.BasePolicy` for more detailed
        explanation.
    """

    def __init__(
        self,
        *,
        actor: torch.nn.Module,
        critic: torch.nn.Module,
        optim: torch.optim.Optimizer,
        dist_fn: Callable[[TDistParams], torch.distributions.Distribution],
        action_space: gym.Space,
        eps_clip: float = 0.2,
        dual_clip: float | None = None,
        value_clip: bool = False,
        advantage_normalization: bool = True,
        recompute_advantage: bool = False,
        vf_coef: float = 0.5,
        ent_coef: float = 0.01,
        max_grad_norm: float | None = None,
        gae_lambda: float = 0.95,
        max_batchsize: int = 256,
        discount_factor: float = 0.99,
        # TODO: rename to return_normalization?
        reward_normalization: bool = False,
        deterministic_eval: bool = False,
        observation_space: gym.Space | None = None,
        action_scaling: bool = True,
        action_bound_method: Literal["clip", "tanh"] | None = "clip",
        lr_scheduler: TLearningRateScheduler | None = None,
    ) -> None:
        assert (
            dual_clip is None or dual_clip > 1.0
        ), f"Dual-clip PPO parameter should greater than 1.0 but got {dual_clip}"

        super().__init__(
            actor=actor,
            critic=critic,
            optim=optim,
            dist_fn=dist_fn,
            action_space=action_space,
            vf_coef=vf_coef,
            ent_coef=ent_coef,
            max_grad_norm=max_grad_norm,
            gae_lambda=gae_lambda,
            max_batchsize=max_batchsize,
            discount_factor=discount_factor,
            reward_normalization=reward_normalization,
            deterministic_eval=deterministic_eval,
            observation_space=observation_space,
            action_scaling=action_scaling,
            action_bound_method=action_bound_method,
            lr_scheduler=lr_scheduler,
        )
        self.eps_clip = eps_clip
        self.dual_clip = dual_clip
        self.value_clip = value_clip
        self.norm_adv = advantage_normalization
        self.recompute_adv = recompute_advantage
        self._actor_critic: ActorCritic

    def process_fn(
        self,
        batch: RolloutBatchProtocol,
        buffer: ReplayBuffer,
        indices: np.ndarray,
    ) -> LogpOldProtocol:
        if self.recompute_adv:
            # buffer input `buffer` and `indices` to be used in `learn()`.
            self._buffer, self._indices = buffer, indices
        batch = self._compute_returns(batch, buffer, indices)
        batch.act = to_torch_as(batch.act, batch.v_s)
        with torch.no_grad():
            batch.logp_old = self(batch).dist.log_prob(batch.act)
        batch: LogpOldProtocol
        return batch

    # TODO: why does mypy complain?
    def learn(  # type: ignore
        self,
        batch: RolloutBatchProtocol,
        batch_size: int,
        repeat: int,
        *args: Any,
        **kwargs: Any,
    ) -> dict[str, list[float]]:
        losses, clip_losses, vf_losses, ent_losses = [], [], [], []
        for step in range(repeat):
            if self.recompute_adv and step > 0:
                batch = self._compute_returns(batch, self._buffer, self._indices)
            for minibatch in batch.split(batch_size, merge_last=True):
                # calculate loss for actor
                dist = self(minibatch).dist
                if self.norm_adv:
                    mean, std = minibatch.adv.mean(), minibatch.adv.std()
                    minibatch.adv = (minibatch.adv - mean) / (std + self._eps)  # per-batch norm
                ratio = (dist.log_prob(minibatch.act) - minibatch.logp_old).exp().float()
                ratio = ratio.reshape(ratio.size(0), -1).transpose(0, 1)
                surr1 = ratio * minibatch.adv
                surr2 = ratio.clamp(1.0 - self.eps_clip, 1.0 + self.eps_clip) * minibatch.adv
                if self.dual_clip:
                    clip1 = torch.min(surr1, surr2)
                    clip2 = torch.max(clip1, self.dual_clip * minibatch.adv)
                    clip_loss = -torch.where(minibatch.adv < 0, clip2, clip1).mean()
                else:
                    clip_loss = -torch.min(surr1, surr2).mean()
                # calculate loss for critic
                value = self.critic(minibatch.obs).flatten()
                if self.value_clip:
                    v_clip = minibatch.v_s + (value - minibatch.v_s).clamp(
                        -self.eps_clip,
                        self.eps_clip,
                    )
                    vf1 = (minibatch.returns - value).pow(2)
                    vf2 = (minibatch.returns - v_clip).pow(2)
                    vf_loss = torch.max(vf1, vf2).mean()
                else:
                    vf_loss = (minibatch.returns - value).pow(2).mean()
                # calculate regularization and overall loss
                ent_loss = dist.entropy().mean()
                loss = clip_loss + self.vf_coef * vf_loss - self.ent_coef * ent_loss
                self.optim.zero_grad()
                loss.backward()
                if self.max_grad_norm:  # clip large gradient
                    nn.utils.clip_grad_norm_(
                        self._actor_critic.parameters(),
                        max_norm=self.max_grad_norm,
                    )
                self.optim.step()
                clip_losses.append(clip_loss.item())
                vf_losses.append(vf_loss.item())
                ent_losses.append(ent_loss.item())
                losses.append(loss.item())

        return {
            "loss": losses,
            "loss/clip": clip_losses,
            "loss/vf": vf_losses,
            "loss/ent": ent_losses,
        }