source/model.py

from __future__ import annotations

import torch
from torch import nn


class DuelingQNetwork(nn.Module):
    def __init__(self, observation_size: int = 17, actions: int = 4):
        super().__init__()
        self.encoder = nn.Sequential(
            nn.Linear(observation_size, 128),
            nn.LayerNorm(128),
            nn.SiLU(),
            nn.Linear(128, 128),
            nn.SiLU(),
        )
        self.value = nn.Sequential(
            nn.Linear(128, 64),
            nn.SiLU(),
            nn.Linear(64, 1),
        )
        self.advantage = nn.Sequential(
            nn.Linear(128, 64),
            nn.SiLU(),
            nn.Linear(64, actions),
        )

    def forward(self, observations: torch.Tensor) -> torch.Tensor:
        features = self.encoder(observations)
        value = self.value(features)
        advantage = self.advantage(features)
        return value + advantage - advantage.mean(dim=-1, keepdim=True)


class MicroPolicy(nn.Module):
    def __init__(self, observation_size: int = 17, actions: int = 4):
        super().__init__()
        self.network = nn.Sequential(
            nn.Linear(observation_size, 32),
            nn.SiLU(),
            nn.Linear(32, 32),
            nn.SiLU(),
            nn.Linear(32, actions),
        )

    def forward(self, observations: torch.Tensor) -> torch.Tensor:
        return self.network(observations)