source/model.py

from __future__ import annotations

import torch
from torch import nn


class FederatedMLP(nn.Module):
    def __init__(self) -> None:
        super().__init__()
        self.network = nn.Sequential(
            nn.Linear(64, 32),
            nn.GELU(),
            nn.Linear(32, 10),
        )

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


def parameter_count(model: nn.Module) -> int:
    return sum(parameter.numel() for parameter in model.parameters())