affinity/model.py

from __future__ import annotations

import torch
from torch import nn


class AffinityRegressor(nn.Module):
    def __init__(
        self,
        input_dim: int,
        hidden_dims: list[int] | tuple[int, ...] = (256, 128, 64),
        dropout: float = 0.2,
    ) -> None:
        super().__init__()
        layers: list[nn.Module] = []
        previous = input_dim
        for hidden in hidden_dims:
            layers.extend(
                [
                    nn.Linear(previous, hidden),
                    nn.LayerNorm(hidden),
                    nn.GELU(),
                    nn.Dropout(dropout),
                ]
            )
            previous = hidden
        layers.append(nn.Linear(previous, 1))
        self.network = nn.Sequential(*layers)

    def forward(self, features: torch.Tensor) -> torch.Tensor:
        return self.network(features).squeeze(-1)