vae_model_architecture.py

import torch.nn as nn
import torch.nn.functional as F

# NOTE: LATENT_DIM doit être le même que celui utilisé pour l'entraînement (128)
class VAE(nn.Module):
    def __init__(self, latent_dim=128):
        super(VAE, self).__init__()
        self.latent_dim = latent_dim
        
        # ENCODEUR
        self.encoder = nn.Sequential(
            nn.Conv2d(3, 32, kernel_size=4, stride=2, padding=1), nn.ReLU(),
            nn.Conv2d(32, 64, kernel_size=4, stride=2, padding=1), nn.ReLU(),
            nn.Conv2d(64, 128, kernel_size=4, stride=2, padding=1), nn.ReLU(),
            nn.Conv2d(128, 256, kernel_size=4, stride=2, padding=1), nn.ReLU(),
            nn.Flatten()
        )
        self.fc_mu = nn.Linear(256 * 4 * 4, latent_dim)
        self.fc_logvar = nn.Linear(256 * 4 * 4, latent_dim)

        # DÉCODEUR
        self.decoder_input = nn.Linear(latent_dim, 256 * 4 * 4)
        self.decoder = nn.Sequential(
            nn.ConvTranspose2d(256, 128, kernel_size=4, stride=2, padding=1), nn.ReLU(),
            nn.ConvTranspose2d(128, 64, kernel_size=4, stride=2, padding=1), nn.ReLU(),
            nn.ConvTranspose2d(64, 32, kernel_size=4, stride=2, padding=1), nn.ReLU(),
            nn.ConvTranspose2d(32, 3, kernel_size=4, stride=2, padding=1),  
            nn.Tanh()
        )

    def decode(self, z):
        h = self.decoder_input(z)
        h = h.view(-1, 256, 4, 4) 
        return self.decoder(h)