# vae_decoder.py import torch.nn as nn from config import IMG_CHANNELS, LATENT_CHANNELS, VAE_BASE_CHANNELS, VAE_CHANNEL_MULT from vae_encoder import ResBlock class VAEDecoder(nn.Module): """32x32x4 latent -> 256x256x3 image. Mirror image of the encoder.""" def __init__(self, latent_channels=LATENT_CHANNELS, out_channels=IMG_CHANNELS, base_channels=VAE_BASE_CHANNELS, channel_mult=VAE_CHANNEL_MULT): super().__init__() chs = [base_channels * m for m in channel_mult][::-1] # deepest first layers = [nn.Conv2d(latent_channels, chs[0], 3, padding=1), ResBlock(chs[0], chs[0])] for i, ch in enumerate(chs): prev = chs[i - 1] if i > 0 else chs[0] layers += [ResBlock(prev, ch), ResBlock(ch, ch)] if i < len(chs) - 1: layers += [nn.Upsample(scale_factor=2, mode='nearest'), nn.Conv2d(ch, ch, 3, padding=1)] # upsample x2 layers += [nn.GroupNorm(32, chs[-1]), nn.SiLU(), nn.Conv2d(chs[-1], out_channels, 3, padding=1)] self.net = nn.Sequential(*layers) def forward(self, z): # No Tanh: it squashes gradients near +-1 (pure white/black, i.e. text strokes). # Clamp to [-1, 1] only when converting to an image for viewing. return self.net(z)