dataeval.utils.models.VAEEncoder¶

class dataeval.utils.models.VAEEncoder(input_shape, latent_dim)¶

Variational Autoencoder Encoder that outputs mean and log-variance.

Parameters:¶

input_shape : tuple[int, int, int]¶

Shape of the input data in CHW format.

latent_dim : int¶

The size of the latent space (encoding dimension).

forward(x)¶

Perform a forward pass through the encoder.

Parameters:¶

x : torch.Tensor¶

Input tensor

Returns:¶

Mean (mu) and log-variance (logvar) of the latent distribution.

Return type:¶

tuple[torch.Tensor, torch.Tensor]