Generative AI Overview

tensoris.models.genai

Generative AI Pre-built Models Package.

Classes

VariationalAutoencoder

Bases: Module

Variational Autoencoder (VAE) for generative image modeling and latent space sampling.

References

Kingma, D. P., & Welling, M. (2014). Auto-Encoding Variational Bayes. International Conference on Learning Representations (ICLR 2014). arXiv: https://arxiv.org/abs/1312.6114

Source code in src/tensoris/models/genai/vae.py

class VariationalAutoencoder(nn.Module):
    """Variational Autoencoder (VAE) for generative image modeling and latent space sampling.

    References:
        Kingma, D. P., & Welling, M. (2014).
        Auto-Encoding Variational Bayes.
        International Conference on Learning Representations (ICLR 2014).
        arXiv: https://arxiv.org/abs/1312.6114
    """

    def __init__(
        self, input_dim: int = 784, hidden_dim: int = 400, latent_dim: int = 20
    ) -> None:
        """Initialize VAE encoder and decoder parameters.

        Args:
            input_dim: Flattened input sample feature dimension.
            hidden_dim: Hidden representation layer dimension.
            latent_dim: Dimensionality of latent Gaussian z space.
        """
        super().__init__()
        self.encoder = VAEEncoderBlock(input_dim, hidden_dim, latent_dim)
        self.decoder = VAEDecoderBlock(latent_dim, hidden_dim, input_dim)

    def encode(self, x: torch.Tensor) -> tuple[torch.Tensor, torch.Tensor]:
        """Encode input into mean and log variance parameters."""
        return self.encoder(x)

    def reparameterize(self, mu: torch.Tensor, logvar: torch.Tensor) -> torch.Tensor:
        """Apply Gaussian reparameterization trick: z = mu + std * epsilon."""
        std = torch.exp(0.5 * logvar)
        eps = torch.randn_like(std)
        return mu + eps * std

    def decode(self, z: torch.Tensor) -> torch.Tensor:
        """Decode latent vector z back to original feature space."""
        return self.decoder(z)

    def forward(
        self, x: torch.Tensor
    ) -> tuple[torch.Tensor, torch.Tensor, torch.Tensor]:
        """Execute full VAE forward encoding, reparameterization, and reconstruction.

        Args:
            x: Input tensor batch of shape (N, input_dim).

        Returns:
            Tuple of (reconstructed_x, mu, logvar).
        """
        mu, logvar = self.encode(x)
        z = self.reparameterize(mu, logvar)
        return self.decode(z), mu, logvar

Methods:

__init__

__init__(input_dim=784, hidden_dim=400, latent_dim=20)

Initialize VAE encoder and decoder parameters.

Parameters:

Name	Type	Description	Default
input_dim	int	Flattened input sample feature dimension.	784
hidden_dim	int	Hidden representation layer dimension.	400
latent_dim	int	Dimensionality of latent Gaussian z space.	20

Source code in src/tensoris/models/genai/vae.py

def __init__(
    self, input_dim: int = 784, hidden_dim: int = 400, latent_dim: int = 20
) -> None:
    """Initialize VAE encoder and decoder parameters.

    Args:
        input_dim: Flattened input sample feature dimension.
        hidden_dim: Hidden representation layer dimension.
        latent_dim: Dimensionality of latent Gaussian z space.
    """
    super().__init__()
    self.encoder = VAEEncoderBlock(input_dim, hidden_dim, latent_dim)
    self.decoder = VAEDecoderBlock(latent_dim, hidden_dim, input_dim)

decode

decode(z)

Decode latent vector z back to original feature space.

Source code in src/tensoris/models/genai/vae.py

def decode(self, z: torch.Tensor) -> torch.Tensor:
    """Decode latent vector z back to original feature space."""
    return self.decoder(z)

encode

encode(x)

Encode input into mean and log variance parameters.

Source code in src/tensoris/models/genai/vae.py

def encode(self, x: torch.Tensor) -> tuple[torch.Tensor, torch.Tensor]:
    """Encode input into mean and log variance parameters."""
    return self.encoder(x)

forward

forward(x)

Execute full VAE forward encoding, reparameterization, and reconstruction.

Parameters:

Name	Type	Description	Default
x	Tensor	Input tensor batch of shape (N, input_dim).	required

Returns:

Type	Description
tuple[Tensor, Tensor, Tensor]	Tuple of (reconstructed_x, mu, logvar).

Source code in src/tensoris/models/genai/vae.py

def forward(
    self, x: torch.Tensor
) -> tuple[torch.Tensor, torch.Tensor, torch.Tensor]:
    """Execute full VAE forward encoding, reparameterization, and reconstruction.

    Args:
        x: Input tensor batch of shape (N, input_dim).

    Returns:
        Tuple of (reconstructed_x, mu, logvar).
    """
    mu, logvar = self.encode(x)
    z = self.reparameterize(mu, logvar)
    return self.decode(z), mu, logvar

reparameterize

reparameterize(mu, logvar)

Apply Gaussian reparameterization trick: z = mu + std * epsilon.

Source code in src/tensoris/models/genai/vae.py

def reparameterize(self, mu: torch.Tensor, logvar: torch.Tensor) -> torch.Tensor:
    """Apply Gaussian reparameterization trick: z = mu + std * epsilon."""
    std = torch.exp(0.5 * logvar)
    eps = torch.randn_like(std)
    return mu + eps * std