MNIST Variational Autoencoder
This repository contains a Variational Autoencoder (VAE) trained on the MNIST handwritten digit dataset.
Model
- Model: Variational Autoencoder (VAE)
- Dataset: MNIST
- Framework: PyTorch
- Latent Dimension: 20
Description
The VAE learns a latent representation of handwritten digits and can reconstruct input images and generate new digit images from the latent space.
Training
The model is trained using:
- Reconstruction Loss
- KL Divergence
The total VAE loss is the combination of reconstruction loss and KL divergence.
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support