Simple, custom-built autoregressive models for generating MNIST images using PyTorch.
There are many, slightly different implementations/architectures of MNIST image generation in this repository. The main purpose of this repository is to explore how to build an effective autoregressive model for MNIST images using convolutional layers and machine learning techniques in general.
- Pytorch
Just run one of the train codes and it will start training the model and at the end, it will spit out sample images of numbers from 0 to 9.
To be written.
train_2:
train_3:
train_5:
train_6:
