Pytorch implementation of DRAW: A Recurrent Neural Network For Image Generation on the MNIST generation task.
| With Attention |
|---|
 |
python3 train.py downloads the MNIST dataset to ./data/mnist and train the DRAW model with attention for both reading and writing. After training, the weights files are written to ./save/weights_final.tar and the generated images are written to ./image/.png
python3 generate.py loads wieghts from save/weights_final.tar and generates images
The weights_final.tar file is trained for 50 epoch with minibatch size 64 on GTX 970 GPU.
- TensorFlow implement: https://github.com/ericjang/draw
- Base repository: https://github.com/chenzhaomin123/draw_pytorch