AutoEncoder implementation in Pytorch trained on the MNIST data.
Trains an AutoEncoder on MNIST data, calculates test error and saves the trained model
python train.py --lr 0.001 --alpha 0.5 --batch_size 32 --epochs 10 --noise_var 0.5 --valid_split 0.22 --loss bce --plot_images
-- plot_kernel --save_dir ./outputs/ --model_file 'mnist_autoencoder.pt
Loads the trained model, samples few images from the dataset and reduces their generated noise and generates new images.
python inference.py ---model_file mnist_autoencoder.pt --save_dir ./outputs/ --noise_var 0.5 --imgs_to_gen 3
--imgs_to_denoise 3
Clone the repository:
git clone [email protected]:vulcans/vulns_sols_matching.git
Package main contributer is Keren Gaiger: [email protected]