PyTorch implementation of Attentive Recurrent Comparators by Shyam et al.
A blog explaining Attentive Recurrent Comparators
This is repository for testing ARC on custom (digits) dataset.
Download our data and unzip two folders into data folder of this repository.
python train.py --cuda
The training should achieve around 85%+ accuracy on test data.
python viz.py --cuda --load name_of_model --same
Run with exactly the same parameters as train.py and specify the model to load. The script dumps images to a directory in visualization. The name of directory is taken from --name parameter if specified, else name is a function of the parameters of network. name_of_model should be in data folder.
python test_model.py --cuda --load name_of_model
name_of_model should be in data folder.