Neural Turing Machines (NTM) is a mamory-augmented recurrent deep learning model. I implemented NTM and tested it in the sequence copying task as described in the paper.
Note that because the project was done in 2017, some usages of pytorch were obsolete, like torch.autograd.Variable. I have modified the code so that it can barely work with Pytorch version >= 1.0.
- Python3
- Pytorch >= 1.0
Directly execute main.py to run the experiment.