Online Real-Time Recurrent Learning Using Sparse Connections and Selective Learning

Paper: link here

ALE policy evaluation dataset: link here

Compilation instructions

Make sure that the following prerequisites are set up in your system:

The code can be compiled by running bash compile.sh in the base directory.

Name		Name	Last commit message	Last commit date
Python_DenseLSTM		Python_DenseLSTM
experiment_configs		experiment_configs
include		include
proto_files		proto_files
pybind_interface		pybind_interface
pytorch_tests		pytorch_tests
src		src
src_py		src_py
.gitignore		.gitignore
CMakeLists.torch.txt		CMakeLists.torch.txt
CMakeLists.txt		CMakeLists.txt
LICENSE		LICENSE
README.md		README.md
atari_experiment.cpp		atari_experiment.cpp
compile.sh		compile.sh
count_configs.cpp		count_configs.cpp
experiment_animal_learning.cpp		experiment_animal_learning.cpp
gradient_gt.py		gradient_gt.py
list_of_inp		list_of_inp
lsm_test_case.cpp		lsm_test_case.cpp
requirements.txt		requirements.txt
train_mnist_bptt.py		train_mnist_bptt.py
train_mnist_constructive.py		train_mnist_constructive.py