Skip to content
Source code of Neural Logic Reinforcement Learning (https://arxiv.org/abs/1904.10729)
Branch: master
Clone or download
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
core remove useless imports (2) Apr 29, 2019
model/ICMLtest init Apr 27, 2019
.gitignore Initial commit Apr 27, 2019
LICENSE Initial commit Apr 27, 2019
Plot.ipynb init Apr 27, 2019
README.md Update README.md Apr 27, 2019
main.py remove useless imports, add windycliffwalking Apr 29, 2019

README.md

Neural Logic Reinforcement Learing

Implementaion of Neural Logic Reinforcement learning and several benchmarks. Neural Logic Reinforcement Learning uses deep reinforcement leanring methods to train a differential indutive logic progamming architecture, obtaining explainable and generalizable policies. Paper accepted by ICML2019.

Enviornments

Developed in python2.7, Linux enviornment.

Dependencies

  • numpy
  • tensorflow

User Guide

  • use main.py to run the experiments
  • --mode= to specify the running mode, can be "train" or "generalize", where generalize means to run a generalization test.
  • --task= to specify the task, can be "stack", "unstack", "on" or "cliffwalking".
  • --algo to specify agent type, can be "DILP", "NN" or "Random"
  • --name to specify the id of this run.
  • for example: python main.py --mode=train --algo=DILP --task=unstack --name=ICMLtest
You can’t perform that action at this time.