Skip to content

Latest commit

 

History

History
39 lines (30 loc) · 1.25 KB

README.md

File metadata and controls

39 lines (30 loc) · 1.25 KB

Mountain Car

Simple Solvers for MountainCar-v0 and MountainCarContinuous-v0 @ gym. Methods including Q-learning, SARSA, Expected-SARSA, DDPG and DQN.

Demo

Testing Environment

  • gym
  • pytorch 1.3.1
  • torchvision 0.4.2

MountainCar-v0

Before run any script, please check out the parameters defined in the script and modify any of them as you please.

Train with Temporal-Difference Method

python TD.py

TODO: Train with DQN Method

Adapted from REINFORCEMENT LEARNING (DQN) TUTORIAL in pytorch tutorials, which originally deals with CartPole Problem.

DQN method has not been run and tested.

python DQN.py

inference with Temporal-Difference Method

python inference.py

MountainCarContinuous-v0

Train with DDPG

Adapted from Deep Reinforcement Learning Algorithms with PyTorch but rewritten in complete pytorch format, and redundant functions are removed.

python train_continuous.py