Skip to content

Latest commit

 

History

History
13 lines (10 loc) · 621 Bytes

README.md

File metadata and controls

13 lines (10 loc) · 621 Bytes

RL_InvertedPendulum

Solving Inverted Pendulum with Reinforce Learning

Implemented

  • A basic system of inverted pendulum and training pipeline.
  • Action value function in form of either discrete table or rbf kernel function.
  • Sarsa & Q-Learning with/without TD-λ and eligibility traces.

Examples

  • run python main.py to train, it will save the results into ./output or some other where determined by your config.
  • run python analysis.py to draw trace during test. Please change path to the checkpoint in the code.

image