Skip to content

Latest commit

 

History

History
22 lines (12 loc) · 630 Bytes

README.md

File metadata and controls

22 lines (12 loc) · 630 Bytes
Raw

Q-Learning

a) A simple tabular Q learning app for OpenAIGym CartPole v1 model


b) A simple deep Q learning app with one layer for OpenAIGym CartPole v1 model
  • It works for at least 400000 steps

c) A simple deep Q learning try with PyTorch for OpenAIGym CartPole v1 model
  • It doesn't work for more than one layer, It cannot be solved.

d) A simple Policy Gradient application for OpenAIGym CartPole v1 model

  • It generally converges and when it converges it works for at least 5000(because I only tried 5000), probably infinite steps. If not converges, run it one more times.