Cartpole problem extended to be solved with DQN (month 21 and 22)
This month we focused on the two following goals:
- Making my RTX GeForce 3080 work successfully. Tensorflow did not work as expected, so Pytorch was chosen as the GPU deep RL framework to use
- Solve the cartpole classic openAiGym proposed problem in RL-Studio now using DQN and making the solution more stable
Due to this integration, the rl-studio project has been refactored. Additionally, a new DQN algorithm and its corresponding cartpole dqn-related problem was added to enable its further refinement.