Skip to content

PPO Dash: Improving Generalization in Deep Reinforcement Learning

Notifications You must be signed in to change notification settings

Sohojoe/ppo-dash

Repository files navigation

PPO-Dash

Code for reproducing the results found in PPO Dash: Improving Generalization in Deep Reinforcement Learning

About

PPO-Dash is a modified version of the PPO algrothem that utalises the following optimizations and best practices:

  • Action Space Reduction
  • Frame Stack Reduction
  • Large Scale Hyperparameters
  • Vector Observations
  • Normalized Observations
  • Reward Hacking
  • Recurrent Memory

PPO-Dash was able to solve the first 10 levels of the Obsticle Tower Enviroment without the need for demonstrations or curosity based algorthemic enhancements.

The version of PPO-Dash in the technical paper, placed 2nd in Round One of the Obsticle Tower Challenge with an average score of 10. We were able to reproduce this score in Round Two of the challenge, with a minor modifiaction (randomizing the themes during in training). We placed 4th overall, with a score of 10.8 with the addition of demonstrations.

Reproducing Results

To reproduce the results listed in the paper and for round one of the competition, see ReproduceRound1

Acknowlegements

This codebase derives from pytorch-a2c-ppo-acktr - #8258f95

Citation

If you use PPO-Dash in your research, we ask that you cite the technical report as a reference.

About

PPO Dash: Improving Generalization in Deep Reinforcement Learning

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published