Skip to content

DrLux/Planpix

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

47 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

This repo is from my Master's degree thesis work develped at Addfor s.p.a
I used PlaNet to prove that model-based DRL can overcome the model-free algorithms in terms of sample efficiency. My implementation of PlaNet is based on the Kaixhin one, but I reach better results. I also experiment with a regularizer based on DAE to reduce the gap between the real and the predicted rewards.
The company asks me to note publish that feature, but you can find all the explanations in my blog article (you can also contact me).

Full trained agent

PlaNet Overview

General overview of Planet model architecture. If you want a full explanation, click on it! Blog_article

Medium Articles

Results

ceetah_planet_vs_ddpg cartpole_planet_vs_ddpg reacher_planet_vs_ddpg walker_planet_vs_ddpg my_planet_vs_soa

Comparison's data are form: Curl: Contrastive unsupervised representations for reinforcement learning. Laskin, M., Srinivas, A., & Abbeel, P. (2020, July)

Requirements

Links

Acknowledgements

References

[1] Learning Latent Dynamics for Planning from Pixels
[2] Overcoming the limits of DRL using a model-based approach

Releases

No releases published

Packages

No packages published

Languages