You're reading from Deep Reinforcement Learning Hands-On A practical and easy-to-follow guide to RL from Q-learning and DQNs to PPO and RLHF

Product type Paperback

Published in Nov 2024

Publisher Packt

ISBN-13 9781835882702

Length 716 pages

Edition 3rd Edition

Languages

Python

Tools

PyTorch

Concepts

Deep Reinforcement Learning

Author (1):

Maxim Lapan

View More author details

Table of Contents (29) Chapters

Preface

1. Part 1 Introduction to RL FREE CHAPTER

2. What Is Reinforcement Learning?

3. OpenAI Gym API and Gymnasium

4. Deep Learning with PyTorch

5. The Cross-Entropy Method

6. Part 2 Value-based methods

7. Tabular Learning and the Bellman Equation

8. Deep Q-Networks

9. Higher-Level RL Libraries

10. DQN Extensions

11. Ways to Speed Up RL

12. Stocks Trading Using RL

13. Part 3 Policy-based methods

14. Policy Gradients

15. Actor-Critic Method: A2C and A3C

16. The TextWorld Environment

17. Web Navigation

18. Part 4 Advanced RL

19. Continous Action Space

20. Trust Region Methods

21. Black-Box Optimizations in RL

22. Advanced Exploration

23. Reinforcement Learning with Human Feedback

24. AlphaGo Zero and MuZero

25. RL in Discrete Optimization

26. Multi-Agent RL

27. Bibliography

28. Index

Results

The final result published in the paper is quite impressive. After 44 hours of training on a machine with three GPUs, the network learned how to solve cubes at the same level as (and sometimes better than) human-crafted solvers. The final model has been compared against the two solvers described earlier: the Kociemba two-stage solver and Korf. The method proposed in the paper is named DeepCube.

To compare efficiency, 640 randomly scrambled cubes were used in all the methods. The depth of the scramble was 1,000 moves. The time limit for the solution was an hour and both the DeepCube and Kociemba solvers were able to solve all of the cubes within the limit. The Kociemba solver is very fast, and its median solution time is just one second, but due to the hardcoded rules implemented in the method, its solutions are not always the shortest ones.

The DeepCube method took much more time, with the median time being about 10 minutes, but it was able to match the length...

The rest of the chapter is locked

You're reading from Deep Reinforcement Learning Hands-On A practical and easy-to-follow guide to RL from Q-learning and DQNs to PPO and RLHF

Table of Contents (29) Chapters

Results

Authors (1)

Personalised recommendations for you

You're reading from Deep Reinforcement Learning Hands-On A practical and easy-to-follow guide to RL from Q-learning and DQNs to PPO and RLHF

Table of Contents (29) Chapters

Results

Unlock this book and the full library FREE for 7 days

Authors (1)

Personalised recommendations for you