https://github.com/redleader962/lecturedirigedrlimplementation

Directed reading on Deep Reinforcement Learning
https://github.com/redleader962/lecturedirigedrlimplementation

actor-critic deep-reinforcement-learning policy-gradient soft-actor-critic tensorflow universite-laval

Last synced: 7 months ago
JSON representation

Directed reading on Deep Reinforcement Learning

• Host: GitHub
• URL: https://github.com/redleader962/lecturedirigedrlimplementation
• Owner: RedLeader962
• Created: 2019-08-14T18:22:40.000Z (about 6 years ago)
• Default Branch: master
• Last Pushed: 2023-12-17T03:29:15.000Z (almost 2 years ago)
• Last Synced: 2025-04-13T09:14:07.520Z (7 months ago)
• Topics: actor-critic, deep-reinforcement-learning, policy-gradient, soft-actor-critic, tensorflow, universite-laval
• Language: Python
• Homepage:
• Size: 62.4 MB
• Stars: 2
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
[![codecov](https://codecov.io/gh/RedLeader962/LectureDirigeDRLimplementation/branch/master/graph/badge.svg)](https://codecov.io/gh/RedLeader962/LectureDirigeDRLimplementation)



Repository for the course IFT-7014: Directed reading on

# Deep Reinforcement Learning 

by **Luc Coupal**,

Université Laval,

Montréal, QC, Canada,

[Luc.Coupal.1@uLaval.ca](Luc.Coupal.1@uLaval.ca) 

#### Under the supervision of:

**Professor Brahim Chaib-draa**,

Directeur du programme de baccalauréat en génie logiciel de l'Université Laval,

Québec, QC, Canada,

[Brahim.Chaib-draa@ift.ulaval.ca](Brahim.Chaib-draa@ift.ulaval.ca)

---

![TaxonomyActorCritic](./visual/TaxonomyDRLgithub.png) 

### Essay on:

- Maximum Entropy RL:

  - **[Soft Actor-Critic](DRLimplementation/SoftActorCritic)**

  - See my **blog post [_Soft Actor-Critic_ part 1: intuition and theoretical aspect](https://redleader962.github.io/blog/2020/SAC-part-1-distillarized/)** for more details on _SAC_ and _MaxEnt-RL_

- Classical RL:

  - **[Actor-Critic](DRLimplementation/ActorCritic)**

  - **[Basic policy gradient](DRLimplementation/BasicPolicyGradient)**

- **[A reflexion on design, architecture and implementation details](https://github.com/RedLeader962/LectureDirigeDRLimplementation/raw/master/Reflexion_on_design_and_architecture_LucCoupal_v1-1.pdf)**

---

![Trained agent in action](video/SAC_video/SAC_gif/SAC_postTraining_testOnHardLunar540p24fps.gif)

[Watch mp4 video - Soft Actor-Critic Post training - Test run on 2X harder LunarLanderContinuous-v2 environment](video/SAC_video/SAC_postTraining_testOnHardLunar540p.mp4) 

---

    

### Install instruction:

1) **Create & activate a new virtual environment** (I recommand using [conda](https://www.anaconda.com/distribution/), ... it's a walk in the park)

    ```bash

    conda create --name myNewVirtualEnvironmentName python=3.7

    conda activate myNewVirtualEnvironmentName

    ```

2) **Clone** the GitHub repository & **install dependencies**:

    ```bash

    git clone https://github.com/RedLeader962/LectureDirigeDRLimplementation.git

    cd LectureDirigeDRLimplementation

    pip install -e .

    ```

    This will automaticaly install those **dependencies** in `myNewVirtualEnvironmentName` :

        'gym>=0.14.0'

        'tensorflow>=1.14.0,<2.0',

        'matplotlib>=3.1.0',

        'numpy>=1.16.4',

        'seaborn>=0.9.0',

        'pytest',

    

3) **Enjoy** DRL script

---