https://github.com/jihoonerd/actor-critic

Vanilla Actor Critic
https://github.com/jihoonerd/actor-critic

actor-critic cartpole-v1 reinforcement-learning

Last synced: 4 months ago
JSON representation

Vanilla Actor Critic

• Host: GitHub
• URL: https://github.com/jihoonerd/actor-critic
• Owner: jihoonerd
• License: mit
• Created: 2021-02-06T11:26:34.000Z (over 4 years ago)
• Default Branch: main
• Last Pushed: 2024-04-17T14:48:26.000Z (about 1 year ago)
• Last Synced: 2025-01-15T17:32:09.412Z (5 months ago)
• Topics: actor-critic, cartpole-v1, reinforcement-learning
• Language: Python
• Homepage:
• Size: 715 KB
• Stars: 2
• Watchers: 4
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Vanilla Actor Critic

This implements a vanilla actor critic architecture.

## Architecture

This is a simple actor critic architecture. Actor and critic networks share shallow two dense layers (1024, 512).

## Demo

![900](assets/eps-900.gif)

## Score graph of `CartPole-v1`

![graph](assets/score_fig.png)

## Environment

* Python 3.8

* Tensorflow 2.4

* Tensorflow Probability 0.12