/DDPG-PER-PEN

The DDPG algorithm incorporates Actor-Critic Deep Learning Agent for solving continuous action reinforcement learning problems.

Primary LanguagePython

DDPG with Parametric Noise Exploration & Prioritized Experience Replay Buffer

This repository implements a DDPG agent with parametric noise for exploration and prioritized experience replay buffer to train the agent faster and better for the openai-gym's "LunarLanderContinuous-v2".

Let's see how fast and better it is!

Agent Profile

  1. DDPG + PNE + PER

  2. Vanilla DDPG (I like Vanilla that is why! :>)

Dependencies

Install dependencies using this,

pip install -r requirements.txt

TODO

- Add adaptive parametric noise for exploration

Contact