naivoder/DreamWeaver

Continual Meta-Reinforcement Learning for Operating System Optimization Tasks with a Shared World Model and Adaptive Compositional Policy Initializations

DreamWeaver

DreamWeaver is an implementation of continual meta-reinforcement learning that builds upon the DreamerV3 model. This repository features a shared world model and adaptive compositional policy initializations for task-specific Actor-Critic networks. The framework is evaluated on the Park environment, which encompasses 12 distinct operating system optimization tasks.

Table of Contents

• Installation
• Usage

Installation

Prerequisites

We recommend using Conda for managing dependencies and environments.

Steps

1. Clone the repository:

   git clone https://github.com/naivoder/DreamWeaver.git
   cd DreamWeaver

2. Create and activate a Conda environment:

   conda create --name dreamweaver python=3.10
   conda activate dreamweaver

3. Install required packages:

   pip install -r requirements.txt

Usage

🚫👷 UNDER CONSTRUCTION!!!