menicefellow / dqn-modelunloading

Model Unloading Optimization using DQN

Project Overview

This project implements a reinforcement learning (RL) solution using Deep Q-Networks (DQN) to optimize the decision of unloading a language model on a server. The goal is to minimize resource usage while ensuring the model is available for user requests, thereby balancing performance and cost.

Features

• Simulation of User Behavior: Mimics user interactions with a server to test different strategies for model loading and unloading.
• Deep Q-Networks (DQN): Utilizes a DQN with separate policy and target networks to stabilize the learning process.
• Performance Tracking: Tracks and prints the reward improvement over episodes to evaluate the learning progress.

Dependencies

To run this project, you will need the following:

• Python 3.x
• PyTorch
• NumPy

You can install the necessary libraries using pip:

pip install torch numpy

Setup

1. Clone the repository to your local machine.
2. Ensure that all required dependencies are installed.
3. Navigate to the project directory.

Running the Code

To start the training process, run the following command in your terminal:

python main.py

Output

The script will output the average rewards every 10 episodes to the console, allowing you to monitor the improvement in decision-making ability of the agent over time.

Contributing

Contributions to this project are welcome. Please fork the repository and submit a pull request with your suggested changes.

License

This project is open-source and available under the APACHE 2.0 License.

Discord Server

Join our Discord server here.

Feeling Generous? 😊

Eager to buy me a cup of 2$ coffe or iced tea?🍵☕ Sure, here is the link: https://ko-fi.com/drnicefellow. Please add a note on which one you want me to drink?