RL-Laboratory🤖

Code for the Reinforcement Learning lab of Reinforcement Learning course, MSc degree in Artificial Intelligence 2025/2026 at the University of Verona.

First Set-Up (Conda)

1. Download Miniconda for your System.

2. Install Miniconda

   • On Linux/Mac
     • Use ./Miniconda3-latest-Linux-{version}.sh to install.
     • sudo apt-get install git (may be required).
   • On Windows
     • Double click the installer to launch.
     • NB: Ensure to install "Anaconda Prompt" and use it for the other steps.

3. Set-Up conda environment:

   • git clone https://github.com/Isla-lab/RL-lab
   • conda env create -f RL-Lab/tools/rl-lab-environment.yml

First Set-Up (Python Virtual Environments)

Python virtual environments users (venv) can avoid the Miniconda installation. The following package should be installed:

• scipy, numpy, gym
• jupyter, matplotlib, tqdm
• tensorflow, keras

Assignments

Following the link to the code snippets for the lessons:

• Lesson 1: Multi-Armed Bandit Slides, Code, Results

• Lesson 2: MDP and Gym Environments Slides, Code, Results

• Lesson 3: Monte Carlo RL methods Slides, Code, Results

• Lesson 4: Temporal difference methods Slides, Code, Results

• Lesson 5: Dyna-Q Slides, Code, Results

• Lesson 6: Tensorflow-PyTorch and Deep Neural Networks Slides, Code, Results

• Lesson 7: Deep Q-Network Slides, Code, Results

• Lesson 8: REINFORCE Slides, Code, Results

• Lesson 9: A2C Slides, Code, Results

• Lesson 10: PPO Slides, Code, Setup and full instructions

• Lesson 11: DDPG Slides, Code, Setup and full instructions

• Lesson 12: SAC Slides, Code, Setup and full instructions

• Lesson Extra: Value/Policy Iteration Slides, Code, Results

Tutorials

This repo includes a set of introductory tutorials to help accomplish the exercises. In detail, we provide the following Jupyter notebook that contains the basic instructions for the lab:

• Tutorial 1 - Gym Environment: Here!
• Tutorial 2 - TensorFlow (Keras): Here!
• Tutorial 3 - PyTorch: Here!

---

Contact information

• Teaching assistant: Riccardo Fidanza - riccardo.fidanza@studenti.univr.it
• Professor: Alberto Castellini - alberto.castellini@univr.it