Why it matters

• New users need clear, practical examples to understand how to use DQN in Fehu.
• A demo with code, explanations, and visualizations will make it easier to learn and experiment.
• Improves documentation and lowers the barrier to adoption for contributors and researchers.

Your task

• Create a Markdown notebook (.mdx) using MDX that demonstrates:
  • How to set up and train a DQN agent on a simple environment.
  • How to visualize training progress (e.g., episode rewards, Q-values).
  • Step-by-step explanations for each part of the process.
• Ensure the demo runs as an MDX document, leveraging existing examples of executable code blocks in .mdx files within the codebase (instead of Quill CLI).

Tips

• Use a simple environment such as CartPole or GridWorld for clarity.
• Combine code snippets and markdown explanations in a readable, tutorial-style format.
• Use available hugin library for visualizations.
• Keep the demo beginner-friendly, well-commented, and easy to follow.

Done when

• The MDX document runs successfully and reproduces the full training + visualization pipeline.
• Each step (setup, training, visualization) is clearly explained and illustrated.
• The demo is easy for new users to follow and experiment with.

Related issue

Builds on and extends the save/load implementation from #106.