Open-source, reproducible visual explainers for AI, software, and technical concepts.
Watch them. Reproduce them. Remix them.
The code and templates are under Apache-2.0. The original videos and soundtracks are licensed under CC BY 4.0. If you build something cool with it, a link back to this project is always appreciated, but not required for the code.
Each explainer is self-contained.
Yourself:
git clone https://github.com/sujee/visual-explainers.git
cd visual-explainers/explainers/neural-network-training
# install the prerequisites listed in its README.md, then:
./render.sh # all deliverables, final quality (4K, takes a while)
./render.sh draft # quick draft previewWith a coding agent: start it in the explainer's folder (or the repository root) and ask:
Read AGENTS.md and README.md. Install the prerequisites and render the final video.
Start a coding agent in the repository root (the new explainer's folder doesn't exist yet) and ask, for example:
Create a new explainer about prefill vs decode.
It starts by writing a brief.md for you to review, then renders drafts (v1, v2, …) for your feedback. Once you approve a version, ask it to prepare the Repro Bundle before committing.
Or start by hand: copy templates/explainer/ to explainers/<name>/ and fill in the placeholders.
All the videos are in one YouTube playlist.
| Explainer | What it teaches | Watch | Built with |
|---|---|---|---|
| neural-network-training | How a network learns - forward pass, error, backpropagation, weight updates | Video · Short · 4K release | Manim (Python) |
| neural-network-training-cartoon | The same explainer in a bright cartoon style, with a bouncy soundtrack | Video · Short · 4K release | Manim (Python) |