Here, the 11th graders are practicing calculating weight gradients by hand in neural networks. This video illustrates the rigor with which the students learn the concepts -- they don't just import a machine learning model from an existing library; rather, they learn how the underlying math works so well that they're able to build the machine learning models from scratch.
Here, the 11th graders are having an enthusiastic discussion about their Space Empires game implementations. They found some weird behavior in the games they coded up and were discussing the causes and brainstorming potential ways to fix it. Towards the end of the clip, they are joking around a bit.
First half (0:00-53:00): The 10th graders present the initial versions of their Space Empires game implementations, which they worked on for about 2 weeks. They're able to speak cogently about their work and are learning how to present their code at a high level (instead of going line-by-line through their code). We note the pros and cons of different students' implementations.
Second half (53:00-74:00): The 11th graders present their tic-tac-toe implementations, which they worked on for just one assignment (2-3 hours). They had already built much more advanced Space Empires games over the course of many months, but the fact that they were able to implement these tic-tac-toe games in just 2-3 hours shows how fluent they've become with coding and design patterns. They are also able to speak cogently about their work and present their code at a high level.