To answer your question at the end there - you are indeed allowed to go back and redo past lessons, once you've completed the course.

Now, starting with the first section - the cylinders around arbitrary minor axes - there are some pretty big issues that stand out to me.

The first of these is the fact that you appear to have made a purposeful, concerted effort to draw all the cylinders such that their side edges remain completely parallel on the page. That is fundamentally incorrect, and it's something we covered way back in Lesson 1 - the core principle of perspective that lines that are parallel in 3D space converge towards a shared vanishing point when drawn in 2D (like on a piece of paper, or a canvas). The only situation where this vanishing point goes "to infinity" (resulting in those lines becoming parallel on the page) is when the form is aligned in a very specific way relative to the viewer - specifically running perpendicular to the direction the viewer is looking. That is something that occurs when we mean for it to intentionally, but something that simply doesn't happen when drawing a bunch of random, freely rotated forms.

In the homework section for this challenge, where the work was actually assigned, I have there written in bold:

Be sure to vary the rate of foreshortening across your cylinders across the set

So at the very least there should have been a much wider range of cylinders with different rates of foreshortening - some more dramatic, with more rapid convergence, and some shallower with more gradual convergence. That said, even then having no convergence at all would be incorrect.

The foreshortening on a form basically gives the viewer a way to determine in rough terms how long the cylinder is. If the foreshortening is more dramatic, there's going to be more distance between the two ends. If the foreshortening is shallower, then there's less distance between the ends. When there's no foreshortening, as in your cylinders, there is no distance between those ends - but this is something we can obviously see to be false, leading to a visual contradiction in your drawing.

Taking it all a step farther, foreshortening doesn't just manifest in the shift from one end being larger in scale, to the farther end being smaller. We also see it in the width of these ends, with the closer end being narrower, and the farther end being wider, as explained here. If there's no foreshortening at all, then there should be no shift of either sort (no shift in scale, no shift in width/degree). They need to be consistent.

Now, fortunately your cylinders in boxes are much better. You're doing a good job of checking the convergences of those line extensions, which is helping you to improve your ability to estimate the proportions of those boxes. After all, if the boxes don't feature a pair of planes that are proportionally square, then the line extensions of the ellipses won't line up with those of the box itself. That is the key to the exercise - by adjusting the way we draw the boxes, the ellipses inside them line up better with the characteristics of circles in 3D space. The whole thing is just designed to make you better at drawing square faces, which is very valuable in the next lesson. You've done a pretty solid job at that, and your estimation of your convergences are coming along well.

So, you basically bungled the first part, and nailed the second one. So I'm going to have you do an additional 50 cylinders around arbitrary minor axes, this time incorporating lots of variation in foreshortening.