Looking at the overall structure of the article, would it make sense to swap the ordering of "Related ideas" and "Feature-wise transformations in the literature" sections?
For deep learning beginners, it may be helpful to first discuss where FiLM has directly been used, and then transition to the broader discussion of what other bigger ideas FiLM relates to.
For deep learning experts, now that the sections are collapsible, they can see right away that things like attention and bilinear transformation are also addressed; these readers can then just jump directly to those sections.
Looking at the overall structure of the article, would it make sense to swap the ordering of "Related ideas" and "Feature-wise transformations in the literature" sections?