The ECCV paper was accepted, so I put a few notes on it. I tried to be as neutral as possible as we do not want to deviate from the initial story-line. Understanding FiLM (rather than making FiLM more complex).
However, I can easily imagine a new section that can be created in the following months to keep updating the article: New FiLM-Based Models. At this points, It may be worth to add a new diagram to better explain multi-hop FiLM.
Note that the paper is not yet on arxiv, and it will be available after the deadline.
For now is it ok to not put an url to the paper? If you do want an url, I can create a link on my personal website with the paper and we change it on the next update.
The ECCV paper was accepted, so I put a few notes on it. I tried to be as neutral as possible as we do not want to deviate from the initial story-line. Understanding FiLM (rather than making FiLM more complex). However, I can easily imagine a new section that can be created in the following months to keep updating the article: New FiLM-Based Models. At this points, It may be worth to add a new diagram to better explain multi-hop FiLM.
Note that the paper is not yet on arxiv, and it will be available after the deadline. For now is it ok to not put an url to the paper? If you do want an url, I can create a link on my personal website with the paper and we change it on the next update.