Some nodes like MergeMem nodes often tend to get very wide due to many inputs. A lot of them are often just top and do not provide more information. When hiding these with the "Hide other subgraph" filter, it removes the top edges but the MergeMem node itself is still the same size (see mergemem.png). This makes it difficult to see what's going on, especially when opening other nodes. It would be nice if nodes would resize themselves if inputs are hidden. We can still see which index a non-hidden input node has by hovering over the edge.