Hi All,
I am new to Relay/TVM world, and I have read a number of papers as well as a number of posts on this forum. One thing that I have not been able to understand why Relay is better than NNVM? I know NNVM is based on the static computational graph and is that all we need for the standard neural networks? Could someone please elaborate on what Relay provides that NNVM can not?
Best.