Alexandre Eichenberger

@jenkins-droid test this please

@philass since you don't have rights to push down code (I think), let me know when you are ready to do so.

@chentong319 ready to approve? If yes, and @philass is happy with the state of the PR, I can merge it. Tx to all

@chentong319 are you ready to approve this PR, or suggest changes? Tx @philass there are some conflicts that need manual attention.

updating and hopefully the AMD ci will come clean

@jenkins-droid test this please

So our shape inference lost track that the input was 1 for the last dim? Or the input to the model function did not specify 1? What was it @tungld...

@tungld do you mind to look into this one? If you think this might be resolved by the newer scheme on shape inference, then let's work on that one instead...

@tungld is it because the signature is only checked if more precise for element wise ops only?

I like the idea to better track of the warnings, and avoiding them as much as possible. LLVM is clearly an example to follow, as compiling 3000 files resulted in...