You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
That we did not realize %10 is a view of %y and %16 is a reference of %10, so we early terminate translation when we see the output %y has been created
We probably won't be able to fully fix this, since the solution is equivalent to torch's own functionalization 馃槀 But we do should find some way to error out, rather than silently produces a wrong Core ML model
The text was updated successfully, but these errors were encountered:
馃悶Describing the bug
This simple model can be converted but gives incorrect result
The issue is in how we parse the TorchScript
That we did not realize
%10
is a view of%y
and%16
is a reference of%10
, so we early terminate translation when we see the output%y
has been createdPotential Solution
We probably won't be able to fully fix this, since the solution is equivalent to torch's own functionalization 馃槀 But we do should find some way to error out, rather than silently produces a wrong Core ML model
The text was updated successfully, but these errors were encountered: