-
Notifications
You must be signed in to change notification settings - Fork 7
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add converter for index.Tensor_hacked_twin #98
Add converter for index.Tensor_hacked_twin #98
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM! Can we also try to add it to the aot_compile test suite for e2e coverage?
%int1 = torch.constant.int 1 | ||
%int4 = torch.constant.int 4 // this is a dtype on arange.... | ||
%int-1 = torch.constant.int -1 | ||
%arange = torch.aten.arange.start_step %int0, %int1, %int1, %int4, %none, %none, %none : !torch.int, !torch.int, !torch.int, !torch.int, !torch.none, !torch.none, !torch.none -> !torch.vtensor<[1],si64> |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
torch.aten.arange.start_step
will remain unchanged right? As we don't handle it in tcp.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Currently this remains unchanged. I was thinking that we should add an optimization pass that removes this when it selects everything, but I have yet to do that
…so gather custom op seems to be required
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM for GatherOp::verify()
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM! Thanks for working on this tricky case.
tcp.custom_op
variants ofTensor_hacked_twin
tcp.gather
andtcp.const
to ensure that is used correctly