mvanhorn opened a new pull request, #19816:
URL: https://github.com/apache/tvm/pull/19816
## Summary
The Relax ONNX frontend's GridSample._impl_v16 converter unconditionally
permutes the grid from ONNX [N,H,W,2] to TVM [N,2,H,W] and calls
image.grid_sample with layout="NCHW". For 5D volumetric inputs ([N,C,D,H,W]
with grid [N,D,H,W,3]) this crashes at permute_dims with an InternalError
('PermuteDims expects the number of input axes to equal the ndim of the input
tensor.
## Changes
In GridSample._impl_v16, read data.struct_info.ndim and dispatch on rank.
For ndim==4, keep the existing permute_dims(grid,[0,3,1,2]) +
grid_sample(layout="NCHW").
Fixes #19688
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
