mirror of https://github.com/llvm/torch-mlir
2b1b0f6e19
The preserve memory specifies that `If any of the input tensors is in channels_last format, operator output should be in channels_last format` and hence can be added as is in aten_empty_like op. |
||
---|---|---|
.. | ||
linalg_on_tensors_backends | ||
test_suite | ||
torchscript | ||
tosa_backends | ||
CMakeLists.txt | ||
__init__.py | ||
utils.py |