mirror of https://github.com/llvm/torch-mlir
282e9b0e64
In practice, this was caught by the way that AOT autograd traces `convolution_backward`. For the unit test, we just repro it with a custom op. |
||
---|---|---|
.. | ||
fx_decomp_util.py | ||
fx_importer.py | ||
onnx_importer.py |