Tong Chen

@amanwithfruit Above is the test case.

I saw upsample beams deprecated. Can we use onnx.resize to express onnx.upsample?

Comment just based on PR description: Is this op used to transform the layout or to mark the layout? It seems to me that the operation of an "onnx.onnx.LayoutTransform" is...

@AlexandreEichenberger I made changes according to your comments.

Agree. Similar to check-onnx-backend with only --EmitONNXBasic?

The special problem with shape inference is that there is a binding for the block arguments and the merging for the joined control flow. Not sure about whether MLIR can...

Shape inference for regions Regions could can from IfOp, LoopOp or functionOp. There be be more than one control flow going into a region. For example, a functionOp may have...

> > 9\. repeat step 2 until S_arguments reaches fixed point > > this repeat is done by the loop in `ONNXOpTransformPass::runOnOperation()`, right? No, we do not have that part...

Can you provide your .onnx file?

The definition of onnx ops in onnx-mlir is imported from onnx package through script onnx-mlir/utils/gen_onnx_mlir.py. Are the customize ops you want to add defined with the same data structure but...