Use ONNX parser to create SequenceMap based models

[jantonguirao]

Signed-off-by: Joaquin Anton janton@nvidia.com

Description

• Changed Batch processing tutorial to use the more intuitive ONNX parser API to represent the model

Motivation and Context

• The motivation is to encourage users to use ONNX parser

[jantonguirao]

@jcwchen This is pending review for a while. Can you please help me find the right person to review it? Thanks

[jcwchen]

@jcwchen This is pending review for a while. Can you please help me find the right person to review it? Thanks

Thank you for the reminder! I will help. I was trying to run this notebook: regarding this line

onnxruntime.InferenceSession("tutorials/pad_to_largest.onnx")

I bumped into: Fail: [ONNXRuntimeError] : 1 : FAIL : Load model from tutorials/pad_to_largest.onnx failed:Node () output arg (max_shape) type inference failed . Is it expected? I was using ORT 1.14.1 + ONNX 1.13.1.

[jantonguirao]

@jcwchen This is pending review for a while. Can you please help me find the right person to review it? Thanks

Thank you for the reminder! I will help. I was trying to run this notebook: regarding this line

onnxruntime.InferenceSession("tutorials/pad_to_largest.onnx")

I bumped into: Fail: [ONNXRuntimeError] : 1 : FAIL : Load model from tutorials/pad_to_largest.onnx failed:Node () output arg (max_shape) type inference failed . Is it expected? I was using ORT 1.14.1 + ONNX 1.13.1.

Right. I forgot to mention that. There was a bug, @gramalingam fixed it here: onnx/onnx#4880
I guess we can wait for the next ONNX release before merging.

[jcwchen]

Right. I forgot to mention that. There was a bug, @gramalingam fixed it here: onnx/onnx#4880
I guess we can wait for the next ONNX release before merging.

Thank you for the info. Although that fix has been included in the fresh ONNX 1.13.1 patch, I guess we still need to wait for ONNX Runtime consuming that ONNX commit. (ONNX Runtime 1.14.1 still uses ONNX 1.13.0 commit)