coremltools.converters.convert gives a RuntimeError during conversion to .mlmodel

[sheshap]

🐞Describe the bug

We have followed the instructions in Model Scripting to convert our custom PyTorch model to .mlmodel. Please find our google colab notebook here

Stack Trace

[/usr/local/lib/python3.7/dist-packages/coremltools/converters/mil/frontend/torch/converter.py](https://localhost:8080/#) in _lower_graph_block(graph)
    377                 attr_name = getattr(node, node.kindOf("name"))("name")
    378 
--> 379                 module = getattr(node_to_module_map[_input], attr_name)
    380                 node_to_module_map[_output] = module
    381 

KeyError: images.7 defined in (%images.7 : __torch__.torchvision.models.detection.image_list.ImageList, %targets.31 : Dict(str, Tensor)[]? = prim::TupleUnpack(%427)
)

To Reproduce

Run our google colab notebook to reproduce the issue.

Model conversion fails in the above Google Colab notebook.

System environment (please complete the following information):

PyTorch version: 1.11.0+cu113 Is debug build: False CUDA used to build PyTorch: 11.3 ROCM used to build PyTorch: N/A

OS: Ubuntu 18.04.5 LTS (x86_64) GCC version: (Ubuntu 7.5.0-3ubuntu1~18.04) 7.5.0 Clang version: 6.0.0-1ubuntu2 (tags/RELEASE_600/final) CMake version: version 3.22.4 Libc version: glibc-2.26

Python version: 3.7.13 (default, Apr 24 2022, 01:04:09) [GCC 7.5.0] (64-bit runtime) Python platform: Linux-5.4.188+-x86_64-with-Ubuntu-18.04-bionic Is CUDA available: True CUDA runtime version: 11.1.105 GPU models and configuration: GPU 0: Tesla T4 Nvidia driver version: 460.32.03 cuDNN version: Probably one of the following: /usr/lib/x86_64-linux-gnu/libcudnn.so.7.6.5 /usr/lib/x86_64-linux-gnu/libcudnn.so.8.0.5 /usr/lib/x86_64-linux-gnu/libcudnn_adv_infer.so.8.0.5 /usr/lib/x86_64-linux-gnu/libcudnn_adv_train.so.8.0.5 /usr/lib/x86_64-linux-gnu/libcudnn_cnn_infer.so.8.0.5 /usr/lib/x86_64-linux-gnu/libcudnn_cnn_train.so.8.0.5 /usr/lib/x86_64-linux-gnu/libcudnn_ops_infer.so.8.0.5 /usr/lib/x86_64-linux-gnu/libcudnn_ops_train.so.8.0.5 HIP runtime version: N/A MIOpen runtime version: N/A Is XNNPACK available: True

Versions of relevant libraries: [pip3] numpy==1.21.6 [pip3] torch==1.11.0+cu113 [pip3] torchaudio==0.11.0+cu113 [pip3] torchsummary==1.5.1 [pip3] torchtext==0.12.0 [pip3] torchvision==0.12.0+cu113 [conda] Could not collect

[TobyRoseman]

@sheshap - I can't even view your notebook. It is not public. Also sharing a notebook to reproduce the problem is not very helpful. Can you please work to create a minimal code example that reproduces the error?

[sheshap]

@TobyRoseman I apologize. I have edited it now. Please access now. The notebook has the code to reproduce the issue.

[TobyRoseman]

It looks like you are using a torch script model. Our support for torch script is experimental. Can you try tracing your model, then try converting the traced model?

[sheshap]

We have an ongoing issue with torch trace. We were suggested to use the torch script here. Can you please check for the torch script?