Multiple Inputs/Outputs with torch.onnx.dynamo_export

[khari998]

🐛 Describe the bug

I am getting an error trying to export a model with multiple inputs and outputs.

 dummy_input = torch.randn(
    1, 100, 400
).to(device)
dummy_layer_one_mask = torch.randint(
    0, 2, (1, 10)
).to(device)
dummy_layer_two_mask = torch.randint(
    0, 2, (1,10)
).to(device)

onnx_model = torch.onnx.dynamo_export(
    actor_model,
    (dummy_input, dummy_layer_one_mask, dummy_layer_two_mask),
)

onnx_model.save(f"{model_directory}/model/actor.onnx")

And for the output of the forward method, my model has 4 outputs.
I am returning two maps that have strings as keys and tensors as values for the first two outputs and two tensors for the last two outputs.

I keep getting the error

439, in dynamo_export
    raise OnnxExporterError(
torch.onnx.OnnxExporterError: Failed to export the model to ONNX. Generating SARIF report at 'report_dynamo_export.sarif'. SARIF is a standard format for the output of static analysis tools. SARIF logs can be loaded in VS Code SARIF viewer extension, or SARIF web viewer (https://microsoft.github.io/sarif-web-component/). Please report a bug on PyTorch Github: https://github.com/pytorch/pytorch/issues

and when I tried dragging the sarif file into the website listed, i get a blank page that says no results found.

Are multiple inputs suppored when converting to onnx with dynamo_export? Are dictionaries supported as outputs? From what I understood, ONNX supports maps as outputs. I have a complex architecture so flattening all of these outputs into a single tensor is not ideal at all for me and I would prefer to keep my four outputs in the data structure they currently are in

Versions

Collecting environment information...
PyTorch version: N/A
Is debug build: N/A
CUDA used to build PyTorch: N/A
ROCM used to build PyTorch: N/A

OS: macOS 14.5 (arm64)
GCC version: Could not collect
Clang version: 15.0.0 (clang-1500.3.9.4)
CMake version: Could not collect
Libc version: N/A

Python version: 3.12.3 (main, Apr 9 2024, 08:09:14) [Clang 15.0.0 (clang-1500.3.9.4)] (64-bit runtime)
Python platform: macOS-14.5-arm64-arm-64bit
Is CUDA available: N/A
CUDA runtime version: Could not collect
CUDA_MODULE_LOADING set to: N/A
GPU models and configuration: Could not collect
Nvidia driver version: Could not collect
cuDNN version: Could not collect
HIP runtime version: N/A
MIOpen runtime version: N/A
Is XNNPACK available: N/A

CPU:
Apple M1

Versions of relevant libraries:
[pip3] No relevant packages
[conda] Could not collect

[justinchuby]

Multiple inputs are supported. You would need

onnx_model = torch.onnx.dynamo_export(
    actor_model,
    dummy_input, dummy_layer_one_mask, dummy_layer_two_mask,
)

(unpack the tuple)