Bug Description
seems like we are missing the type Dict in shape analysis when we parse the input types.
reported error:
terminate called after throwing an instance of 'torch_tensorrt::Error'
what(): [Error thrown at core/partitioning/shape_analysis.cpp:90] Expected to find type Dict(str, Tensor) for value encoder_outputs.1 but get nothing.
Expected behavior
Can be fixed quickly once we support the Dict type parsing in shape analysis.