I am trying to use export custom model to a executorch

Hi I am trying to export a customer model to executorch, I was able to convert the model to a graph representation using the following code

import torch
from torch.export import export, ExportedProgram
from executorch.exir import EdgeProgramManager, to_edge

variant='edgeface_xxs_q'
model = torch.hub.load('otroshi/edgeface', variant, source='github', pretrained=True)
model.eval()

example_args = (torch.randn(1, 3, 112, 112),)
aten_dialect: ExportedProgram = export(model, example_args)
print(aten_dialect)

But getting an error while lowering to the Edge Dialect

edge_program: EdgeProgramManager = to_edge(aten_dialect)
print("Edge Dialect Graph")
print(edge_program.exported_program())

Error:

Traceback (most recent call last):
  File "/my-disk/bob.paper.tbiom2023_edgeface/test.py", line 51, in <module>
    edge_program: EdgeProgramManager = to_edge(aten_dialect)
  File "/home/dinusha/anaconda3/envs/executorch/lib/python3.10/site-packages/executorch/exir/program/_program.py", line 1142, in to_edge
    program = program.run_decompositions(_default_decomposition_table())
  File "/home/dinusha/anaconda3/envs/executorch/lib/python3.10/site-packages/torch/export/exported_program.py", line 114, in wrapper
    return fn(*args, **kwargs)
  File "/home/dinusha/anaconda3/envs/executorch/lib/python3.10/site-packages/torch/export/exported_program.py", line 1003, in run_decompositions
    return _decompose_exported_program(
  File "/home/dinusha/anaconda3/envs/executorch/lib/python3.10/site-packages/torch/export/exported_program.py", line 617, in _decompose_exported_program
    gm, new_graph_signature = _decompose_and_get_gm_with_new_signature_constants(
  File "/home/dinusha/anaconda3/envs/executorch/lib/python3.10/site-packages/torch/export/exported_program.py", line 410, in _decompose_and_get_gm_with_new_signature_constants
    gm, graph_signature = aot_export_module(
  File "/home/dinusha/anaconda3/envs/executorch/lib/python3.10/site-packages/torch/_functorch/aot_autograd.py", line 1246, in aot_export_module
    fx_g, metadata, in_spec, out_spec = _aot_export_function(
  File "/home/dinusha/anaconda3/envs/executorch/lib/python3.10/site-packages/torch/_functorch/aot_autograd.py", line 1480, in _aot_export_function
    fx_g, meta = create_aot_dispatcher_function(
  File "/home/dinusha/anaconda3/envs/executorch/lib/python3.10/site-packages/torch/_functorch/aot_autograd.py", line 522, in create_aot_dispatcher_function
    return _create_aot_dispatcher_function(
  File "/home/dinusha/anaconda3/envs/executorch/lib/python3.10/site-packages/torch/_functorch/aot_autograd.py", line 623, in _create_aot_dispatcher_function
    fw_metadata = run_functionalized_fw_and_collect_metadata(
  File "/home/dinusha/anaconda3/envs/executorch/lib/python3.10/site-packages/torch/_functorch/_aot_autograd/collect_metadata_analysis.py", line 150, in inner
    assert all(isinstance(a, tuple(KNOWN_TYPES)) for a in flat_args)
AssertionError

any help on this would be highly appreciated

Thank you

cc @JacobSzwejbka @angelayi

Start with the provided test.py reproduction at the to_edge call on line 51, then trace torch.export.ExportedProgram.run_decompositions through executorch.exir.to_edge to the shown assertion. Reproduce the failure with the edgeface_xxs_q model and determine whether the model can be lowered; done means documenting a validated resolution or compatibility limitation.