ConstraintViolationError

Hi,
I just upgraded to executorch 0.4 and ran my code which previously failed as described in
https://github.com/pytorch/executorch/issues/1350

Now it fails with
ConstraintViolationError

Please find the error log below.

Please let me know if you need additional information.

Thanks

ERROR LOG

Traceback (most recent call last):
  File "/home/adonnini1/anaconda3/envs/executorch/lib/python3.10/site-packages/torch/export/_trace.py", line 670, in _export_to_aten_ir
    produce_guards_callback(gm)
  File "/home/adonnini1/anaconda3/envs/executorch/lib/python3.10/site-packages/torch/export/_trace.py", line 1655, in _produce_guards_callback
    return produce_guards_and_solve_constraints(
  File "/home/adonnini1/anaconda3/envs/executorch/lib/python3.10/site-packages/torch/_export/non_strict_utils.py", line 287, in produce_guards_and_solve_constraints
    raise constraint_violation_error
  File "/home/adonnini1/anaconda3/envs/executorch/lib/python3.10/site-packages/torch/_export/non_strict_utils.py", line 270, in produce_guards_and_solve_constraints
    shape_env.produce_guards(
  File "/home/adonnini1/anaconda3/envs/executorch/lib/python3.10/site-packages/torch/fx/experimental/symbolic_shapes.py", line 3788, in produce_guards
    raise ConstraintViolationError(
torch.fx.experimental.symbolic_shapes.ConstraintViolationError: L['args'][0][1].size()[1] = 12 is not equal to L['args'][0][0].size()[1] = 7

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/home/adonnini1/Development/ContextQSourceCode/NeuralNetworks/adonnini-trajectory-prediction-transformers-masterContextQ/train-minimum.py", line 443, in <module>
    pre_autograd_aten_dialect = torch.export._trace._export(
  File "/home/adonnini1/anaconda3/envs/executorch/lib/python3.10/site-packages/torch/export/_trace.py", line 1017, in wrapper
    raise e
  File "/home/adonnini1/anaconda3/envs/executorch/lib/python3.10/site-packages/torch/export/_trace.py", line 990, in wrapper
    ep = fn(*args, **kwargs)
  File "/home/adonnini1/anaconda3/envs/executorch/lib/python3.10/site-packages/torch/export/exported_program.py", line 114, in wrapper
    return fn(*args, **kwargs)
  File "/home/adonnini1/anaconda3/envs/executorch/lib/python3.10/site-packages/torch/export/_trace.py", line 1880, in _export
    export_artifact = export_func(  # type: ignore[operator]
  File "/home/adonnini1/anaconda3/envs/executorch/lib/python3.10/site-packages/torch/export/_trace.py", line 1683, in _non_strict_export
    aten_export_artifact = _to_aten_func(  # type: ignore[operator]
  File "/home/adonnini1/anaconda3/envs/executorch/lib/python3.10/site-packages/torch/export/_trace.py", line 672, in _export_to_aten_ir
    raise UserError(UserErrorType.CONSTRAINT_VIOLATION, str(e))  # noqa: B904
torch._dynamo.exc.UserError: L['args'][0][1].size()[1] = 12 is not equal to L['args'][0][0].size()[1] = 7

Start with the export call at train-minimum.py line 443 and trace the reported paths in torch/export/_trace.py, torch/_export/non_strict_utils.py, and torch/fx/experimental/symbolic_shapes.py. Compare the failure with issue 1350 and obtain a minimal reproducible example, since the issue provides no model, project file, or test. Done would require a confirmed cause and a project-level fix or documented resolution for the 12-versus-7 shape constraint.