Error occurs when I use the tools/export.py

[zyqhnu]

Opencd is cool. And I try to use the ‘export.py to transfer .pth model into onnx.
with this:
python3 tools/export.py --config ./work_dirs/changer_ex_eva_256x256_levircd/changer_ex_eva_256x256_levircd.py --checkpoint ./work_dirs/changer_ex_eva_256x256_levircd/best_mIoU_iter_29000.pth

the error occurs

INFO:opencd:Namespace(config='/root/cd_test/rs-pangu-cd/code/work_dirs/changer_ex_eva_256x256_levircd/changer_ex_eva_256x256_levircd.py', checkpoint='/root/cd_test/rs-pangu-cd/code/work_dirs/changer_ex_eva_256x256_levircd/best_mIoU_iter_29000.pth', inputs=['1,3,256,256', '1,3,256,256'], model_name='model.onnx')
INFO:root:Shape of rope freq: torch.Size([324, 64])

Traceback (most recent call last):
  File "/root/cd_test/rs-pangu-cd/code/tools/export.py", line 65, in <module>
    main(args)
  File "/root/cd_test/rs-pangu-cd/code/tools/export.py", line 41, in main
    torch.onnx.export(
  File "/usr/local/lib64/python3.9/site-packages/torch/onnx/__init__.py", line 305, in export
    return utils.export(model, args, f, export_params, verbose, training,
  File "/usr/local/lib64/python3.9/site-packages/torch/onnx/utils.py", line 118, in export
    _export(model, args, f, export_params, verbose, training, input_names, output_names,
  File "/usr/local/lib64/python3.9/site-packages/torch/onnx/utils.py", line 719, in _export
    _model_to_graph(model, args, verbose, input_names,
  File "/usr/local/lib64/python3.9/site-packages/torch/onnx/utils.py", line 503, in _model_to_graph
    graph = _optimize_graph(graph, operator_export_type,
  File "/usr/local/lib64/python3.9/site-packages/torch/onnx/utils.py", line 232, in _optimize_graph
    graph = torch._C._jit_pass_onnx(graph, operator_export_type)
  File "/usr/local/lib64/python3.9/site-packages/torch/onnx/__init__.py", line 354, in _run_symbolic_function
    return utils._run_symbolic_function(*args, **kwargs)
  File "/usr/local/lib64/python3.9/site-packages/torch/onnx/utils.py", line 1038, in _run_symbolic_function
    return torch._C._jit_onnx_convert_pattern_from_subblock(block, n, env)
  File "/usr/local/lib64/python3.9/site-packages/torch/onnx/__init__.py", line 354, in _run_symbolic_function
    return utils._run_symbolic_function(*args, **kwargs)
  File "/usr/local/lib64/python3.9/site-packages/torch/onnx/utils.py", line 1061, in _run_symbolic_function
    return symbolic_fn(g, *inputs, **attrs)
  File "/usr/local/lib64/python3.9/site-packages/torch/onnx/symbolic_opset11.py", line 114, in index_put
    if indices_list[idx_].type().scalarType() == "Bool":
RuntimeError: rINTERNAL ASSERT FAILED at "/root/pytorch/aten/src/ATen/core/jit_type_base.h":540, please report a bug to PyTorch.

How can I solve it?

Environment:
torch 1.11.0
mmcv 2.1.0
mmdet 3.2.0
mmengine 0.10.3
mmsegmentation 1.2.2

[GreatV]

hi @zyqhnu, This looks like an internal pytorch error.