pytorch导出onnx格式时报错，RuntimeError:

将pytorch模型.pth导出为onnx格式时报错，源代码如下：

错误信息如下

Traceback (most recent call last):

  File "111.py", line 17, in <module>

    torch.onnx.export(model,dummy_input,"yolov3.onnx",input_names=['images'],output_names=['outTensors'],export_params=True,training=False)

  File "/home/xyj/.local/lib/python3.6/site-packages/torch/onnx/__init__.py", line 25, in export

    return utils.export(*args, **kwargs)

  File "/home/xyj/.local/lib/python3.6/site-packages/torch/onnx/utils.py", line 131, in export

    strip_doc_string=strip_doc_string)

  File "/home/xyj/.local/lib/python3.6/site-packages/torch/onnx/utils.py", line 363, in _export

    _retain_param_name, do_constant_folding)

  File "/home/xyj/.local/lib/python3.6/site-packages/torch/onnx/utils.py", line 278, in _model_to_graph

    _disable_torch_constant_prop=_disable_torch_constant_prop)

  File "/home/xyj/.local/lib/python3.6/site-packages/torch/onnx/utils.py", line 188, in _optimize_graph

    graph = torch._C._jit_pass_onnx(graph, operator_export_type)

  File "/home/xyj/.local/lib/python3.6/site-packages/torch/onnx/__init__.py", line 50, in _run_symbolic_function

    return utils._run_symbolic_function(*args, **kwargs)

  File "/home/xyj/.local/lib/python3.6/site-packages/torch/onnx/utils.py", line 589, in _run_symbolic_function

    return fn(g, *inputs, **attrs)

  File "/home/xyj/.local/lib/python3.6/site-packages/torch/onnx/symbolic.py", line 130, in wrapper

    args = [_parse_arg(arg, arg_desc) for arg, arg_desc in zip(args, arg_descriptors)]

  File "/home/xyj/.local/lib/python3.6/site-packages/torch/onnx/symbolic.py", line 130, in <listcomp>

    args = [_parse_arg(arg, arg_desc) for arg, arg_desc in zip(args, arg_descriptors)]

  File "/home/xyj/.local/lib/python3.6/site-packages/torch/onnx/symbolic.py", line 90, in _parse_arg

    raise RuntimeError("Failed to export an ONNX attribute, "

RuntimeError: Failed to export an ONNX attribute, since it's not constant, please try to make things (e.g., kernel size) static if possible

解决方案：将pytorch版本从1.1.0降为1.0.1

先查看此时的pytorch版本

然后安装1.0.1版本

sudo pip3 install torch==1.0.1-fhttps://download.pytorch.org/whl/cu90/stable

并查看版本信息

然后再次运行就不会出现上述错误