尝试使用ONNX转换此pytorch模型时出现此错误。我搜索过github，这个错误以前在1.1.0版本中出现过，但显然已经纠正了。现在我使用的是火炬1.4.0。(python 3.6.9(，我看到了这个错误。
File "/usr/local/lib/python3.6/dist-packages/torch/onnx/init.py", line 148, in export
strip_doc_string, dynamic_axes, keep_initializers_as_inputs)
File "/usr/local/lib/python3.6/dist-packages/torch/onnx/utils.py", line 66, in export
dynamic_axes=dynamic_axes, keep_initializers_as_inputs=keep_initializers_as_inputs)
File "/usr/local/lib/python3.6/dist-packages/torch/onnx/utils.py", line 416, in _export
fixed_batch_size=fixed_batch_size)
File "/usr/local/lib/python3.6/dist-packages/torch/onnx/utils.py", line 296, in _model_to_graph
fixed_batch_size=fixed_batch_size, params_dict=params_dict)
File "/usr/local/lib/python3.6/dist-packages/torch/onnx/utils.py", line 135, in _optimize_graph
graph = torch._C._jit_pass_onnx(graph, operator_export_type)
File "/usr/local/lib/python3.6/dist-packages/torch/onnx/init.py", line 179, in _run_symbolic_function
return utils._run_symbolic_function(*args, **kwargs)
File "/usr/local/lib/python3.6/dist-packages/torch/onnx/utils.py", line 657, in _run_symbolic_function
return op_fn(g, *inputs, **attrs)
File "/usr/local/lib/python3.6/dist-packages/torch/onnx/symbolic_helper.py", line 128, in wrapper
args = [_parse_arg(arg, arg_desc) for arg, arg_desc in zip(args, arg_descriptors)]
File "/usr/local/lib/python3.6/dist-packages/torch/onnx/symbolic_helper.py", line 128, in
args = [_parse_arg(arg, arg_desc) for arg, arg_desc in zip(args, arg_descriptors)]
File "/usr/local/lib/python3.6/dist-packages/torch/onnx/symbolic_helper.py", line 81, in _parse_arg
"', since it's not constant, please try to make "
RuntimeError: Failed to export an ONNX attribute 'onnx::Gather', since it's not constant, please try to make things (e.g., kernel size) static if possible
如何修复？我也尝试了最新的夜间构建，同样的错误出现了。
我的代码：
from model import BiSeNet
import torch.onnx
import torch
net = BiSeNet(19)
net.cuda()
net.load_state_dict(torch.load('/content/drive/My Drive/Collab/fp/res/cp/79999_iter.pth'))
net.eval()
dummy = torch.rand(1,3,512,512).cuda()
torch.onnx.export(net, dummy, "Model.onnx", input_names=["image"], output_names=["output"])
就在引发运行时错误之前，我将print (v.node ())添加到了symbolic_helper.py中，以查看错误的原因。
这是输出：%595 : Long() = onnx::Gather[axis=0](%592, %594) # /content/drive/My Drive/Collab/fp/model.py:111:0
模型.py中的111行是：avg = F.avg_pool2d(feat32, feat32.size()[2:])
这个来源表明pytorch中的tensor.size方法不能被onnx识别，需要修改为常数。
我过去在使用导出时也有类似的错误
torch.onnx.export(model, x, ONNX_FILE_PATH)
我通过指定opset_version来修复它，如下所示：
torch.onnx.export(model, x, ONNX_FILE_PATH, opset_version = 11)
对于那些从谷歌搜索中找到这个问题并获得Unable to cast from non-held to held instance (T& to Holder) (compile in debug mode for type information)的人，可以尝试添加operator_export_type=torch.onnx.OperatorExportTypes.ONNX_ATEN_FALLBACK(如这里所述(，如下所示：
torch.onnx.export(model, input, "output-name.onnx", export_params=True, opset_version=12, operator_export_type=torch.onnx.OperatorExportTypes.ONNX_ATEN_FALLBACK)
这修复了"；持有的实例"；我的问题。
更改实例
x = F.avg_pool2d(x, x.shape[2:])
至
x_shape = [int(s) for s in x.shape[2:]]
x = F.avg_pool2d(x, x_shape)
使得对avg_pool2d的输入形状是恒定的[k,k]而不是这里提到的torch.Size([k, k])。
您可以：
print(feat32.size()[2:])
并替换：
F.avg_pool2d(feat32, feat32.size()[2:])
带有：
F.avg_pool2d(feat32, your_print_contant_result)
我也遇到过同样的问题。
F.在我的情况下不支持adaptive_avg_pool2d。你必须尝试其他操作。
我希望这能有所帮助。
感谢