Failed to export onnx model

Code version (Git Hash) and PyTorch version

st-gcn master@e7024ac and Pytorch ‘1.1.0a0+828a6a3’

Dataset used

Demo

Expected behavior

Successfully export onnx model

Actual behavior

root@p4station:/workspace# python main.py demo --openpose openpose/build/ --device 1 /workspace/processor/io.py:39: YAMLLoadWarning: calling yaml.load() without Loader=… is deprecated, as the default Loader is unsafe. Please read https://msg.pyyaml.org/load for full details. default_arg = yaml.load(f) Starting OpenPose demo… Auto-detecting all available GPUs… Detected 1 GPU(s), using 1 of them starting at GPU 0. Starting thread(s)… OpenPose demo successfully finished. Total time: 47.039486 seconds. Pose estimation complete.

Network forwad… Prediction result: skateboarding Done. /workspace/net/utils/tgcn.py:58: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can’t record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs! assert A.size(0) == self.kernel_size Traceback (most recent call last): File “main.py”, line 31, in <module> p.start() File “/workspace/processor/demo.py”, line 85, in start torch.onnx.export(self.model, dummy_input, “st-gcn_kinetics-skeleton.onnx”) File “/opt/conda/lib/python3.6/site-packages/torch/onnx/init.py”, line 24, in export return utils.export(*args, **kwargs) File “/opt/conda/lib/python3.6/site-packages/torch/onnx/utils.py”, line 108, in export _retain_param_name=_retain_param_name) File “/opt/conda/lib/python3.6/site-packages/torch/onnx/utils.py”, line 315, in _export _retain_param_name) File “/opt/conda/lib/python3.6/site-packages/torch/onnx/utils.py”, line 245, in _model_to_graph graph = _optimize_graph(graph, operator_export_type) File “/opt/conda/lib/python3.6/site-packages/torch/onnx/utils.py”, line 164, in _optimize_graph graph = torch._C._jit_pass_onnx(graph, operator_export_type) File “/opt/conda/lib/python3.6/site-packages/torch/onnx/init.py”, line 49, in _run_symbolic_function return utils._run_symbolic_function(*args, **kwargs) File “/opt/conda/lib/python3.6/site-packages/torch/onnx/utils.py”, line 550, in _run_symbolic_function n.kindOf(“value”))) RuntimeError: Unsupported prim::Constant kind: s. Send a bug report.

Steps to reproduce the behavior

dummy_input = torch.randn(1, 3, 300, 18, 2, device=‘cuda’) torch.onnx.export(self.model, dummy_input, “st-gcn_kinetics-skeleton.onnx”)

Other comments

[nfeng0105]

@lxy5513 I replaced einsum through other supported ops,

`#x = torch.einsum(‘nkctv,kvw->nctw’, (x, A))

x = x.permute(0, 2, 3, 1, 4).contiguous() n, c, t, k, v = x.size() k, v, w = A.size() x = x.view(n * c * t, k * v) A = A.view(k * v, w) x = torch.mm(x, A) x = x.view(n, c, t, w) `

[RobotTim]

@nfeng0105 @lxy5513 hi sir, have you ever export onnx successfully. Could you share the export script for us pls!!!. This problem bothered me several days!!!