export导出onnx运行出错

[CYlcbw]

在使用export.py导出onnx的时候，在生成onnx的时候报错（提示在调用 embedding 时，有至少两个设备（cuda:0 和 cpu）存在不一致），请问有没有大佬知道怎么解决啊
具体报错是｛(torch) root@mfkahpxnfnicaehc-snow-6c68c5f989-jvxzb:/data/coding/SenseVoice# python export.py
Downloading Model to directory: /root/.cache/modelscope/hub/iic/SenseVoiceSmall
2025-02-14 18:51:14,969 - modelscope - WARNING - Using branch: master as version is unstable, use with caution
Loading remote code successfully: model
Traceback (most recent call last):
File "/data/coding/SenseVoice/export.py", line 29, in
export_dir = export_utils.export(model=rebuilt_model, **kwargs)
File "/data/coding/SenseVoice/utils/export_utils.py", line 17, in export
_onnx(
File "/data/coding/SenseVoice/utils/export_utils.py", line 43, in _onnx
torch.onnx.export(
File "/data/miniconda/envs/torch/lib/python3.10/site-packages/torch/onnx/utils.py", line 516, in export
_export(
File "/data/miniconda/envs/torch/lib/python3.10/site-packages/torch/onnx/utils.py", line 1612, in _export
graph, params_dict, torch_out = _model_to_graph(
File "/data/miniconda/envs/torch/lib/python3.10/site-packages/torch/onnx/utils.py", line 1134, in _model_to_graph
graph, params, torch_out, module = _create_jit_graph(model, args)
File "/data/miniconda/envs/torch/lib/python3.10/site-packages/torch/onnx/utils.py", line 1010, in _create_jit_graph
graph, torch_out = _trace_and_get_graph_from_model(model, args)
File "/data/miniconda/envs/torch/lib/python3.10/site-packages/torch/onnx/utils.py", line 914, in _trace_and_get_graph_from_model
trace_graph, torch_out, inputs_states = torch.jit._get_trace_graph(
File "/data/miniconda/envs/torch/lib/python3.10/site-packages/torch/jit/_trace.py", line 1310, in _get_trace_graph
outs = ONNXTracedModule(
File "/data/miniconda/envs/torch/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/data/miniconda/envs/torch/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
return forward_call(*args, **kwargs)
File "/data/miniconda/envs/torch/lib/python3.10/site-packages/torch/jit/_trace.py", line 138, in forward
graph, out = torch._C._create_graph_by_tracing(
File "/data/miniconda/envs/torch/lib/python3.10/site-packages/torch/jit/_trace.py", line 129, in wrapper
outs.append(self.inner(*trace_inputs))
File "/data/miniconda/envs/torch/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/data/miniconda/envs/torch/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
return forward_call(*args, **kwargs)
File "/data/miniconda/envs/torch/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1522, in _slow_forward
result = self.forward(*input, **kwargs)
File "/data/coding/SenseVoice/export_meta.py", line 32, in export_forward
language_query = self.embed(language.to(speech.device)).unsqueeze(1)
File "/data/miniconda/envs/torch/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/data/miniconda/envs/torch/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
return forward_call(*args, **kwargs)
File "/data/miniconda/envs/torch/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1522, in _slow_forward
result = self.forward(*input, **kwargs)
File "/data/miniconda/envs/torch/lib/python3.10/site-packages/torch/nn/modules/sparse.py", line 163, in forward
return F.embedding(
File "/data/miniconda/envs/torch/lib/python3.10/site-packages/torch/nn/functional.py", line 2264, in embedding
return torch.embedding(weight, input, padding_idx, scale_grad_by_freq, sparse)
RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cpu! (when checking argument for argument index in method wrapper_CUDA__index_select)｝