TypeError: forward() missing 1 required positional argument: 'x_body'

TypeError: forward() missing 1 required positional argument: ‘x_body’.

#But I already have ‘x_body’ in the network definition.

Question

I use the torch.onnx.export() function to convert the.pth format model to.onnx format. I converted the two resNet models to the.onNx format and it worked fine. Now I have a problem transforming the model defined in my following code. This model takes the outputs of the two RESNet models as parameter inputs.

Traceback

Traceback (most recent call last): File “E:/pythonProject/Pytorch2TFLite/PyTorch2ONNX.py”, line 61, in <module> torch.onnx.export(model_emotic, File “D:\Anaconda3\lib\site-packages\torch-1.9.0-py3.8-win-amd64.egg\torch\onnx_init_.py”, line 275, in export return utils.export(model, args, f, export_params, verbose, training, File “D:\Anaconda3\lib\site-packages\torch-1.9.0-py3.8-win-amd64.egg\torch\onnx\utils.py”, line 88, in export _export(model, args, f, export_params, verbose, training, input_names, output_names, File “D:\Anaconda3\lib\site-packages\torch-1.9.0-py3.8-win-amd64.egg\torch\onnx\utils.py”, line 689, in _export _model_to_graph(model, args, verbose, input_names, File “D:\Anaconda3\lib\site-packages\torch-1.9.0-py3.8-win-amd64.egg\torch\onnx\utils.py”, line 458, in _model_to_graph graph, params, torch_out, module = _create_jit_graph(model, args, File “D:\Anaconda3\lib\site-packages\torch-1.9.0-py3.8-win-amd64.egg\torch\onnx\utils.py”, line 422, in _create_jit_graph graph, torch_out = _trace_and_get_graph_from_model(model, args) File “D:\Anaconda3\lib\site-packages\torch-1.9.0-py3.8-win-amd64.egg\torch\onnx\utils.py”, line 373, in _trace_and_get_graph_from_model torch.jit._get_trace_graph(model, args, strict=False, _force_outplace=False, _return_inputs_states=True) File “D:\Anaconda3\lib\site-packages\torch-1.9.0-py3.8-win-amd64.egg\torch\jit_trace.py”, line 1160, in _get_trace_graph outs = ONNXTracedModule(f, strict, _force_outplace, return_inputs, _return_inputs_states)(*args, **kwargs) File “D:\Anaconda3\lib\site-packages\torch-1.9.0-py3.8-win-amd64.egg\torch\nn\modules\module.py”, line 1051, in _call_impl return forward_call(*input, **kwargs) File “D:\Anaconda3\lib\site-packages\torch-1.9.0-py3.8-win-amd64.egg\torch\jit_trace.py”, line 127, in forward graph, out = torch._C._create_graph_by_tracing( File “D:\Anaconda3\lib\site-packages\torch-1.9.0-py3.8-win-amd64.egg\torch\jit_trace.py”, line 118, in wrapper outs.append(self.inner(*trace_inputs)) File “D:\Anaconda3\lib\site-packages\torch-1.9.0-py3.8-win-amd64.egg\torch\nn\modules\module.py”, line 1051, in _call_impl return forward_call(*input, **kwargs) File “D:\Anaconda3\lib\site-packages\torch-1.9.0-py3.8-win-amd64.egg\torch\nn\modules\module.py”, line 1039, in _slow_forward result = self.forward(*input, **kwargs) TypeError: forward() missing 1 required positional argument: ‘x_body’

My code

import os.path as osp
import numpy as np
import onnx
import onnxruntime as ort
import torch
import torchvision
import torch.nn as nn

#加入model_emotic1.pth:初始化
class Emotic(nn.Module):
    ''' Emotic Model
        转换model_body1.pth和model_context1.pth不需要这个类。
        转换model_emotic1.pth时，需要加此类。
    '''

    def __init__(self, num_context_features, num_body_features):
        super(Emotic, self).__init__()
        self.num_context_features = num_context_features
        self.num_body_features = num_body_features
        self.fc1 = nn.Linear((self.num_context_features + num_body_features), 256)
        self.bn1 = nn.BatchNorm1d(256)
        self.d1 = nn.Dropout(p=0.5)
        self.fc_cat = nn.Linear(256, 26)
        self.fc_cont = nn.Linear(256, 3)
        self.relu = nn.ReLU()

    '''forward是自动调用的,如：model=Emotic();y=model(x,y)，则会调用网络模型定义的forward()方法【而不是model.forward(x,y)】。
        即,当把定义的网络模型当作函数调用的时候，就自动调用定义的网络模型的forward()方法。
        是通过nn.Module的__call__方法调用的，就相当于调用了模型，就是直接调用它的forward函数，y=model(x)，这个x就是直接传入到forward函数的x参数。
    '''
    def forward(self, x_context, x_body):   #定义前向传播
        context_features = x_context.view(-1, self.num_context_features)
        body_features = x_body.view(-1, self.num_body_features)
        fuse_features = torch.cat((context_features, body_features),
                                  1)  # torch.cat()函数输入了context_features和body_features。torch.cat是将两个张量（tensor）拼接在一起，cat是concatnate的意思，即拼接，联系在一起。使用torch.cat((A,B),dim)时，除拼接维数dim数值可不同外其余维数数值需相同，方能对齐。dim=1，横着拼接。
        fuse_out = self.fc1(fuse_features)
        fuse_out = self.bn1(fuse_out)
        fuse_out = self.relu(fuse_out)
        fuse_out = self.d1(fuse_out)
        cat_out = self.fc_cat(fuse_out)
        cont_out = self.fc_cont(fuse_out)
        return cat_out, cont_out


test_arr=np.random.randn(10,3,224,224).astype(np.float32)   #astype():数组的副本，转换为指定类型
dummy_input=torch.tensor(test_arr)
body_test_arr=np.random.randn(10,3,128,128).astype(np.float32)
body_dummy_input=torch.tensor(body_test_arr)

model_body=torch.load("./models/model_body1.pth")
model_context=torch.load("./models/model_context1.pth")
pred_context=model_context(torch.from_numpy(test_arr))
pred_body=model_body(torch.from_numpy(body_test_arr))

model_emotic=torch.load("./models/model_emotic1.pth")
model_emotic.eval()
torch_output=model_emotic(pred_context,pred_body)  #torch.from_numpy():从numpy.ndarray创建Tensor

input_names=["input"]
output_names=["cat","cont"]
torch.onnx.export(model_emotic,
                  dummy_input,
                  "./models/model_emotic1.onnx",
                  verbose=False,    #如果verbose指定了，将输出正在导出的跟踪的调试描述。默认：false
                  input_names=input_names,  #按顺序分配给图的输入节点的名称
                  output_names=output_names)    #按顺序分配给图的输入节点的名称

model=onnx.load("./models/model_emotic1.onnx")
ort_session=ort.InferenceSession("./models/model_emotic1.onnx")
onnx_outputs=ort_session.run(None,{'input':test_arr})
print('Export ONNX!')

I hope you can solve my confusion as soon as possible. Thank you very much!

[RyanShun-511]

@jcwchen Thank you very much for your reply. I have solved the problem. Sorry for the trouble caused to you.

[Leonardo0325]

I have the same problem,look this. TypeError:forward() takes 1 positional argument but 2 were given.