[Pytorch Android] Two input is possible?

Mobile
1

Hi, I have two input tensor like below.

    input_blob_size = (1, 3, 1, 540, 960)
    print(input_blob_size)
    input_tensor = torch.randn(input_blob_size)
    input_tensor2 = torch.randn(input_blob_size)
    traced_model = torch.jit.trace(model, (input_tensor, input_tensor2), strict=False)
    optimized_traced__model = optimize_for_mobile(traced_model)

but I got this error…

optimized_traced__model = optimize_for_mobile(traced_model)

File “/home/hsm/anaconda3/envs/reproducibleresearch/lib/python3.6/site-packages/torch/utils/mobile_optimizer.py”, line 65, in optimize_for_mobile
preserved_methods_str)
RuntimeError: Method ‘forward’ is not defined.

I want to use two inputs, rather than one input.
Can I do it?

This is my model snipet.

class ResNet50_forward2(torch.nn.Module):
    def __init__(self):
        super(ResNet50_forward2, self).__init__()
        self.batch = nn.Sequential(*list(models.mobilenet_v3_small(pretrained=True).children())[0])

    def forward(self, pixel, pixel2):
        raw_read = torch.cat((pixel, pixel2), 0)
        print("DL1", raw_read.shape) #batch x chan x fr x w x h
        return raw_read