Description
Hi! I'm trying to load the models in C++ too, but I'm having the following error:
terminate called after throwing an instance of 'torch::jit::ErrorReport' what():
aten::_convolution(Tensor input, Tensor weight, Tensor? bias, int[] stride, int[] padding, int[] dilation, bool transposed, int[] output_padding, int groups, bool benchmark, bool deterministic, bool cudnn_enabled) -> (Tensor):
Expected at most 12 arguments but found 13 positional arguments.
_ Serialized File "code/torch/torch/nn/modules/conv.py", line 8
def forward(self: torch.torch.nn.modules.conv.Conv2d,
input: Tensor) -> Tensor:
input0 = torch.convolution(input, self.weight, None, [2, 2], [3, 3], [1, 1], False, [0, 0], 1, False, False, True, True)
~~~~~~~~~~~~~~~~~~ <--- HERE
return input0
I used the model trained in cars example (jupyternotebook)
ENCODER = 'se_resnext50_32x4d'
ENCODER_WEIGHTS = 'imagenet'
CLASSES = ['car']
ACTIVATION = 'sigmoid' # could be None for logits or 'softmax2d' for multicalss segmentation
DEVICE = 'cuda'
model = smp.FPN(
encoder_name=ENCODER,
encoder_weights=ENCODER_WEIGHTS,
classes=len(CLASSES),
activation=ACTIVATION,
)
preprocessing_fn = smp.encoders.get_preprocessing_fn(ENCODER, ENCODER_WEIGHTS)
Then I saved it using torch.jit.save as following :
input_img = torch.rand(1, 3, 256, 256).to(DEVICE)
best_model = torch.load('./best_model.pth')
best_model.eval()
traced_script_module = torch.jit.trace(best_model, input_img)
traced_script_module.save('./best_model.pt')
When I load the model in c++ using torch::jit::load(model_path.string()); it result in an error.
I'm using Pytorch 1.71