Error during PTQ

[Darianek]

Hello everyone,

currently, I'm trying to speed up the inference of yolop. I tried several different containers from NGC with torch-tensorrt support but I always get the same error. When I use Hardswish with return x * F.hardtanh(x + 3, 0., 6.) / 6. # for torchscript, CoreML and ONNX I'm getting the error: '__torch__.lib.models.common.Hardswish' object has no attribute or method '__add__'. Did you forget to initialize an attribute in __init__()? When I try to use Hardswish with return x * F.hardsigmoid(x) # for torchscript and CoreML I'm getting: Expected a value of type 'Tensor' for argument 'input' but instead found type '__torch__.lib.models.common.Hardswish. Any solutions?