I have print all the the box_objectness,it is 1e-5 or 1e-6 ,very samll.Is it right?

this code id ncnn or tensorrt cpp code. box_objectness is samller than 0.001? IS it right??
float box_objectness = data[basic_pos+4];

1.20168e-06
9.85814e-07
1.1292e-07
2.29236e-06
9.9316e-07
1.07553e-07
5.76235e-09
9.71645e-09
1.51367e-10
3.59336e-06
4.66548e-07
2.84124e-08
2.82141e-07
2.16512e-07
1.12944e-06
1.0082e-07
6.90443e-07
4.16271e-07
6.637e-06
2.00837e-06
5.78883e-06
5.79828e-06
1.58486e-08
3.98419e-09
3.6147e-08
3.37897e-07
2.81759e-06
1.53196e-06
4.82851e-08
3.40777e-06
1.37554e-06
8.34869e-08
7.64393e-09
1.18429e-08
1.30057e-09
2.45159e-06
3.15884e-07
7.48874e-09
5.06838e-07
4.64801e-08
1.29585e-07
9.24573e-10
1.00373e-07
7.96577e-08
3.95333e-06
1.89665e-07
1.6945e-06
2.70703e-06
5.5672e-08
6.93358e-08
6.68422e-08
8.03648e-08
1.75489e-06
1.4784e-06
1.8257e-07
1.13432e-06
3.50879e-07
6.07953e-09
3.96261e-10
1.26475e-08
7.96639e-08
4.24208e-08
2.70764e-08
2.50077e-08
4.87679e-06
5.07077e-08
2.80433e-08
1.35693e-09
5.9227e-09
6.7103e-07
1.27091e-07
6.59451e-08
2.52134e-06
7.37795e-07
2.57831e-07
1.57848e-07
4.05893e-08
5.61519e-08
1.01321e-06
3.54634e-07
1.29778e-07
1.76346e-08
1.29838e-07

Start by tracing the ncnn or TensorRT C++ path that supplies data[basic_pos+4] and compare the raw output with the model or export configuration. Done means documenting whether these objectness values are expected and, if not, providing a reproducible model and inference setup.