something about cpu explosion

hi, Giovanni, I just re-implemented the MG-2D on my dataset and met with some other bug for cpu explosion.

When I used MG for pre-training, I found that the utilization rate of the CPU would increase strangely, but the utilization rate of the GPU fluctuated greatly, and the buzzer of the machine PC would alarm soon. After several rounds of debugging, I found that it was this piece of code that caused the CPU utilization to sharply increase to 4000%+!!!!
https://github.com/MrGiovanni/ModelsGenesis/blob/13932523d1754a3180301448f4bfd5274a1c9f41/pytorch/utils.py#L45
I don't know why, but after changing np.dot in function : nonlinear_transformation and related calculations to torch, the CPU utilization has returned to normal (200%+).

Maybe this is a special case in my machine, and it has no reference value.
Thanks. :-)

Start at pytorch/utils.py line 45, in nonlinear_transformation, and inspect the np.dot call and related calculations. Reproduce the MG pre-training case if possible, then compare the CPU and GPU utilization with the reported torch-based calculations. Done means the transformation no longer causes extreme CPU utilization while preserving the expected pre-training behavior.