TLT3 NHWC error on 940MX, error while training on TLT3

940MX has compute capability of 5.0.
The same issue as TLT Detectnet TrafficCamNet training not working - #9 by Morganh
Training on Custom Dataset using TLT
Error with tlt train in official Jupyter notebook TLT 3.0 - #5 by Morganh
Training Peoplent on custom data - #20 by abhigoku10