Add or remove lines below (one item per line), then classify with any or all of the models below to compare results and speed.
PyTorch Results
ONNX (fp32) Results
ONNX (int8) Results