E convolution output_padding only suport [0,0] for now, but meet [0,0]

#1

when perform rknn.load_pytorch(model=model, input_size_list=input_size_list), encounter the following error messages:

W Channels(360) of input node: input.4 > 128, mean/std values will be set to default 0/1.
W Please do pre-processing manually before inference.
E convolution output_padding only suport [0,0] for now, but meet [0,0]

Questions:

  1. what does “Channels(360) of input node: input.4 > 128, mean/std values will be set to default 0/1.” mean? The input tensor is [3,2,180,360]. It is 2 2D image [3,180,360] stacked into a 3D tensor [3,2,180,360] for conv3d operation in [C,D,H,W] format. How to change the rknn.config(channel_mean_value, reorder_channel) to reflect this format?
  2. what does “convolution output_padding only suport [0,0] for now, but meet [0,0]”? it sournds no error! It supports [0,0] output_padding and it met [0,0] padding. So what’s the error?

Thank you very much for your help.