mxnet to pytorch (No module named 'MainModel')

[kevinzezel]

Hi,

I used this commands to convert mxnet resnet50 model to pytorch (https://www.dropbox.com/s/53ftnlarhyrpkg2/retinaface-R50.zip?dl=0):

python -m mmdnn.conversion._script.convertToIR -f mxnet -n R50-symbol.json -w R50-0000.params -d resnet50 --inputShape 3,224,224

[14:31:27] src/nnvm/legacy_json_util.cc:209: Loading symbol saved by previous version v1.3.0. Attempting to upgrade... [14:31:27] src/nnvm/legacy_json_util.cc:217: Symbol successfully upgraded! /opt/conda/lib/python3.6/site-packages/mxnet/module/base_module.py:55: UserWarning: You created Module with Module(..., label_names=['softmax_label']) but input with name 'softmax_label' is not found in symbol.list_arguments(). Did you mean one of: data warnings.warn(msg) Warning: MXNet Parser has not supported operator null with name data. Warning: convert the null operator with name [data] into input layer. IR network structure is saved as [resnet50.json]. IR network structure is saved as [resnet50.pb]. IR weights are saved as [resnet50.npy].

python -m mmdnn.conversion._script.IRToCode -f pytorch --IRModelPath resnet50.pb --dstModelPath kit_imagenet.py --IRWeightPath resnet50.npy -dw kit_pytorch.npy  

Parse file [resnet50.pb] with binary format successfully. Target network code snippet is saved as [kit_imagenet.py]. Target weights are saved as [kit_pytorch.npy].

python -m mmdnn.conversion.examples.pytorch.imagenet_test --dump resnet50Full.pth -n kit_imagenet.py -w kit_pytorch.npy

PyTorch model file is saved as [resnet50Full.pth], generated by [kit_imagenet.py.py] and [kit_pytorch.npy].

However when I import the model in pytorch I get this error:

>>> import torch
>>> import numpy as np
>>> model_address = 'resnet50Full.pth'
>>> device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
>>> model = torch.load(model_address).to(device)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/opt/conda/lib/python3.6/site-packages/torch/serialization.py", line 592, in load
    return _legacy_load(opened_file, map_location, pickle_module, **pickle_load_args)
  File "/opt/conda/lib/python3.6/site-packages/torch/serialization.py", line 772, in _legacy_load
    result = unpickler.load()
ModuleNotFoundError: No module named 'MainModel'

mxnet -> 1.6.0 torch -> 1.5.1 torchvision -> 0.6.0a0 mmdnn -> 0.3.0

Machine settings: Ubuntu Linux 18.04 LTS GPUs: 2x Nvidia Tesla V100 32gb CPUs: 2x Intel Xeon E5-2690 v4 RAM: 512GB

Regards, Kevin

[linmajia]

@kevinzezel , thank you very much for the feedback. Please refer to this FAQ, and add the following code:

import imp
MainModel = imp.load_source('MainModel', "kit_imagenet.py")

before

model = torch.load(model_address).to(device)

[kevinzezel]

Thank you!!