AttributeError: 'BatchNorm2d' object has no attribute 'num_batches_tracked'

[zhenglongyu]

I'm using pytorch0.3.1 with python3.6, cudnn-7.1.3 and cudatoolkit-8.0 and had some problems. I have googled and know that it is because the pytorch version I used is different from that is used to train the model.

The log is below: Start testing ...... Traceback (most recent call last): File "/mnt/E/zhenglongyu/pycharm-community-2018.1.4/helpers/pydev/pydev_run_in_console.py", line 52, in run_file pydev_imports.execfile(file, globals, locals) # execute the script File "/mnt/E/zhenglongyu/pycharm-community-2018.1.4/helpers/pydev/_pydev_imps/_pydev_execfile.py", line 18, in execfile exec(compile(contents+"\n", file, 'exec'), glob, loc) File "/mnt/E/zhenglongyu/PlaneMatch/test.py", line 77, in feature1 = model(x1, x4, x7, x10, x13, x16, x19, x22) File "/mnt/E/zhenglongyu/pytorch_conda/lib/python3.6/site-packages/torch/nn/modules/module.py", line 477, in call result = self.forward(*input, **kwargs) File "/mnt/E/zhenglongyu/PlaneMatch/network_planematch.py", line 172, in forward x1 = self.bn1(x1) File "/mnt/E/zhenglongyu/pytorch_conda/lib/python3.6/site-packages/torch/nn/modules/module.py", line 477, in call result = self.forward(*input, **kwargs) File "/mnt/E/zhenglongyu/pytorch_conda/lib/python3.6/site-packages/torch/nn/modules/batchnorm.py", line 57, in forward self.num_batches_tracked += 1 File "/mnt/E/zhenglongyu/pytorch_conda/lib/python3.6/site-packages/torch/nn/modules/module.py", line 518, in getattr type(self).name, name)) AttributeError: 'BatchNorm2d' object has no attribute 'num_batches_tracked' PyDev console: starting. Python 3.6.6 |Anaconda, Inc.| (default, Oct 9 2018, 12:34:16) [GCC 7.3.0] on linux QXcbConnection: Failed to initialize XRandr Qt: XKEYBOARD extension not present on the X server.

So could you tell me the pytorch version you used to train the model

[yifeishi]

Did you solve the problem? I would recommend you to first switch to python 2.7 and see if that works.

[elbashandy]

I got the same error but not in training the model.. just testing it.

I realized I forgot to add:

model.eval()