Exception in main training loop: module 'cupy.cudnn' has no attribute 'pooling_forward'

Hi, When I run chainer/examples/imagenet/train_imagenet_data_parallel.py, I got this error.

/home/dnn/.pyenv/versions/ymd/bin/python3 /home/dnn/ymd/chainer/examples/imagenet/train_imagenet_data_parallel.py train.txt val.txt
/home/dnn/ymd/chainer/chainer/training/updaters/multiprocess_parallel_updater.py:164: UserWarning: optimizer.lr is changed to 0.00125 by MultiprocessParallelUpdater for new batch size.
format(optimizer.lr))
Exception in main training loop: module 'cupy.cudnn' has no attribute 'pooling_forward'
Traceback (most recent call last):
  File "/home/dnn/ymd/chainer/chainer/training/trainer.py", line 315, in run
    update()
  File "/home/dnn/ymd/chainer/chainer/training/updaters/standard_updater.py", line 165, in update
    self.update_core()
  File "/home/dnn/ymd/chainer/chainer/training/updaters/multiprocess_parallel_updater.py", line 237, in update_core
    loss = _calc_loss(self._master, batch)
  File "/home/dnn/ymd/chainer/chainer/training/updaters/multiprocess_parallel_updater.py", line 271, in _calc_loss
    return model(*in_arrays)
  File "/home/dnn/ymd/chainer/chainer/link.py", line 242, in __call__
    out = forward(*args, **kwargs)
  File "/home/dnn/ymd/chainer/examples/imagenet/nin.py", line 28, in forward
    h = F.max_pooling_2d(F.relu(self.mlpconv1(x)), 3, stride=2)
  File "/home/dnn/ymd/chainer/chainer/functions/pooling/max_pooling_2d.py", line 383, in max_pooling_2d
    return func.apply((x,))[0]
  File "/home/dnn/ymd/chainer/chainer/function_node.py", line 263, in apply
    outputs = self.forward(in_data)
  File "/home/dnn/ymd/chainer/chainer/function_node.py", line 369, in forward
    return self.forward_gpu(inputs)
  File "/home/dnn/ymd/chainer/chainer/functions/pooling/max_pooling_2d.py", line 64, in forward_gpu
    return super(MaxPooling2D, self).forward_gpu(x)
  File "/home/dnn/ymd/chainer/chainer/functions/pooling/pooling_2d.py", line 54, in forward_gpu
    cudnn.pooling_forward(
Will finalize trainer extensions and updater before reraising the exception.
Process _Worker-3:

my environment is

• Chainer version: 5.0.0
• CuPy version: 6.0.0a1
• OS/Platform: Ubuntu18.04
• CUDA/cuDNN version 10.0/7.4.1

Any help would be useful, thanks.

[beam2d]

I hope the problem has been resolved, but just a note: Chainer does not prefer any specific Python environment management tools over the others. If you still have an installation problem with pyenv, do not hesitate to reopen it or open a new issue. Thanks!

[fiarabbit]

I apologize that I unintentionally led the reader not to use a specific python environment.

For the compensation, I show a small reproduction code of the situation which reported by @crook52 (i.e. the reason of my inference). I just guessed that something like this might have happened for his situation.

$ pyenv install 3.6.0 3.6.1
$ pyenv global 3.6.0
$ git clone <chainer-repository>
$ pip install -e <chainer-repository>
$ pyenv local 3.6.1
$ pip install chainer==v5.0.0
$ $PYENV_ROOT/versions/3.6.0/bin/python -c "import chainer;print(chainer.__versions__)"
6.0.0b2
$ python
>>> import chainer
>>> chainer.print_runtime_info()
Platform: Linux-4.4.0-135-generic-x86_64-with-debian-stretch-sid
Chainer: 5.0.0
NumPy: 1.16.1
CuPy: Not Available
iDeep: Not Available