ChainerのConvolution2DをGPU上で動かせません

前提・実現したいこと

Chainerの畳み込み層をGPU上で動かしたい

発生している問題・エラーメッセージ

Traceback (most recent call last):
  File "test.py", line 9, in <module>
    y = l(x)
  File "/home/user/.local/lib/python3.6/site-packages/chainer/links/connection/convolution_2d.py", line 171, in __call__
    groups=self.groups)
  File "/home/user/.local/lib/python3.6/site-packages/chainer/functions/connection/convolution_2d.py", line 577, in convolution_2d
    y, = fnode.apply(args)
  File "/home/user/.local/lib/python3.6/site-packages/chainer/function_node.py", line 235, in apply
    ', '.join(str(type(x)) for x in in_data)))
TypeError: incompatible array types are mixed in the forward input (Convolution2DFunction).
Actual: <class 'cupy.core.core.ndarray'>, <class 'numpy.ndarray'>, <class 'numpy.ndarray'>

該当のソースコード

python
1import chainer 
2import chainer.links as L
3from chainer import cuda 
4
5xp = cuda.cupy 
6
7x  = xp.arange(1 * 3 * 10 * 10, dtype=xp.float32).reshape(1, 3, 10, 10)
8l = L.Convolution2D(3, 7, 5)
9y = l(x)
10print(y.shape)