<chainercv-ssd-azure>物体検出の学習時エラー（azureのみ）

現在、SSDを用いた物体検出を実施しようとこのサイトを参考に学習を進めています。
上記サイトのコードをコピーして実行したところ、google colaboratoryでは学習できましたが、
azure上で実行するとtrainer.run()時に下記エラーが出ます。
なにとぞご教授よろしくお願いします。

python
1Exception in main training loop: merge_sort: failed to synchronize: unspecified launch failure
2Traceback (most recent call last):
3  File "/usr/local/lib/python3.6/dist-packages/chainer/training/trainer.py", line 316, in run
4    update()
5  File "/usr/local/lib/python3.6/dist-packages/chainer/training/updaters/standard_updater.py", line 175, in update
6    self.update_core()
7  File "/usr/local/lib/python3.6/dist-packages/chainer/training/updaters/standard_updater.py", line 187, in update_core
8    optimizer.update(loss_func, *in_arrays)
9  File "/usr/local/lib/python3.6/dist-packages/chainer/optimizer.py", line 825, in update
10    loss = lossfun(*args, **kwds)
11  File "/usr/local/lib/python3.6/dist-packages/chainer/link.py", line 294, in __call__
12    out = forward(*args, **kwargs)
13  File "<ipython-input-7-393c695393be>", line 18, in forward
14    mb_locs, mb_confs, gt_mb_locs, gt_mb_labels, self.k)
15  File "/usr/local/lib/python3.6/dist-packages/chainercv/links/model/ssd/multibox_loss.py", line 91, in multibox_loss
16    hard_negative = _hard_negative(conf_loss.array, positive, k)
17  File "/usr/local/lib/python3.6/dist-packages/chainercv/links/model/ssd/multibox_loss.py", line 19, in _hard_negative
18    rank = (x * (positive - 1)).argsort(axis=1).argsort(axis=1)
19  File "cupy/core/core.pyx", line 627, in cupy.core.core.ndarray.argsort
20  File "cupy/core/core.pyx", line 644, in cupy.core.core.ndarray.argsort
21  File "cupy/core/_routines_sorting.pyx", line 101, in cupy.core._routines_sorting._ndarray_argsort
22  File "cupy/cuda/thrust.pyx", line 135, in cupy.cuda.thrust.argsort
23Will finalize trainer extensions and updater before reraising the exception.
24
25RuntimeErrorTraceback (most recent call last)
26<ipython-input-17-041e2033e90a> in <module>
27----> 1 trainer.run()
28
29/usr/local/lib/python3.6/dist-packages/chainer/training/trainer.py in run(self, show_loop_exception_msg)
30    347                         f.write('Traceback (most recent call last):\n')
31    348                         traceback.print_tb(sys.exc_info()[2])
32--> 349             six.reraise(*exc_info)
33    350         finally:
34    351             for _, entry in extensions:
35
36/usr/local/lib/python3.6/dist-packages/six.py in reraise(tp, value, tb)
37    691             if value.__traceback__ is not tb:
38    692                 raise value.with_traceback(tb)
39--> 693             raise value
40    694         finally:
41    695             value = None
42
43/usr/local/lib/python3.6/dist-packages/chainer/training/trainer.py in run(self, show_loop_exception_msg)
44    314                 self.observation = {}
45    315                 with reporter.scope(self.observation):
46--> 316                     update()
47    317                     for name, entry in extensions:
48    318                         if entry.trigger(self):
49
50/usr/local/lib/python3.6/dist-packages/chainer/training/updaters/standard_updater.py in update(self)
51    173 
52    174         """
53--> 175         self.update_core()
54    176         self.iteration += 1
55    177 
56
57/usr/local/lib/python3.6/dist-packages/chainer/training/updaters/standard_updater.py in update_core(self)
58    185 
59    186         if isinstance(in_arrays, tuple):
60--> 187             optimizer.update(loss_func, *in_arrays)
61    188         elif isinstance(in_arrays, dict):
62    189             optimizer.update(loss_func, **in_arrays)
63
64/usr/local/lib/python3.6/dist-packages/chainer/optimizer.py in update(self, lossfun, *args, **kwds)
65    823         if lossfun is not None:
66    824             use_cleargrads = getattr(self, '_use_cleargrads', True)
67--> 825             loss = lossfun(*args, **kwds)
68    826             if use_cleargrads:
69    827                 self.target.cleargrads()
70
71/usr/local/lib/python3.6/dist-packages/chainer/link.py in __call__(self, *args, **kwargs)
72    292             # forward is implemented in the child classes
73    293             forward = self.forward  # type: ignore
74--> 294         out = forward(*args, **kwargs)
75    295 
76    296         # Call forward_postprocess hook
77
78<ipython-input-7-393c695393be> in forward(self, imgs, gt_mb_locs, gt_mb_labels)
79     16         mb_locs, mb_confs = self.model(imgs)
80     17         loc_loss, conf_loss = multibox_loss(
81---> 18             mb_locs, mb_confs, gt_mb_locs, gt_mb_labels, self.k)
82     19         loss = loc_loss * self.alpha + conf_loss
83     20 
84
85/usr/local/lib/python3.6/dist-packages/chainercv/links/model/ssd/multibox_loss.py in multibox_loss(mb_locs, mb_confs, gt_mb_locs, gt_mb_labels, k, comm)
86     89 
87     90         conf_loss = _elementwise_softmax_cross_entropy(mb_confs, gt_mb_labels)
88---> 91         hard_negative = _hard_negative(conf_loss.array, positive, k)
89     92         conf_loss *= xp.logical_or(
90     93             positive, hard_negative).astype(conf_loss.dtype)
91
92/usr/local/lib/python3.6/dist-packages/chainercv/links/model/ssd/multibox_loss.py in _hard_negative(x, positive, k)
93     17 
94     18 def _hard_negative(x, positive, k):
95---> 19     rank = (x * (positive - 1)).argsort(axis=1).argsort(axis=1)
96     20     hard_negative = rank < (positive.sum(axis=1) * k)[:, np.newaxis]
97     21     return hard_negative
98
99cupy/core/core.pyx in cupy.core.core.ndarray.argsort()
100
101cupy/core/core.pyx in cupy.core.core.ndarray.argsort()
102
103cupy/core/_routines_sorting.pyx in cupy.core._routines_sorting._ndarray_argsort()
104
105cupy/cuda/thrust.pyx in cupy.cuda.thrust.argsort()
106
107RuntimeError: merge_sort: failed to synchronize: unspecified launch failure