RuntimeError: cuda runtime error (8) : invalid device function at /pytorch/torch/lib/THC/generic/THCTensorIndex.cu:648

[GYM13]

Traceback (most recent call last):
File "check_model.py", line 101, in
scores, boxes, net_conv, im_scale = im_detect(net, im)
File "../tools/../lib/model/test.py", line 104, in im_detect
_, scores, bbox_pred, rois, net_conv = net.test_image(blobs['data'], blobs['im_info'])
File "../tools/../lib/nets/network.py", line 571, in test_image
self.forward(image, im_info, None, None, mode='TEST')
File "../tools/../lib/nets/network.py", line 541, in forward
net_conv, rois, cls_prob, bbox_pred, mask_prob = self._predict()
File "../tools/../lib/nets/network.py", line 468, in _predict
rois = self._region_proposal(net_conv)
File "../tools/../lib/nets/network.py", line 256, in _region_proposal
rois, _ = self._proposal_layer(rpn_cls_prob, rpn_bbox_pred)
File "../tools/../lib/nets/network.py", line 94, in _proposal_layer
self._feat_stride, self._anchors, self._num_anchors)
File "../tools/../lib/layer_utils/proposal_layer.py", line 61, in proposal_layer
proposals = proposals[keep, :]
File "/usr/lib64/python2.7/site-packages/torch/autograd/variable.py", line 78, in getitem
return Index.apply(self, key)
File "/usr/lib64/python2.7/site-packages/torch/autograd/_functions/tensor.py", line 87, in forward
result = i.index(ctx.index)

[winterest]

Have you figured it out?

[Millielele]

cuda8.0 pytorch0.4.0 maybe work