line 61 in TrainChain Class in chainercv/examples/fpn/train_multi.py ,
I think
x = np.zeros(
(len(imgs), 3, pad_size[0], pad_size[1]), dtype=np.float32)
should be changed to
x = self.xp.zeros(
(len(imgs), 3, pad_size[0], pad_size[1]), dtype=np.float32) because when model is gpu-mode, the model use cupy instead of using numpy.
Am I right?
Hello,
line 61 in TrainChain Class in chainercv/examples/fpn/train_multi.py ,
I think x = np.zeros( (len(imgs), 3, pad_size[0], pad_size[1]), dtype=np.float32) should be changed to x = self.xp.zeros( (len(imgs), 3, pad_size[0], pad_size[1]), dtype=np.float32) because when model is gpu-mode, the model use cupy instead of using numpy. Am I right?