gs-ren / shufflenet-gluon Goto Github PK

定义：def channel_shuffle(data, groups):
data = mx.sym.reshape(data, shape=(0, -4, groups, -1, -2))
data = mx.sym.swapaxes(data, 1, 2)
data = mx.sym.reshape(data, shape=(0, -3, -2))
return data
测试数据：blk = Shufflenetv2(3)
blk.initialize()
x = nd.random.uniform(shape=(4, 3, 6, 6))
blk(x).shape
问题：AssertionError Traceback (most recent call last)
in ()
2 blk.initialize()
3 x = nd.random.uniform(shape=(4, 3, 6, 6))
----> 4 blk(x).shape

D:\anaconda\envs\gluon\lib\site-packages\mxnet\gluon\block.py in call(self, *args)
411 def call(self, *args):
412 """Calls forward. Only accepts positional arguments."""
--> 413 return self.forward(*args)
414
415 def forward(self, *args):

in forward(self, x)
48
49 x = nd.concat(x_left,x_right, dim=1)
---> 50 x = channel_shuffle(x,2)
51 return x

in channel_shuffle(data, groups)
1 def channel_shuffle(data, groups):
----> 2 data = mx.sym.reshape(data, shape=(0, -4, groups, -1, -2))
3 data = mx.sym.swapaxes(data, 1, 2)
4 data = mx.sym.reshape(data, shape=(0, -3, -2))
5 return data

D:\anaconda\envs\gluon\lib\site-packages\mxnet\symbol\register.py in reshape(data, shape, reverse, target_shape, keep_highest, name, attr, out, **kwargs)

AssertionError: Argument data must be Symbol instances, but got
[[[[ 6.13965929e-01 6.79090619e-01 5.35632789e-01 8.64281416e-01
5.89909971e-01 7.50844240e-01]
提问：请问这是因为我的测试数据是随机生成的自然数的原因吗？您有遇到过相关问题吗？谢谢