[GitHub] hsddlz opened a new issue #10065: Error in operator reshape

[GitBox <gi...@apache.org>]

hsddlz opened a new issue #10065: Error in operator reshape 
URL: https://github.com/apache/incubator-mxnet/issues/10065
 
 
   when I bind the data .It told me like this:
   
   simple_bind error. Arguments:
   data: (32, 1L, 32L, 286L)
   softmax_label: (32, 21L)
   Error in operator reshape0: [19:25:11] src/operator/tensor/./matrix_op-inl.h:157: Check failed: oshape.Size() == dshape.Size() (17920 vs. 573440) Target shape size is different to source. Target: [35,512]
   Source: [32,128,4,35]
   
   my ndarray data shape is [DataDesc[data,(32, 1L, 32L, 286L),<type 'numpy.float32'>,NCHW]]
   
   this is my code 
   `stride=(1,1)
   bn_mom=0.9
   fix_gamma=False
   eps=2e-5
   rnn_length =35
   rnn_dimen=512
   rnn_size = 256
   
   
   x = mx.sym.Variable("data")
   label = mx.sym.Variable('label')
   for i, n_cnn in enumerate([3, 4, 6]):
       for j in range(n_cnn):
           x = mx.symbol.Convolution(data=x, num_filter=32*2**i, kernel=(3, 3), stride=stride, pad=(1,1))
           x = mx.symbol.BatchNorm(data=x, fix_gamma=fix_gamma, eps=eps, momentum=bn_mom)
           x = mx.symbol.Activation(data=x, act_type='relu')
       
       x = mx.sym.Pooling(data=x, kernel=(2, 2), stride=(2, 2), pool_type='max')
   x=mx.symbol.reshape(x,shape=(rnn_length, rnn_dimen))`
   
   how can I fix it.thx

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
users@infra.apache.org


With regards,
Apache Git Services