ERROR (theano.gof.opt): Optimization failure due to: constant_folding

问题

In Neural Networks and Deep Learning, there's an object called network3 (which is a PY file, written for python 2.7 and theano 0.7). I modified it to run with python 3.6 and theano 1.0.3. However, when I run the following code:

import network3
from network3 import Network
from network3 import ConvPoolLayer , FullyConnectedLayer , SoftmaxLayer

training_data , validation_data , test_data = network3.load_data_shared()

mini_batch_size = 10

net = Network([FullyConnectedLayer(n_in=784, n_out=100),
               SoftmaxLayer(n_in=100, n_out=10)], mini_batch_size)

net.SGD(training_data , 60, mini_batch_size , 0.1, validation_data , test_data)

It returned a lot of errors that have similar pattern like this:

ERROR (theano.gof.opt): Optimization failure due to: constant_folding

ERROR (theano.gof.opt): node: MakeVector{dtype='int64'}(TensorConstant{10}, TensorConstant{100})

AttributeError: module 'numpy.core.multiarray' has no attribute '_get_ndarray_c_version'

Then the last paragraph in the output was:

AttributeError: ('The following error happened while compiling the node', Elemwise{Composite{(i0 * (i1 + i2))}}(TensorConstant{10}, TensorConstant{1}, <TensorType(int64, scalar)>), '\n', "module 'numpy.core.multiarray' has no attribute '_get_ndarray_c_version'")

I think line 1379 caused the error:

   1298                                  libraries=self.libraries(),
   1299                                  header_dirs=self.header_dirs(),
-> 1300                                  c_compiler=self.c_compiler(),
   1301                                  )
   1302 

~\AppData\Local\Continuum\Anaconda3\lib\site-packages\theano\gof\cc.py in cmodule_key_(self, fgraph, no_recycling, compile_args, libraries, header_dirs, insert_config_hash, c_compiler)
   1377         # DynamicModule always add the include <numpy/arrayobject.h>
   1378         sig.append('NPY_ABI_VERSION=0x%X' %
-> 1379                    np.core.multiarray._get_ndarray_c_version())
   1380         if c_compiler:
   1381             sig.append('c_compiler_str=' + c_compiler.version_str())

AttributeError: ('The following error happened while compiling the node', Elemwise{Composite{(i0 * (i1 + i2))}}(TensorConstant{10}, TensorConstant{1}, <TensorType(int64, scalar)>), '\n', "module 'numpy.core.multiarray' has no attribute '_get_ndarray_c_version'")

I have googled the issues but couldn't find relevant solution. Could you please help?

Thanks in advance.

回答1:

What is your numpy version? Open a python terminal and run the following lines from numpy import __version__ and then print(__version__). In the numpy version (1.14.2) that I am using this function is well defined.

Sometimes a package update can depreciate a function or even remove a function but also propose a new function to replace the older one. I don't know if that is the case but be always aware of this possibility.

来源：https://stackoverflow.com/questions/55024382/error-theano-gof-opt-optimization-failure-due-to-constant-folding

标签

python

deep-learning

conv-neural-network

theano