Tensorflow c++, issue with tensors and feeding

Question

I am very new to TensorFlow C++ API and trying to just build a very simple graph in Python and load/test it in C++ API. Here is the python code to create the graph:

with tf.Session() as sess: 
   a = tf.placeholder(tf.float32, shape=[2,2], name='a')
   b = tf.placeholder(tf.float32, shape=[2,2], name='b')
   c = tf.matmul(a, b, name="c")

   sess.run(tf.initialize_all_variables())

tf.train.write_graph(sess.graph_def, 'models/', 'graph.pb', as_text=False)

and here is the C code to load and run the graph:

Tensor a(DT_FLOAT, TensorShape({2,2}));

Tensor b(DT_FLOAT, TensorShape({2,2}));
std::vector<std::pair<string, tensorflow::Tensor>> inputs = {
  { "a", a },
  { "b", b },
};

std::vector<tensorflow::Tensor> outputs;

status = session->Run(inputs, {"c"}, {}, &outputs);

However I get this error message:

./tensorflow/core/framework/tensor.h:500] Check failed: 1 == NumElements() (1 vs. 4)Must have a one element tensor

What could be the problem? I noticed if I define my tensors in he python and C++ as [1,1], it goes through without any problem!

Hamed MP · Accepted Answer · 2016-03-11 10:04:57Z

1

Take a look at here, I tried to explain everything completely To summarize you should use 'freeze_graph' file to make trained variables constant and then import it in your C++ file.

answered Mar 11, 2016 at 10:04

Hamed MP

5,5033 gold badges41 silver badges38 bronze badges

Sign up to request clarification or add additional context in comments.

Collectives™ on Stack Overflow

Tensorflow c++, issue with tensors and feeding

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related