Tutorial 3 "Tesorflow" is broken

[schnapper79]

I run the tutorial notebooks in a docker as proposed in the ReadMe. Everything works fine, except tutorial no 3.

when running Cell 4

# input images, shape (N, 784)
xs = tf.placeholder(tf.float64)

# weight matrix parameter for soft-max regression, shape (784, 10)
# initialize to a zeros matrix generated by Julia.
init_W = zeros(Float64, 784, 10)
W = tf.Variable(init_W)
    
# bias vector parameter for soft-max regression, shape (10,)
# initialize to a zeros vector generated by Julia.
init_b = zeros(Float64, 10)
b = tf.Variable(init_b)

# probabilities for each class, shape (N, 10)
probs = nn.softmax(tf.add(tf.matmul(xs, W), b), axis=1);

I get the following errormessage:

PyError ($(Expr(:escape, :(ccall(#= /root/.julia/packages/PyCall/0jMpb/src/pyfncall.jl:44 =# @pysym(:PyObject_Call), PyPtr, (PyPtr, PyPtr, PyPtr), o, pyargsptr, kw))))) <class 'TypeError'>
TypeError("unsupported operand type(s) for -: 'NoneType' and 'int'",)
  File "/venv/lib/python3.5/site-packages/tensorflow/python/util/deprecation.py", line 507, in new_func
    return func(*args, **kwargs)
  File "/venv/lib/python3.5/site-packages/tensorflow/python/ops/nn_ops.py", line 2903, in softmax
    return _softmax(logits, gen_nn_ops.softmax, axis, name)
  File "/venv/lib/python3.5/site-packages/tensorflow/python/ops/nn_ops.py", line 2833, in _softmax
    is_last_dim = (dim == -1) or (dim == shape.ndims - 1)


Stacktrace:
 [1] pyerr_check at /root/.julia/packages/PyCall/0jMpb/src/exception.jl:60 [inlined]
 [2] pyerr_check at /root/.julia/packages/PyCall/0jMpb/src/exception.jl:64 [inlined]
 [3] macro expansion at /root/.julia/packages/PyCall/0jMpb/src/exception.jl:84 [inlined]
 [4] __pycall!(::PyObject, ::Ptr{PyCall.PyObject_struct}, ::PyObject, ::PyObject) at /root/.julia/packages/PyCall/0jMpb/src/pyfncall.jl:44
 [5] _pycall!(::PyObject, ::PyObject, ::Tuple{PyObject}, ::Int64, ::PyObject) at /root/.julia/packages/PyCall/0jMpb/src/pyfncall.jl:29
 [6] _pycall!(::PyObject, ::PyObject, ::Tuple{PyObject}, ::Base.Iterators.Pairs{Symbol,Int64,Tuple{Symbol},NamedTuple{(:axis,),Tuple{Int64}}}) at /root/.julia/packages/PyCall/0jMpb/src/pyfncall.jl:11
 [7] #call#89 at /root/.julia/packages/PyCall/0jMpb/src/pyfncall.jl:89 [inlined]
 [8] (::getfield(PyCall, Symbol("#kw#PyObject")))(::NamedTuple{(:axis,),Tuple{Int64}}, ::PyObject, ::PyObject) at ./none:0
 [9] top-level scope at In[4]:13

As all cells before this one work I guess the installation of tensorflow into the docker container was successful. Any Ideas?

[mckeown12]

It appears that tf2.0 is installed in the container. That api no longer uses placeholders. It would be nice to see this tutorial reworked with tf2.0