gcn.py 报错。

[tide1994cc]

TypeError: in converted code:

/home/jupyter-tide1994/tide/embedding/GraphNeuralNetwork-master/gnn/gcn.py:73 call *

TypeError: Input 'pred' of 'Switch' Op has type float32 that does not match expected type of bool.

def call(self, inputs, training=None, **kwargs):
    features, A = inputs
    features = self.dropout(features, training=training)
    output = tf.matmul(tf.sparse_tensor_dense_matmul(A, features), self.kernel)
    if self.bias:
        output += self.bias
    act = self.activation(output)

    act._uses_learning_phase = features._uses_learning_phase
    return act

[shenweichen]

在tf1.4下运行试试

[tide1994cc]

还是 不行。。。

[Youskrpig]

if self.bias 后面加上 is not None

[shazhongcheng]

if self.bias 后面加上 is not None

I think it should change self.bias to self.use_bias and remove act._uses_learning_phase = features._uses_learning_phase

And then , i meet error ValueError: setting an array element with a sequence.
And don't how to deal with it.
I only run success in run_gat_cora.py
I also meet error Unsupported feed type in run_graphsage_cora
And don't know how to deal with it.

I see the code in book is pytorch, but why on the github, it turns out TensorFlow??

[shazhongcheng]

好吧 大家都是中国人，作者是不是不打算回应后续问题了呢?
跪求作者解惑！

[shazhongcheng]

书上代码是pytorch版的，而且看起来更加易懂，为什么要改成不好debug的TensorFlow版的呢？而且在更改过程中，好像还因此出现了问题，作者确定可以运行成功吗？
跪求作者大大解惑！

[Billy1900]

if self.bias 后面加上 is not None

I think it should change self.bias to self.use_bias and remove act._uses_learning_phase = features._uses_learning_phase

And then , i meet error ValueError: setting an array element with a sequence.
And don't how to deal with it.
I only run success in run_gat_cora.py
I also meet error Unsupported feed type in run_graphsage_cora
And don't know how to deal with it.

I see the code in book is pytorch, but why on the github, it turns out TensorFlow??

I can run the run_graphsage_cora and run_gat_cora.py with the implementation of tf 1.14.0
but I also met the error in run_gcn_cora.py, the info is following:

Traceback (most recent call last):
  File "run_gcn_cora.py", line 32, in 
    feature_less=FEATURE_LESS, )
  File "/root/Lnq/GraphNeuralNetwork/gnn/gcn.py", line 113, in GCN
    h = GraphConvolution(n_hidden, activation=activation, dropout_rate=dropout_rate, l2_reg=l2_reg)([h,Adj])
  File "/usr/local/miniconda3/envs/tf14/lib/python3.5/site-packages/tensorflow/python/keras/engine/base_layer.py", line 634, in __call__
    outputs = call_fn(inputs, *args, **kwargs)
  File "/usr/local/miniconda3/envs/tf14/lib/python3.5/site-packages/tensorflow/python/autograph/impl/api.py", line 149, in wrapper
    raise e.ag_error_metadata.to_exception(type(e))
TypeError: in converted code:

    /root/Lnq/GraphNeuralNetwork/gnn/gcn.py:74 call
        if self.bias:
    /usr/local/miniconda3/envs/tf14/lib/python3.5/site-packages/tensorflow/python/ops/resource_variable_ops.py:777 __bool__
        return bool(self.read_value())
    /usr/local/miniconda3/envs/tf14/lib/python3.5/site-packages/tensorflow/python/framework/ops.py:690 __bool__
        raise TypeError("Using a `tf.Tensor` as a Python `bool` is not allowed. "

I notice maybe there are errors about the number of parameters in function GCN, but I don't know how to deal with it.

[lishuaibeijing0689]

gcn.py :
[Line 72]output = tf.matmul(tf.matmul(A, features), self.kernel) # fix by ls
[Line 73] if self.use_bias: # fix by ls
[Line 95] Adj = Input(shape=(adj_dim,)) # fix by ls

I fix the bug ValueError: setting an array element with a sequence. like this.
@shazhongcheng