I’m creating a numerical prediction model with a single output node. Whenever I add the output node to the model, I get this error: “Output dimensions does not match with the provided output shape”. This happens with any number of hidden layers or with none at all, as seen in the attached image.
First, what does this mean? Second, how do I adjust the model to avoid this?
The output layer is the layer which you just need to connect at the end of the model.
Its dimension is the dimension of your output.
Since you want to build a model with 1 classes its size is (None,1).
If you want to build NN with just 1 hidden layer then you need to add 2 dense layer.
1st layer will be your hidden layer with any number of neuron and next dense layer will be your output layer which must have 1 as the output dimension and function as softmax since you are doing binary classification.
Thanks! I have fixed that error and I can proceed to Training.
However, when starting Training, I get an error “‘NoneType’ object has no attribute ‘symbol’”.
Can you please help again?
Deeply regret for late reply.
Can you please share the dataset details, logs and model configuration? So that we can assist you on this error.