[keras] common errors

TypeError: Output tensors to a Model must be Keras tensors.
layers connection is wrong

AttributeError: 'NoneType' object has no attribute 'summary'
your model funtion forget to return model

in fit_generator batch_size = x[0].shape[0] IndexError: list index out of range
八成是输入的generator有问题，检查yield

ValueError: None values not supported.
出现在fit函数中，可以追溯到optimizer的问题，一般是某个操作导致梯度断了，或者定义了可训练的参数，但实际上并没有将参数用到网络中