ValueError: Tensor("w1-tu:0", shape=(49, 50), dtype=float32_ref) must be from the same graph as Tensor("Add:0", shape=(?, 49), dtype=float32).
Explanation :
This can happen if you forget to run a sess.run() on initializing tempvalues for your variables.
This error could be much clearer to avoid confusions with 2 different graphs, even if it is somewhat true.
See TensorFlow issue submit
ValueError: Tensor("w1-tu:0", shape=(49, 50), dtype=float32_ref) must be from the same graph as Tensor("Add:0", shape=(?, 49), dtype=float32).
Explanation : This can happen if you forget to run a sess.run() on initializing tempvalues for your variables. This error could be much clearer to avoid confusions with 2 different graphs, even if it is somewhat true. See TensorFlow issue submit