Closed Kovax007 closed 2 years ago
Issue is when usign tf.cast to calculate any parameter, that it wants to create with the keras.metadatafile, converting the values with float() or using numpy solves the issue. as tf.cast creates EagerTensor object
Loading the tfprocess.py with a yaml, than using the model.save('somepath') or either tf.keras.models.save_model() function does not work and return with the error: TypeError: Unable to serialize 0.319471538066864 to JSON. Unrecognized type <class 'tensorflow.python.framework.ops.EagerTensor'>.