I followed this timeseries classification Transformer model example in keras.
When I trained, and valided in one file, It works well.
But, when I saved that model, and import model to test file using
loadModel = tf.keras.models.load_model(baseModelPath+modelName)
occured error.
error shown like that...
---> 50 loadModel = tf.keras.models.load_model(baseModelPath+modelName)
and follows some other modules... and last error is
/usr/local/lib/python3.10/dist-packages/keras/layers/attention/multi_head_attention.py in from_config(cls, config)
307 # If the layer has a different build() function from the Keras default,
308 # we need to trigger the customized build to create weights.
--> 309 query_shape = config.pop("query_shape")
310 key_shape = config.pop("key_shape")
311 value_shape = config.pop("value_shape")
KeyError: 'query_shape'
why this problem occured? If saved module is problem, then train code is problem. So, I think if module is problem, then train isn't work correctly. But, train and validation are really works well. So I really don't know what's the problem in this case.
I'm using google colab. Thanks for helping.