Sehen Sie sich Keynotes, Produktsitzungen, Workshops und mehr in Google I / O an. Siehe Wiedergabeliste


Strip clustering wrappers from the model.

Used in the notebooks

Used in the guide

Once a model has been clustered, this method can be used to restore the original model with the clustered weights.

Only sequential and functional models are supported for now.

model A tf.keras.Model instance with clustered layers.

A keras model with clustering wrappers removed.

ValueError if the model is not a tf.keras.Model instance.
NotImplementedError if the model is a subclass model.


orig_model = tf.keras.Model(inputs, outputs)
clustered_model = cluster_weights(orig_model)
exported_model = strip_clustering(clustered_model)

The exported_model and the orig_model have the same structure.