hi everyone can we give me what’s the difference between the keras policy and embedded policy in features.is embedded policy quickly learns the dialog follow then keras policy or it has some addition features.notes i am not asking about the what is keras policy and embedded policy i m just ask about features of embedded policy compare to keras policy. thanks for the future reply
KerasPolicy only featurizes a set number of dialogue turns from the training data (this parameter is called
max_history and its default is
3, but can be configured).
EmbeddingPolicy takes the longest dialogue in your story data and creates feature vectors which are padded to this length.
Did this answer your question? There’s more info in the docs
thanks for the reply i will check the docs