Currently, the DSLstmPredictor only consists of a an Extrapolator Network, which takes a full Featureset (Sequence of characters+categories) as input. However, to enable free autocompletion, a second network is needed to predict token categories from a string.
Currently, the DSLstmPredictor only consists of a an Extrapolator Network, which takes a full Featureset (Sequence of characters+categories) as input. However, to enable free autocompletion, a second network is needed to predict token categories from a string.