rasa.nlu.selectors.response_selector
ResponseSelector Objects
Response selector using supervised embeddings.
The response selector embeds user inputs and candidate response into the same space. Supervised embeddings are trained by maximizing similarity between them. It also provides rankings of the response that did not "win".
The supervised response selector needs to be preceded by
a featurizer in the pipeline.
This featurizer creates the features used for the embeddings.
It is recommended to use CountVectorsFeaturizer
that
can be optionally preceded by SpacyNLP
and SpacyTokenizer
.
Based on the starspace idea from: https://arxiv.org/abs/1709.03856.
However, in this implementation the mu
parameter is treated differently
and additional hidden layers are added together with dropout.
required_components
Components that should be included in the pipeline before this component.
get_default_config
The component's default config (see parent class for full docstring).
__init__
Declare instance variables with default values.
Arguments:
config
- Configuration for the component.model_storage
- Storage which graph components can use to persist and load themselves.resource
- Resource locator for this component which can be used to persist and load itself from themodel_storage
.execution_context
- Information about the current graph run.index_label_id_mapping
- Mapping between label and index used for encoding.entity_tag_specs
- Format specification all entity tags.model
- Model architecture.all_retrieval_intents
- All retrieval intents defined in the data.responses
- All responses defined in the data.model_storage
0 - Ifmodel_storage
1 loads the model with pre-trained weights, otherwise initializes it with random weights.model_storage
2 - Sizes of the sparse features the model was trained on.
label_key
Returns label key.
label_sub_key
Returns label sub_key.
model_class
Returns model class.
preprocess_train_data
Prepares data for training.
Performs sanity checks on training data, extracts encodings for labels.
Arguments:
training_data
- training data to preprocessed.
process
Selects most like response for message.
Arguments:
messages
- List containing latest user message.
Returns:
List containing the message augmented with the most likely response, the associated intent_response_key and its similarity to the input.
persist
Persist this model into the passed directory.
load
Loads the trained model from the provided directory.
DIET2BOW Objects
DIET2BOW transformer implementation.
DIET2DIET Objects
Diet 2 Diet transformer implementation.
batch_loss
Calculates the loss for the given batch.
Arguments:
batch_in
- The batch.
Returns:
The loss of the given batch.
batch_predict
Predicts the output of the given batch.
Arguments:
batch_in
- The batch.
Returns:
The output to predict.