Warning: This document is for an old version of Rasa. The latest version is 1.10.7.

Components

Note

For clarity, we have renamed the pre-defined pipelines to reflect what they do rather than which libraries they use as of Rasa NLU 0.15. The tensorflow_embedding pipeline is now called supervised_embeddings, and spacy_sklearn is now known as pretrained_embeddings_spacy. Please update your code if you are using these.

Note

We deprecated all pre-defined pipeline templates. Take a look at Choosing a Pipeline to decide on what components you should use in your configuration file.

This is a reference of the configuration options for every built-in component in Rasa NLU. If you want to build a custom component, check out Custom NLU Components.

Word Vector Sources

MitieNLP

Short

MITIE initializer

Outputs

Nothing

Requires

Nothing

Description

Initializes mitie structures. Every mitie component relies on this, hence this should be put at the beginning of every pipeline that uses any mitie components.

Configuration

The MITIE library needs a language model file, that must be specified in the configuration:

pipeline:
- name: "MitieNLP"
  # language model to load
  model: "data/total_word_feature_extractor.dat"

For more information where to get that file from, head over to installing MITIE.

SpacyNLP

Short

spaCy language initializer

Outputs

Nothing

Requires

Nothing

Description

Initializes spaCy structures. Every spaCy component relies on this, hence this should be put at the beginning of every pipeline that uses any spaCy components.

Configuration

Language model, default will use the configured language. If the spaCy model to be used has a name that is different from the language tag ("en", "de", etc.), the model name can be specified using this configuration variable. The name will be passed to spacy.load(name).

pipeline:
- name: "SpacyNLP"
  # language model to load
  model: "en_core_web_md"

  # when retrieving word vectors, this will decide if the casing
  # of the word is relevant. E.g. `hello` and `Hello` will
  # retrieve the same vector, if set to `false`. For some
  # applications and models it makes sense to differentiate
  # between these two words, therefore setting this to `true`.
  case_sensitive: false

For more information on how to obtain the spaCy models, head over to installing SpaCy.

HFTransformersNLP

Short

HuggingFace’s Transformers based pre-trained language model initializer

Outputs

Nothing

Requires

Nothing

Description

Initializes specified pre-trained language model from HuggingFace’s Transformers library. The component applies language model specific tokenization and featurization to compute sequence and sentence level representations for each example in the training data. Include LanguageModelTokenizer and LanguageModelFeaturizer to utilize the output of this component for downstream NLU models.

Note

To use HFTransformersNLP component, install Rasa Open Source with pip install rasa[transformers].

Configuration
pipeline:
  - name: HFTransformersNLP

    # Name of the language model to use
    model_name: "bert"

    # Shortcut name to specify architecture variation of the above model. Full list of supported architectures
    # can be found at https://huggingface.co/transformers/pretrained_models.html . If left empty, it uses the
    # default model architecture that original transformers library loads
    model_weights: "bert-base-uncased"

#    +----------------+--------------+-------------------------+
#    | Language Model | Parameter    | Default value for       |
#    |                | "model_name" | "model_weights"         |
#    +----------------+--------------+-------------------------+
#    | BERT           | bert         | bert-base-uncased       |
#    +----------------+--------------+-------------------------+
#    | GPT            | gpt          | openai-gpt              |
#    +----------------+--------------+-------------------------+
#    | GPT-2          | gpt2         | gpt2                    |
#    +----------------+--------------+-------------------------+
#    | XLNet          | xlnet        | xlnet-base-cased        |
#    +----------------+--------------+-------------------------+
#    | DistilBERT     | distilbert   | distilbert-base-uncased |
#    +----------------+--------------+-------------------------+
#    | RoBERTa        | roberta      | roberta-base            |
#    +----------------+--------------+-------------------------+

Tokenizers

Tokenizers split text into tokens. If you want to split intents into multiple labels, e.g. for predicting multiple intents or for modeling hierarchical intent structure, use these flags with any tokenizer:

  • intent_tokenization_flag indicates whether to tokenize intent labels or not. Set it to True, so that intent labels are tokenized.

  • intent_split_symbol sets the delimiter string to split the intent labels, default is underscore (_).

    Note

    All tokenizer add an additional token __CLS__ to the end of the list of tokens when tokenizing text and responses.

WhitespaceTokenizer

Short

Tokenizer using whitespaces as a separator

Outputs

tokens for user messages, responses (if present), and intents (if specified)

Requires

Nothing

Description

Creates a token for every whitespace separated character sequence.

Configuration

Make the tokenizer case insensitive by adding the case_sensitive: False option, the default being case_sensitive: True.

pipeline:
- name: "WhitespaceTokenizer"
  # Flag to check whether to split intents
  "intent_tokenization_flag": False
  # Symbol on which intent should be split
  "intent_split_symbol": "_"
  # Text will be tokenized with case sensitive as default
  "case_sensitive": True

JiebaTokenizer

Short

Tokenizer using Jieba for Chinese language

Outputs

tokens for user messages, responses (if present), and intents (if specified)

Requires

Nothing

Description

Creates tokens using the Jieba tokenizer specifically for Chinese language. It will only work for the Chinese language.

Note

To use JiebaTokenizer you need to install Jieba with pip install jieba.

Configuration

User’s custom dictionary files can be auto loaded by specifying the files’ directory path via dictionary_path. If the dictionary_path is None (the default), then no custom dictionary will be used.

pipeline:
- name: "JiebaTokenizer"
  dictionary_path: "path/to/custom/dictionary/dir"
  # Flag to check whether to split intents
  "intent_tokenization_flag": False
  # Symbol on which intent should be split
  "intent_split_symbol": "_"

MitieTokenizer

Short

Tokenizer using MITIE

Outputs

tokens for user messages, responses (if present), and intents (if specified)

Requires

MitieNLP

Description

Creates tokens using the MITIE tokenizer.

Configuration
pipeline:
- name: "MitieTokenizer"
  # Flag to check whether to split intents
  "intent_tokenization_flag": False
  # Symbol on which intent should be split
  "intent_split_symbol": "_"

SpacyTokenizer

Short

Tokenizer using spaCy

Outputs

tokens for user messages, responses (if present), and intents (if specified)

Requires

SpacyNLP

Description

Creates tokens using the spaCy tokenizer.

Configuration
pipeline:
- name: "SpacyTokenizer"
  # Flag to check whether to split intents
  "intent_tokenization_flag": False
  # Symbol on which intent should be split
  "intent_split_symbol": "_"

ConveRTTokenizer

Short

Tokenizer using ConveRT model.

Outputs

tokens for user messages, responses (if present), and intents (if specified)

Requires

Nothing

Description

Creates tokens using the ConveRT tokenizer. Must be used whenever the ConveRTFeaturizer is used.

Configuration

Make the tokenizer case insensitive by adding the case_sensitive: False option, the default being case_sensitive: True.

pipeline:
- name: "ConveRTTokenizer"
  # Flag to check whether to split intents
  "intent_tokenization_flag": False
  # Symbol on which intent should be split
  "intent_split_symbol": "_"
  # Text will be tokenized with case sensitive as default
  "case_sensitive": True

LanguageModelTokenizer

Short

Tokenizer from pre-trained language models

Outputs

tokens for user messages, responses (if present), and intents (if specified)

Requires

HFTransformersNLP

Description

Creates tokens using the pre-trained language model specified in upstream HFTransformersNLP component. Must be used whenever the LanguageModelFeaturizer is used.

Configuration
pipeline:
- name: "LanguageModelTokenizer"
  # Flag to check whether to split intents
  "intent_tokenization_flag": False
  # Symbol on which intent should be split
  "intent_split_symbol": "_"

Text Featurizers

Text featurizers are divided into two different categories: sparse featurizers and dense featurizers. Sparse featurizers are featurizers that return feature vectors with a lot of missing values, e.g. zeros. As those feature vectors would normally take up a lot of memory, we store them as sparse features. Sparse features only store the values that are non zero and their positions in the vector. Thus, we save a lot of memory and are able to train on larger datasets.

By default all featurizers will return a matrix of length (number-of-tokens x feature-dimension). So, the returned matrix will have a feature vector for every token. This allows us to train sequence models. However, the additional token at the end (e.g. __CLS__) contains features for the complete utterance. This feature vector can be used in any non-sequence model. The corresponding classifier can therefore decide what kind of features to use.

MitieFeaturizer

Short

Creates a vector representation of user message and response (if specified) using the MITIE featurizer.

Outputs

dense_features for user messages and responses

Requires

MitieNLP

Type

Dense featurizer

Description

Creates features for entity extraction, intent classification, and response classification using the MITIE featurizer.

Note

NOT used by the MitieIntentClassifier component. But can be used by any component later in the pipeline that makes use of dense_features.

Configuration

The sentence vector, i.e. the vector of the __CLS__ token, can be calculated in two different ways, either via mean or via max pooling. You can specify the pooling method in your configuration file with the option pooling. The default pooling method is set to mean.

pipeline:
- name: "MitieFeaturizer"
  # Specify what pooling operation should be used to calculate the vector of
  # the __CLS__ token. Available options: 'mean' and 'max'.
  "pooling": "mean"

SpacyFeaturizer

Short

Creates a vector representation of user message and response (if specified) using the spaCy featurizer.

Outputs

dense_features for user messages and responses

Requires

SpacyNLP

Type

Dense featurizer

Description

Creates features for entity extraction, intent classification, and response classification using the spaCy featurizer.

Configuration

The sentence vector, i.e. the vector of the __CLS__ token, can be calculated in two different ways, either via mean or via max pooling. You can specify the pooling method in your configuration file with the option pooling. The default pooling method is set to mean.

pipeline:
- name: "SpacyFeaturizer"
  # Specify what pooling operation should be used to calculate the vector of
  # the __CLS__ token. Available options: 'mean' and 'max'.
  "pooling": "mean"

ConveRTFeaturizer

Short

Creates a vector representation of user message and response (if specified) using ConveRT model.

Outputs

dense_features for user messages and responses

Requires

ConveRTTokenizer

Type

Dense featurizer

Description

Creates features for entity extraction, intent classification, and response selection. Uses the default signature to compute vector representations of input text.

Note

Since ConveRT model is trained only on an English corpus of conversations, this featurizer should only be used if your training data is in English language.

Note

To use ConveRTFeaturizer you need to install additional TensorFlow libraries (tensorflow_text and tensorflow_hub). You should do a pip install of Rasa with pip install rasa[convert] to install those.

Configuration
pipeline:
- name: "ConveRTFeaturizer"

LanguageModelFeaturizer

Short

Creates a vector representation of user message and response (if specified) using a pre-trained language model.

Outputs

dense_features for user messages and responses

Requires

HFTransformersNLP

Type

Dense featurizer

Description

Creates features for entity extraction, intent classification, and response selection. Uses the pre-trained language model specified in upstream HFTransformersNLP component to compute vector representations of input text.

Note

Please make sure that you use a language model which is pre-trained on the same language corpus as that of your training data.

Configuration

Include HFTransformersNLP and LanguageModelTokenizer components before this component. Use LanguageModelTokenizer to ensure tokens are correctly set for all components throughout the pipeline.

pipeline:
- name: "LanguageModelFeaturizer"

RegexFeaturizer

Short

Creates a vector representation of user message using regular expressions.

Outputs

sparse_features for user messages and tokens.pattern

Requires

tokens

Type

Sparse featurizer

Description

Creates features for entity extraction and intent classification. During training the RegexFeaturizer creates a list of regular expressions defined in the training data format. For each regex, a feature will be set marking whether this expression was found in the input, which will later be fed into intent classifier / entity extractor to simplify classification (assuming the classifier has learned during the training phase, that this set feature indicates a certain intent / entity). Regex features for entity extraction are currently only supported by the CRFEntityExtractor and the DIETClassifier components!

Configuration
pipeline:
- name: "RegexFeaturizer"

CountVectorsFeaturizer

Short

Creates bag-of-words representation of user messages, intents, and responses.

Outputs

sparse_features for user messages, intents, and responses

Requires

tokens

Type

Sparse featurizer

Description

Creates features for intent classification and response selection. Creates bag-of-words representation of user message, intent, and response using sklearn’s CountVectorizer. All tokens which consist only of digits (e.g. 123 and 99 but not a123d) will be assigned to the same feature.

Configuration

See sklearn’s CountVectorizer docs for detailed description of the configuration parameters.

This featurizer can be configured to use word or character n-grams, using analyzer config parameter. By default analyzer is set to word so word token counts are used as features. If you want to use character n-grams, set analyzer to char or char_wb.

Note

Option ‘char_wb’ creates character n-grams only from text inside word boundaries; n-grams at the edges of words are padded with space. This option can be used to create Subword Semantic Hashing.

Note

For character n-grams do not forget to increase min_ngram and max_ngram parameters. Otherwise the vocabulary will contain only single letters.

Handling Out-Of-Vacabulary (OOV) words:

Note

Enabled only if analyzer is word.

Since the training is performed on limited vocabulary data, it cannot be guaranteed that during prediction an algorithm will not encounter an unknown word (a word that were not seen during training). In order to teach an algorithm how to treat unknown words, some words in training data can be substituted by generic word OOV_token. In this case during prediction all unknown words will be treated as this generic word OOV_token.

For example, one might create separate intent outofscope in the training data containing messages of different number of OOV_token s and maybe some additional general words. Then an algorithm will likely classify a message with unknown words as this intent outofscope.

Note

This featurizer creates a bag-of-words representation by counting words, so the number of OOV_token in the sentence might be important.

  • OOV_token set a keyword for unseen words; if training data contains OOV_token as words in some messages, during prediction the words that were not seen during training will be substituted with provided OOV_token; if OOV_token=None (default behaviour) words that were not seen during training will be ignored during prediction time;

  • OOV_words set a list of words to be treated as OOV_token during training; if a list of words that should be treated as Out-Of-Vacabulary is known, it can be set to OOV_words instead of manually changing it in trainig data or using custom preprocessor.

Note

Providing OOV_words is optional, training data can contain OOV_token input manually or by custom additional preprocessor. Unseen words will be substituted with OOV_token only if this token is present in the training data or OOV_words list is provided.

Sharing Vocabulary between user message and labels:

Note

Enabled only if use_shared_vocab is True

Build a common vocabulary set between tokens in labels and user message.

pipeline:
- name: "CountVectorsFeaturizer"
  # whether to use a shared vocab
  "use_shared_vocab": False,
  # whether to use word or character n-grams
  # 'char_wb' creates character n-grams only inside word boundaries
  # n-grams at the edges of words are padded with space.
  analyzer: 'word'  # use 'char' or 'char_wb' for character
  # the parameters are taken from
  # sklearn's CountVectorizer
  # regular expression for tokens
  token_pattern: r'(?u)\b\w\w+\b'
  # remove accents during the preprocessing step
  strip_accents: None  # {'ascii', 'unicode', None}
  # list of stop words
  stop_words: None  # string {'english'}, list, or None (default)
  # min document frequency of a word to add to vocabulary
  # float - the parameter represents a proportion of documents
  # integer - absolute counts
  min_df: 1  # float in range [0.0, 1.0] or int
  # max document frequency of a word to add to vocabulary
  # float - the parameter represents a proportion of documents
  # integer - absolute counts
  max_df: 1.0  # float in range [0.0, 1.0] or int
  # set ngram range
  min_ngram: 1  # int
  max_ngram: 1  # int
  # limit vocabulary size
  max_features: None  # int or None
  # if convert all characters to lowercase
  lowercase: true  # bool
  # handling Out-Of-Vacabulary (OOV) words
  # will be converted to lowercase if lowercase is true
  OOV_token: None  # string or None
  OOV_words: []  # list of strings

LexicalSyntacticFeaturizer

Short

Creates lexical and syntactic features for a user message to support entity extraction.

Outputs

sparse_features for user messages

Requires

tokens

Type

Sparse featurizer

Description

Creates features for entity extraction. Moves with a sliding window over every token in the user message and creates features according to the configuration (see below). As a default configuration is present, you don’t need to specify a configuration.

Configuration

You can configure what kind of lexical and syntactic features the featurizer should extract. The following features are available:

# ==============  ==========================================================================================
# Feature Name    Description
# ==============  ==========================================================================================
# BOS             Checks if the token is at the beginning of the sentence.
# EOS             Checks if the token is at the end of the sentence.
# low             Checks if the token is lower case.
# upper           Checks if the token is upper case.
# title           Checks if the token starts with an uppercase character and all remaining characters are
#                 lowercased.
# digit           Checks if the token contains just digits.
# prefix5         Take the first five characters of the token.
# prefix2         Take the first two characters of the token.
# suffix5         Take the last five characters of the token.
# suffix3         Take the last three characters of the token.
# suffix2         Take the last two characters of the token.
# suffix1         Take the last character of the token.
# pos             Take the Part-of-Speech tag of the token (``SpacyTokenizer`` required).
# pos2            Take the first two characters of the Part-of-Speech tag of the token
#                 (``SpacyTokenizer`` required).
# ==============  ==========================================================================================

As the featurizer is moving over the tokens in a user message with a sliding window, you can define features for previous tokens, the current token, and the next tokens in the sliding window. You define the features as a [before, token, after] array. If you want to define features for the token before, the current token, and the token after, your features configuration would look like this:

pipeline:
- name: "LexicalSyntacticFeaturizer":
  "features": [
    ["low", "title", "upper"],
    [
      "BOS",
      "EOS",
      "low",
      "upper",
      "title",
      "digit",
    ],
    ["low", "title", "upper"],
  ]

This configuration is also the default configuration.

Note

If you want to make use of pos or pos2 you need to add SpacyTokenizer to your pipeline.

Intent Classifiers

Intent classifiers assign one of the intents defined in the domain file to incoming user messages.

MitieIntentClassifier

Short

MITIE intent classifier (using a text categorizer)

Outputs

intent

Requires

tokens for user message

Output-Example
{
    "intent": {"name": "greet", "confidence": 0.98343}
}
Description

This classifier uses MITIE to perform intent classification. The underlying classifier is using a multi-class linear SVM with a sparse linear kernel (see MITIE trainer code).

Configuration
pipeline:
- name: "MitieIntentClassifier"

SklearnIntentClassifier

Short

Sklearn intent classifier

Outputs

intent and intent_ranking

Requires

dense_features for user messages

Output-Example
{
    "intent": {"name": "greet", "confidence": 0.78343},
    "intent_ranking": [
        {
            "confidence": 0.1485910906220309,
            "name": "goodbye"
        },
        {
            "confidence": 0.08161531595656784,
            "name": "restaurant_search"
        }
    ]
}
Description

The sklearn intent classifier trains a linear SVM which gets optimized using a grid search. It also provides rankings of the labels that did not “win”. The SklearnIntentClassifier needs to be preceded by a dense featurizer in the pipeline. This dense featurizer creates the features used for the classification.

Configuration

During the training of the SVM a hyperparameter search is run to find the best parameter set. In the config, you can specify the parameters that will get tried.

pipeline:
- name: "SklearnIntentClassifier"
  # Specifies the list of regularization values to
  # cross-validate over for C-SVM.
  # This is used with the ``kernel`` hyperparameter in GridSearchCV.
  C: [1, 2, 5, 10, 20, 100]
  # Specifies the kernel to use with C-SVM.
  # This is used with the ``C`` hyperparameter in GridSearchCV.
  kernels: ["linear"]

EmbeddingIntentClassifier

Warning

EmbeddingIntentClassifier is deprecated and should be replaced by DIETClassifier. See migration guide for more details.

Short

Embedding intent classifier for intent classification

Outputs

intent and intent_ranking

Requires

dense_features and/or sparse_features for user messages, and optionally the intent

Output-Example
{
    "intent": {"name": "greet", "confidence": 0.78343},
    "intent_ranking": [
        {
            "confidence": 0.1485910906220309,
            "name": "goodbye"
        },
        {
            "confidence": 0.08161531595656784,
            "name": "restaurant_search"
        }
    ]
}
Description

The EmbeddingIntentClassifier embeds user inputs and intent labels into the same space. Supervised embeddings are trained by maximizing similarity between them. This algorithm is based on StarSpace. However, in this implementation the loss function is slightly different and additional hidden layers are added together with dropout. This algorithm also provides similarity rankings of the labels that did not “win”.

Note

If during prediction time a message contains only words unseen during training and no Out-Of-Vacabulary preprocessor was used, an empty intent None is predicted with confidence 0.0. This might happen if you only use the CountVectorsFeaturizer with a word analyzer as featurizer. If you use the char_wb analyzer, you should always get an intent with a confidence value > 0.0.

Configuration

The following hyperparameters can be set:

  • neural network’s architecture:

    • hidden_layers_sizes.text sets a list of hidden layer sizes before the embedding layer for user inputs, the number of hidden layers is equal to the length of the list.

    • hidden_layers_sizes.label sets a list of hidden layer sizes before the embedding layer for intent labels, the number of hidden layers is equal to the length of the list.

    • share_hidden_layers if set to True, shares the hidden layers between user inputs and intent label.

  • training:

    • batch_size sets the number of training examples in one forward/backward pass, the higher the batch size, the more memory space you’ll need.

    • batch_strategy sets the type of batching strategy, it should be either sequence or balanced.

    • epochs sets the number of times the algorithm will see training data, where one epoch equals one forward pass and one backward pass of all the training examples.

    • random_seed if set you will get reproducible training results for the same inputs.

    • learning_rate sets the initial learning rate of the optimizer.

  • embedding:

    • dense_dimension.text sets the dense dimensions for user inputs to use for sparse tensors if no dense features are present.

    • dense_dimension.label sets the dense dimensions for intent labels to use for sparse tensors if no dense features are present.

    • embedding_dimension sets the dimension of embedding space.

    • number_of_negative_examples sets the number of incorrect intent labels. The algorithm will minimize their similarity to the user input during training.

    • similarity_type sets the type of the similarity, it should be either auto, cosine or inner, if auto, it will be set depending on loss_type, inner for softmax, cosine for margin.

    • loss_type sets the type of the loss function, it should be either softmax or margin.

    • ranking_length defines the number of top confidences over which to normalize ranking results if loss_type: "softmax". To turn off normalization set it to 0.

    • maximum_positive_similarity controls how similar the algorithm should try to make embedding vectors for correct intent labels, used only if loss_type is set to margin.

    • maximum_negative_similarity controls maximum negative similarity for incorrect intents, used only if loss_type is set to margin.

    • use_maximum_negative_similarity if true the algorithm only minimizes maximum similarity over incorrect intent labels, used only if loss_type is set to margin.

    • scale_loss if true the algorithm will downscale the loss for examples where correct label is predicted with high confidence, used only if loss_type is set to softmax.

  • regularization:

    • regularization_constant sets the scale of L2 regularization. Higher values will result in more regularization.

    • negative_margin_scale sets the scale of how important is to minimize the maximum similarity between embeddings of different intent labels.

    • drop_rate sets the dropout rate, it should be between 0 and 1, e.g. drop_rate=0.1 would drop out 10% of input units.

    • weight_sparsity sets the sparsity of the weght kernels in dense layers.

    • use_sparse_input_dropout specifies whether to apply dropout to sparse tensors or not.

Note

For cosine similarity maximum_positive_similarity and maximum_negative_similarity should be between -1 and 1.

Note

There is an option to use linearly increasing batch size. The idea comes from https://arxiv.org/abs/1711.00489. In order to do it pass a list to batch_size, e.g. "batch_size": [64, 256] (default behaviour). If constant batch_size is required, pass an int, e.g. "batch_size": 64.

Note

Parameter maximum_negative_similarity is set to a negative value to mimic the original starspace algorithm in the case maximum_negative_similarity = maximum_positive_similarity and use_maximum_negative_similarity = False. See starspace paper for details.

Default values:

pipeline:
- name: "EmbeddingIntentClassifier"
    # ## Architecture of the used neural network
    # Hidden layer sizes for layers before the embedding layers for user message
    # and labels.
    # The number of hidden layers is equal to the length of the corresponding
    # list.
    "hidden_layers_sizes": {"text": [256, 128], "label": []}
    # Whether to share the hidden layer weights between user message and labels.
    "share_hidden_layers": False
    # ## Training parameters
    # Initial and final batch sizes:
    # Batch size will be linearly increased for each epoch.
    "batch_size": [64, 256]
    # Strategy used when creating batches.
    # Can be either 'sequence' or 'balanced'.
    "batch_strategy": "balanced"
    # Number of epochs to train
    "epochs": 300
    # Set random seed to any 'int' to get reproducible results
    "random_seed": None
    # Initial learning rate for the optimizer
    "learning_rate": 0.001
    # ## Parameters for embeddings
    # Dimension size of embedding vectors
    "embedding_dimension": 20
    # Default dense dimension to use if no dense features are present.
    "dense_dimension": {"text": 512, "label": 20}
    # The number of incorrect labels. The algorithm will minimize
    # their similarity to the user input during training.
    "number_of_negative_examples": 20
    # Type of similarity measure to use, either 'auto' or 'cosine' or 'inner'.
    "similarity_type": "auto"
    # The type of the loss function, either 'softmax' or 'margin'.
    "loss_type": "softmax"
    # Number of top actions to normalize scores for loss type 'softmax'.
    # Set to 0 to turn off normalization.
    "ranking_length": 10
    # Indicates how similar the algorithm should try to make embedding vectors
    # for correct labels.
    # Should be 0.0 < ... < 1.0 for 'cosine' similarity type.
    "maximum_positive_similarity": 0.8
    # Maximum negative similarity for incorrect labels.
    # Should be -1.0 < ... < 1.0 for 'cosine' similarity type.
    "maximum_negative_similarity": -0.4
    # If 'True' the algorithm only minimizes maximum similarity over
    # incorrect intent labels, used only if 'loss_type' is set to 'margin'.
    "use_maximum_negative_similarity": True
    # Scale loss inverse proportionally to confidence of correct prediction
    "scale_loss": True
    # ## Regularization parameters
    # The scale of regularization
    "regularization_constant": 0.002
    # The scale of how important is to minimize the maximum similarity
    # between embeddings of different labels.
    "negative_margin_scale": 0.8
    # Sparsity of the weights in dense layers
    "weight_sparsity": 0.8
    # Dropout rate for encoder
    "drop_rate": 0.2
    # If 'True' apply dropout to sparse tensors
    "use_sparse_input_dropout": False
    # ## Evaluation parameters
    # How often calculate validation accuracy.
    # Small values may hurt performance, e.g. model accuracy.
    "evaluate_every_number_of_epochs": 20
    # How many examples to use for hold out validation set
    # Large values may hurt performance, e.g. model accuracy.
    "evaluate_on_number_of_examples": 0

KeywordIntentClassifier

Short

Simple keyword matching intent classifier, intended for small, short-term projects.

Outputs

intent

Requires

Nothing

Output-Example
{
    "intent": {"name": "greet", "confidence": 1.0}
}
Description

This classifier works by searching a message for keywords. The matching is case sensitive by default and searches only for exact matches of the keyword-string in the user message. The keywords for an intent are the examples of that intent in the NLU training data. This means the entire example is the keyword, not the individual words in the example.

Note

This classifier is intended only for small projects or to get started. If you have few NLU training data, you can take a look at the recommended pipelines in Choosing a Pipeline.

Configuration
pipeline:
- name: "KeywordIntentClassifier"
  case_sensitive: True

Selectors

ResponseSelector

Short

Response Selector

Outputs

A dictionary with key as direct_response_intent and value containing response and ranking

Requires

dense_features and/or sparse_features for user messages and response

Output-Example
{
    "response_selector": {
      "faq": {
        "response": {"confidence": 0.7356462617, "name": "Supports 3.5, 3.6 and 3.7, recommended version is 3.6"},
        "ranking": [
            {"confidence": 0.7356462617, "name": "Supports 3.5, 3.6 and 3.7, recommended version is 3.6"},
            {"confidence": 0.2134543431, "name": "You can ask me about how to get started"}
        ]
      }
    }
}
Description

Response Selector component can be used to build a response retrieval model to directly predict a bot response from a set of candidate responses. The prediction of this model is used by Retrieval Actions. It embeds user inputs and response labels into the same space and follows the exact same neural network architecture and optimization as the DIETClassifier.

Note

If during prediction time a message contains only words unseen during training and no Out-Of-Vacabulary preprocessor was used, an empty intent None is predicted with confidence 0.0. This might happen if you only use the CountVectorsFeaturizer with a word analyzer as featurizer. If you use the char_wb analyzer, you should always get an intent with a confidence value > 0.0.

Configuration

The algorithm includes all the hyperparameters that DIETClassifier uses. In addition, the component can also be configured to train a response selector for a particular retrieval intent.

  • retrieval_intent sets the name of the intent for which this response selector model is trained. Default is None, i.e. the model is trained for all retrieval intents.

Default values:

pipeline:
  - name: "ResponseSelector"
    # ## Architecture of the used neural network
    # Hidden layer sizes for layers before the embedding layers for user message
    # and labels.
    # The number of hidden layers is equal to the length of the corresponding
    # list.
    hidden_layers_sizes: {"text": [256, 128], "label": [256, 128]}
    # Whether to share the hidden layer weights between input words and responses
    "share_hidden_layers": False
    # Number of units in transformer
    "transformer_size": None
    # Number of transformer layers
    "number_of_transformer_layers": 0
    # Number of attention heads in transformer
    "number_of_attention_heads": 4
    # If 'True' use key relative embeddings in attention
    "use_key_relative_attention": False
    # If 'True' use key relative embeddings in attention
    "use_value_relative_attention": False
    # Max position for relative embeddings
    "max_relative_position": None
    # Use a unidirectional or bidirectional encoder.
    "unidirectional_encoder": False
    # ## Training parameters
    # Initial and final batch sizes:
    # Batch size will be linearly increased for each epoch.
    "batch_size": [64, 256]
    # Strategy used when creating batches.
    # Can be either 'sequence' or 'balanced'.
    "batch_strategy": "balanced"
    # Number of epochs to train
    "epochs": 300
    # Set random seed to any 'int' to get reproducible results
    "random_seed": None
    # Initial learning rate for the optimizer
    "learning_rate": 0.001
    # ## Parameters for embeddings
    # Dimension size of embedding vectors
    "embedding_dimension": 20
    # Default dense dimension to use if no dense features are present.
    "dense_dimension": {"text": 512, "label": 512}
    # The number of incorrect labels. The algorithm will minimize
    # their similarity to the user input during training.
    "number_of_negative_examples": 20
    # Type of similarity measure to use, either 'auto' or 'cosine' or 'inner'.
    "similarity_type": "auto"
    # The type of the loss function, either 'softmax' or 'margin'.
    "loss_type": "softmax"
    # Number of top actions to normalize scores for loss type 'softmax'.
    # Set to 0 to turn off normalization.
    "ranking_length": 10
    # Indicates how similar the algorithm should try to make embedding vectors
    # for correct labels.
    # Should be 0.0 < ... < 1.0 for 'cosine' similarity type.
    "maximum_positive_similarity": 0.8
    # Maximum negative similarity for incorrect labels.
    # Should be -1.0 < ... < 1.0 for 'cosine' similarity type.
    "maximum_negative_similarity": -0.4
    # If 'True' the algorithm only minimizes maximum similarity over
    # incorrect intent labels, used only if 'loss_type' is set to 'margin'.
    "use_maximum_negative_similarity": True
    # Scale loss inverse proportionally to confidence of correct prediction
    "scale_loss": True
    # ## Regularization parameters
    # The scale of regularization
    "regularization_constant": 0.002
    # Sparsity of the weights in dense layers
    "weight_sparsity": 0.8
    # The scale of how important is to minimize the maximum similarity
    # between embeddings of different labels.
    "negative_margin_scale": 0.8
    # Dropout rate for encoder
    "drop_rate": 0.2
    # Dropout rate for attention
    "drop_rate_attention": 0
    # If 'True' apply dropout to sparse tensors
    "use_sparse_input_dropout": False
    # ## Evaluation parameters
    # How often calculate validation accuracy.
    # Small values may hurt performance, e.g. model accuracy.
    "evaluate_every_number_of_epochs": 20
    # How many examples to use for hold out validation set
    # Large values may hurt performance, e.g. model accuracy.
    "evaluate_on_number_of_examples": 0
    # ## Selector config
    # If 'True' random tokens of the input message will be masked and the model
    # should predict those tokens.
    "use_masked_language_model": False
    # Name of the intent for which this response selector is to be trained
    "retrieval_intent: None

Entity Extractors

Entity extractors extract entities, such as person names or locations, from the user message.

MitieEntityExtractor

Short

MITIE entity extraction (using a MITIE NER trainer)

Outputs

entities

Requires

MitieNLP and tokens

Output-Example
{
    "entities": [{
        "value": "New York City",
        "start": 20,
        "end": 33,
        "confidence": null,
        "entity": "city",
        "extractor": "MitieEntityExtractor"
    }]
}
Description

MitieEntityExtractor uses the MITIE entity extraction to find entities in a message. The underlying classifier is using a multi class linear SVM with a sparse linear kernel and custom features. The MITIE component does not provide entity confidence values.

Configuration
pipeline:
- name: "MitieEntityExtractor"

SpacyEntityExtractor

Short

spaCy entity extraction

Outputs

entities

Requires

SpacyNLP

Output-Example
{
    "entities": [{
        "value": "New York City",
        "start": 20,
        "end": 33,
        "confidence": null,
        "entity": "city",
        "extractor": "SpacyEntityExtractor"
    }]
}
Description

Using spaCy this component predicts the entities of a message. spaCy uses a statistical BILOU transition model. As of now, this component can only use the spaCy builtin entity extraction models and can not be retrained. This extractor does not provide any confidence scores.

Configuration

Configure which dimensions, i.e. entity types, the spaCy component should extract. A full list of available dimensions can be found in the spaCy documentation. Leaving the dimensions option unspecified will extract all available dimensions.

pipeline:
- name: "SpacyEntityExtractor"
  # dimensions to extract
  dimensions: ["PERSON", "LOC", "ORG", "PRODUCT"]

EntitySynonymMapper

Short

Maps synonymous entity values to the same value.

Outputs

Modifies existing entities that previous entity extraction components found.

Requires

Nothing

Description

If the training data contains defined synonyms, this component will make sure that detected entity values will be mapped to the same value. For example, if your training data contains the following examples:

[
    {
      "text": "I moved to New York City",
      "intent": "inform_relocation",
      "entities": [{
        "value": "nyc",
        "start": 11,
        "end": 24,
        "entity": "city",
      }]
    },
    {
      "text": "I got a new flat in NYC.",
      "intent": "inform_relocation",
      "entities": [{
        "value": "nyc",
        "start": 20,
        "end": 23,
        "entity": "city",
      }]
    }
]

This component will allow you to map the entities New York City and NYC to nyc. The entity extraction will return nyc even though the message contains NYC. When this component changes an existing entity, it appends itself to the processor list of this entity.

Configuration
pipeline:
- name: "EntitySynonymMapper"

CRFEntityExtractor

Short

Conditional random field (CRF) entity extraction

Outputs

entities

Requires

tokens and dense_features (optional)

Output-Example
{
    "entities": [{
        "value": "New York City",
        "start": 20,
        "end": 33,
        "entity": "city",
        "confidence": 0.874,
        "extractor": "CRFEntityExtractor"
    }]
}
Description

This component implements a conditional random fields (CRF) to do named entity recognition. CRFs can be thought of as an undirected Markov chain where the time steps are words and the states are entity classes. Features of the words (capitalisation, POS tagging, etc.) give probabilities to certain entity classes, as are transitions between neighbouring entity tags: the most likely set of tags is then calculated and returned.

Note

If POS features are used (pos or pos2), you need to have SpacyTokenizer in your pipeline.

Note

If “pattern” features are used, you need to have RegexFeaturizer in your pipeline.

Configuration

CRFEntityExtractor has a list of default features to use. However, you can overwrite the default configuration. The following features are available:

# ==============  ==========================================================================================
# Feature Name    Description
# ==============  ==========================================================================================
# low             Checks if the token is lower case.
# upper           Checks if the token is upper case.
# title           Checks if the token starts with an uppercase character and all remaining characters are
#                 lowercased.
# digit           Checks if the token contains just digits.
# prefix5         Take the first five characters of the token.
# prefix2         Take the first two characters of the token.
# suffix5         Take the last five characters of the token.
# suffix3         Take the last three characters of the token.
# suffix2         Take the last two characters of the token.
# suffix1         Take the last character of the token.
# pos             Take the Part-of-Speech tag of the token (``SpacyTokenizer`` required).
# pos2            Take the first two characters of the Part-of-Speech tag of the token
#                 (``SpacyTokenizer`` required).
# pattern         Take the patterns defined by ``RegexFeaturizer``.
# bias            Add an additional "bias" feature to the list of features.
# ==============  ==========================================================================================

As the featurizer is moving over the tokens in a user message with a sliding window, you can define features for previous tokens, the current token, and the next tokens in the sliding window. You define the features as [before, token, after] array.

Additional you can set a flag to determine whether to use the BILOU tagging schema or not.

  • BILOU_flag determines whether to use BILOU tagging or not. Default True.

pipeline:
- name: "CRFEntityExtractor"
    # BILOU_flag determines whether to use BILOU tagging or not.
    # More rigorous however requires more examples per entity
    # rule of thumb: use only if more than 100 egs. per entity
    "BILOU_flag": True
    # crf_features is [before, token, after] array with before, token,
    # after holding keys about which features to use for each token,
    # for example, 'title' in array before will have the feature
    # "is the preceding token in title case?"
    # POS features require SpacyTokenizer
    # pattern feature require RegexFeaturizer
    "features": [
        ["low", "title", "upper"],
        [
            "bias",
            "low",
            "prefix5",
            "prefix2",
            "suffix5",
            "suffix3",
            "suffix2",
            "upper",
            "title",
            "digit",
            "pattern",
        ],
        ["low", "title", "upper"],
    ]
    # The maximum number of iterations for optimization algorithms.
    "max_iterations": 50
    # weight of the L1 regularization
    "L1_c": 0.1
    # weight of the L2 regularization
    "L2_c": 0.1

DucklingHTTPExtractor

Short

Duckling lets you extract common entities like dates, amounts of money, distances, and others in a number of languages.

Outputs

entities

Requires

Nothing

Output-Example
{
    "entities": [{
        "end": 53,
        "entity": "time",
        "start": 48,
        "value": "2017-04-10T00:00:00.000+02:00",
        "confidence": 1.0,
        "extractor": "DucklingHTTPExtractor"
    }]
}
Description

To use this component you need to run a duckling server. The easiest option is to spin up a docker container using docker run -p 8000:8000 rasa/duckling.

Alternatively, you can install duckling directly on your machine and start the server.

Duckling allows to recognize dates, numbers, distances and other structured entities and normalizes them. Please be aware that duckling tries to extract as many entity types as possible without providing a ranking. For example, if you specify both number and time as dimensions for the duckling component, the component will extract two entities: 10 as a number and in 10 minutes as a time from the text I will be there in 10 minutes. In such a situation, your application would have to decide which entity type is be the correct one. The extractor will always return 1.0 as a confidence, as it is a rule based system.

Configuration

Configure which dimensions, i.e. entity types, the duckling component should extract. A full list of available dimensions can be found in the duckling documentation. Leaving the dimensions option unspecified will extract all available dimensions.

pipeline:
- name: "DucklingHTTPExtractor"
  # url of the running duckling server
  url: "http://localhost:8000"
  # dimensions to extract
  dimensions: ["time", "number", "amount-of-money", "distance"]
  # allows you to configure the locale, by default the language is
  # used
  locale: "de_DE"
  # if not set the default timezone of Duckling is going to be used
  # needed to calculate dates from relative expressions like "tomorrow"
  timezone: "Europe/Berlin"
  # Timeout for receiving response from http url of the running duckling server
  # if not set the default timeout of duckling http url is set to 3 seconds.
  timeout : 3

Combined Entity Extractors and Intent Classifiers

DIETClassifier

Short

Dual Intent Entity Transformer (DIET) used for intent classification and entity extraction

Outputs

entities, intent and intent_ranking

Requires

dense_features and/or sparse_features for user message and, and optionally the intent

Output-Example
{
    "intent": {"name": "greet", "confidence": 0.8343},
    "intent_ranking": [
        {
            "confidence": 0.385910906220309,
            "name": "goodbye"
        },
        {
            "confidence": 0.28161531595656784,
            "name": "restaurant_search"
        }
    ],
    "entities": [{
        "end": 53,
        "entity": "time",
        "start": 48,
        "value": "2017-04-10T00:00:00.000+02:00",
        "confidence": 1.0,
        "extractor": "DIETClassifier"
    }]
}
Description

DIET (Dual Intent and Entity Transformer) is a multi-task architecture for intent classification and entity recognition. The architecture is based on a transformer which is shared for both tasks. A sequence of entity labels is predicted through a Conditional Random Field (CRF) tagging layer on top of the transformer output sequence corresponding to the input sequence of tokens. The transformer output for the __CLS__ token and intent labels are embedded into a single semantic vector space. We use the dot-product loss to maximize the similarity with the target label and minimize similarities with negative samples.

Note

If during prediction time a message contains only words unseen during training and no Out-Of-Vacabulary preprocessor was used, an empty intent None is predicted with confidence 0.0. This might happen if you only use the CountVectorsFeaturizer with a word analyzer as featurizer. If you use the char_wb analyzer, you should always get an intent with a confidence value > 0.0.

Configuration

The following hyperparameters can be set:

  • neural network’s architecture:

    • hidden_layers_sizes.text sets a list of hidden layer sizes before the embedding layer for user inputs, the number of hidden layers is equal to the length of the list.

    • hidden_layers_sizes.label sets a list of hidden layer sizes before the embedding layer for intent labels, the number of hidden layers is equal to the length of the list.

    • share_hidden_layers if set to True, shares the hidden layers between user inputs and intent label.

    • transformer_size sets the size of the transformer.

    • number_of_transformer_layers sets the number of transformer layers to use.

    • number_of_attention_heads sets the number of attention heads to use.

    • unidirectional_encoder specifies whether to use a unidirectional or bidirectional encoder.

    • use_key_relative_attention if true use key relative embeddings in attention.

    • use_value_relative_attention if true use key relative embeddings in attention.

    • max_relative_position sets the max position for relative embeddings.

  • training:

    • batch_size sets the number of training examples in one forward/backward pass, the higher the batch size, the more memory space you’ll need.

    • batch_strategy sets the type of batching strategy, it should be either sequence or balanced.

    • epochs sets the number of times the algorithm will see training data, where one epoch equals one forward pass and one backward pass of all the training examples.

    • random_seed if set you will get reproducible training results for the same inputs.

    • learning_rate sets the initial learning rate of the optimizer.

  • embedding:

    • dense_dimension.text sets the dense dimensions for user inputs to use for sparse tensors if no dense features are present.

    • dense_dimension.label sets the dense dimensions for intent labels to use for sparse tensors if no dense features are present.

    • embedding_dimension sets the dimension of embedding space.

    • number_of_negative_examples sets the number of incorrect intent labels. The algorithm will minimize their similarity to the user input during training.

    • similarity_type sets the type of the similarity, it should be either auto, cosine or inner, if auto, it will be set depending on loss_type, inner for softmax, cosine for margin.

    • loss_type sets the type of the loss function, it should be either softmax or margin.

    • ranking_length defines the number of top confidences over which to normalize ranking results if loss_type: "softmax". To turn off normalization set it to 0.

    • maximum_positive_similarity controls how similar the algorithm should try to make embedding vectors for correct intent labels, used only if loss_type is set to margin.

    • maximum_negative_similarity controls maximum negative similarity for incorrect intents, used only if loss_type is set to margin.

    • use_maximum_negative_similarity if true the algorithm only minimizes maximum similarity over incorrect intent labels, used only if loss_type is set to margin.

    • scale_loss if true the algorithm will downscale the loss for examples where correct label is predicted with high confidence, used only if loss_type is set to softmax.

  • regularization:

    • regularization_constant sets the scale of L2 regularization. Higher values will result in more regularization.

    • negative_margin_scale sets the scale of how important is to minimize the maximum similarity between embeddings of different intent labels.

    • drop_rate sets the dropout rate, it should be between 0 and 1, e.g. drop_rate=0.1 would drop out 10% of input units.

    • drop_rate_attention sets the dropout rate for attention, it should be between 0 and 1, e.g. drop_rate_attention=0.1 would drop out 10% of input units.

    • weight_sparsity sets the sparsity of weight kernels in dense layers.

    • use_sparse_input_dropout specifies whether to apply dropout to sparse tensors or not.

  • model configuration:

    • use_masked_language_model specifies whether to apply masking or not.

    • intent_classification indicates whether intent classification should be performed or not.

    • entity_recognition indicates whether entity recognition should be performed or not.

    • BILOU_flag determines whether to use BILOU tagging or not.

Note

For cosine similarity maximum_positive_similarity and maximum_negative_similarity should be between -1 and 1.

Note

There is an option to use linearly increasing batch size. The idea comes from https://arxiv.org/abs/1711.00489. In order to do it pass a list to batch_size, e.g. "batch_size": [64, 256] (default behaviour). If constant batch_size is required, pass an int, e.g. "batch_size": 64.

Note

Parameter maximum_negative_similarity is set to a negative value to mimic the original starspace algorithm in the case maximum_negative_similarity = maximum_positive_similarity and use_maximum_negative_similarity = False. See starspace paper for details.

Default values:

pipeline:
- name: "DIETClassifier"
    # ## Architecture of the used neural network
    # Hidden layer sizes for layers before the embedding layers for user message
    # and labels.
    # The number of hidden layers is equal to the length of the corresponding
    # list.
    "hidden_layers_sizes": {TEXT: [], LABEL: []}
    # Whether to share the hidden layer weights between user message and labels.
    "share_hidden_layers": False
    # Number of units in transformer
    "transformer_size": 256
    # Number of transformer layers
    "number_of_transformer_layers": 2
    # Number of attention heads in transformer
    "number_of_attention_heads": 4
    # If 'True' use key relative embeddings in attention
    "use_key_relative_attention": False
    # If 'True' use key relative embeddings in attention
    "use_value_relative_attention": False
    # Max position for relative embeddings
    "max_relative_position": None
    # Max sequence length
    "maximum_sequence_length": 256
    # Use a unidirectional or bidirectional encoder.
    "unidirectional_encoder": False
    # ## Training parameters
    # Initial and final batch sizes:
    # Batch size will be linearly increased for each epoch.
    "batch_size": [64, 256]
    # Strategy used when creating batches.
    # Can be either 'sequence' or 'balanced'.
    "batch_strategy": "balanced"
    # Number of epochs to train
    "epochs": 300
    # Set random seed to any 'int' to get reproducible results
    "random_seed": None
    # Initial learning rate for the optimizer
    "learning_rate": 0.001
    # ## Parameters for embeddings
    # Dimension size of embedding vectors
    "embedding_dimension": 20
    # Default dense dimension to use if no dense features are present.
    "dense_dimension": {TEXT: 512, LABEL: 20}
    # The number of incorrect labels. The algorithm will minimize
    # their similarity to the user input during training.
    "number_of_negative_examples": 20
    # Type of similarity measure to use, either 'auto' or 'cosine' or 'inner'.
    "similarity_type": "auto"
    # The type of the loss function, either 'softmax' or 'margin'.
    "loss_type": "softmax"
    # Number of top actions to normalize scores for loss type 'softmax'.
    # Set to 0 to turn off normalization.
    "ranking_length": 10
    # Indicates how similar the algorithm should try to make embedding vectors
    # for correct labels.
    # Should be 0.0 < ... < 1.0 for 'cosine' similarity type.
    "maximum_positive_similarity": 0.8
    # Maximum negative similarity for incorrect labels.
    # Should be -1.0 < ... < 1.0 for 'cosine' similarity type.
    "maximum_negative_similarity": -0.4
    # If 'True' the algorithm only minimizes maximum similarity over
    # incorrect intent labels, used only if 'loss_type' is set to 'margin'.
    "use_maximum_negative_similarity": True
    # Scale loss inverse proportionally to confidence of correct prediction
    "scale_loss": True
    # ## Regularization parameters
    # The scale of regularization
    "regularization_constant": 0.002
    # The scale of how important is to minimize the maximum similarity
    # between embeddings of different labels.
    "negative_margin_scale": 0.8
    # Sparsity of the weights in dense layers
    "weight_sparsity": 0.8
    # Dropout rate for encoder
    "drop_rate": 0.2
    # Dropout rate for attention
    "drop_rate_attention": 0
    # If 'True' apply dropout to sparse tensors
    "use_sparse_input_dropout": True
    # ## Evaluation parameters
    # How often calculate validation accuracy.
    # Small values may hurt performance, e.g. model accuracy.
    "evaluate_every_number_of_epochs": 20
    # How many examples to use for hold out validation set
    # Large values may hurt performance, e.g. model accuracy.
    "evaluate_on_number_of_examples": 0
    # ## Model config
    # If 'True' intent classification is trained and intent predicted.
    "intent_classification": True
    # If 'True' named entity recognition is trained and entities predicted.
    "entity_recognition": True
    # If 'True' random tokens of the input message will be masked and the model
    # should predict those tokens.
    "use_masked_language_model": False
    # 'BILOU_flag' determines whether to use BILOU tagging or not.
    # If set to 'True' labelling is more rigorous, however more
    # examples per entity are required.
    # Rule of thumb: you should have more than 100 examples per entity.
    "BILOU_flag": True