Version: 2.0.x

rasa.utils.train_utils

normalize

normalize(values: np.ndarray, ranking_length: Optional[int] = 0) -> np.ndarray

Normalizes an array of positive numbers over the top ranking_length values. Other values will be set to 0.

update_similarity_type

update_similarity_type(config: Dict[Text, Any]) -> Dict[Text, Any]

If SIMILARITY_TYPE is set to 'auto', update the SIMILARITY_TYPE depending on the LOSS_TYPE.

Arguments:

  • config - model configuration
  • Returns - updated model configuration

align_token_features

align_token_features(list_of_tokens: List[List[Token]], in_token_features: np.ndarray, shape: Optional[Tuple] = None) -> np.ndarray

Align token features to match tokens.

ConveRTTokenizer, LanguageModelTokenizers might split up tokens into sub-tokens. We need to take the mean of the sub-token vectors and take that as token vector.

Arguments:

  • list_of_tokens - tokens for examples
  • in_token_features - token features from ConveRT
  • shape - shape of feature matrix

Returns:

Token features.

update_evaluation_parameters

update_evaluation_parameters(config: Dict[Text, Any]) -> Dict[Text, Any]

If EVAL_NUM_EPOCHS is set to -1, evaluate at the end of the training.

Arguments:

  • config - model configuration
  • Returns - updated model configuration

load_tf_hub_model

load_tf_hub_model(model_url: Text) -> Any

Load model from cache if possible, otherwise from TFHub

check_deprecated_options

check_deprecated_options(config: Dict[Text, Any]) -> Dict[Text, Any]

If old model configuration parameters are present in the provided config, replace them with the new parameters and log a warning.

Arguments:

  • config - model configuration
  • Returns - updated model configuration