notice
This is unreleased documentation for Rasa Open Source Documentation Master/Unreleased version.
For the latest released documentation, see the latest version (2.2.x).
rasa.utils.train_utils
normalize
Normalizes an array of positive numbers over the top ranking_length
values.
Other values will be set to 0.
update_similarity_type
If SIMILARITY_TYPE is set to 'auto', update the SIMILARITY_TYPE depending on the LOSS_TYPE.
Arguments:
config
- model configurationReturns
- updated model configuration
align_token_features
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 examplesin_token_features
- token features from ConveRTshape
- shape of feature matrix
Returns:
Token features.
update_evaluation_parameters
If EVAL_NUM_EPOCHS is set to -1, evaluate at the end of the training.
Arguments:
config
- model configurationReturns
- updated model configuration
load_tf_hub_model
Load model from cache if possible, otherwise from TFHub
check_deprecated_options
Update the config according to changed config params.
If old model configuration parameters are present in the provided config, replace them with the new parameters and log a warning.
Arguments:
config
- model configurationReturns
- updated model configuration
check_core_deprecated_options
Update the core config according to changed config params.
If old model configuration parameters are present in the provided config, replace them with the new parameters and log a warning.
Arguments:
config
- model configurationReturns
- updated model configuration
entity_label_to_tags
Convert the output predictions for entities to the actual entity tags.
Arguments:
model_predictions
- the output predictions using the entity tag indicesentity_tag_specs
- the entity tag specificationsbilou_flag
- if 'True', the BILOU tagging schema was usedprediction_index
- the index in the batch of predictions to use for entity extraction
Returns:
A map of entity tag type, e.g. entity, role, group, to actual entity tags and confidences.
override_defaults
Override default config with the given config.
We cannot use dict.update
method because configs contain nested dicts.
Arguments:
defaults
- default configcustom
- user config containing new parameters
Returns:
updated config
init_split_entities
Initialise the behaviour for splitting entities by comma (or not).
Returns:
Defines desired behaviour for splitting specific entity types and default behaviour for splitting any entity types for which no behaviour is defined.