TED Entities Recognition

Rasa Open Source
1

During the Core Model Training phase, when the TED policy trainig starts, among the debugging logs there is this one:

2021-08-10 08:57:39 DEBUG rasa.core.policies.ted_policy - Entity recognition cannot be performed, set 'entity_recognition' config parameter to 'False'.

I am also using a DIET component, and I annotated my stories with entities information, like:

  steps:
  - intent: color_comment
    entities:
    - color: blue
      role: favorite
  - action: action_set_favorite_color
  - slot_was_set:
    - favorite_color: blue
  - action: action_answer_color_comment
  - checkpoint: favorite_color_not_red_2

When is it possible to train TED for entity recognition? There is some additional info to add to stories or configuration?

TED:

- name: TEDPolicy
    max_history: 15
    constrain_similarities: True
    epochs: 400
    e2e_confidence_threshold: 0.9
    entity_recognition: True
    evaluate_on_number_of_examples: 0
    evaluate_every_number_of_epochs: 25
    tensorboard_log_directory: "./tensorboard"
    tensorboard_log_level: "epoch"
    random_seed: 43
    model_confidence: softmax

DIET:

- name: DIETClassifier
    constrain_similarities: True
    epochs: 200
    use_masked_language_model: True
    BILOU_flag: True
    evaluate_every_number_of_epochs: 20
    random_seed: 43
    tensorboard_log_directory: .\tensorboard

Rasa Version:

Rasa Version      :         2.8.0
Minimum Compatible Version: 2.8.0
Rasa SDK Version  :         2.8.0
Rasa X Version    :         None
Python Version    :         3.8.0
Operating System  :         Windows-10-10.0.19041-SP0
Python Path       :         c:\users\a.porporato\anaconda3\envs\rasatestenv\python.exe

Thanks.

2

Hi @a-porporato

Have you registered your entities in domain.yml

Also, while training your nlu, have you assigned these entities to any of the two examples of your nlu intents. Thanks

Best Regards, Ravi