Unload a rasa model from memory

Rasa Open Source
1

I am using the below code to load a rasa model in memory.

from rasa.nlu.model import Interpreter

model = Interpreter.load(nlu_model)

Is there something similar to easily unload model from the memory and quickly release all the memory.

Currently I am storing the references to each loaded model in a python dictionary data structure and using the below code to delete the model

models = {} // This keeps  references to all the loaded models
del models['model_id']

Looks like , this doesn’t completely releases all the memory and I am sure there must be some better way of unloading model from the memory.

Using the below nlu pipeline to train the model :

pipeline : 
   - name: LanguageModelTokenizer
     intent_tokenization_flag : False
   - name: LanguageModelFeaturizer
     model_name : "distilbert"
     model_weights : "distilbert-base-uncased"
   - name: DIETClassifier
     epochs: 300
     entity_recognition: False
     checkpoint_model: True
     random_seed: 42
     evaluate_every_number_of_epochs: 100
  policies : Null
2

I don’t believe so. I think your option is to load a new model or exit.