3

I am new for Rasa and stackoverflow,

I have trained a RASA NLU model with the Default rasa config.yml file. with 3 intent and per intent 3 example and total 2 entities.

I have use below commend for train model

config = 'confing.yml'
training_files = './data/'
domain = 'domain.yml'
output = './models/

    import rasa
    rasa.train(domain, config, [training_files], output, fixed_model_name='model_name')

When i have load a RASA NLU model with the following code

    from rasa.core.agent import Agent
    Agent.load(model_path='Model_path')

It's take 1.5 to 2 minute for load model in variable.

Is there anyway that this can be mitigated ? or I am doing something wrong ? Because I want to serve these models on demand, which requires a faster load time.

Any help would be highly appreciated. Thanks in advance!

Sagar jagani
  • 31
  • 2

1 Answers1

1

One of the many things it depends on is the overall pipeline configuration. for eg. If you are using a language model featurizer and you have bert-base-uncased, it'll take slightly more time to load than say compared to distilbert-base-uncased. Similarly if you use a ConveRT featurizer, It'll take a lot less time to load. However, I have found when using the exact same pipeline, Rasa3.x takes more time to load compared to Rasa2.x

ManishL7
  • 11
  • 1
  • Hello @ManishL7, Thanks for your answer. I am new to rasa. **I attached my config.yml file screenshot.** Please check and let me know what changes I need to make in my config file. https://prnt.sc/utOIwwWL-OQm – Sagar jagani Oct 23 '22 at 09:30