Keep Deep Learning Model Loaded in Memory in Django Project

Question

I have a deep learning model and I have added it in my django web REST API. I want model to be always loaded in memory. Curretly when user request , model loads in memory and then released after some time. I want it to constantly loaded in ram so that the response time of my api got better. I have deployed API on Windows Server 2019.

Answer 1

You can load it from your apps.py file.

In your django app (Lets' call your app myapp), you may have a file called apps.py in the same level as your models.py, views.py, etc

You override the ready() method in there.

Example in myapp/apps.py

from django.apps import AppConfig

loaded_model = None

class MyAppConfig(AppConfig):
    default_auto_field = 'django.db.models.BigAutoField'
    name = 'myapp'

    def ready(self):
        global loaded_model
        loaded_model = <load your model here>  

Update in your app in INSTALLED_APPS settings in settings.py like this 'myapp.apps.MyAppConfig'

You can then use this in your views.py like this

from .apps import loaded_model

def my_view(request):
    result = loaded_model.<your_model_method>