lyst / lightfm

A Python implementation of LightFM, a hybrid recommendation algorithm.
Apache License 2.0
4.66k stars 679 forks source link

How to deploy model using node js #669

Open s-bh opened 1 year ago

s-bh commented 1 year ago

I created a recommendation system using lightFM on Google Colab. I want to deploy this model using Node JS. I read this article and tried the method recommended here. It recommended using tensorflow.js. For that the model needed to be saved and this is where the problem arose since I wrote my model using lightFM.

This gave rise to following error: ` AttributeError Traceback (most recent call last)

in ----> 1 model.save("model.h5") AttributeError: 'LightFM' object has no attribute 'save' ` What is the alternative to model.save in lightfm? How can we deploy model written in lightfm using node js?
np-n commented 12 months ago

You can save the model using the serialization library called pickle and load the model too.

import pickle
from lightfm import LightFM

# instantiate model
# train your LightFM model here

# Save the trained model to a pickle file
with open("./model_file.pkl", "wb") as file:
    pickle.dump(model, file)

and load the model:

import pickle
from lightfm import LightFM

# read the pickle file
with open("./model_file.pkl", "rb") as file:
    model = pickle.load(file)

I hope this will help to load and save the model. Thank you!