Reputation: 1
I am receiving an error message while trying to load a pickled ML model for a data project. How to resolve it?
I have the correct file path and already confirmed the path "/home/jovyan/work", and loaded the pickle before.
I don't know if it is a cloud issue, but when I comment out the writing of the pickle alone, or if I comment out the writing of the pickle and the fitting of the model also... I cannot simply load the pickle for some reason. Which is the whole reason I am pickling the model to not have to fit the model every time.
These are the functions i am using:
def write_pickle(path, model_object, save_as:str):
with open(path + save_as + ".pickle", "wb") as to_write:
pickle.dump(model_object, to_write)
def read_pickle(path, saved_model_name:str):
with open(path + saved_model_name + '.pickle', 'rb') as to_read:
model = pickle.load(to_read)
return model
These functions have worked with the file path and everything, but when i comment out the writing of the pickle or comment out the fitting of the ML model.... I am getting a FileNotFoundError. I don't know what the problem could be. Without me being able to figure this out, it seems that it is useless for me to even be pickling the model.
Upvotes: -1
Views: 72
Reputation: 13
If you comment out your writing then ensure that the model has been already pickled and present in the specified directory which you can check by:
import os
print(os.listdir("home/jovyan/work"))
Also, since its a "FileNotFoundError" and if you're running this in different environment while you mentioned before that it can be a cloud issue....it can be because you might be using the relative path rather than absolute path.
Upvotes: 0