Giters

jmschrei / apricot

apricot implements submodular optimization for the purpose of selecting subsets of massive data sets to train machine learning models quickly. See the documentation page: https://apricot-select.readthedocs.io/en/latest/index.html

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

CustomSelection creates multiple copies of my submodular function method

itskalvik opened this issue · comments

commented

Hi, I have a submodular function that is implemented as a python class. My class has a method that is passed to a CustomSelection object as follows

subfn= CustomSubmodular(arg1, arg2, ...)
model = CustomSelection(5, subfn.submodular_function)

But when I optimize this model, it creates new instances of my CustomSubmodular class instead of using the provided instance. Is there a way around this? I suppose I can use the function_kwds parameter of CustomSelection to make a submodular function that is not defined as a class, but I would like to avoid it if possible.

commented

The problem was my class was using joblib to parallelize a task and it was making a copy of my class instance, not apricot.