Cuda Out of Memory

[AmoghTiwari]

This is a follow up issue of Issue #9. I am also using a 1080Ti machine similar to issue #9. Please refer to issue #9 for more details).

The steps mentioned there weren't working directly as the repository has moved ahead from the file structure which existed when the above solution was provided.

[AmoghTiwari]

Here's what I did to resolve the issue.

1. Go to the point in the repository when the above-mentioned commit with a solution was made. While doing this, I didn't want to lose track of the most recent version of the repo too. So I created a new branch named branch1 which connected to the repo at the required commit version. command

2. Make the changes as suggested by geopavlakos.

3. You might run into the following error error (full stack trace at end of the issue). This error comes because linear_assignment_ has been deprecated in the recent versions of sklearn. To resolve this, go to the file where this error shows up. And then do the following: A. replace from sklearn.utils.linear_assignment_ import linear_assignment with from scipy.optimize import linear_sum_assignment as linear_assignment. B. Add the following line indices = np.array(list(zip(*indices))) after the indices = linear_assignment(cost_matrix) line. [Reference] (https://stackoverflow.com/questions/57369848/how-do-i-resolve-use-scipy-optimize-linear-sum-assignment-instead/57992848#57992848)

I was able to resolve the memory issue post this.