vame.community_videos cannot generate community videos for cohort == True.
It only generates videos for communities that were calculated individually, per each session (cohort == False).
This is partially because of how the community labels are saved by vame.community, depending whether cohort is true or false.
The cohort-level labels of communities are being calculated, but saved in the file results/communit_cohort/.../cohort_community_label.npy, where the labels for all videos are concatenated together.
Should we adapt vame.community to save the cohort-level labels of communities for each individual session as well? E.g. under
The function vame.community_videos (in fact, the get_cluster_vid function) always searches in that sessions-level path to generate the community videos.
vame.community_videos
cannot generate community videos for cohort == True. It only generates videos for communities that were calculated individually, per each session (cohort == False).This is partially because of how the community labels are saved by
vame.community
, depending whether cohort is true or false.If cohort is True:
If cohort is False:
The cohort-level labels of communities are being calculated, but saved in the file
results/communit_cohort/.../cohort_community_label.npy
, where the labels for all videos are concatenated together. Should we adaptvame.community
to save the cohort-level labels of communities for each individual session as well? E.g. underThe function
vame.community_videos
(in fact, theget_cluster_vid
function) always searches in that sessions-level path to generate the community videos.@katiekly @DrSRMiller your thoughts?