machine learning2 min read
PBT doesn’t make grokking faster, but it can escape a hyperparameter range where nothing groks [P]
I ran Population Based Training over learning rate and weight decay on grokking (modular addition mod 97, 35% train split, small MLP, 20 members, 20k steps, full batch).
By /u/Virtual-Message-9739Reddit - r/ML
Read the full story
Originally published by Reddit - r/ML on Oct 26, 2026.
Comments
Sign in to join the discussion.