You're welcome, and thank you for your post also :). I posted an updated version of my experiment, which (hopefully) improves the logic of my prior experiment, at https://www.reddit.com/r/MachineLearning/comments/k2n3yv/d_an_experiment_that_shows_that_gpt3_can_plan/.