BTW thank you for drawing my attention to the parallelization penalty crux; I hadn’t realized how sensitive the model was to that parameter before! Very important to think more about this. ETA: Curious if you have thoughts on my confused question here.
OK, seems we mostly agree then.
BTW thank you for drawing my attention to the parallelization penalty crux; I hadn’t realized how sensitive the model was to that parameter before! Very important to think more about this. ETA: Curious if you have thoughts on my confused question here.