Thanks Rupert, both for being persistent and for being patient. And thanks for the meme, made me giggle.
I can’t make any promises… BUT - could you help me identify a clade that I could use for a comparison experiment? Something constrained enough that I could do an A vs B experiment training up models with and without hybrid taxa, and see how they compare? Ideally it would have less than 1,000 taxa.
Then I could design & schedule that experiment and if everything checks out, I could present it to the team and see if there’s consensus on making this change going forward for the full trains.
Apologies if you’ve already done this work and are just catching me up - I have had covid for the past two weeks and am coming out of a mental cloud, and I’ve also been off working on other stuff (3.0 feature extractor, some changes to how we do geo modeling, etc) so I’m not really up to date on the thread. Thanks for your patience.