1,195 karma · joined April 19, 2010
Using generated training data is a good way to ensure that the training includes things that are too obvious to appear in normal writing. (Such as "there are zero giraffes in this photo.") This paper describes the limits of using transformer-generated data to train other transformers.
If you were able to answer this definitively, then you wouldn't need the 4 intermediate models.
https://www.wolframalpha.com/input?i=world+water+area+%2F+wo...
[1] https://en.wikipedia.org/wiki/Betteridge%27s_law_of_headline...
The windsorized mean reduces the weight of outliers, so it can measure "how many outliers are there?" instead of "how extreme are the outliers?"
cough mysterious 1:4:9 monolith cough
Would a modicum of caching in GitHub Actions libraries not make this problem go away for all hosts in this category?