Hey, all! I’ve got a conceptual question I’m struggling with.
I’m creating a regression model with tabular data, and noted that my “y” values in the source data range from .0667 to ~4200. When I call a random batch after creating a data loader, however, I noticed that many of the “Target” values are zero. Does the data object somehow scale the dependent variable when creating the “Target” column?
There’s indeed normalisation in TabularDataset, see doc :
Note that the replacement from categories to codes as well as the normalization of continuous variables are automatically done in a
TabularDataset.
THANK GOD. Let’s set aside my embarrassment in how I missed that line in the docs I’ve been re-reading and just say THANKS for the assist, @PierreO!
No worries, I nearly missed it myself 
Next step would have been to take a look at the source code to see what’s going on, which I would encourage you to do if you don’t find answers in the docs on the forum.