Screenshot of Annotation Diff tool has wrong AS "ML-results" instead
of ML.

Maybe both exercises should use the same output AS name.

Before the open-ended hands-on exercise starts, add a slide about the
routine for editing the config, reinitializing the PR, training, &
testing.

The worksheet has the wrong combinations of name, feature, position, &
range for the items in the config file.  Maybe eliminate the worksheet
& replace it with a slide of suggestions, since I always forget to
print the worksheets in advance anyway, & the students are happy to
scribble on bits of paper.

Changing from PAUM to SVM would make the threshold/probability values
more interesting.  (Gen says PAUM's probabilities are pants.  I've
observed a suspiciously high frequency of 1.0 values.)

I've been telling people the outputAS doesn't matter in evaluation
mode.  This may be wrong!  Investigate & correct the slides.