By default we adopt Cartesian product logic when combining parameters. This behavior can be adjusted either grouping parameters together:

n_df: (100,2), (1000, 4)

Or apply @FILTER:

n: 100, 1000
df: 2, 4
@FILTER: (n = 100 and df = 2) or (n = 1000 and df = 4)

A proposed syntax is:

(n, df): (100, 2), (1000, 4)

I will implement this behavior if deemed reasonable.