How to validate every subset

[mpg-data]

I have 2 columns a and b, each one with several values. Turns out that for a certain value of a, all values in b must be the same.

I attempted to do this

pa.Check(check_fn = lambda x: x.nunique() == 1, groupby = "a")

But I get an error and would need to provide the specific values of a (there are thousands of it)

Is there any way to do this?