What do the the gls() weight options represent?

Question

I am fitting models to my data and would like to run gls() with weights because I have mildly non-normal residuals and slight heteroscedasticity.

gls1 <- g