rms Predict question

library(rms)
library(ggplot2)
ddiamonds <- datadist(diamonds)
options(datadist = "ddiamonds")
fit.ols <- ols(price ~ carat * cut, diamonds)

## can specify a range on a continuous predictor 
Predict(fit.ols, carat = seq(2,4,.25))

## can specify a single level for a categorical predictor
Predict(fit.ols, cut = "Good")

## can specify multiple levels for a categorical predictor
Predict(fit.ols, cut = c("Fair", "Good"))

## Is there a way to collapse across multiple levels of a categorical predictor?  
## Would like to see the effect of carat, averaging over all cuts