Best approach to take a random sample of geo spatial data in R

I am trying to take a relatively small sample, maybe 5-10 samples by city of a country. As you can see I am working with Pakistan, and I have a very large data set, maybe 50,000 samples of geo spatial data. I want 5-10 random samples from each city. The sample data set below is only 2 samples. Any help would be greatly appreciated.

sampledataset <- matrix(c("ChIJ-11JNWkFGTkRY-jDNnTGwnQ","HEALTH LINK PHARMACY","3a Sultan Ahmed Road", "c(1005348.08594025, 3500185.38218226)", "ChIJ-aPvOavJIDkRMgJGa4ub284", "Baloch Pharmacy", "Thowa Dhurnal Road, Dhurnal", "c(790797.725041541, 3634310.02159564)"), ncol = 4, byrow = TRUE)

colnames(sampledataset) <- (c("place_id","name", "vicinity", "geometry"))