find list of higgest value in row and get colum name of them

Here is an option with apply to do this for every row in df.

apply(df, 1, function(x) names(head(sort(x, decreasing = TRUE), 3)))

#      [,1] 
#[1,] "taz"
#[2,] "gaz"
#[3,] "paz"

If you need both the value and column name you may return a list.

apply(df, 1, function(x) {
  x <- head(sort(x, decreasing = TRUE), 3)
  list(value = as.numeric(x), name = names(x))
})

#[[1]]
#[[1]]$value
#[1] 14 12  4

#[[1]]$name
#[1] "taz" "gaz" "paz"

If your data.frame is really only one row, you could use

library(dplyr)
library(tidyr)

df %>% 
  pivot_longer(everything()) %>% 
  slice_max(value, n = 3)

This returns

# A tibble: 3 x 2
  name  value
  <chr> <dbl>
1 taz      14
2 gaz      12
3 paz       4

Otherwise you either have to do some filtering or use a row number combined with some grouping.

We may use stack with order in base R

out <- stack(df)[2:1]
head(out[order(-out$values),], 3)
  ind values
5 taz     14
4 gaz     12
6 paz      4

We can try this

> df[order(-unlist(df))[1:3]]
  taz gaz paz
1  14  12   4