My expression data looks like where g1 is gene, s1 and s2 are samples

  Symbols       s1    s2  
    g1          2.3   4.5  
    g1          4.5   2.3

I have used limma::avereps to get unique gene symbols, however, i got unique gene symbols but the expression values was not averaged. The result is like

Symbols        s1   s2
g1             2.3  4.5

Please do suggest me!!

Since your input x contains numbers and symbols, it is not a (numeric) matrix, and therefore not compatible with avereps(). The behavior you observed is documented on the avereps help pages.

> # Generate some data
> x <-  matrix(51:82, 8)
> colnames(x) <- c("S1","S2","S3","S4")
> #check
> is.numeric(x)
[1] TRUE
> 
> # define gene symbols, and combine with expression data
> Symbols <-  c("b","a","a","c","c","b","b","b")
> x2 <- cbind(x, Symbols)
> #check
> is.numeric(x2)
[1] FALSE
> 
> # Still try:
> avereps(x=x2, ID=x2[,"Symbols"])
     S1   S2   S3   S4   Symbols
[1,] "51" "59" "67" "75" "b"    
[2,] "52" "60" "68" "76" "a"    
[3,] "54" "62" "70" "78" "c"    
> # note that only first row/instance of genes (letters) are kept.
> # No averages are calculated!
> 
> # This works (using separate inputs for avereps!):
> avereps(x=x, ID=Symbols)
    S1   S2   S3   S4
b 55.5 63.5 71.5 79.5
a 52.5 60.5 68.5 76.5
c 54.5 62.5 70.5 78.5
>