Here's another awk that mantains the order.

Example

$ awk '/[0-9]+/{a+=$1;next}a{print a;a=0}{printf($1FS);a=0}END{print a}' file
abc 123
dev 456
hij 1232

Explanation

/[0-9]+/{a+=$1;next}:When a number is detected as the content of the record, it's value is accumulated into a var, then next is used to stop further processing and pass the flow to the next record.

a{print a;a=0}: ONLY when the counter is not null we print a value that correspond to the previous word and initialize it.

{printf($1FS);a=0}: Print current record and the separator avoiding the carriage return. This is applied to all text records.

END{print a}: Show the final counter that will remain after the last record.

It's possible to do this by using awk and arrays:

$ awk '{if($1!~/^[0-9]+$/){cur=$1}else{sums[cur]+=$1}}END{for(el in sums){printf("%s %d\n",el,sums[el])}}' Filevalue.txt 
hij 1232
dev 456
abc 123

Here is the same code but written to the file sum.awk and with comments:

# this block will be executed for every line in the file
{
    # if current processing line is not a number, then it's a name of the next
    # group, let's save it to cur variable
    if ($1 !~ /^[0-9]+$/) {
        cur = $1
    } else {
        # here we're summing values
        sums[cur] += $1
    }
}
# this block will be executed at the end of file processing, here we're
# printing array with sums
END {
    for (el in sums) {
        printf("%s %d\n", el, sums[el])
    }
}

Use it like this:

$ awk -f sum.awk Filevalue.txt 
hij 1232
dev 456
abc 123

The only downside of using awk here is that it doesn't preserve keys order.