Wrong normalizing data in libsvm

Question

I'm learning about libsvm.

Now, I have 25 training text files in RGB format, and want to normalize the data to the range of 0 to 1

I know the libsvm format is Lable 1:value 2:value ...

but no matter what I do, it always gives me a wong format.

My data

  R    G    B
1 1:39 2:8 3:100
1 1:34 2:3 3:67

I typed in cmd svm-scale -l 0 -u 1 train.txt > out.scale

The Output

1 1:1 2:1 3:1 
1 

What not ouput 1 1:0.xxx 2:0.xxx 3:0.xxx ...?

I've been searching information for several days, and then I have some questions.

1. How can I correctly normalize 0 to 1?

2. what exactly is the purpose of the label?

3. I have 25 training text . Should I copy all of them into one text file and then transform them?