scaling features in structured prediction

[Darryl]

Hello,

Thanks for creating this great library. Its been really easy to use on a structured prediction problem that I am dealing with. I had a question regarding the use of feature scaling prior to fitting a pystruct model. 

I know that SVM's can benefit from feature scaling. Is this also the case for structured SVMs? I am tagging tokens in a string of text where the token is converted into an array of numerical features of uniform length. I could potentially use the feature values for every token in the dataset to scale each feature to a [0,1] or [-1,1] scale. Would this be beneficial? I did a quick search online to see if anyone had asked this question previously but did not any resources addressing this issue.

Thanks,

Darryl