Do you support Cost Sensitive One vs. All logistic regression?

[dhruvghulati-zz]

I am looking to have a wrapper that essentially places my existing numpy 2D arrays and unicode vector labels into the VW format and allows me to perform fast cost sensitive categorical predictions:

https://github.com/JohnLangford/vowpal_wabbit/wiki/Cost-Sensitive-One-Against-All-%28csoaa%29-multi-class-example