Dr-Fairness: Dynamic Data Ratio Adjustment for Fair Training on Real and Generated Data - View it on GitHub