Whenever you're ready, walk me through how you'd approach this.
You're training supervised learning models and notice that training performance keeps improving while validation performance stalls or drops.
How do you approach the problem of overfitting in machine learning?