Achieving high performance during training only to see poor results during testing is a common challenge in machine learning.