Label your training data as a fragmented converter model. You then use the tagged data to fine-tune a small monitor model.

Continue reading “Towards Data Science”