Comment by mostdataisnice

SQLite is literally a backend for MLflow, so the argument being made really is that you should just use SQL when you can, which is kind of adjacent to any criticisms of MLflow