The gap between researchers and labelers should be as small as possible

Comments