Conceptlabel count distribution / 标签频次表
Label Count Distribution
A diagnostic method for detecting scorer bias by counting the frequency of each class label in the scorer's outputs. It requires no manual annotation or additional model calls and can be embedded in an evaluation pipeline as a pre-check.
Timeline (last 90 days)
Oct 20
Looking at the label count per category can serve as a zero-cost diagnostic for detecting scorer class collapse: if the compliant category count is abnormally low or even zero while adjacent categories are abnormally inflated, it indicates the scorer may be collapsing.
Unverified50%