[KongchangAI]
Concepthuman alignment

Human Alignment

An evaluation attribute that measures the correlation or consistency between model judgments and human annotation preferences. It is commonly quantified using metrics such as accuracy and Kendall rank correlation coefficient.

Timeline (last 90 days)

Nov 5

An evaluator has three independent properties: repeatability (whether results are stable under the same conditions), presentation robustness (whether results are unaffected by irrelevant presentation changes), and human alignment (whether evaluation results are consistent with human preferences), and there is no entailment relationship among the three.

Unverified50%

All Facts (1)

Source Articles