jxzhangjhu/Awesome-LLM-Uncertainty-Reliability-Robustness

The paper “Calibrate Before Use: ...” is not a model calibration related work

AmourWaltz opened this issue · 0 comments

I've read the paper "Calibrate Before Use: Improving Few-Shot Performance of Language Models" under ## Calibration Module. Actually, the "calibrate" in this paper is more likely to "correct" or "adjust", not related to the conventional concept of model calibration.