Improving Few-Shot Performance of Language Models via Nearest Neighbor Calibration

Open in new window