Prototypical Calibration for Few-shot Learning of Language Models. (arXiv:2205.10183v1 [cs.CL])

In-context learning of GPT-like models has been recognized as fragile across
different hand-crafted templates, and demonstration permutations. In this work,
we propose prototypical calibration to adaptively learn a more robust decision
boundary for zero- and few-shot classification, instead of greedy decoding.
Concretely, our method first adopts Gaussian mixture distribution to estimate
the prototypical clusters for all categories. Then we assign each cluster to
the corresponding label by solving a weighted bipartite matching problem. Given
an example, its prediction is calibrated …

arxiv few-shot learning language language models learning