
The author proposes a complexity-based prompt selection approach for sequence tagging tasks to improve few-shot learning capabilities of pretrained language models. By aligning the syntactico-semantic complexity of examples with test sentences, significant performance gains are achieved.


coremsg

enhancing-few-shot-example-selection-with-complexity-based-metrics


Enhancing Few-Shot Example Selection with Complexity-Based Metrics


title_rewrite


Pretrained language models benefit from complexity-based prompt selection for improved few-shot performance.


few-shot-example-selection-metrics-design


Few-Shot Example Selection Metrics Design



Pretrained language models benefit from complexity-based prompt selection for improved few-shot learning performance.
