GitHub

SeDPO: Learning to Rank for In-Context Example Retrieval

A novel approach to in-context learning example selection/ordering through the extension of DPO. Experimental results demonstrate the top-1 performance across 4 NLP ...