Beyond Recall Saturation: Scaling Hard-Negative Reranker Training for Industrial Legal Retrieval
It is shown that, under a sufficiently large but practical candidate pool, embedding recall exhibits diminishing returns, motivating a closer examination of reranker quality, and support evaluating the retrieval and reranking stages separately when selecting an optimization target for industrial retrieval systems.
Yun-Han Li, Ming-Jie Xie, Ze-Yang Shi et al.
· 0 citations