咨询关于 Rerank Overlap 相关的问题

netease-youdao / BCEmbedding

Netease Youdao's open-source embedding and reranker models for RAG products.

Apache License 2.0

1.3k stars 85 forks source link

Closed sherlcok314159 closed 1 month ago

sherlcok314159 commented 1 month ago

请问在学术论文领域，rerank overlap 设置多少比较合适呢？而且感觉是不是按照完整的句子单位，比如3-5个来替代比较硬性的 tokens 数量会更好呢？

shenlei1020 commented 1 month ago

可以根据自己的业务场景试，没统一标准