jina-reranker-v3：文檔重排序中的最後但非最晚交互

摘要

jina-reranker-v3 是一款擁有 0.6B 參數的多語言文件重排序模型，它引入了一種新穎的「最後但非遲延交互」機制。與 ColBERT 等遲延交互模型不同，後者先進行獨立編碼再進行多向量匹配，我們的方法則是在同一上下文窗口內對查詢和文件進行因果自注意力運算，從而能在從每個文件的最後一個標記提取上下文嵌入之前，實現豐富的跨文件交互。這種緊湊的架構在 BEIR 基準測試中達到了 61.94 nDCG@10 的頂尖性能，同時體積僅為生成式列表重排序模型的十分之一。

English

jina-reranker-v3 is a 0.6B parameter multilingual document reranker that introduces a novel last but not late interaction. Unlike late interaction models such as ColBERT that perform separate encoding followed by multi-vector matching, our approach conducts causal self-attention between query and documents within the same context window, enabling rich cross-document interactions before extracting contextual embeddings from the last token of each document. This compact architecture achieves state-of-the-art BEIR performance with 61.94 nDCG@10 while being ten times smaller than generative listwise rerankers.

jina-reranker-v3：文檔重排序中的最後但非最晚交互

jina-reranker-v3: Last but Not Late Interaction for Document Reranking

摘要

Support