This is a small cross-encoder reranker model trained specifically for the Japanese language. It is designed to be used in retrieval pipelines to rerank candidate documents or passages based on relevance to a query. Built with the sentence-transformers library, it is lightweight and effective for Japanese information retrieval tasks.
Japanese Reranker Cross Encoder Small sits in PulseGate's Foundation models & chat category. It focuses on improving search relevance for Japanese-language queries by reranking retrieved documents using a specialized cross-encoder model. Japanese Reranker Cross Encoder Small is an open-source project aimed at developers building Japanese search or RAG applications. The project is open source (CC-BY-SA-4.0). It runs on the web and API.
Behind Japanese Reranker Cross Encoder Small is hotchpotch, and the product first shipped in 2022. The project is developed in the open on GitHub with 346 stars. Among its 3 catalogued features are Text Reranking, cross-Encoder, and Japanese Language.
Latest indexed changes and source events
hotchpotch/japanese-reranker-cross-encoder-small-v1 verified by the PulseGate indexer