Ruri Reranker Large
Ruri Reranker is a general-purpose Japanese reranking model based on the Sentence Transformers architecture, specifically designed for Japanese text relevance ranking tasks.
Downloads 2,538
Release Time : 8/20/2024
Model Overview
This model is a Japanese text reranker capable of scoring the relevance of query-document pairs, suitable for reranking results in information retrieval systems.
Model Features
High-Performance Japanese Reranking
Outperforms in multiple Japanese benchmarks, achieving leading scores particularly on the JQaRA, JaCWIR, and MIRACL datasets.
Based on Sentence Transformers
Utilizes the CrossEncoder architecture, optimized for query-document pair relevance scoring.
Large Model Scale
Boasts 337 million parameters (with 303 million in the embedding layer), providing stronger semantic understanding capabilities.
Model Capabilities
Query-document relevance scoring
Information retrieval result reranking
Japanese text understanding
Use Cases
Information Retrieval
Search Engine Result Reranking
Reranks search engine results to improve the ranking of the most relevant outcomes.
Achieves an nDCG@10 score of 77.1 on the JQaRA dataset.
QA System Answer Ranking
Reranks candidate answers returned by a QA system based on relevance.
Achieves an nDCG@10 score of 96.1 on the MIRACL dataset.
Featured Recommended AI Models