R

Reranker Ms Marco MiniLM L6 V2 Gooaq Bce

Developed by ayushexel
This is a cross-encoder model fine-tuned from cross-encoder/ms-marco-MiniLM-L6-v2, developed using the sentence-transformers library. It can compute scores for text pairs and is suitable for text re-ranking and semantic search.
Downloads 15
Release Time : 3/30/2025

Model Overview

This model is a MiniLM-based cross-encoder specifically designed for text re-ranking tasks. It improves search result quality by calculating relevance scores between queries and documents.

Model Features

Efficient Re-ranking
Specially optimized for re-ranking tasks, significantly improving search result relevance.
Multi-dataset Validation
Validated on multiple datasets including GooAQ, MSMARCO, NFCorpus, and NQ with stable performance.
Long Text Processing
Supports sequences up to 512 tokens, suitable for longer queries and documents.

Model Capabilities

Text Relevance Scoring
Search Result Re-ranking
Semantic Search Optimization

Use Cases

Information Retrieval
Search Engine Result Optimization
Re-rank initial search results to improve the ranking of relevant documents.
Achieved NDCG@10 score of 0.6822 on GooAQ development set.
Q&A Systems
Rank candidate answers by relevance to select the best match.
Achieved NDCG@10 score of 0.5091 on NanoNQ dataset.
Healthcare
Medical Q&A Matching
Match user medical questions with professional medical answers.
As shown in examples, can accurately identify medical explanations related to left arm pain.
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
© 2025AIbase