R

Reranker ModernBERT Base Gooaq Bce

Developed by tomaarsen
This is a cross-encoder model fine-tuned from ModernBERT-base for text re-ranking and semantic search tasks.
Downloads 483
Release Time : 3/20/2025

Model Overview

This model is a cross-encoder based on ModernBERT-base, specifically designed for computing similarity scores between text pairs, suitable for re-ranking tasks in information retrieval.

Model Features

Long Text Processing Capability
Supports sequences up to 8192 tokens, suitable for processing long texts
Efficient Re-ranking
Optimized for text re-ranking tasks, performs well on the GooAQ dataset
Binary Classification
Trained using binary cross-entropy loss, outputs relevance scores for text pairs

Model Capabilities

Text Similarity Calculation
Information Retrieval Result Re-ranking
Semantic Search

Use Cases

Information Retrieval
Search Engine Result Re-ranking
Re-rank initial retrieval results to improve relevance
Achieves NDCG@10 of 0.7713 on GooAQ development set
Question Answering System
Rank candidate answers by relevance
Achieves NDCG@10 of 0.4630 on NanoNQ dataset
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
Š 2025AIbase