R

Reranker Msmarco ModernBERT Base Lambdaloss

Developed by tomaarsen
This is a cross-encoder model fine-tuned from ModernBERT-base, designed for calculating scores of text pairs, suitable for text re-ranking and semantic search tasks.
Downloads 89
Release Time : 3/17/2025

Model Overview

The model is based on the ModernBERT-base architecture, trained on the msmarco dataset using the sentence-transformers library, specifically for calculating similarity scores of text pairs. It can be applied to scenarios such as information retrieval and Q&A systems.

Model Features

Efficient Text Re-ranking
Capable of quickly calculating similarity scores for text pairs, effectively improving the ranking quality of retrieval systems.
Large Sequence Length Support
Supports sequences up to 8192 tokens, making it suitable for processing long texts.
High Performance Metrics
Performs excellently on multiple evaluation datasets, such as achieving an ndcg@10 of 0.7251 on NanoMSMARCO_R100.

Model Capabilities

Text similarity calculation
Information retrieval result re-ranking
Q&A system answer ranking
Semantic search

Use Cases

Information Retrieval
Search Engine Result Re-ranking
Re-rank search engine results to improve the ranking of relevant documents.
Achieves a map of 0.6768 on the MSMARCO dataset.
Q&A Systems
Answer Relevance Ranking
Score candidate answers for relevance and select the most relevant ones.
Achieves an mrr@10 of 0.7402 on the NanoNQ_R100 dataset.
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
© 2025AIbase