ComingUp ComingUp
KRLabsOrg/verbatim-rag-modern-bert-v2

KRLabsOrg/verbatim-rag-modern-bert-v2

Jun 10, 2026 AI & Machine Learning
bert language model nlp passage-retrieval rag

Gallery

KRLabsOrg/verbatim-rag-modern-bert-v2

About

KRLabsOrg/verbatim-rag-modern-bert-v2 is a ModernBERT-based model designed for Retrieval-Augmented Generation (RAG) systems. It specializes in verbatim passage retrieval, extracting exact text segments to provide precise context for language models. This second iteration offers improved accuracy and efficiency for information extraction tasks.

Comments (5)

Ryan Hayes Ryan Hayes 2 months ago

does verbatim extraction miss stuff that chunking with overlap would catch?

Danyka DuBuque Danyka DuBuque 2 months ago

benchmarks vs bge-m3 for exact passage retrieval?

Jazmin O'Keefe Jazmin O'Keefe 2 months ago

Verbatim extraction breaks down on noisy production data.

Vince Farrell Vince Farrell 2 months ago

what chunk sizes were benchmarked for the verbatim extraction?

Madisyn Watsica Madisyn Watsica 2 months ago

verbatim retrieval market seems pretty small vs semantic search competitors