Skip to content

Introducing the Ettin Reranker Family

67
Academic
Chimera Difficulty Score
a synthesis of Flesch-Kincaid, Coleman-Liau, SMOG, and Dale-Chall readability metrics
Introducing the Ettin Reranker Family TL;DR Today I'm releasing six new Sentence Transformers CrossEncoder rerankers, state-of-the-art at their respective sizes, built on top of the Ettin ModernBERT encoders, together with the data and full training recipe that produced them: cross-encoder/ettin-reranker-17m-v1 cross-encoder/ettin-reranker-32m-v1 cross-encoder/ettin-reranker-68m-v1 cross-encoder/e...
The release of the Ettin Reranker Family represents a significant advancement in the field of search systems, particularly in the retrieve-then-rerank pipeline. The models are designed to improve the accuracy of search results by reordering the top-K candidates retrieved by an embedding model. The use of a distillation recipe from the mixedbread-ai/mxbai-rerank-large-v2 model ensures that the smaller models inherit the calibration and judgment of the larger teacher model, resulting in high accur...