mteb-sickr-approximately-10000-pairs

IN premise — summaries/2026/08/24/muennighoff-2022-mteb-s2-task-imbalance-tasks-in-mteb-have-a-differ.md

Created 2026-08-25T02:58:19+00:00

SICK-R contains approximately 10,000 sentence pairs focused on compositional semantics (lexical, syntactic, and semantic variation).

Summary

SICK-R is a moderately sized dataset of roughly ten thousand sentence pairs, each one engineered so that only a small word swap, structural shift, or semantic change distinguishes the two sentences. This makes it a targeted tool for checking whether a model actually understands how meaning shifts at the component level, rather than just matching surface patterns.