BTZSC: A Benchmark for Zero-Shot Text Classification Across Cross-Encoders, Embedding Models, Rerankers and LLMs

Ilias Aarab · 2026

Read BTZSC: A Benchmark for Zero-Shot Text Classification Across Cross-Encoders, Embedding Models, Rerankers and LLMs by Ilias Aarab (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-11991.txt.

Description

Aarab builds BTZSC, a benchmark for zero-shot text classification that needs no task-specific labels. It covers 22 public datasets across sentiment, topic, intent, and emotion tasks. Four model families face off in the tests: embedding models, rerankers, instruction-tuned LLMs, and NLI cross-encoders. Modern rerankers take the top scores, while strong embedding models close much of the gap with the best speed trade-off. NLI cross-encoders level off as they grow, and the full test code is public.

Catalog record

CREATORIlias Aarab
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE116483
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.11991

Browse the full library · About the Tower of Babel