Holmes: A Benchmark to Assess the Linguistic Competence of Language Models

TL;DR

Holmes is a comprehensive benchmark that evaluates language models' unconscious linguistic understanding across various phenomena using probing techniques.

Contribution

It introduces Holmes, a benchmark with extensive datasets and analysis methods to disentangle linguistic competence from other cognitive abilities.

Findings

Model size correlates with linguistic competence.

Architecture and instruction tuning significantly affect performance.

FlashHolmes reduces computation while maintaining accuracy.

Abstract

We introduce Holmes, a new benchmark designed to assess language models (LMs) linguistic competence - their unconscious understanding of linguistic phenomena. Specifically, we use classifier-based probing to examine LMs' internal representations regarding distinct linguistic phenomena (e.g., part-of-speech tagging). As a result, we meet recent calls to disentangle LMs' linguistic competence from other cognitive abilities, such as following instructions in prompting-based evaluations. Composing Holmes, we review over 270 probing studies and include more than 200 datasets to assess syntax, morphology, semantics, reasoning, and discourse phenomena. Analyzing over 50 LMs reveals that, aligned with known trends, their linguistic competence correlates with model size. However, surprisingly, model architecture and instruction tuning also significantly influence performance, particularly in morphology and syntax. Finally, we propose FlashHolmes, a streamlined version that reduces the computation load while maintaining high-ranking precision.