AI Science Tracker
EN

Encoding and Propagation of Grammatical Number in Large Language Models

University of Minnesota · 5 Nov 2026 · computer science

Discovery

Encoding and Propagation of Grammatical Number in Large Language Models

Source check: Source checked. Not independent scientific confirmation.

Finding / tested result: Sparse autoencoders trained on Pythia-70M identified features associated with plural subjects and verb generation; intervening on those features changed number agreement in model outputs. Similar patterns were reported for Gemma-2-2B and Llama-3.1-8B, for the tested English constructions.

Technology: Sparse autoencoders and activation interventions in language models

Publication: Preprint · Test context: direct test. Catalog categories, not an evidence grade.

AI contribution
Sparse autoencoders identify language-model features associated with grammatical number; activation interventions test the features’ effects on verb agreement.
Limitation / still unknown
The study evaluates the reported models and datasets only; broader generalization and real-world deployment outcomes were not established.
Evidence sources and annotation context

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

Evidence sources: https://arxiv.org/abs/2611.07283v1 · https://arxiv.org/html/2611.07283v1

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 22 Nov 2026. Review scope: Primary result-bearing material for the stated method, result and evaluation, with bibliographic version information..

No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.

Editorial assessment: Impact 2/5 · Validation 3/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Open in the full catalog for filters, charts and related discoveries. Browse computer science discoveries.