English

ToMMeR -- Efficient Entity Mention Detection from Large Language Models

Computation and Language 2026-04-21 v2 Artificial Intelligence

Abstract

Identifying which text spans refer to entities - mention detection - is both foundational for information extraction and a known performance bottleneck. We introduce ToMMeR, a lightweight model (<300K parameters) probing mention detection capabilities from early LLM layers. Across 13 NER benchmarks, ToMMeR achieves 93% recall zero-shot, with an estimated 90% precision under a human-calibrated LLM-judge protocol, showing that ToMMeR rarely produces spurious predictions despite high recall. Cross-model analysis reveals that diverse architectures (14M-15B parameters) converge on similar mention boundaries (DICE >75%), confirming that mention detection emerges naturally from language modeling. When extended with span classification heads, ToMMeR achieves competitive NER performance (80-87% F1 on standard benchmarks). Our work provides evidence that structured entity representations exist in early transformer layers and can be efficiently recovered with minimal parameters.

Keywords

Cite

@article{arxiv.2510.19410,
  title  = {ToMMeR -- Efficient Entity Mention Detection from Large Language Models},
  author = {Victor Morand and Nadi Tomeh and Josiane Mothe and Benjamin Piwowarski},
  journal= {arXiv preprint arXiv:2510.19410},
  year   = {2026}
}

Comments

Accepted at ACL2026 - Code: https://github.com/VictorMorand/llm2ner

R2 v1 2026-07-01T06:59:25.496Z