LMEnt is an open suite of entity-annotated Wikipedia data, 12 pretrained LMs, and an entity-QID retrieval index that substantially outperforms string-based search and enables per-step knowledge attribution.
In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (V olume 1: Long Pa- pers), pages 9802–9822, Toronto, Canada
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
LMEnt: A Suite for Analyzing Knowledge in Language Models from Pretraining Data to Representations
LMEnt is an open suite of entity-annotated Wikipedia data, 12 pretrained LMs, and an entity-QID retrieval index that substantially outperforms string-based search and enables per-step knowledge attribution.