Do Language Models Know the Way to Rome?

Bastien Liétard, Mostafa Abdou, Anders Søgaard


Abstract
The global geometry of language models is important for a range of applications, but language model probes tend to evaluate rather local relations, for which ground truths are easily obtained. In this paper we exploit the fact that in geography, ground truths are available beyond local relations. In a series of experiments, we evaluate the extent to which language model representations of city and country names are isomorphic to real-world geography, e.g., if you tell a language model where Paris and Berlin are, does it know the way to Rome? We find that language models generally encode limited geographic information, but with larger models performing the best, suggesting that geographic knowledge can be induced from higher-order co-occurrence statistics.
Anthology ID:
2021.blackboxnlp-1.40
Volume:
Proceedings of the Fourth BlackboxNLP Workshop on Analyzing and Interpreting Neural Networks for NLP
Month:
November
Year:
2021
Address:
Punta Cana, Dominican Republic
Editors:
Jasmijn Bastings, Yonatan Belinkov, Emmanuel Dupoux, Mario Giulianelli, Dieuwke Hupkes, Yuval Pinter, Hassan Sajjad
Venue:
BlackboxNLP
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
510–517
Language:
URL:
https://aclanthology.org/2021.blackboxnlp-1.40
DOI:
10.18653/v1/2021.blackboxnlp-1.40
Bibkey:
Cite (ACL):
Bastien Liétard, Mostafa Abdou, and Anders Søgaard. 2021. Do Language Models Know the Way to Rome?. In Proceedings of the Fourth BlackboxNLP Workshop on Analyzing and Interpreting Neural Networks for NLP, pages 510–517, Punta Cana, Dominican Republic. Association for Computational Linguistics.
Cite (Informal):
Do Language Models Know the Way to Rome? (Liétard et al., BlackboxNLP 2021)
Copy Citation:
PDF:
https://aclanthology.org/2021.blackboxnlp-1.40.pdf