Symbols and mental programs: a hypothesis about human singularity

S Dehaene, F Al Roumi, Y Lakretz, S Planton… - Trends in Cognitive …, 2022 - cell.com
Natural language is often seen as the single factor that explains the cognitive singularity of
the human species. Instead, we propose that humans possess multiple internal languages …

Beyond the imitation game: Quantifying and extrapolating the capabilities of language models

A Srivastava, A Rastogi, A Rao, AAM Shoeb… - arxiv preprint arxiv …, 2022 - arxiv.org
Language models demonstrate both quantitative improvement and new qualitative
capabilities with increasing scale. Despite their potentially transformative impact, these new …

Does localization inform editing? surprising differences in causality-based localization vs. knowledge editing in language models

P Hase, M Bansal, B Kim… - Advances in Neural …, 2024 - proceedings.neurips.cc
Abstract Language models learn a great quantity of factual information during pretraining,
and recent work localizes this information to specific model weights like mid-layer MLP …

Masked language modeling and the distributional hypothesis: Order word matters pre-training for little

K Sinha, R Jia, D Hupkes, J Pineau, A Williams… - arxiv preprint arxiv …, 2021 - arxiv.org
A possible explanation for the impressive performance of masked language model (MLM)
pre-training is that such models have learned to represent the syntactic structures prevalent …

A taxonomy and review of generalization research in NLP

D Hupkes, M Giulianelli, V Dankers, M Artetxe… - Nature Machine …, 2023 - nature.com
The ability to generalize well is one of the primary desiderata for models of natural language
processing (NLP), but what 'good generalization'entails and how it should be evaluated is …

Is Ockham's razor losing its edge? New perspectives on the principle of model parsimony

M Dubova, S Chandramouli, G Gigerenzer… - Proceedings of the …, 2025 - pnas.org
The preference for simple explanations, known as the parsimony principle, has long guided
the development of scientific theories, hypotheses, and models. Yet recent years have seen …

State-of-the-art generalisation research in NLP: a taxonomy and review

D Hupkes, M Giulianelli, V Dankers, M Artetxe… - arxiv preprint arxiv …, 2022 - arxiv.org
The ability to generalise well is one of the primary desiderata of natural language
processing (NLP). Yet, what'good generalisation'entails and how it should be evaluated is …

Causal analysis of syntactic agreement mechanisms in neural language models

M Finlayson, A Mueller, S Gehrmann, S Shieber… - arxiv preprint arxiv …, 2021 - arxiv.org
Targeted syntactic evaluations have demonstrated the ability of language models to perform
subject-verb agreement given difficult contexts. To elucidate the mechanisms by which the …

On the proper role of linguistically oriented deep net analysis in linguistic theorising

M Baroni - Algebraic structures in natural language, 2022 - taylorfrancis.com
A lively research field has recently emerged that uses experimental methods to probe the
linguistic behaviour of modern deep networks. While work in this tradition often reports …

Language models as models of language

R Millière - arxiv preprint arxiv:2408.07144, 2024 - arxiv.org
This chapter critically examines the potential contributions of modern language models to
theoretical linguistics. Despite their focus on engineering goals, these models' ability to …