Linear probes on LM hidden states detect grammaticality better than string probabilities, generalize to human benchmarks and other languages, and correlate weakly with likelihood.
InProceedings of the 2022 Conference on Empirical Methods in Natu- ral Language Processing, pages 4606–4634, Abu Dhabi, United Arab Emirates
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Implicit Representations of Grammaticality in Language Models
Linear probes on LM hidden states detect grammaticality better than string probabilities, generalize to human benchmarks and other languages, and correlate weakly with likelihood.