Underspecification in Language Modeling Tasks: A Causality-Informed Study of Gendered Pronoun Resolution

arXiv:2210.00131v4 Announce Type: replace
Abstract: Modern language modeling tasks are often underspecified: for a given token prediction, many words may satisfy the user's intent of producing natural language at inference time, however only one word will minimize the task's loss function at training time. We introduce a simple causal mechanism to describe the role underspecification plays in the generation of spurious correlations. Despite its simplicity, our causal model directly informs the development of two lightweight black-box evaluation methods, that we …

arxiv causality cs.ai cs.cl language modeling study tasks type