Poster
What do you learn from context? Probing for sentence structure in contextualized word representations
Ian Tenney · Patrick Xia · Berlin Chen · Alex Wang · Adam Poliak · Tom McCoy · Najoung Kim · Benjamin Van Durme · Sam Bowman · Dipanjan Das · Ellie Pavlick
Great Hall BC #52
Keywords: [ word embeddings ] [ natural language processing ] [ interpretability ] [ transfer learning ]
Contextualized representation models such as ELMo (Peters et al., 2018a) and BERT (Devlin et al., 2018) have recently achieved state-of-the-art results on a diverse array of downstream NLP tasks. Building on recent token-level probing work, we introduce a novel edge probing task design and construct a broad suite of sub-sentence tasks derived from the traditional structured NLP pipeline. We probe word-level contextual representations from four recent models and investigate how they encode sentence structure across a range of syntactic, semantic, local, and long-range phenomena. We find that existing models trained on language modeling and translation produce strong representations for syntactic phenomena, but only offer comparably small improvements on semantic tasks over a non-contextual baseline.
Live content is unavailable. Log in and register to view live content