papers AI Learner
The Github is limit! Click to go to the new site.

Grammatical Analysis of Pretrained Sentence Encoders with Acceptability Judgments

2019-01-11
Alex Warstadt, Samuel R. Bowman

Abstract

Recent pretrained sentence encoders achieve state of the art results on language understanding tasks, but does this mean they have implicit knowledge of syntactic structures? We introduce a grammatically annotated development set for the Corpus of Linguistic Acceptability (CoLA; Warstadt et al., 2018), which we use to investigate the grammatical knowledge of three pretrained encoders, including the popular OpenAI Transformer (Radford et al., 2018) and BERT (Devlin et al., 2018). We fine-tune these encoders to do acceptability classification over CoLA and compare the models’ performance on the annotated analysis set. Some phenomena, e.g. modification by adjuncts, are easy to learn for all models, while others, e.g. long-distance movement, are learned effectively only by models with strong overall performance, and others still, e.g. morphological agreement, are hardly learned by any model.

Abstract (translated by Google)
URL

http://arxiv.org/abs/1901.03438

PDF

http://arxiv.org/pdf/1901.03438


Similar Posts

Comments