Probing What Different NLP Tasks Teach Machines about Function Word Comprehension
arXiv:1904.11544
Abstract
We introduce a set of nine challenge tasks that test for the understanding of function words. These tasks are created by structurally mutating sentences from existing datasets to target the comprehension of specific types of function words (e.g., prepositions, wh-words). Using these probing tasks, we explore the effects of various pretraining objectives for sentence encoders (e.g., language modeling, CCG supertagging and natural language inference (NLI)) on the learned representations. Our results show that pretraining on language modeling performs the best on average across our probing tasks, supporting its widespread use for pretraining state-of-the-art NLP models, and CCG supertagging and NLI pretraining perform comparably. Overall, no pretraining objective dominates across the board, and our function word probing tasks highlight several intuitive differences between pretraining objectives, e.g., that NLI helps the comprehension of negation.
arXiv commentAccepted to *SEM 2019 (revised submission). Corresponding authors: Najoung Kim ([email protected]), Ellie Pavlick ([email protected])
References in corpus (7)
- On the Convergence of Adam and Beyond
- Bidirectional Attention Flow for Machine Comprehension
- Skip-Thought Vectors
- GLUE: A Multi-Task Benchmark and Analysis Platform for Natural Language Understanding
- The Natural Language Decathlon: Multitask Learning as Question Answering
- Discourse-Based Objectives for Fast Unsupervised Sentence Representation Learning
- DisSent: Sentence Representation Learning from Explicit Discourse Relations