The RepEval 2017 Shared Task: Multi-Genre Natural Language Inference with Sentence Representations

TL;DR

This paper reports on the RepEval 2017 Shared Task, evaluating neural network models for sentence representation on the MultiNLI corpus, showing that advanced models outperform baselines and learn domain-independent sentence meanings.

Contribution

It introduces a shared task for evaluating neural sentence representations on MultiNLI and demonstrates that models with residual BiLSTMs achieve state-of-the-art results.

Findings

Best model achieved 74.5% accuracy on genre-matched test set.

Models learned domain-independent sentence representations.

All submitted systems outperformed baseline models.

Abstract

This paper presents the results of the RepEval 2017 Shared Task, which evaluated neural network sentence representation learning models on the Multi-Genre Natural Language Inference corpus (MultiNLI) recently introduced by Williams et al. (2017). All of the five participating teams beat the bidirectional LSTM (BiLSTM) and continuous bag of words baselines reported in Williams et al.. The best single model used stacked BiLSTMs with residual connections to extract sentence features and reached 74.5% accuracy on the genre-matched test set. Surprisingly, the results of the competition were fairly consistent across the genre-matched and genre-mismatched test sets, and across subsets of the test data representing a variety of linguistic phenomena, suggesting that all of the submitted systems learned reasonably domain-independent representations for sentence meaning.