Dynamic Coattention Networks For Question Answering

Code

• github.com/BAJUKA/SQuAD-NLP
  tf★ 0
• github.com/Lou1sM/AdvancedML-Project-Dynamic-Coattention-Networks
  tf★ 0
• github.com/wasimusu/MachineRC
  pytorch★ 0
• github.com/andreiilie1/dynamic_coattention_networks
  tf★ 0
• github.com/lmn-extracts/dcn_plus
  tf★ 0
• github.com/Lou1sM/AML-Project
  tf★ 0

Abstract

Several deep learning models have been proposed for question answering. However, due to their single-pass nature, they have no way to recover from local maxima corresponding to incorrect answers. To address this problem, we introduce the Dynamic Coattention Network (DCN) for question answering. The DCN first fuses co-dependent representations of the question and the document in order to focus on relevant parts of both. Then a dynamic pointing decoder iterates over potential answer spans. This iterative procedure enables the model to recover from initial local maxima corresponding to incorrect answers. On the Stanford question answering dataset, a single DCN model improves the previous state of the art from 71.0% F1 to 75.9%, while a DCN ensemble obtains 80.4% F1.

Tasks

Benchmark Results

Reproductions