• Publications
  • Influence
Deep contextualized word representations
TLDR
We introduce a new type of deep contextualized word representation that models both (1) complex characteristics of word use (e.g., syntax and semantics), and (2) how these uses vary across linguistic contexts (i.e., to model polysemy). Expand
  • 4,590
  • 943
  • PDF
Ask Me Anything: Dynamic Memory Networks for Natural Language Processing
TLDR
We introduce the dynamic memory network (DMN), a neural network architecture which processes input sequences and questions, forms episodic memories, and generates relevant answers. Expand
  • 823
  • 83
  • PDF
Deep Unordered Composition Rivals Syntactic Methods for Text Classification
TLDR
We present a simple deep neural network that competes with and, in some cases, outperforms such models on sentiment analysis and factoid question answering tasks while taking only a fraction of the training time. Expand
  • 517
  • 78
  • PDF
QuAC : Question Answering in Context
TLDR
We present QuAC, a dataset for Question Answering in Context that contains 14K information-seeking QA dialogs (100K questions in total). Expand
  • 210
  • 56
  • PDF
Adversarial Example Generation with Syntactically Controlled Paraphrase Networks
TLDR
We propose syntactically controlled paraphrase networks (SCPNs) and use them to generate adversarial examples that improve the robustness of pretrained models to syntactic variation when used to augment their training data. Expand
  • 213
  • 28
  • PDF
A Neural Network for Factoid Question Answering over Paragraphs
TLDR
We introduce a recursive neural network (rnn) model that can reason over such input by modeling textual compositionality. Expand
  • 272
  • 16
  • PDF
Search-based Neural Structured Learning for Sequential Question Answering
TLDR
We propose a novel dynamic neural semantic parsing framework trained using a weakly supervised reward-guided search that is designed for solving sequential question answering. Expand
  • 91
  • 14
  • PDF
Unsupervised Latent Tree Induction with Deep Inside-Outside Recursive Autoencoders
TLDR
We introduce the deep inside-outside recursive autoencoder (DIORA), a fully-unsupervised method for discovering syntax that simultaneously learns representations for constituents within the induced tree. Expand
  • 34
  • 12
  • PDF
Pathologies of Neural Models Make Interpretation Difficult
TLDR
We use input reduction, a process that iteratively removes the least important word from the input while maintaining the model’s prediction. Expand
  • 97
  • 11
  • PDF
The Amazing Mysteries of the Gutter: Drawing Inferences Between Panels in Comic Book Narratives
TLDR
We present the COMICS dataset, which contains over 1.2 million panels drawn from almost 4,000 publicly-available comic books published during the “Golden Age” of American comics. Expand
  • 35
  • 8
  • PDF