• Publications
  • Influence
Learning from Explanations with Neural Execution Tree
TLDR
We propose a novel Neural Execution Tree (NExT) framework to augment training data for text classification using NL explanations. Expand
Looking Beyond Label Noise: Shifted Label Distribution Matters in Distantly Supervised Relation Extraction
TLDR
We study the problem what limits the performance of DS-trained neural models, conduct thorough analyses, and identify a factor that can influence the performance greatly, shifted label distribution. Expand
Teaching Machine Comprehension with Compositional Explanations
TLDR
We use learnable neural modules and soft logic to handle linguistic variation and overcome sparse coverage; the modules are jointly optimized with the MRC model to improve final performance. Expand
Refining Neural Networks with Compositional Explanations
TLDR
Neural networks are prone to learning spurious correlations from biased datasets, and are thus vulnerable when making inferences in a new target domain. Expand
LEAN-LIFE: A Label-Efficient Annotation Framework Towards Learning from Explanation
TLDR
We introduce LEAN-LIFE, a web-based, Label-Efficient AnnotatioN framework for sequence labeling and classification tasks, with an easy-to-use UI that not only allows an annotator to provide the needed labels for a task but also enables LearnIng From Explanations for each labeling decision. Expand
Zero-shot Learning by Generating Task-specific Adapters
TLDR
We introduce HYPTER, a framework that improves zero-shot transferability of text-to-text transformers by training a hypernetwork to generate task-specific adapters from task descriptions. Expand
Studying Strategically: Learning to Mask for Closed-book QA
TLDR
We learn a masking policy to extract spans that are likely to be tested, using supervision from the downstream task itself, then deploy the learned policy during intermediate pre-training. Expand
Modeling Content Interaction in Information Diffusion with Pre-trained Sentence Embedding
TLDR
In this paper, we focus on how multiple pieces of information interaction with each other in information diffusion process, which verifies the general behavior of users. Expand
CrossFit: A Few-shot Learning Challenge for Cross-task Generalization in NLP
TLDR
We introduce CROSSFIT, a task setup for studying cross-task few-shot learning ability, which standardizes seen/unseen task splits, data access during different learning stages, and evaluation protocols. Expand
On the Influence of Masking Policies in Intermediate Pre-training
TLDR
We investigate the effects of using heuristic, directly supervised, and meta-learned MLM policies for intermediate pretraining, on eight selected tasks across three categories (closed-book QA, knowledgeintensive language tasks, and abstractive summarization). Expand
...
1
2
...