Skip to content

Latest commit

 

History

History
14 lines (8 loc) · 717 Bytes

README.md

File metadata and controls

14 lines (8 loc) · 717 Bytes

Contrapositive Local Inference

This repository contains the code and resources for the contrapositive inference model presented in the following paper:

Omid Kashefi, Rebecca Hwa, 2021. "Contrapositive Local Class Inference." in EMNLP Workshop on Noisy User-generated Text (W-NUT).

The model is implemented as a transfer function on top of ctrl-gen model (Hu et al., 2017), modified with an additional conciseness loss constraint to ensure the class transference with minimal changes to the original greater-context.

Usage

python3 model\contrapositive-tf.py --config config.[task]