Skip to content

ajaech/calm

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

CALM

Context Aware Language Models

Code for building language models that adapt to different contexts. This code was originally written to support the experiments in the paper Improving Context Aware Language Models. It has since been modified to support experiments for the paper Low-Rank RNN Adaptation for Context-Aware Language Modeling (to appear in TACL). Read the paper for a complete description of the model.

The main idea is that metadata or other context information can be used to adapt or control a language model. The adaptation takes place by allowing a context embedding to transform the weights of the recurrent layer of the model. We call this model the FactorCell model.

I will work on documenting the code more. Send me a message if you want some help getting started.

About

Context Aware Language Models

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published