This is the official repo for the paper "Inference-Time Policy Adapters (IPA): Tailoring Extreme-Scale LMs without Fine-tuning" (EMNLP 2023)
We suggest using conda to setup environment. You need to first replace prefix in environment.yml with your home path. With conda installed, create an environment called quark with:
conda env create -f environment.yml
The main branch contains policy adapter for constrained generation task. We put the other three tasks, detoxification, open-ended generation and dialogue generation, in the toxicity, open_ended and dialogue branch separately.
If you use this codebase in your work, please consider citing our paper:
@article{Lu2023InferenceTimePA,
title={Inference-Time Policy Adapters (IPA): Tailoring Extreme-Scale LMs without Fine-tuning},
author={Ximing Lu and Faeze Brahman and Peter West and Jaehun Jang and Khyathi Raghavi Chandu and Abhilasha Ravichander and Lianhui Qin and Prithviraj Ammanabrolu and Liwei Jiang and Sahana Ramnath and Nouha Dziri and Jillian R. Fisher and Bill Yuchen Lin and Skyler Hallinan and Xiang Ren and Sean Welleck and Yejin Choi},
journal={ArXiv},
year={2023},
volume={abs/2305.15065},
url={https://api.semanticscholar.org/CorpusID:258865629}
}