This repository contains the original code used to run the Plug and Play Language Model (PPLM).
It has also been integrated into the 🤗/Transformers repository.
Authors: Sumanth Dathathri, Andrea Madotto, Janice Lan, Jane Hung, Eric Frank, Piero Molino, Jason Yosinski, and Rosanne Liu
PPLM allows a user to flexibly plug in one or more tiny attribute models representing the desired steering objective into a large, unconditional language model (LM). The method has the key property that it uses the LM as is—no training or fine-tuning is required—which enables researchers to leverage best-in-class LMs even if they do not have the extensive hardware required to train them.
Paper link: https://arxiv.org/abs/1912.02164
Blog link: https://eng.uber.com/pplm
Colab link to test it out without any setup: https://colab.research.google.com/drive/1Ux0Z4-ruiVtJ6jUk98uk6FqfvGHCOYL3
pip install -r requirements.txt
@article{dathathri2019plug,
title={Plug and Play Language Models: a Simple Approach to Controlled Text Generation},
author={Sumanth Dathathri and Andrea Madotto and Janice Lan and Jane Hung and Eric Frank and Piero Molino and Jason Yosinski and Rosanne Liu},
journal={arXiv preprint arXiv:1912.02164},
year={2019},
}
python run_pplm.py -B military --cond_text "The potato" --length 50 --gamma 1.5 --num_iterations 3 --num_samples 10 --stepsize 0.03 --window_length 5 --kl_scale 0.01 --gm_scale 0.99 --colorama --sample
-
Increase
--stepsize
to intensify topic control, and decrease its value to soften the control.--stepsize 0
recovers the original uncontrolled GPT-2 model. -
If the language being generated is repetitive (For e.g. "science science experiment experiment"), there are several options to consider:
a) Reduce the--stepsize
b) Increase--kl_scale
(the KL-loss coefficient) or decrease--gm_scale
(the gm-scaling term)
c) Add--grad-length xx
where xx is an (integer <= length, e.g.--grad-length 30
).
python run_pplm.py -D sentiment --class_label 2 --cond_text "My dog died" --length 50 --gamma 1.0 --num_iterations 10 --num_samples 10 --stepsize 0.04 --kl_scale 0.01 --gm_scale 0.95 --sample
-
Increase
--stepsize
to intensify topic control, and decrease its value to soften the control.--stepsize 0
recovers the original uncontrolled GPT-2 model. -
Use
--class_label 3
for negative, and--class_label 2
for positive