Meta-Learning the Difference: Preparing Large Language Models for Efficient Adaptation

Zejiang Hou, Julian Salazar, George Polovets


Abstract
Large pretrained language models (PLMs) are often domain- or task-adapted via finetuning or prompting. Finetuning requires modifying all of the parameters and having enough data to avoid overfitting while prompting requires no training and few examples but limits performance. Instead, we prepare PLMs for data- and parameter-efficient adaptation by learning to learn the difference between general and adapted PLMs. This difference is expressed in terms of model weights and sublayer structure through our proposed dynamic low-rank reparameterization and learned architecture controller. Experiments on few-shot dialogue completion, low-resource abstractive summarization, and multi-domain language modeling show improvements in adaptation time and performance over direct finetuning or preparation via domain-adaptive pretraining. Ablations show our task-adaptive reparameterization (TARP) and model search (TAMS) components individually improve on other parameter-efficient transfer like adapters and structure-learning methods like learned sparsification.
Anthology ID:
2022.tacl-1.72
Volume:
Transactions of the Association for Computational Linguistics, Volume 10
Month:
Year:
2022
Address:
Cambridge, MA
Editors:
Brian Roark, Ani Nenkova
Venue:
TACL
SIG:
Publisher:
MIT Press
Note:
Pages:
1249–1265
Language:
URL:
https://aclanthology.org/2022.tacl-1.72
DOI:
10.1162/tacl_a_00517
Bibkey:
Cite (ACL):
Zejiang Hou, Julian Salazar, and George Polovets. 2022. Meta-Learning the Difference: Preparing Large Language Models for Efficient Adaptation. Transactions of the Association for Computational Linguistics, 10:1249–1265.
Cite (Informal):
Meta-Learning the Difference: Preparing Large Language Models for Efficient Adaptation (Hou et al., TACL 2022)
Copy Citation:
PDF:
https://aclanthology.org/2022.tacl-1.72.pdf
Video:
 https://aclanthology.org/2022.tacl-1.72.mp4