Large Product Key Memory for Pretrained Language Models

Gyuwan Kim, Tae Hwan Jung


Abstract
Product key memory (PKM) proposed by Lample et al. (2019) enables to improve prediction accuracy by increasing model capacity efficiently with insignificant computational overhead. However, their empirical application is only limited to causal language modeling. Motivated by the recent success of pretrained language models (PLMs), we investigate how to incorporate large PKM into PLMs that can be finetuned for a wide variety of downstream NLP tasks. We define a new memory usage metric, and careful observation using this metric reveals that most memory slots remain outdated during the training of PKM-augmented models. To train better PLMs by tackling this issue, we propose simple but effective solutions: (1) initialization from the model weights pretrained without memory and (2) augmenting PKM by addition rather than replacing a feed-forward network. We verify that both of them are crucial for the pretraining of PKM-augmented PLMs, enhancing memory utilization and downstream performance. Code and pretrained weights are available at https://github.com/clovaai/pkm-transformers.
Anthology ID:
2020.findings-emnlp.362
Volume:
Findings of the Association for Computational Linguistics: EMNLP 2020
Month:
November
Year:
2020
Address:
Online
Venue:
Findings
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
4060–4069
Language:
URL:
https://aclanthology.org/2020.findings-emnlp.362
DOI:
10.18653/v1/2020.findings-emnlp.362
Bibkey:
Cite (ACL):
Gyuwan Kim and Tae Hwan Jung. 2020. Large Product Key Memory for Pretrained Language Models. In Findings of the Association for Computational Linguistics: EMNLP 2020, pages 4060–4069, Online. Association for Computational Linguistics.
Cite (Informal):
Large Product Key Memory for Pretrained Language Models (Kim & Jung, Findings 2020)
Copy Citation:
PDF:
https://aclanthology.org/2020.findings-emnlp.362.pdf
Video:
 https://slideslive.com/38939424
Data
CoLAGLUEPG-19SQuADSST