Subformer: A Parameter Reduced TransformerDownload PDF

28 Sept 2020 (modified: 05 May 2023)ICLR 2021 Conference Withdrawn SubmissionReaders: Everyone
Keywords: transformers, sequence modeling, machine translation, efficiency
Abstract: The advent of the Transformer can arguably be described as a driving force behind many of the recent advances in natural language processing. However, despite their sizeable performance improvements, as recently shown, the model is severely over-parameterized, being parameter inefficient and computationally expensive to train. Inspired by the success of parameter-sharing in pre-trained deep contextualized word representation encoders, we explore parameter-sharing methods in Transformers, with a specific focus on encoder-decoder models for sequence-to-sequence tasks such as Machine Translation. We perform an analysis of different parameter sharing/reduction methods and develop the Subformer, a parameter efficient Transformer-based model which combines the newly proposed Sandwich-style parameter sharing technique and self-attentive embedding factorization (SAFE). Experiments on machine translation, abstractive summarization, and language modeling show that the Subformer can outperform the Transformer even when using significantly fewer parameters. On the WMT'14 English-German test set, we show we can perform equally well, and even sometimes outperform (+0.1 BLEU score) the Transformer-base model while using 40% fewer parameters. We also perform equally well as Transformer-big with 40% fewer parameters, achieve performance within 0.1 BLEU with 70% fewer parameters, and outperform the model by 0.7 BLEU with 12M fewer parameters. We also outperform the standard Transformer-XL model, achieving a significant 3.6 lower perplexity with 37% fewer parameters.
One-sentence Summary: The Subformer combines a novel weight-sharing technique with a novel approach to embedding factorization for training parameter-efficient Transformers with larger capacity.
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics
Reviewed Version (pdf): https://openreview.net/references/pdf?id=0j9QztAzhx
18 Replies

Loading