Practical Computational Power of Linear Transformers and Their Recurrent and Self-Referential Extensions

Autor:	Irie, Kazuki, Csordás, Róbert, Schmidhuber, Jürgen
Rok vydání:	2023
Předmět:	Computer Science - Machine Learning
Druh dokumentu:	Working Paper
Popis:	Recent studies of the computational power of recurrent neural networks (RNNs) reveal a hierarchy of RNN architectures, given real-time and finite-precision assumptions. Here we study auto-regressive Transformers with linearised attention, a.k.a. linear Transformers (LTs) or Fast Weight Programmers (FWPs). LTs are special in the sense that they are equivalent to RNN-like sequence processors with a fixed-size state, while they can also be expressed as the now-popular self-attention networks. We show that many well-known results for the standard Transformer directly transfer to LTs/FWPs. Our formal language recognition experiments demonstrate how recently proposed FWP extensions such as recurrent FWPs and self-referential weight matrices successfully overcome certain limitations of the LT, e.g., allowing for generalisation on the parity problem. Our code is public. Comment: Accepted to EMNLP 2023 (short paper)
Databáze:	arXiv
Externí odkaz:	http://arxiv.org/abs/2310.16076 Zobrazit plný text záznamu View this record from Arxiv