Randomized Positional Encodings Boost Length Generalization of TransformersDownload PDF


16 Oct 2022 (modified: 05 May 2023)ACL ARR 2022 October Blind SubmissionReaders: Everyone
Keywords: length generalization, positional encodings, algorithmic reasoning
Abstract: Transformers have impressive generalization capabilities on tasks with a fixed context length. However, they fail to generalize to sequences of arbitrary length, even for seemingly simple tasks such as duplicating a string. Moreover, simply training on longer sequences is inefficient due to the quadratic computation complexity of the global attention mechanism. In this work, we demonstrate that this failure mode is linked to the fact that positional encodings are out-of-distribution for longer sequences (even for relative encodings) and introduce a novel family of positional encodings that can overcome this problem. Concretely, our randomized positional encoding scheme simulates the positions of longer sequences and randomly selects an ordered subset to fit the sequence's length. Our large-scale empirical evaluation of 6000 models across 15 algorithmic reasoning tasks shows that our method allows Transformers to generalize to sequences of unseen length (increasing test accuracy by 11.0% on average).
Paper Type: short
Research Area: Machine Learning for NLP
0 Replies