Towards Understanding Inductive Bias in Transformers: A View From Infinity

Autor:	Lavie, Itay, Gur-Ari, Guy, Ringel, Zohar
Rok vydání:	2024
Předmět:	Computer Science - Machine Learning Condensed Matter - Disordered Systems and Neural Networks Statistics - Machine Learning
Zdroj:	Proceedings of the 41 st International Conference on Machine Learning, Vienna, Austria. PMLR 235, 2024
Druh dokumentu:	Working Paper
Popis:	We study inductive bias in Transformers in the infinitely over-parameterized Gaussian process limit and argue transformers tend to be biased towards more permutation symmetric functions in sequence space. We show that the representation theory of the symmetric group can be used to give quantitative analytical predictions when the dataset is symmetric to permutations between tokens. We present a simplified transformer block and solve the model at the limit, including accurate predictions for the learning curves and network outputs. We show that in common setups, one can derive tight bounds in the form of a scaling law for the learnability as a function of the context length. Finally, we argue WikiText dataset, does indeed possess a degree of permutation symmetry. Comment: ICML 2024
Databáze:	arXiv
Externí odkaz:	http://arxiv.org/abs/2402.05173 Zobrazit plný text záznamu View this record from Arxiv