Can I read Towards Understanding Inductive Bias in Transformers: A View From Infinity on EtoBox?

Towards Understanding Inductive Bias in Transformers: A View From Infinity by Lavie, Itay; Gur-Ari, Guy; Ringel, Zohar is a scholarly article available to read on EtoBox.

What is Towards Understanding Inductive Bias in Transformers: A View From Infinity about?

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.

Author

Lavie, Itay; Gur-Ari, Guy; Ringel, Zohar

Published

2024

Language

EN