Improving generalization in large langue model by learning prefix subspaces

LOUIS FALISSARD, Vincent Guigue, Laure Soulier

Submission Type: Regular Long Paper

Submission Track: Efficient Methods for NLP

Submission Track 2: Machine Learning for NLP

Keywords: Deep learning, parameter efficient fine-tuning, prefix-tuning, subspace learning, natural language processing

TL;DR: Learning simplexes of prefixes for large language models

Abstract:

This article focuses on large language models (LLMs) fine-tuning in the scarce data regime (also known as "few-shot learning setting"). We propose a method to increase the generalization capabilities of LLMs based on neural network subspaces.

This optimization method, recently introduced in computer vision, aims to improve model generalization by identifying wider local optima through the joint optimization of an entire simplex of models in parameter space.

Although this property would be highly beneficial in the context of training large language models in the “few-shot learning” setting, its adaptation to massive, pretrained transformers poses some challenges. First, their considerable number of parameters make it difficult to train several model jointly, and second, their deterministic parameter initialisation schemes make them unfit to the subspace method as originaly proposed.
We show in this paper that its application to "Parameter Efficient Fine-Tuning" (PEFT) methods, however, is relatively natural, and we propose to apply it to prefix-tuning, by learning entire simplexes of continous prefixes.

We test our method on a variant of the GLUE benchmark adapted to the few-shot learning setting, and show that both our contributions (learning prefix simplexes, and non-deterministic validation metric inference) jointly lead to a gain in average performances compared to state of the art methods.

Submission Number: 4063