Why Do Pretrained Language Models Help in Downstream Tasks? An Analysis of Head and Prompt Tuning

TL;DR

This paper provides a theoretical framework linking pretraining and downstream NLP tasks using latent variable models, analyzing head and prompt tuning, and demonstrating conditions under which these methods succeed, supported by synthetic experiments.

Contribution

It introduces a generative model-based analysis of head and prompt tuning, revealing conditions for successful downstream task adaptation and comparing their effectiveness.

Findings

Head tuning works under certain non-degeneracy conditions.

Prompt tuning requires weaker conditions for guarantees.

Memory-augmented models recover task information more effectively.

Abstract

Pretrained language models have achieved state-of-the-art performance when adapted to a downstream NLP task. However, theoretical analysis of these models is scarce and challenging since the pretraining and downstream tasks can be very different. We propose an analysis framework that links the pretraining and downstream tasks with an underlying latent variable generative model of text -- the downstream classifier must recover a function of the posterior distribution over the latent variables. We analyze head tuning (learning a classifier on top of the frozen pretrained model) and prompt tuning in this setting. The generative model in our analysis is either a Hidden Markov Model (HMM) or an HMM augmented with a latent memory component, motivated by long-term dependencies in natural language. We show that 1) under certain non-degeneracy conditions on the HMM, simple classification heads can solve the downstream task, 2) prompt tuning obtains downstream guarantees with weaker non-degeneracy conditions, and 3) our recovery guarantees for the memory-augmented HMM are stronger than for the vanilla HMM because task-relevant information is easier to recover from the long-term memory. Experiments on synthetically generated data from HMMs back our theoretical findings.