Instructive Decoding: Instruction-Tuned Large Language Models are Self-Refiner from Noisy Instructions

TL;DR

Instructive Decoding enhances instruction-tuned language models by using noisy instructions to improve response accuracy without additional training, effectively refining outputs through contrastive logit adjustments.

Contribution

This paper introduces Instructive Decoding, a novel contrastive decoding method that improves instruction-tuned models' responses using noisy instructions without extra parameters.

Findings

Significant performance improvements across models and tasks.

'Opposite' noisy instructions yield the largest gains.

Method works without additional parameter updates.

Abstract

While instruction-tuned language models have demonstrated impressive zero-shot generalization, these models often struggle to generate accurate responses when faced with instructions that fall outside their training set. This paper presents Instructive Decoding (ID), a simple yet effective approach that augments the efficacy of instruction-tuned models. Specifically, ID adjusts the logits for next-token prediction in a contrastive manner, utilizing predictions generated from a manipulated version of the original instruction, referred to as a noisy instruction. This noisy instruction aims to elicit responses that could diverge from the intended instruction yet remain plausible. We conduct experiments across a spectrum of such noisy instructions, ranging from those that insert semantic noise via random words to others like 'opposite' that elicit the deviated responses. Our approach achieves considerable performance gains across various instruction-tuned models and tasks without necessitating any additional parameter updates. Notably, utilizing 'opposite' as the noisy instruction in ID, which exhibits the maximum divergence from the original instruction, consistently produces the most significant performance gains across multiple models and tasks.