PromptIntern: Saving Inference Costs by Internalizing Recurrent Prompt during Large Language Model Fine-tuning

TL;DR

PromptIntern introduces a novel fine-tuning approach that internalizes recurrent prompts into model parameters, significantly reducing inference costs and token usage while maintaining performance on NL2Code tasks.

Contribution

It presents a new method to embed prompt knowledge directly into model parameters, decreasing reliance on lengthy prompts during inference.

Findings

Reduces input tokens by over 90%

Speeds up inference by 4.2 times

Cuts inference costs by 88.3%

Abstract

Recent advances in fine-tuning large language models (LLMs) have greatly enhanced their usage in domain-specific tasks. Despite the success, fine-tuning continues to rely on repeated and lengthy prompts, which escalate computational expenses, require more resources, and lead to slower inference. In this paper, we present a novel approach, PromptIntern, which internalizes prompt knowledge during model fine-tuning to achieve efficient inference and save costs. Instead of compressing the prompts for a vanilla model, PromptIntern aims to embed the recurrent prompt directly into the model parameters. We design a fine-tuning pipeline that includes instruction template compression, few-shot example absorption, and a progressive internalization strategy, effectively diminishing the need for intricate prompts during inference. Comprehensive experiments on challenging NL2Code tasks demonstrate that our method reduces input tokens by more than 90%, accelerates inference by 4.2 times, and reduces monetary inference costs by 88.3%.