RNR: Teaching Large Language Models to Follow Roles and Rules

TL;DR

This paper introduces RNR, a data generation pipeline that enhances large language models' ability to follow complex roles and rules, ensuring safer and more aligned interactions without sacrificing general instruction-following performance.

Contribution

We propose RNR, an automated method to generate diverse role and rule data for training LLMs, significantly improving their adherence to complex system prompts.

Findings

Over 25% increase in rule adherence pass-rate.

No regression on standard instruction-following benchmarks.

Improved performance on role and rule following benchmarks.

Abstract

Instruction fine-tuning (IFT) elicits instruction following capabilities and steers the behavior of large language models (LLMs) via supervised learning. However, existing models trained on open-source IFT datasets only have the ability to follow instructions from users, and often fail to follow complex role and rules specified by developers, a.k.a. system prompts. The ability to follow these roles and rules is essential for deployment, as it ensures that the model safely interacts with users within developer defined guidelines. To improve such role and rule following ability, we propose \model, an automated data generation pipeline that generates diverse roles and rules from existing IFT instructions, along with corresponding responses. This data can then be used to train models that follow complex system prompts. The models are evaluated on our newly created benchmarks for role and rule following ability, as well as standard instruction-following benchmarks and general NLP tasks. Our framework significantly improves role and rule following capability in LLMs, as evidenced by over 25% increase in pass-rate on rule adherence, i.e. following all requirements, in our experiments with the Alpaca and Ultrachat datasets. Moreover, our models achieves this increase without any regression on popular instruction following benchmarks.