Refining Positive and Toxic Samples for Dual Safety Self-Alignment of LLMs with Minimal Human Interventions

TL;DR

This paper introduces PT-ALIGN, a safety self-alignment method for LLMs that automatically refines positive and toxic samples with minimal human intervention, improving safety without sacrificing helpfulness.

Contribution

The paper presents a novel approach that leverages LLMs to automatically generate and refine safety-related training samples, reducing manual annotation efforts and enhancing safety alignment.

Findings

PT-ALIGN effectively improves safety alignment across 9 open-source LLMs.

The method maintains helpfulness and usefulness while enhancing safety.

Iterative sample refinement requires fewer than 50 human annotations.

Abstract

Recent AI agents, such as ChatGPT and LLaMA, primarily rely on instruction tuning and reinforcement learning to calibrate the output of large language models (LLMs) with human intentions, ensuring the outputs are harmless and helpful. Existing methods heavily depend on the manual annotation of high-quality positive samples, while contending with issues such as noisy labels and minimal distinctions between preferred and dispreferred response data. However, readily available toxic samples with clear safety distinctions are often filtered out, removing valuable negative references that could aid LLMs in safety alignment. In response, we propose PT-ALIGN, a novel safety self-alignment approach that minimizes human supervision by automatically refining positive and toxic samples and performing fine-grained dual instruction tuning. Positive samples are harmless responses, while toxic samples deliberately contain extremely harmful content, serving as a new supervisory signals. Specifically, we utilize LLM itself to iteratively generate and refine training instances by only exploring fewer than 50 human annotations. We then employ two losses, i.e., maximum likelihood estimation (MLE) and fine-grained unlikelihood training (UT), to jointly learn to enhance the LLM's safety. The MLE loss encourages an LLM to maximize the generation of harmless content based on positive samples. Conversely, the fine-grained UT loss guides the LLM to minimize the output of harmful words based on negative samples at the token-level, thereby guiding the model to decouple safety from effectiveness, directing it toward safer fine-tuning objectives, and increasing the likelihood of generating helpful and reliable content. Experiments on 9 popular open-source LLMs demonstrate the effectiveness of our PT-ALIGN for safety alignment, while maintaining comparable levels of helpfulness and usefulness.