REVIEW 3 cited by
Legend: Leveraging Representation Engineering to Annotate Safety Margin for Preference Datasets
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
Signed reviews
read the original abstract
The success of the reward model in distinguishing between responses with subtle safety differences depends critically on the high-quality preference dataset, which should capture the fine-grained nuances of harmful and harmless responses. This motivates the need to develop a dataset involving preference margins, which accurately quantify how harmless one response is compared to another. In this paper, we take the first step to propose an effective and cost-efficient framework to promote the margin-enhanced preference dataset development. Our framework, Legend, Leverages representation engineering to annotate preference datasets. It constructs the specific direction within the LLM's embedding space that represents safety. By leveraging this safety direction, Legend can then leverage the semantic distances of paired responses along this direction to annotate margins automatically. We experimentally demonstrate our effectiveness in both reward modeling and harmless alignment for LLMs. Legend also stands out for its efficiency, requiring only the inference time rather than additional training. This efficiency allows for easier implementation and scalability, making Legend particularly valuable for practical applications in aligning LLMs with safe conversations.
Forward citations
Cited by 3 Pith papers
-
Icon$^{2}$: Aligning Large Language Models Using Self-Synthetic Preference Data via Inherent Regulation
ICON2 uses representation-space steering to generate preference data from the model itself, improving alignment benchmarks and cutting cost.
-
On Almost Surely Safe Alignment of Large Language Models at Inference-Time
An inference-time beam-search method with a safety-state tracker and latent critic enforces a user-supplied safety cost model, with an almost-sure guarantee only relative to that model.
-
Cross-model Transferability among Large Language Models on the Platonic Representations of Concepts
A single linear transformation aligns concept representations between different LLMs, so steering vectors transfer across models and even from small to large models.
Discussion (0). Continue with ORCID to comment.