mlsanityData to 7 Oct 2026

Self and Cross-Model Distillation for LLMs: Effective Methods for Refusal Pattern Alignment

arXiv 2406.11285 ↗First posted 17 Jun 2024cs.CR cs.CLcited by 3

On the map: LLM alignment / RLHF / preference optimisation → Safety alignment and jailbreak attacks

Abstract (arXiv)

Large Language Models (LLMs) like OpenAI's GPT series, Anthropic's Claude, and Meta's LLaMa have shown remarkable capabilities in text generation. However, their susceptibility to toxic prompts presents significant security challenges. This paper investigates alignment techniques, including Supervised Fine-Tuning (SFT) and Reinforcement Learning from Human Feedback (RLHF), to mitigate these risks. We conduct an empirical study on refusal patterns across nine LLMs, revealing that models with uniform refusal patterns, such as Claude3, exhibit higher security. Based on these findings, we propose self-distilling and cross-model distilling methods to enhance LLM security. Our results show that these methods significantly improve refusal rates and reduce unsafe content, with cross-model distilling achieving refusal rates close to Claude3's 94.51%. These findings underscore the potential of distillation-based alignment in securing LLMs against toxic prompts.

Key facts. Cited by 3 papers in the map (data through 7 Oct 2026). Placed in Safety alignment and jailbreak attacks, one of 10 directions in LLM alignment / RLHF / preference optimisation. Builds on 11 of its 21 references that are in the map. First posted on arXiv 17 Jun 2024.

Builds on

11 of 21 references.

References · 11

  1. 2020Language Models are Few-Shot Learnerscited by 32,199
  2. 2023LLaMA: Open and Efficient Foundation Language Modelscited by 13,958
  3. 2023Prompt Injection attack against LLM-integrated Applicationscited by 568
  4. 2023Jailbreaking ChatGPT via Prompt Engineering: An Empirical Studycited by 478
  5. 2023Fine-Grained Human Feedback Gives Better Rewards for Language Model Trainingcited by 399
  6. 2023Pretraining Language Models with Human Preferencescited by 245
  7. 2023How Abilities in Large Language Models are Affected by Supervised Fine-tuning Data Compositioncited by 222
  8. 2023PromptRobust: Towards Evaluating the Robustness of Large Language Models on Adversarial Promptscited by 218
  9. 2023Latent Jailbreak: A Benchmark for Evaluating Text Safety and Output Robustness of Large Language Modelscited by 56
  10. 2023Adversarial attacks and defenses in explainable artificial intelligence: A surveycited by 47
  11. 2023A Comprehensive Overview of Backdoor Attacks in Large Language Models within Communication Networkscited by 38

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 1; 2026: 2.

3 papers.

  1. 2026OpenRT: An Open-Source Red Teaming Framework for Multimodal LLMscited by 18
  2. 2025Characterizing Selective Refusal Bias in Large Language Modelscited by 4
  3. 2026Visual-Noise Guided In-Context Distillation for Multimodal Large Language Model Unlearningcited by 3

Similar, not cited

Close in wording, no citation either way

  1. 2025Improving LLM Safety Alignment with Dual-Objective Optimization
  2. 2026LLMs Can Unlearn Refusal with Only 1,000 Benign Samples
  3. 2025From Threat to Tool: Leveraging Refusal-Aware Injection Attacks for Safety Alignment
  4. 2026Fail-Closed Alignment for Large Language Models
  5. 2024OR-Bench: An Over-Refusal Benchmark for Large Language Models
  6. 2026Deactivating Refusal Triggers: Understanding and Mitigating Overrefusal in Safety Alignment
  7. 2025FalseReject: A Resource for Improving Contextual Safety and Mitigating Over-Refusals in LLMs via Structured Reasoning
  8. 2024Mind the Inconspicuous: Revealing the Hidden Weakness in Aligned LLMs' Refusal Boundaries