mlsanityData to 7 Oct 2026

Purifying Large Language Models by Ensembling a Small Language Model

arXiv 2402.14845 ↗First posted 19 Feb 2024cs.CL cs.AI cs.LGcited by 25

On the map: LLM safety and security → Machine unlearning

Abstract (arXiv)

The emerging success of large language models (LLMs) heavily relies on collecting abundant training data from external (untrusted) sources. Despite substantial efforts devoted to data cleaning and curation, well-constructed LLMs have been reported to suffer from copyright infringement, data poisoning, and/or privacy violations, which would impede practical deployment of LLMs. In this study, we propose a simple and easily implementable method for purifying LLMs from the negative effects caused by uncurated data, namely, through ensembling LLMs with benign and small language models (SLMs). Aside from theoretical guarantees, we perform comprehensive experiments to empirically confirm the efficacy of ensembling LLMs with SLMs, which can effectively preserve the performance of LLMs while mitigating issues such as copyright infringement, data poisoning, and privacy violations.

Key facts. Cited by 25 papers in the map (data through 7 Oct 2026). Placed in Machine unlearning, one of 12 directions in LLM safety and security. Builds on 28 of its 48 references that are in the map. First posted on arXiv 19 Feb 2024.

Builds on

28 of 48 references.

Closest ancestors · 1

Same direction, newest first

  1. 2023Who's Harry Potter? Approximate Unlearning in LLMs3 Oct

Other references · 27

  1. 2020Language Models are Few-Shot Learnerscited by 32,199
  2. 2023LLaMA: Open and Efficient Foundation Language Modelscited by 13,958
  3. 2021Evaluating Large Language Models Trained on Codecited by 8,053
  4. 2020Scaling Laws for Neural Language Modelscited by 6,411
  5. 2022PaLM: Scaling Language Modeling with Pathwayscited by 5,304
  6. 2018Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challengecited by 4,710
  7. 2019PIQA: Reasoning about Physical Commonsense in Natural Languagecited by 2,818
  8. 2022BLOOM: A 176B-Parameter Open-Access Multilingual Language Modelcited by 1,968
  9. 2023Pythia: A Suite for Analyzing Large Language Models Across Training and Scalingcited by 1,734
  10. 2022Fast Inference from Transformers via Speculative Decodingcited by 1,615
  11. 2022LaMDA: Language Models for Dialog Applicationscited by 1,218
  12. 2016The LAMBADA dataset: Word prediction requiring a broad discourse contextcited by 960
15 more
  1. 2023StarCoder: may the source be with you!cited by 790
  2. 2017Crowdsourcing Multiple Choice Science Questionscited by 742
  3. 2023LLM-Blender: Ensembling Large Language Models with Pairwise Ranking and Generative Fusioncited by 550
  4. 2024Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Trainingcited by 478
  5. 2022The Stack: 3 TB of permissively licensed source codecited by 354
  6. 2020ONION: A Simple and Effective Defense Against Textual Backdoor Attackscited by 231
  7. 2019Detecting AI Trojans Using Meta Neural Analysiscited by 192
  8. 2023A Recipe for Watermarking Diffusion Modelscited by 136
  9. 2020You Autocomplete Me: Poisoning Vulnerabilities in Neural Code Completioncited by 112
  10. 2023On Provable Copyright Protection for Generative Modelscited by 97
  11. 2021BadPre: Task-agnostic Backdoor Attacks to Pre-trained NLP Foundation Modelscited by 76
  12. 2023SILO Language Models: Isolating Legal Risk In a Nonparametric Datastorecited by 71
  13. 2023Bag of Tricks for Training Data Extraction from Language Modelscited by 67
  14. 2022Differentially Private Optimization on Large Model at Small Costcited by 65
  15. 2021A Survey on Ensemble Learning under the Era of Deep Learningcited by 47

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 8; 2025: 13; 2026: 4.

25 papers.

Sort:

By area:

  1. 2025Harnessing Multiple Large Language Models: A Survey on LLM Ensemblecited by 86
  2. 2025Safety at Scale: A Comprehensive Survey of Large Model and Agent Safetycited by 63
  3. 2024Pack of LLMs: Model Fusion at Test-Time via Perplexity Optimizationcited by 56
  4. 2024Merge, Ensemble, and Cooperate! A Survey on Collaborative Strategies in the Era of Large Language Modelscited by 43
  5. 2024Ensembling Large Language Models with Process Reward-Guided Tree Search for Better Complex Reasoningcited by 38
  6. 2025Embodied-R: Collaborative Framework for Activating Embodied Spatial Reasoning in Foundation Models via Reinforcement Learningcited by 37
  7. 2025Fast Large Language Model Collaborative Decoding via Speculationcited by 12
  8. 2024Large Language Model Sentinel: LLM Agent for Adversarial Purificationcited by 10
  9. 2024Speculative Coreset Selection for Task-Specific Fine-tuningcited by 10
  10. 2025TagRouter: Learning Route to LLMs through Tags for Open-Domain Text Generation Taskscited by 9
  11. 2024Hit the Sweet Spot! Span-Level Ensemble for Large Language Modelscited by 8
  12. 2025Ensemble Learning for Large Language Models in Text and Code Generation: A Surveycited by 8
13 more
  1. 2025LLMBoost: Make Large Language Models Stronger with Boostingcited by 8
  2. 2024Copyright-Protected Language Generation via Adaptive Model Fusioncited by 6
  3. 2025Position: Enough of Scaling LLMs! Lets Focus on Downscalingcited by 5
  4. 2025A Survey on Collaborating Small and Large Language Models for Performance, Cost-effectiveness, Cloud-edge Privacy, and Trustworthinesscited by 5
  5. 2024Strong Copyright Protection for Language Models via Adaptive Model Fusioncited by 4
  6. 2025Revisiting RAG Ensemble: A Theoretical and Mechanistic Analysis of Multi-RAG System Collaborationcited by 2
  7. 2025LLM Chemistry Estimation for Multi-LLM Recommendationcited by 2
  8. 2026Ensembling Language Models with Sequential Monte Carlocited by 2
  9. 2026Rethinking LLM Ensembling from the Perspective of Mixture Modelscited by 1
  10. 2025Invoke Interfaces Only When Needed: Adaptive Invocation for Large Language Models in Question Answeringcited by 0
  11. 2025Are AI Coders Snitches? An Empirical Study of Pretraining Data Detection on Code Large Language Modelscited by 0
  12. 2026Provably Protecting Fine-Tuned LLMs from Training Data Extraction while Preserving Utilitycited by 0
  13. 2026Trade-offs in Ensembling, Merging and Routing Among Parameter-Efficient Expertscited by 0