mlsanityData to 14 Oct 2026

Composing Parameter-Efficient Modules with Arithmetic Operations

arXiv 2306.14870 ↗First posted 26 Jun 2023cs.CLcited by 156

Not on the map. Why?

Cited by 33 in LLM safety and security, but placed by its own abstract, which didn’t fit.

Abstract (arXiv)

As an efficient alternative to conventional full finetuning, parameter-efficient finetuning (PEFT) is becoming the prevailing method to adapt pretrained language models. In PEFT, a lightweight module is learned on each dataset while the underlying pretrained language model remains unchanged, resulting in multiple compact modules representing diverse skills when applied to various domains and tasks. In this paper, we propose to compose these parameter-efficient modules through linear arithmetic operations in the weight space, thereby integrating different module capabilities. Specifically, we first define addition and negation operators for the module, and then further compose these two basic operators to perform flexible arithmetic. Our approach requires no additional training and enables highly flexible module composition. We apply different arithmetic operations to compose the parameter-efficient modules for (1) distribution generalization, (2) multi-tasking, (3) unlearning, and (4) domain transfer. Additionally, we extend our approach to detoxify Alpaca-LoRA, the latest instruction-tuned large language model based on LLaMA. Empirical results demonstrate that our approach produces new and effective parameter-efficient modules that significantly outperform existing ones across all settings.

Key facts. Cited by 156 papers in the map (data through 14 Oct 2026). Builds on 36 of its 51 references that are in the map. First posted on arXiv 26 Jun 2023.

Builds on

36 of 51 references.

References · 36

  1. 2017Attention Is All You Needcited by 60,361
  2. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,399
  3. 2023GPT-4 Technical Reportcited by 18,396
  4. 2019Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformercited by 14,599
  5. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,380
  6. 2023LLaMA: Open and Efficient Foundation Language Modelscited by 14,132
  7. 2019RoBERTa: A Robustly Optimized BERT Pretraining Approachcited by 13,372
  8. 2023Judging LLM-as-a-Judge with MT-Bench and Chatbot Arenacited by 8,626
  9. 2022PaLM: Scaling Language Modeling with Pathwayscited by 5,326
  10. 2018GLUE: A Multi-Task Benchmark and Analysis Platform for Natural Language Understandingcited by 4,899
  11. 2019Parameter-Efficient Transfer Learning for NLPcited by 4,236
  12. 2016SQuAD: 100,000+ Questions for Machine Comprehension of Textcited by 4,148
24 more
  1. 2021Prefix-Tuning: Optimizing Continuous Prompts for Generationcited by 3,915
  2. 2016Pointer Sentinel Mixture Modelscited by 3,096
  3. 2022Scaling Instruction-Finetuned Language Modelscited by 3,024
  4. 2022Self-Instruct: Aligning Language Models with Self-Generated Instructionscited by 2,724
  5. 2017A Broad-Coverage Challenge Corpus for Sentence Understanding through Inferencecited by 2,682
  6. 2015Character-level Convolutional Networks for Text Classificationcited by 2,330
  7. 2021Pre-train, Prompt, and Predict: A Systematic Survey of Prompting Methods in Natural Language Processingcited by 2,234
  8. 2021Multitask Prompted Training Enables Zero-Shot Task Generalizationcited by 1,470
  9. 2022Model soups: averaging weights of multiple fine-tuned models improves accuracy without increasing inference timecited by 1,396
  10. 2022Editing Models with Task Arithmeticcited by 1,120
  11. 2017SemEval-2017 Task 1: Semantic Textual Similarity - Multilingual and Cross-lingual Focused Evaluationcited by 1,001
  12. 2022Few-Shot Parameter-Efficient Fine-Tuning is Better and Cheaper than In-Context Learningcited by 896
  13. 2021Towards a Unified View of Parameter-Efficient Transfer Learningcited by 872
  14. 2020AdapterFusion: Non-Destructive Task Composition for Transfer Learningcited by 788
  15. 2021Merging Models with Fisher-Weighted Averagingcited by 621
  16. 2022Git Re-Basin: Merging Models modulo Permutation Symmetriescited by 471
  17. 2021Compacter: Efficient Low-Rank Hypercomplex Adapter Layerscited by 460
  18. 2019Nuanced Metrics for Measuring Unintended Bias with Real Data for Text Classificationcited by 404
  19. 2022Dataless Knowledge Fusion by Merging Weights of Language Modelscited by 387
  20. 2020What is being transferred in transfer learning?cited by 368
  21. 2022PromptSource: An Integrated Development Environment and Repository for Natural Language Promptscited by 289
  22. 2023AdapterSoup: Weight Averaging to Improve Generalization of Pretrained Language Modelscited by 84
  23. 2023Modular Deep Learningcited by 83
  24. 2022Exploring Mode Connectivity for Pre-trained Language Modelscited by 26

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2023: 15; 2024: 67; 2025: 49; 2026: 25.

156 papers.

Sort:

By area:

  1. 2023Language Models are Super Mario: Absorbing Abilities from Homologous Models as a Free Lunchcited by 694
  2. 2023Directly Fine-Tuning Diffusion Models on Differentiable Rewardscited by 448
  3. 2023LoraHub: Efficient Cross-Task Generalization via Dynamic LoRA Compositioncited by 364
  4. 2024Rethinking Machine Unlearning for Large Language Modelscited by 257
  5. 2024Model Merging in LLMs, MLLMs, and Beyond: Methods, Theories, Applications and Opportunitiescited by 254
  6. 2023AdaMerging: Adaptive Model Merging for Multi-Task Learningcited by 252
  7. 2023Parameter-Efficient Fine-Tuning Methods for Pretrained Language Models: A Critical Review and Assessmentcited by 230
  8. 2023LCM-LoRA: A Universal Stable-Diffusion Acceleration Modulecited by 211
  9. 2024ReFT: Representation Finetuning for Language Modelscited by 197
  10. 2024Mixture of LoRA Expertscited by 181
  11. 2024Latent Adversarial Training Improves Robustness to Persistent Harmful Behaviors in LLMscited by 173
  12. 2024A Comprehensive Study of Knowledge Editing for Large Language Modelscited by 163
38 more
  1. 2024Eight Methods to Evaluate Robust Unlearning in LLMscited by 163
  2. 2024Large Language Model Unlearning via Embedding-Corrupted Promptscited by 116
  3. 2024Twin-Merging: Dynamic Integration of Modular Expertise in Model Mergingcited by 116
  4. 2024EMR-Merging: Tuning-Free High-Performance Model Mergingcited by 110
  5. 2024SOUL: Unlocking the Power of Second-Order Optimization for LLM Unlearningcited by 104
  6. 2024Representation Surgery for Multi-Task Model Mergingcited by 96
  7. 2024Reversing the Forget-Retain Objectives: An Efficient LLM Unlearning Framework from Logit Differencecited by 88
  8. 2024Multi-LoRA Composition for Image Generationcited by 80
  9. 2024Towards Modular LLMs by Building and Reusing a Library of LoRAscited by 72
  10. 2024Parameter Competition Balancing for Model Mergingcited by 72
  11. 2024A Survey on LoRA of Large Language Modelscited by 67
  12. 2023Parameter Efficient Multi-task Model Fusion with Partial Linearizationcited by 65
  13. 2024Learning to Route Among Specialized Experts for Zero-Shot Generalizationcited by 65
  14. 2024FuseChat: Knowledge Fusion of Chat Modelscited by 55
  15. 2023Chat Vector: A Simple Approach to Equip LLMs with Instruction Following and Model Alignment in New Languagescited by 54
  16. 2024LoRA Soups: Merging LoRAs for Practical Skill Composition Taskscited by 52
  17. 2025LLM-based Agents Suffer from Hallucinations: A Survey of Taxonomy, Methods, and Directionscited by 50
  18. 2024Challenging Forgets: Unveiling the Worst-Case Forget Sets in Machine Unlearningcited by 44
  19. 2024CorDA: Context-Oriented Decomposition Adaptation of Large Language Models for Task-Aware Parameter-Efficient Fine-tuningcited by 44
  20. 2024A safety realignment framework via subspace-oriented model fusion for large language modelscited by 41
  21. 2025LLM Unlearning via Neural Activation Redirectioncited by 41
  22. 2025Bring Reason to Vision: Understanding Perception and Reasoning through Model Mergingcited by 40
  23. 2024LoRA-Flow: Dynamic LoRA Fusion for Large Language Models in Generative Taskscited by 39
  24. 2024CLEAR: Character Unlearning in Textual and Visual Modalitiescited by 39
  25. 2023Knowledge Unlearning for LLMs: Tasks, Methods, and Challengescited by 38
  26. 2023Concrete Subspace Learning based Interference Elimination for Multi-task Model Fusioncited by 37
  27. 2024AdaMergeX: Cross-Lingual Transfer with Large Language Models via Adaptive Adapter Mergingcited by 34
  28. 2024Gradient Routing: Masking Gradients to Localize Computation in Neural Networkscited by 34
  29. 2023Separate the Wheat from the Chaff: Model Deficiency Unlearning via Parameter-Efficient Module Operationcited by 33
  30. 2024How to Continually Adapt Text-to-Image Diffusion Models for Flexible Customization?cited by 33
  31. 2024Training-Free Pretrained Model Mergingcited by 32
  32. 2024Lottery Ticket Adaptation: Mitigating Destructive Interference in LLMscited by 31
  33. 2023Language and Task Arithmetic with Parameter-Efficient Layers for Zero-Shot Summarizationcited by 28
  34. 2025CAT Merging: A Training-Free Approach for Resolving Conflicts in Model Mergingcited by 23
  35. 2024Configurable Foundation Models: Building LLMs from a Modular Perspectivecited by 22
  36. 2024Closed-form merging of parameter-efficient modules for Federated Continual Learningcited by 22
  37. 2024Layer Swapping for Zero-Shot Cross-Lingual Transfer in Large Language Modelscited by 21
  38. 2025Improved Representation Steering for Language Modelscited by 21

50 most cited of 156 All 156 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2023Towards Better Parameter-Efficient Fine-Tuning for Large Language Models: A Position Paper
  2. 2025DiaBlo: Diagonal Blocks Are Sufficient For Finetuning
  3. 2024Light-PEFT: Lightening Parameter-Efficient Fine-Tuning via Early Pruning
  4. 2026PEML: Parameter-efficient Multi-Task Learning with Optimized Continuous Prompts
  5. 2024Step-by-Step Unmasking for Parameter-Efficient Fine-tuning of Large Language Models
  6. 2024Let the Expert Stick to His Last: Expert-Specialized Fine-Tuning for Sparse Architectural Large Language Models
  7. 2026GPart: End-to-End Isometric Fine-Tuning via Global Parameter Partitioning
  8. 2024LLM Augmented LLMs: Expanding Capabilities through Composition