mlsanityData to 14 Oct 2026

EasyEdit2: An Easy-to-use Steering Framework for Editing Large Language Models

arXiv 2504.15133 ↗First posted 21 Apr 2025cs.CL cs.AI cs.CVcited by 7

On the map: Mechanistic interpretability → Activation steering methods

Abstract (arXiv)

In this paper, we introduce EasyEdit2, a framework designed to enable plug-and-play adjustability for controlling Large Language Model (LLM) behaviors. EasyEdit2 supports a wide range of test-time interventions, including safety, sentiment, personality, reasoning patterns, factuality, and language features. Unlike its predecessor, EasyEdit2 features a new architecture specifically designed for seamless model steering. It comprises key modules such as the steering vector generator and the steering vector applier, which enable automatic generation and application of steering vectors to influence the model's behavior without modifying its parameters. One of the main advantages of EasyEdit2 is its ease of use-users do not need extensive technical knowledge. With just a single example, they can effectively guide and adjust the model's responses, making precise control both accessible and efficient. Empirically, we report model steering performance across different LLMs, demonstrating the effectiveness of these techniques. We have released the source code on GitHub at https://github.com/zjunlp/EasyEdit along with a demonstration notebook. In addition, we provide a demo video at https://www.youtube.com/watch?v=AkfoiPfp5rQ for a quick introduction.

Key facts. Cited by 7 papers in the map (data through 14 Oct 2026). Placed in Activation steering methods, one of 13 directions in Mechanistic interpretability. Builds on 41 of its 67 references that are in the map. First posted on arXiv 21 Apr 2025.

Builds on

41 of 67 references.

Closest ancestors · 7

Same direction, newest first

  1. 2025Steering Large Language Model Activations in Sparse Spaces28 Feb
  2. 2025A Unified Understanding and Evaluation of Steering Methods4 Feb
  3. 2024Semantics-Adaptive Activation Intervention for LLMs via Dynamic Steering Vectors16 Oct
  4. 2024Improving Instruction-Following in Language Models through Activation Steering15 Oct
  5. 2024Personalized Steering of Large Language Models: Versatile Steering Vectors Through Bi-directional Preference Optimization28 May
  6. 2023Steering Llama 2 via Contrastive Activation Addition9 Dec
  7. 2023Steering Language Models With Activation Engineering20 Aug

Other references · 34

  1. 2024Gemma 2: Improving Open Language Models at a Practical Sizecited by 1,998
  2. 2023Sparse Autoencoders Find Highly Interpretable Features in Language Modelscited by 1,420
  3. 2022Model soups: averaging weights of multiple fine-tuned models improves accuracy without increasing inference timecited by 1,396
  4. 2023Representation Engineering: A Top-Down Approach to AI Transparencycited by 1,336
  5. 2020RealToxicityPrompts: Evaluating Neural Toxic Degeneration in Language Modelscited by 1,255
  6. 2023TIES-Merging: Resolving Interference When Merging Modelscited by 818
  7. 2019Plug and Play Language Models: A Simple Approach to Controlled Text Generationcited by 799
  8. 2023Language Models are Super Mario: Absorbing Abilities from Homologous Models as a Free Lunchcited by 694
  9. 2024Parameter-Efficient Fine-Tuning for Large Models: A Comprehensive Surveycited by 656
  10. 2021DExperts: Decoding-Time Controlled Text Generation with Experts and Anti-Expertscited by 441
  11. 2023Trustworthy LLMs: a Survey and Guideline for Evaluating Large Language Models' Alignmentcited by 401
  12. 2021FUDGE: Controlled Text Generation With Future Discriminatorscited by 376
22 more
  1. 2023Editing Large Language Models: Problems, Methods, and Opportunitiescited by 372
  2. 2023Emergent Linear Representations in World Models of Self-Supervised Sequence Modelscited by 319
  3. 2017Toward Controlled Generation of Textcited by 272
  4. 2022Aging with GRACE: Lifelong Model Editing with Discrete Key-Value Adaptorscited by 271
  5. 2025AxBench: Steering LLMs? Even Simple Baselines Outperform Sparse Autoencoderscited by 218
  6. 2024Arcee's MergeKit: A Toolkit for Merging Large Language Modelscited by 194
  7. 2024Do I Know This Entity? Knowledge Awareness and Hallucinations in Language Modelscited by 115
  8. 2023Language Model Behavior: A Comprehensive Surveycited by 92
  9. 2024Improving Steering Vectors by Targeting Sparse Autoencoder Featurescited by 92
  10. 2024Detoxifying Large Language Models via Knowledge Editingcited by 87
  11. 2023Word Embeddings Are Steers for Language Modelscited by 70
  12. 2025SEAL: Steerable Reasoning Calibration of Large Language Models for Freecited by 68
  13. 2025International AI Safety Reportcited by 67
  14. 2024Steering Knowledge Selection Behaviours in LLMs via SAE-Based Representation Engineeringcited by 63
  15. 2024Applying sparse autoencoders to unlearn knowledge in language modelscited by 59
  16. 2025Taxonomy, Opportunities, and Challenges of Representation Engineering for Large Language Modelscited by 34
  17. 2025Steering off Course: Reliability Challenges in Steering Language Modelscited by 34
  18. 2025Steering Dialogue Dynamics for Robustness against Multi-turn Jailbreaking Attackscited by 20
  19. 2024Can sparse autoencoders be used to decompose and interpret steering vectors?cited by 19
  20. 2025Knowledge Updating? No More Model Editing! Just Selective Contextual Reasoningcited by 16
  21. 2025Position: Editing Large Language Models Poses Serious Safety Riskscited by 15
  22. 2025SAKE: Steering Activations for Knowledge Editingcited by 13

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 5; 2026: 2.

7 papers.

  1. 2025MemOS: A Memory OS for AI Systemcited by 138
  2. 2025MemOS: An Operating System for Memory-Augmented Generation (MAG) in Large Language Modelscited by 73
  3. 2025Large Language Model Sourcing: A Surveycited by 5
  4. 2025Text2Mem: A Unified Memory Operation Language for Memory Operating Systemcited by 3
  5. 2026Composer Vector: Style-steering Symbolic Music Generation in a Latent Spacecited by 3
  6. 2025EtCon: Edit-then-Consolidate for Reliable Knowledge Editingcited by 1
  7. 2026xMIx: High-Performance Serving-Time Platform for Mechanistic Interpretability Appscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025EasySteer: A Unified Framework for High-Performance and Extensible LLM Steering
  2. 2026Does Steering Break Your Model? A Multi-Dimensional Evaluation Suite for LLM Steering Methods
  3. 2025Steering When Necessary: Flexible Steering Large Language Models with Backtracking
  4. 2025Self-Improving Model Steering
  5. 2025To Steer or Not to Steer? Mechanistic Error Reduction with Abstention for Language Models
  6. 2025Dialz: A Python Toolkit for Steering Vectors
  7. 2025Model Whisper: Steering Vectors Unlock Large Language Models' Potential in Test-time
  8. 2026Steer2Edit: From Activation Steering to Component-Level Editing