mlsanityData to 14 Oct 2026

Locate, Steer, and Improve: A Practical Survey of Actionable Mechanistic Interpretability in Large Language Models

arXiv 2601.14004 ↗First posted 20 Jan 2026cs.CLcited by 27

On the map: Mechanistic interpretability → Knowledge and representations inside LLMs

Abstract (arXiv)

Mechanistic Interpretability (MI) has emerged as a vital approach to demystify the opaque decision-making of Large Language Models (LLMs). However, existing reviews primarily treat MI as an observational science, summarizing analytical insights while lacking a systematic framework for actionable intervention. To bridge this gap, we present a practical survey structured around the pipeline: "Locate, Steer, and Improve." We formally categorize Localizing (diagnosis) and Steering (intervention) methods based on specific Interpretable Objects to establish a rigorous intervention protocol. Furthermore, we demonstrate how this framework enables tangible improvements in Alignment, Capability, and Efficiency, effectively operationalizing MI as an actionable methodology for model optimization. The curated paper list of this work is available at https://github.com/rattlesnakey/Awesome-Actionable-MI-Survey.

Key facts. Cited by 27 papers in the map (data through 14 Oct 2026). Placed in Knowledge and representations inside LLMs, one of 13 directions in Mechanistic interpretability. Builds on 307 of its 394 references that are in the map. First posted on arXiv 20 Jan 2026.

Builds on

307 of 394 references.

Closest ancestors · 7

Same direction, newest first

  1. 2025A Review of Developmental Interpretability in Large Language Models19 Aug
  2. 2025How do Large Language Models Understand Relevance? A Mechanistic Interpretability Perspective10 Apr
  3. 2025LogitLens4LLMs: Extending Logit Lens Analysis to Modern Large Language Models24 Feb
  4. 2024Towards Faithful Natural Language Explanations: A Study Using Activation Patching in Large Language Models18 Oct
  5. 2024Unlocking the Future: Exploring Look-Ahead Planning Mechanistic Interpretability in Large Language Models23 Jun
  6. 2023Towards Best Practices of Activation Patching in Language Models: Metrics and Methods27 Sep
  7. 2023Eliciting Latent Predictions from Transformers with the Tuned Lens14 Mar

Other references · 300

  1. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,380
  2. 2023Visual Instruction Tuningcited by 7,869
  3. 2023BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Modelscited by 6,137
  4. 2021RoFormer: Enhanced Transformer with Rotary Position Embeddingcited by 4,703
  5. 2022Flamingo: a Visual Language Model for Few-Shot Learningcited by 4,386
  6. 2017Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layercited by 3,509
  7. 2017Axiomatic Attribution for Deep Networkscited by 3,384
  8. 2021Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsitycited by 3,073
  9. 2022Locating and Editing Factual Associations in GPTcited by 2,691
  10. 2023Efficient Streaming Language Models with Attention Sinkscited by 2,186
  11. 2017Learning Important Features Through Propagating Activation Differencescited by 1,458
  12. 2023Sparse Autoencoders Find Highly Interpretable Features in Language Modelscited by 1,420
38 more
  1. 2022SmoothQuant: Accurate and Efficient Post-Training Quantization for Large Language Modelscited by 1,386
  2. 2016Understanding intermediate layers using linear classifier probescited by 1,353
  3. 2020Transformer Feed-Forward Layers Are Key-Value Memoriescited by 1,336
  4. 2023AWQ: Activation-aware Weight Quantization for LLM Compression and Accelerationcited by 1,298
  5. 2022Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 smallcited by 1,165
  6. 2018Sanity Checks for Saliency Mapscited by 1,156
  7. 2022Editing Models with Task Arithmeticcited by 1,120
  8. 2017SmoothGrad: removing noise by adding noisecited by 1,109
  9. 2023Steering Llama 2 via Contrastive Activation Additioncited by 1,085
  10. 2019BERT Rediscovers the Classical NLP Pipelinecited by 1,059
  11. 2024Refusal in Language Models Is Mediated by a Single Directioncited by 1,032
  12. 2022Mass-Editing Memory in a Transformercited by 1,001
  13. 2023Steering Language Models With Activation Engineeringcited by 922
  14. 2021Probing Classifiers: Promises, Shortcomings, and Advancescited by 876
  15. 2019Analyzing Multi-Head Self-Attention: Specialized Heads Do the Heavy Lifting, the Rest Can Be Prunedcited by 856
  16. 2023Progress measures for grokking via mechanistic interpretabilitycited by 839
  17. 2023TIES-Merging: Resolving Interference When Merging Modelscited by 818
  18. 2023Towards Automated Circuit Discovery for Mechanistic Interpretabilitycited by 724
  19. 2019Designing and Interpreting Probes with Control Taskscited by 640
  20. 2024QuaRot: Outlier-Free 4-Bit Inference in Rotated LLMscited by 634
  21. 2022Mixture-of-Experts with Expert Choice Routingcited by 623
  22. 2021Knowledge Neurons in Pretrained Transformerscited by 601
  23. 2024KVQuant: Towards 10 Million Context Length LLM Inference with KV Cache Quantizationcited by 591
  24. 2022Transformer Feed-Forward Layers Build Predictions by Promoting Concepts in the Vocabulary Spacecited by 532
  25. 2023Dissecting Recall of Factual Associations in Auto-Regressive Language Modelscited by 528
  26. 2018What you can cram into a single vector: Probing sentence embeddings for linguistic propertiescited by 457
  27. 2024Sparse Feature Circuits: Discovering and Editing Interpretable Causal Graphs in Language Modelscited by 435
  28. 2024Mechanistic Interpretability for AI Safety -- A Reviewcited by 422
  29. 2025Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attentioncited by 407
  30. 2024Gemma Scope: Open Sparse Autoencoders Everywhere All At Once on Gemma 2cited by 394
  31. 2024ShortGPT: Layers in Large Language Models are More Redundant Than You Expectcited by 383
  32. 2023DoLa: Decoding by Contrasting Layers Improves Factuality in Large Language Modelscited by 376
  33. 2025Persona Vectors: Monitoring and Controlling Character Traits in Language Modelscited by 345
  34. 2023Explainability for Large Language Models: A Surveycited by 333
  35. 2024LayerSkip: Enabling Early Exit Inference and Self-Speculative Decodingcited by 307
  36. 2024Jumping Ahead: Improving Reconstruction Fidelity with JumpReLU Sparse Autoencoderscited by 300
  37. 2024DuoAttention: Efficient Long-Context LLM Inference with Retrieval and Streaming Headscited by 288
  38. 2024Do Llamas Work in English? On the Latent Language of Multilingual Transformerscited by 281

50 most cited of 300 All 300 (JSON)

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 27.

27 papers.

Sort:

By area:

  1. 2026Qwen-Scope: Turning Sparse Features into Development Tools for Large Language Modelscited by 21
  2. 2026Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigationcited by 11
  3. 2026Why Steering Works: Toward a Unified View of Language Model Parameter Dynamicscited by 9
  4. 2026Residual Decoding: Mitigating Hallucinations in Large Vision-Language Models via History-Aware Residual Guidancecited by 5
  5. 2026Emergent Analogical Reasoning in Transformerscited by 5
  6. 2026The Cylindrical Representation Hypothesis for Language Model Steeringcited by 5
  7. 2026OScaR: The Occam's Razor for Extreme KV Cache Quantization in LLMs and Beyondcited by 5
  8. 2026SnapMLA: Efficient Long-Context MLA Decoding via Hardware-Aware FP8 Quantized Pipeliningcited by 4
  9. 2026PSF-Med: Measuring and Explaining Paraphrase Sensitivity in Medical Vision Language Modelscited by 4
  10. 2026Navigating by Old Maps: The Pitfalls of Static Mechanistic Localization in LLM Post-Trainingcited by 4
  11. 2026The Labyrinth and the Thread: Rethinking Regularizations in Sequential Knowledge Editing for Large Language Modelscited by 2
  12. 2026Massive Activations in Hybrid Linear Attention Large Language Models: Pre-Attention Spikes and Inter-Spike Plateauscited by 2
15 more
  1. 2026DataShield: Safety-degrading Data Filtering for LLM Benign Instruction Fine-Tuningcited by 1
  2. 2026Reading the Gate, Not the Interference: Output-Side Interference Measurement Does Not Track Merge Collapsecited by 1
  3. 2026From Attribution to Action: A Human-Centered Application of Activation Steeringcited by 0
  4. 2026Freeze Deep, Train Shallow: Interpretable Layer Allocation for Continued Pre-Trainingcited by 0
  5. 2026Latent Reward Steering: An Adaptive Inference-Time Framework that Implicitly Promotes Cognitive Behaviors in Reasoning LLMscited by 0
  6. 2026Temporal Preference Concepts and their Functions in a Large Language Modelcited by 0
  7. 2026Shared Semantics, Divergent Mechanisms: Unsupervised Feature Discovery by Aligning Semantics and Mechanismscited by 0
  8. 2026TIPS: Topological Ill-Posedness Probing and Steering in Large Language Modelscited by 0
  9. 2026Towards Mechanistically Understanding Why Memorized Knowledge Fails to Generalize in Large Language Model Finetuningcited by 0
  10. 2026CircuitKIT : Circuit Discovery, Evaluation, and Application Toolkit for Mechanistic Interpretabilitycited by 0
  11. 2026SAEVerbalizer: Generating Explanations for Sparse Autoencoder Features via Representation Verbalizationcited by 0
  12. 2026Beyond Static Interpretability: Anticipating Post-SFT Mechanisms from Pre-SFT Parameters for Better Tuningcited by 0
  13. 2026Steering Under Compression: Dose-Response, Capability Cost, and Failure Asymmetry in Quantized LLMscited by 0
  14. 2026Don't Judge an LLM Only by Its Activations: Discovering Suppressed Safety Features via Counterfactual Activation Potentialcited by 0
  15. 2026Matching Object or Relation? Tracing Abstract Reasoning Inside VLMscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Mechanistic Interpretability for Large Language Model Alignment: Progress, Challenges, and Future Directions
  2. 2024Towards Unifying Interpretability and Control: Evaluation via Intervention
  3. 2024A Practical Review of Mechanistic Interpretability for Transformer-Based Language Models
  4. 2025Toward a Theory of Generalizability in LLM Mechanistic Interpretability Research
  5. 2024Towards Uncovering How Large Language Model Works: An Explainability Perspective
  6. 2024Mechanistic?
  7. 2024Extracting Interpretable Task-Specific Circuits from Large Language Models for Faster Inference
  8. 2025Mechanistic Indicators of Understanding in Large Language Models