Mechanistic Interpretability

Follow this topic Ask about this topic

The field seeks to explain how a model’s internal computations produce particular outputs, rather than relying only on input-output behavior. Researchers use sparse autoencoders and transcoders to identify features, and probes, attribution graphs, activation patching, ablation, and steering to trace or test their roles. Studies examine language-model tasks such as factual recall, reasoning, and refusal, as well as vision, audio, and scientific models. A central challenge is that a signal can be easy to decode without being causally used, while circuit explanations and interventions can vary across prompts, models, and analysis choices.

Latest papers

391 papers63 papers in the last four weeks, up 75% on the four weeks before. 0.4% of all new papers.

Beneath the Surface of Chains-of-Thought: A Mechanistic Interpretation of Reasoning Operations in LLMs
Sep 4, 2026Seogyeong Jeong, Jaehui Hwang, Dongyoon Han +3LLM Interpretability LLM Reasoning
KAIST · NAVER AI Lab
How Do Language Models Represent and Use Phonological Information for Allomorph Selection?
Sep 4, 2026Sangwoo Kim, Sangah LeeRepresentation Geometry in Language Models Mechanistic Interpretability
Department of Linguistics Seoul National University
Tracing Audio Grounding and Answer Selection in Audio LLMs
Sep 4, 2026Hyebin Cho, Suho Yoo, Jihoo Jung +1Audio-Language Models Audio QA
Korea Advanced Institute of Science and Technology, South Korea
The Anatomy of an ASR Hallucination
Sep 3, 2026Hamees Sayed, Apoorv Singh, Kumar Aman +1Robust ASR Conformer
Smallest AI · Fast Code AI
Beyond Shallow Alignment: How Post-Training Methods Determine Refusal Circuits And Steering Robustness
Sep 3, 2026Hoang Cuong Nguyen, Mark Dras, Usman NaseemMechanistic Interpretability Circuits in Language Models
Macquarie University, Sydney, Australia
ObserverBench: Testing Mechanistic Estimates for Intervention and Control
Sep 2, 2026Vijay ErramilliMechanistic Interpretability AI Safety Evaluation
Knowing, and Saying It Only When Asked: LLM Endognostics and the Schizognosis of Minerva-7B
Sep 2, 2026Fabrizio Davide, Francesco CollovaLanguage Model Auditing Mechanistic Interpretability
Independent Researcher, Bacoli (Napoli), Italy
Encoded but Disconnected: Decomposing Vision-Language Model Failures under a Patching Null
Sep 1, 2026Genpei ZhangVLM Interpretability Mechanistic Interpretability
University of Wisconsin–Madison
Beyond Scores: Understanding LLM-as-a-Judge Mechanisms in Summarization Evaluation
Sep 1, 2026Himil Vasava, Ming JiangLLM-as-a-Judge Language Model Generation Evaluation
University of Wisconsin-Madison
Separating Syntax from Language: A Mechanistic Account of Translation in Multilingual LLMs
Sep 1, 2026Mikhail Sonkin, Tanja Baeumel, Daniil Gurgurov +2Multilingual Language Models Mechanistic Interpretability
Saarland University · University of Göttingen · German Research Center for Artificial Intelligence (DFKI) +1
S^3martCirc: Self-supervised Smart Circuit Discovery
Sep 1, 2026Wendy Zheng, Yinhan He, Liang Wu +1LLM Interpretability Mechanistic Interpretability
University of Virginia · Nokia
Heard but Not Heeded: Paralinguistic Information Encoding and Loss in Audio-Language Models
Sep 1, 2026Bhuvan Koduru, Dareen Safar B Alharthi, Rita Singh +1Audio-Language Model Evaluation Audio-Language Models
Language Technologies Institute Carnegie Mellon University
MURANO: Design, Run, and Reproduce Mechanistic Interpretability Experiments as Composable Pipelines
Aug 31, 2026Alireza Bayat Makou, Emirhan Böge, Phu Gia Hoang +5LLM Interpretability Interpretable ML
Ubiquitous Knowledge Processing (UKP) Lab, Technical University of Darmstadt, Germany · Cluster of Excellence “Reasonable Artificial Intelligence” (RAI), hessian.AI, Germany · National Research Center for Applied Cybersecurity ATHENE, Germany +1
Interpreting and Steering for Safe and Correct Code Generation
Aug 30, 2026Hao Yan, Ziyu YaoMechanistic Interpretability LLM Security
Department of Computer Science George Mason University, Fairfax, VA
When Safety Speaks a Language: A Mechanistic Analysis of Safety-Language Identity Entanglement in LLMs
Aug 30, 2026Apoorva Upadhyaya, Sandipan SikdarLLM Safety Alignment Mechanistic Interpretability
L3S Research Center, Leibniz Universität Hannover, Germany
Dissecting Hierarchical Reasoning Models: A Mechanistic Study
Aug 30, 2026Leo Raphael Rodrigues, Jian KangLLM Reasoning Mechanistic Interpretability
Mohamed bin Zayed University of Artificial Intelligence
From Detection to Refusal: Safer LLMs via Circuit-Guided Weight Scaling
Aug 30, 2026Kuan-Lin Chu, Chung-En Sun, Tsui-Wei WengLLM Safety Refusal Behavior in Language Models
Computer Science and Engineering, University of California San Diego · Halıcıoğlu Data Science Institute, University of California San Diego
A Causal Model for Locating and Unlocking Sandbagging in Model Organisms
Aug 29, 2026Hong Kiat Tan, Linh Le, David Williams-KingLanguage Model Auditing Causal Interventions in Language Models
University of California, Los Angeles · Lida Safety · ERA
A Unifying Perspective on Language Model Representations: From Filler-Role Structure to Mechanistic Interpretability
Aug 29, 2026Zhang Enyan, R. Thomas McCoyLLM Interpretability Mechanistic Interpretability
Department of Computer Science Yale University New Haven, CT · Department of Linguistics & Wu Tsai Institute Yale University New Haven, CT
LongGuard: Mechanistic Analysis and Training-Free Mitigation of Long-Context Failure in Safety Guardrails
Aug 27, 2026Ziyang Chen, Xing Wu, Songlin HuLanguage Model Safety Evaluation Long-Context Language Model Evaluation
Institute of Information Engineering, Chinese Academy of Sciences, Beijing, China · School of Cyber Security, University of Chinese Academy of Sciences, Beijing, China
Circuit Condensation: Post-Training that Concentrates a Behavior's Causal Circuit
Aug 27, 2026Sai Adith Senthil KumarMechanistic Interpretability Circuits in Language Models
George Mason University
How Do Linear Probes Emerge? A Circuit-Tracing Framework with Concept-Targeted Attribution
Aug 27, 2026Vedant Palit, Florent Draye, Terry Jingchen Zhang +2Mechanistic Interpretability Feature Attribution
Jinesis Lab, University of Toronto & Vector Institute · EuroSafeAI · Max Planck Institute for Intelligent Systems, Tübingen, Germany +2
Interpreting Latent Protein Language Model Features with Geometric Annotations
Aug 26, 2026Siddharth Setlur, Djordje Mihajlovic, Darrick LeeMechanistic Interpretability Protein Language Models
School of Mathematics University of Edinburgh, UK
Finding and using interpretable latents in a neutrino foundation model with sparse autoencoders
Aug 26, 2026Raphaël Bonnet-Guerrini, Johann Ioannou-Nikolaides, Inar Timiryasov +1Mechanistic Interpretability Uncertainty Quantification
Computer Science Department, Università degli Studi di Milano, Milano, Italy · INFN Sezione di Milano, Milano, Italy · Niels Bohr Institute, University of Copenhagen, Copenhagen, Denmark
RACE: Scalable Statistical Estimation of Functional Consistency in LLM Neurons
Aug 25, 2026Runyu Wang, Bo Liu, Xiaxin Zhang +6Transformer Interpretability LLM Interpretability
School of Transportation and Civil Engineering, Nantong University · Chongqing University of Post and Telecommunications · China Southern Power Grid Company Limited +1
Beyond Static Interpretability: Anticipating Post-SFT Mechanisms from Pre-SFT Parameters for Better Tuning
Aug 25, 2026Hang Chen, Jiaying Zhu, Wenya WangMechanistic Interpretability Supervised Fine-Tuning
College of Computing and Data Science Nanyang Technological University · School of Computer Science and Engineering The Chinese University of Hong Kong
Can We Read the Mind of an Audio LLM? A Verbalizable, Multilingual Middle-Layer Workspace
Aug 25, 2026Jiajun Fan, Jingyuan Li, Prashanth Gurunath Shivakumar +8Audio-Language Models LLM Interpretability
Amazon AGI Foundations · University of Illinois Urbana-Champaign
Mechanistic Circuit Identification for Controllable Data Generation
Aug 25, 2026Nakyung Lee, Sangwoo Hong, Jungwoo LeeSynthetic Data Generation for Language Models Controllable Text Generation
Department of Electrical and Computer Engineering, Seoul National University · Department of Computer Science and Engineering, Konkuk University
What AstroPT knows about galaxies, and what that can teach us about LLMs
Aug 23, 2026UniverseTBD, :, Kshitij Duraphe +3Mechanistic Interpretability Language Model Probing
IUCAA Pune, India
Align, Unify, Suppress, Route: A Coherentist View of Transformer Computation
Aug 22, 2026Nura Aljaafari, Andre FreitasTransformer Interpretability Mechanistic Interpretability
University of Manchester, United Kingdom · Idiap Research Institute, Switzerland

Sep 4, 2026J/K move · Enter open · S save

Beneath the Surface of Chains-of-Thought: A Mechanistic Interpretation of Reasoning Operations in LLMs

Seogyeong Jeong, Jaehui Hwang, Dongyoon Han, Geonmo Gu+2

KAIST · NAVER AI Lab

LLM Interpretability LLM Reasoning CoT Reasoning Representation Geometry in Language Models Mechanistic Interpretability

Open paper AskPDF arXiv

Reasoning in large language models unfolds through diverse functional operations, such as problem formulation, goal decomposition, and deduction. Although these operations are explicitly distinguished in text, little is known about how they are geometrically organized in representation spaces. To this end, we investigate whether distinct reasoning operations exhibit corresponding geometric structure in hidden representations. We find that operations are separable in held-out representations, with separability peaking in middle layers, and verify that this structure is not explained by lexical or positional confounds. Across layers, token-wise operation-alignment becomes more distributed over spans, while identical surface tokens are represented differently depending on the operation of its surrounding chunk. Attention-masking interventions further show that operation-aligned representations at chunk onset depend on preceding reasoning context. Consequently, our work demonstrates that language models maintain representational correspondence between linguistic reasoning expressions and their internal geometric structures. Code and project materials are available at https://github.com/naver-ai/beneath-cot.