mlsanityData to 13 Oct 2026

Discovering Variable Binding Circuitry with Desiderata

arXiv 2307.03637 ↗First posted 7 Jul 2023cs.AIcited by 32

Not on the map. Why?

Abstract (arXiv)

Recent work has shown that computation in language models may be human-understandable, with successful efforts to localize and intervene on both single-unit features and input-output circuits. Here, we introduce an approach which extends causal mediation experiments to automatically identify model components responsible for performing a specific subtask by solely specifying a set of textitdesiderata, or causal attributes of the model components executing that subtask. As a proof of concept, we apply our method to automatically discover shared textitvariable binding circuitry in LLaMA-13B, which retrieves variable values for multiple arithmetic tasks. Our method successfully localizes variable binding to only 9 attention heads (of the 1.6k) and one MLP in the final token's residual stream.

Key facts. Cited by 32 papers in the map (data through 13 Oct 2026). Builds on 13 of its 18 references that are in the map. First posted on arXiv 7 Jul 2023.

Builds on

13 of 18 references.

References · 13

  1. 2014Adam: A Method for Stochastic Optimizationcited by 37,093
  2. 2023LLaMA: Open and Efficient Foundation Language Modelscited by 14,072
  3. 2022Locating and Editing Factual Associations in GPTcited by 2,650
  4. 2022Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 smallcited by 1,145
  5. 2022Discovering Latent Knowledge in Language Models Without Supervisioncited by 750
  6. 2023Towards Automated Circuit Discovery for Mechanistic Interpretabilitycited by 711
  7. 2017Learning Sparse Neural Networks through L0 Regularizationcited by 574
  8. 2022Emergent World Representations: Exploring a Sequence Model Trained on a Synthetic Taskcited by 421
  9. 2023Finding Alignments Between Interpretable Causal Variables and Distributed Neural Representationscited by 201
  10. 2022Toward Transparent AI: A Survey on Interpreting the Inner Structures of Deep Neural Networkscited by 127
  11. 2021Inducing Causal Structure for Interpretable Neural Networkscited by 108
  12. 2023Interpretability at Scale: Identifying Causal Mechanisms in Alpacacited by 102
1 more
  1. 2023Characterizing Manipulation from AI Systemscited by 43

Built on by

Cited by year (2026 to 13 Oct)

Citing papers per year: 2023: 3; 2024: 6; 2025: 9; 2026: 14.

32 papers.

Sort:

By area:

  1. 2024Mechanistic Interpretability for AI Safety -- A Reviewcited by 420
  2. 2024Fine-Tuning Enhances Existing Mechanisms: A Case Study on Entity Trackingcited by 142
  3. 2024RAVEL: Evaluating Interpretability Methods on Disentangling Language Model Representationscited by 78
  4. 2025MIB: A Mechanistic Interpretability Benchmarkcited by 50
  5. 2024Recurrent Neural Networks Learn to Store and Generate Sequences using Non-Linear Representationscited by 45
  6. 2025Language Models use Lookbacks to Track Beliefscited by 36
  7. 2023Interpretability Illusions in the Generalization of Simplified Modelscited by 25
  8. 2026Manifold Steering Reveals the Shared Geometry of Neural Network Representation and Behaviorcited by 25
  9. 2024The Quest for the Right Mediator: Surveying Mechanistic Interpretability Through the Lens of Causal Mediation Analysiscited by 24
  10. 2025Mixing Mechanisms: How Language Models Retrieve Bound Entities In-Contextcited by 24
  11. 2023Grokking Group Multiplication with Cosetscited by 22
  12. 2025How Do Transformers Learn Variable Binding in Symbolic Programs?cited by 20
20 more
  1. 2023Look Before You Leap: A Universal Emergent Decomposition of Retrieval Tasks in Language Modelscited by 18
  2. 2026From Directions to Regions: Decomposing Activations in Language Models via Local Geometrycited by 11
  3. 2025Interpretation Meets Safety: A Survey on Interpretation Methods and Tools for Improving LLM Safetycited by 10
  4. 2025How Causal Abstraction Underpins Computational Explanationcited by 10
  5. 2025LLMs Process Lists With General Filter Headscited by 7
  6. 2024Philosophy of Cognitive Science in the Age of Deep Learningcited by 5
  7. 2026Do Language Models Track Entities Across State Changes?cited by 5
  8. 2025Neuroplasticity and Corruption in Model Mechanisms: A Case Study Of Indirect Object Identificationcited by 3
  9. 2026The Dual Mechanisms of Spatial Variable Binding in Vision-Language Modelscited by 3
  10. 2026Data-driven Circuit Discovery for Interpretability of Language Modelscited by 2
  11. 2026Mechanistic origins of catastrophic forgetting: why RL preserves circuits better than SFT?cited by 2
  12. 2025Constructive Circuit Amplification: Improving Math Reasoning in LLMs via Targeted Sub-Network Updatescited by 1
  13. 2026Test, then Route: How Language Models Execute In-Context Conditional Rules Across Models and Languagescited by 1
  14. 2026Who Says What: Symbolic Trimodal Binding Mechanisms in Audio-Visual LLMscited by 1
  15. 2026Cultural Binding Heads in Language Modelscited by 0
  16. 2026Assign and Add: A Mechanistic Study of Compositional Arithmeticcited by 0
  17. 2026Multi-component Causal Tracing in Large Language Modelscited by 0
  18. 2026Vision-Language Models are Fragile Multilingual Associatorscited by 0
  19. 2026Syntax vs. Semantics: How Transformers Learn Deep Dependenciescited by 0
  20. 2026Causal and Interpretable Structures in LLM Compositional Taskscited by 0