mlsanityData to 14 Oct 2026

GPT-2 Through the Lens of Vector Symbolic Architectures

arXiv 2412.07947 ↗First posted 10 Dec 2024cs.LG cs.AIcited by 3

On the map: Mechanistic interpretability → Circuits and mechanistic analysis

Abstract (arXiv)

Understanding the general priniciples behind transformer models remains a complex endeavor. Experiments with probing and disentangling features using sparse autoencoders (SAE) suggest that these models might manage linear features embedded as directions in the residual stream. This paper explores the resemblance between decoder-only transformer architecture and vector symbolic architectures (VSA) and presents experiments indicating that GPT-2 uses mechanisms involving nearly orthogonal vector bundling and binding operations similar to VSA for computation and communication between layers. It further shows that these principles help explain a significant portion of the actual neural weights.

Key facts. Cited by 3 papers in the map (data through 14 Oct 2026). Placed in Circuits and mechanistic analysis, one of 13 directions in Mechanistic interpretability. Builds on 21 of its 28 references that are in the map. First posted on arXiv 10 Dec 2024.

Builds on

21 of 28 references.

Closest ancestors · 1

Same direction, newest first

  1. 2022Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 small1 Nov

Other references · 20

  1. 2017Attention Is All You Needcited by 60,361
  2. 2013Distributed Representations of Words and Phrases and their Compositionalitycited by 3,474
  3. 2016Understanding intermediate layers using linear classifier probescited by 1,353
  4. 2019BERT Rediscovers the Classical NLP Pipelinecited by 1,059
  5. 2024Refusal in Language Models Is Mediated by a Single Directioncited by 1,032
  6. 2022Toy Models of Superpositioncited by 971
  7. 2023The Linear Representation Hypothesis and the Geometry of Large Language Modelscited by 716
  8. 2023The Geometry of Truth: Emergent Linear Structure in Large Language Model Representations of True/False Datasetscited by 712
  9. 2019What do you learn from context? Probing for sentence structure in contextualized word representationscited by 438
  10. 2022Emergent World Representations: Exploring a Sequence Model Trained on a Synthetic Taskcited by 431
  11. 2024The Faiss librarycited by 376
  12. 2023Finding Neurons in a Haystack: Case Studies with Sparse Probingcited by 334
8 more
  1. 2020Understanding the Role of Individual Units in a Deep Neural Networkcited by 255
  2. 2023How does GPT-2 compute greater-than?: Interpreting mathematical abilities in a pre-trained language modelcited by 202
  3. 2023Linearity of Relation Decoding in Transformer Language Modelscited by 200
  4. 2018What Is One Grain of Sand in the Desert? Analyzing Individual Neurons in Deep NLP Modelscited by 124
  5. 2020Analyzing Individual Neurons in Pre-trained Language Modelscited by 97
  6. 2024On the Origins of Linear Representations in Large Language Modelscited by 79
  7. 2021Vector Symbolic Architectures as a Computing Framework for Emerging Hardwarecited by 35
  8. 2015Representing Objects, Relations, and Sequencescited by 24

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 1; 2026: 2.

3 papers.

  1. 2025Hyperdimensional Probe: Decoding LLM Representations via Vector Symbolic Architecturescited by 3
  2. 2026Sparse Inter-Layer Dependencies of Transformer FFN Neuronscited by 1
  3. 2026The Emergent Symbolic Structure of Artificial Neural Networkscited by 1

Similar, not cited

Close in wording, no citation either way

  1. 2026From Correlation to Cause: A Five-Stage Methodology for Feature Analysis in Transformer Language Models
  2. 2026Change the Product, Keep the Parameters: Associative Algebra Layers for Transformers
  3. 2026Causal Dimensionality of Transformer Representations: Measurement, Scaling, and Layer Structure
  4. 2025Beyond Components: Singular Vector-Based Interpretability of Transformer Circuits
  5. 2023A Survey of Techniques for Optimizing Transformer Inference
  6. 2022SpeedLimit: Neural Architecture Search for Quantized Transformer Models
  7. 2026Transformer Field Theory: A Response-Theoretic Approach to Mechanistic Interpretability
  8. 2025Activation Transport Operators