mlsanityData to 14 Oct 2026

Towards Vision-Language Mechanistic Interpretability: A Causal Tracing Tool for BLIP

arXiv 2308.14179 ↗First posted 27 Aug 2023cs.CL cs.AI cs.CVcited by 62

On the map: Mechanistic interpretability → Interpreting vision and multimodal models

Abstract (arXiv)

Mechanistic interpretability seeks to understand the neural mechanisms that enable specific behaviors in Large Language Models (LLMs) by leveraging causality-based methods. While these approaches have identified neural circuits that copy spans of text, capture factual knowledge, and more, they remain unusable for multimodal models since adapting these tools to the vision-language domain requires considerable architectural changes. In this work, we adapt a unimodal causal tracing tool to BLIP to enable the study of the neural mechanisms underlying image-conditioned text generation. We demonstrate our approach on a visual question answering dataset, highlighting the causal relevance of later layer representations for all tokens. Furthermore, we release our BLIP causal tracing tool as open source to enable further experimentation in vision-language mechanistic interpretability by the community. Our code is available at https://github.com/vedantpalit/Towards-Vision-Language-Mechanistic-Interpretability.

Key facts. Cited by 62 papers in the map (data through 14 Oct 2026). Placed in Interpreting vision and multimodal models, one of 13 directions in Mechanistic interpretability. Builds on 32 of its 45 references that are in the map. First posted on arXiv 27 Aug 2023.

Builds on

32 of 45 references.

References · 32

  1. 2020Language Models are Few-Shot Learnerscited by 32,419
  2. 2014Microsoft COCO: Common Objects in Contextcited by 17,608
  3. 2022BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generationcited by 4,322
  4. 2022Locating and Editing Factual Associations in GPTcited by 2,691
  5. 2022Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 smallcited by 1,165
  6. 2019BERT Rediscovers the Classical NLP Pipelinecited by 1,059
  7. 2022Mass-Editing Memory in a Transformercited by 1,001
  8. 2022Toy Models of Superpositioncited by 971
  9. 2022In-context Learning and Induction Headscited by 964
  10. 2023Progress measures for grokking via mechanistic interpretabilitycited by 839
  11. 2023Towards Automated Circuit Discovery for Mechanistic Interpretabilitycited by 724
  12. 2022Winoground: Probing Vision and Language Models for Visio-Linguistic Compositionalitycited by 529
20 more
  1. 2018What you can cram into a single vector: Probing sentence embeddings for linguistic propertiescited by 457
  2. 2021Causal Abstractions of Neural Networkscited by 422
  3. 2023Finding Neurons in a Haystack: Case Studies with Sparse Probingcited by 334
  4. 2013Direct and Indirect Effectscited by 280
  5. 2018Analysis Methods in Neural Language Processing: A Surveycited by 277
  6. 2016Fine-grained Analysis of Sentence Embeddings Using Auxiliary Prediction Taskscited by 216
  7. 2023Causal Abstraction: A Theoretical Foundation for Mechanistic Interpretabilitycited by 205
  8. 2023Localizing Model Behavior with Path Patchingcited by 195
  9. 2023Birth of a Transformer: A Memory Viewpointcited by 161
  10. 2023A Toy Model of Universality: Reverse Engineering How Networks Learn Group Operationscited by 144
  11. 2015Exploring Models and Data for Image Question Answeringcited by 139
  12. 2023Interpretability at Scale: Identifying Causal Mechanisms in Alpacacited by 103
  13. 2018Under the Hood: Using Diagnostic Classifiers to Investigate and Improve how Language Models Track Agreement Informationcited by 97
  14. 2020Behind the Scene: Revealing the Secrets of Pre-trained Vision-and-Language Modelscited by 90
  15. 2022Linear Connectivity Reveals Generalization Strategiescited by 56
  16. 2021A Review on Explainability in Multimodal Deep Neural Netscited by 46
  17. 2022MultiViz: Towards Visualizing and Understanding Multimodal Modelscited by 36
  18. 2023Scale Alone Does not Improve Mechanistic Interpretability in Vision Modelscited by 18
  19. 2021How Transferable are Reasoning Patterns in VQA?cited by 16
  20. 2023Semantic Composition in Visually Grounded Language Modelscited by 1

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2023: 1; 2024: 14; 2025: 23; 2026: 24.

62 papers.

Sort:

By area:

  1. 2024Mechanistic Interpretability for AI Safety -- A Reviewcited by 422
  2. 2024Devils in Middle Layers of Large Vision-Language Models: Interpreting, Detecting and Mitigating Object Hallucinations via Attention Lenscited by 143
  3. 2024Towards Interpreting Visual Information Processing in Vision-Language Modelscited by 122
  4. 2024Interpreting and Editing Vision-Language Representations to Mitigate Hallucinationscited by 104
  5. 2024Cross-modal Information Flow in Multimodal Large Language Modelscited by 67
  6. 2025CF-VLM:CounterFactual Vision-Language Fine-tuningcited by 59
  7. 2024Explainable and Interpretable Multimodal Large Language Models: A Comprehensive Surveycited by 52
  8. 2024A Concept-Based Explainability Framework for Large Multimodal Modelscited by 39
  9. 2025Visual Large Language Models for Generalized and Specialized Applicationscited by 36
  10. 2025A Survey on Mechanistic Interpretability for Multi-Modal Foundation Modelscited by 34
  11. 2025Token Activation Map to Visually Explain Multimodal LLMscited by 32
  12. 2024What Do VLMs NOTICE? A Mechanistic Interpretability Pipeline for Gaussian-Noise-free Text-Image Corruption and Evaluationcited by 30
38 more
  1. 2025How Visual Representations Map to Language Feature Space in Multimodal LLMscited by 25
  2. 2024Finding Visual Task Vectorscited by 24
  3. 2025Causal Tracing of Object Representations in Large Vision Language Models: Mechanistic Interpretability and Hallucination Mitigationcited by 23
  4. 2024Decomposing and Editing Predictions by Modeling Model Computationcited by 19
  5. 2024Beyond the Doors of Perception: Vision Transformers Represent Relations Between Objectscited by 13
  6. 2024Performance Gap in Entity Knowledge Extraction Across Modalities in Vision Language Modelscited by 12
  7. 2025When Seeing Overrides Knowing: Disentangling Knowledge Conflicts in Vision-Language Modelscited by 11
  8. 2026Arbitration Failure, Not Perceptual Blindness: How Vision-Language Models Resolve Visual-Linguistic Conflictscited by 11
  9. 2025Map the Flow: Revealing Hidden Pathways of Information in VideoLLMscited by 10
  10. 2025From Behavioral Performance to Internal Competence: Interpreting Vision-Language Models with VLM-Lenscited by 9
  11. 2025V-SEAM: Visual Semantic Editing and Attention Modulating for Causal Interpretability of Vision-Language Modelscited by 8
  12. 2025Understanding Multi-View Transformerscited by 8
  13. 2025Unboxing the Black Box: Mechanistic Interpretability for Algorithmic Understanding of Neural Networkscited by 6
  14. 2026Interpretable Debiasing of Vision-Language Models for Social Fairnesscited by 6
  15. 2025Beyond Spurious Signals: Debiasing Multimodal Large Language Models via Counterfactual Inference and Adaptive Expert Routingcited by 5
  16. 2025REMA: A Unified Reasoning Manifold Framework for Interpreting Large Language Modelcited by 5
  17. 2026Causal Tracing of Audio-Text Fusion in Large Audio Language Modelscited by 5
  18. 2024Who Brings the Frisbee: Probing Hidden Hallucination Factors in Large Vision-Language Model via Causality Analysiscited by 4
  19. 2026Where Does Vision Meet Language? Understanding and Refining Visual Fusion in MLLMs via Contrastive Attentioncited by 3
  20. 2026Vision-Default, Prior-Override: Causal Mechanisms of Perception-Knowledge Conflict in Vision-Language Modelscited by 3
  21. 2025Causality Is Key to Understand and Balance Multiple Goals in Trustworthy ML and Foundation Modelscited by 2
  22. 2025Self-Aware Safety Augmentation: Leveraging Internal Semantic Understanding to Enhance Safety in Vision-Language Modelscited by 2
  23. 2025BASIC: Boosting Visual Alignment with Intrinsic Refined Embeddings in Multimodal Large Language Modelscited by 2
  24. 2025Reading Images Like Texts: Sequential Image Understanding in Vision-Language Modelscited by 2
  25. 2025Exploring the Effectiveness and Interpretability of Texts in LLM-based Time Series Modelscited by 1
  26. 2025Multimodal Function Vectors for Visual Relationscited by 1
  27. 2026Towards Understanding Multimodal Fine-Tuning: Spatial Featurescited by 1
  28. 2026Mechanisms of Object Localization in Vision-Language Modelscited by 1
  29. 2026What Do They See? Interpreting Complex Road Scenarios Through the Eyes of Vision-Language-Action Models for Safe and Trustworthy Autonomous Vehicle Learningcited by 1
  30. 2023Trustworthy Large Models in Vision: A Surveycited by 0
  31. 2024Identifying Sub-networks in Neural Networks via Functionally Similar Representationscited by 0
  32. 2025IsoNet: Causal Analysis of Multimodal Transformers for Neuromuscular Gesture Classificationcited by 0
  33. 2025Decoding the Multimodal Maze: A Systematic Review on the Adoption of Explainability in Multimodal Attention-based Modelscited by 0
  34. 2025Interpret, prune and distill Donut : towards lightweight VLMs for VQA on documentcited by 0
  35. 2026A Vision for Multisensory Intelligence: Sensing, Science, and Synergycited by 0
  36. 2026Responsible AI for General-Purpose Systems: Overview, Challenges, and A Path Forwardcited by 0
  37. 2026Mind the Gap No More: Achieving Zero-Gap Multimodal Integration via One Tokenizercited by 0
  38. 2026NAST: Improving Negation Handling in Medical Vision-Language Models through Negation-Aware Selective Trainingcited by 0

50 most cited of 62 All 62 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2026Representation as a Bottleneck for Mechanistic Interpretability: The Manifestation Unit Protocol
  2. 2024Understanding Multimodal LLMs: the Mechanistic Interpretability of Llava in Visual Question Answering
  3. 2026Transcoders Trace Visual Grounding and Hallucinations in Vision-Language Models
  4. 2025Prisma: An Open Source Toolkit for Mechanistic Interpretability in Vision and Video
  5. 2026MechaTerp-TRACE: A Novel Approach for Component Ablation Analysis in Language Models
  6. 2026The Abstraction Gap in Vision-Language Causal Reasoning
  7. 2024Causal Graphical Models for Vision-Language Compositional Understanding
  8. 2025Emergence of Text Readability in Vision Language Models