[2211.00593] Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 small

Abstract:Research in mechanistic interpretability seeks to explain behaviors of machine learning models in terms of their internal components. However, most previous work either focuses on simple behaviors in small models, or describes complicated behaviors in larger models with broad strokes. In this work, we bridge this gap by presenting an explanation for how GPT-2 small performs a natural language task called indirect object identification (IOI). Our explanation encompasses 26 attention heads grouped into 7 main classes, which we discovered using a combination of interpretability approaches relying on causal interventions. To our knowledge, this investigation is the largest end-to-end attempt at reverse-engineering a natural behavior "in the wild" in a language model. We evaluate the reliability of our explanation using three quantitative criteria--faithfulness, completeness and minimality. Though these criteria support our explanation, they also point to remaining gaps in our understanding. Our work provides evidence that a mechanistic understanding of large ML models is feasible, opening opportunities to scale our understanding to both larger models and more complex tasks.

Abstract:Research in mechanistic interpretability seeks to explain behaviors of machine learning models in terms of their internal components. However, most previous work either focuses on simple behaviors in small models, or describes complicated behaviors in larger models with broad strokes. In this work, we bridge this gap by presenting an explanation for how GPT-2 small performs a natural language task called indirect object identification (IOI). Our explanation encompasses 26 attention heads grouped into 7 main classes, which we discovered using a combination of interpretability approache

related reading

• [2211.00593] Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 smallarxiv.org
• Transformer Circuits Threadtransformer-circuits.pub
• Tracing the thoughts of a large language model \ Anthropicanthropic.com
• An Extremely Opinionated Annotated List of My Favourite Mechanistic Interpretability Papers v2 — AI Alignment Forumalignmentforum.org
• On the Biology of a Large Language Modeltransformer-circuits.pub
• TransformerLens Documentationtransformerlensorg.github.io
• TransformerLens Documentationtransformerlensorg.github.io
• An Extremely Opinionated Annotated List of My Favourite Mechanistic Interpretability Papers v2 — AI Alignment Forumalignmentforum.org
• In-context Learning and Induction Headstransformer-circuits.pub
• Prediction, Explanation, or Over-interpretation?elena-baixy.github.io
• Chapter 1: Transformer Interpretability - ARENAlearn.arena.education
• Language models can explain neurons in language modelsopenaipublic.blob.core.windows.net