Attribution Graphs
Attribution graphs are a mechanistic interpretability technique developed by Anthropic that traces the internal "circuits" a large language model uses to turn a specific prompt into a specific output.
Anthropic
Explore Anthropic through related topics and the articles other pages reference most.
Explore articles
Browse subtopics: Interpretability
Articles that also belong to these categories. Counts cover all of Anthropic.
All articles (110)Large Language Models (31)AI Models (28)AI Tools & Products (21)AI Safety (16)AI Agents (14)Developer Tools (10)AI Research (9)Interpretability (8)People (7)AI Alignment (6)AI Policy & Regulation (5)AI for Science (5)Artificial Intelligence (5)Enterprise AI (5)Reasoning Models (4)AI Code Generation (3)AI Companies (3)AI Incidents & Controversies (3)AI Infrastructure (3)Natural Language Processing (3)AI Ethics (2)AI Inference (2)Information Retrieval (2)Open Source AI (2)Data Centers (1)Defense AI (1)Drug Discovery (1)Education AI (1)Finance AI (1)Generative AI (1)Google (1)Machine Learning (1)Mathematics (1)Model Evaluation (1)Multimodal AI (1)OpenAI (1)
Showing 1-8 of 8 articles
Christopher Olah
Christopher Olah (commonly Chris Olah) is a Canadian machine learning researcher, a co-founder of Anthropic, and the researcher most often credited with founding mechanistic interpretability
Crosscoder
A crosscoder is a mechanistic interpretability tool, introduced by Anthropic in October 2024, that generalizes the sparse autoencoder (SAE) and the transcoder by learning a single shared dictionary of sparse…
Golden Gate Claude
Golden Gate Claude was a temporary, research-oriented public demonstration released by Anthropic on May 23, 2024
On the Biology of a Large Language Model
On the Biology of a Large Language Model is a mechanistic interpretability paper published by Anthropic on March 27, 2025, in the Transformer Circuits Thread.
Scaling Monosemanticity
Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnet is the May 21
Towards Monosemanticity
Towards Monosemanticity is an October 2023 mechanistic interpretability paper from Anthropic that used a sparse autoencoder to decompose the internal activations of a small language model into thousands of…
Toy Models of Superposition
Toy Models of Superposition is a September 2022 mechanistic interpretability paper from Anthropic that shows how a neural network can represent more features than it has dimensions by packing them into…