Interpretability with Sparse Autoencoders (Colab exercises) — LessWrong

Update (13th October 2024) - these exercises have been significantly expanded on. Now there are 2 exercise sets: the first one dives deeply into theo…

x Interpretability with Sparse Autoencoders (Colab exercises) — LessWrong Sparse Autoencoders (SAEs) Exercises / Problem-Sets Interpretability (ML & AI) Superposition AI Frontpage 83 Interpretability with Sparse Autoencoders (Colab exercises) by CallumMcDougall 29th Nov 2023 AI Alignment Forum 4 min read 9 83 Ω 32 Update (13th October 2024) - these exercises have been significantly expanded on. Now there are 2 exercise sets: the first one dives deeply into theoretical topics related to superposition, while the second one (much larger) includes a streamlined version of the first one, as well as

related reading

• An Intuitive Explanation of Sparse Autoencoders for LLM Interpretability | Adam Karvonenadamkarvonen.github.io
• A gentle introduction to sparse autoencoders — LessWronglesswrong.com
• Towards Monosemanticity: Decomposing Language Models With Dictionary Learningtransformer-circuits.pub
• An Intuitive Explanation of Sparse Autoencoders for Mechanistic Interpretability of LLMs — LessWronglesswrong.com
• A Survey on Sparse Autoencoders: Interpreting the Internal Mechanisms of Large Language Modelsarxiv.org
• A gentle introduction to sparse autoencodersnickjiang.substack.com
• pdfopenreview.net
• Toy Models of Superpositiontransformer-circuits.pub
• Matryoshka Sparse Autoencoders — LessWronglesswrong.com
• Sparse Autoencoders Find Highly Interpretable Features in Language Modelsarxiv.org
• Open Source Replication & Commentary on Anthropic's Dictionary Learning Paper — LessWronglesswrong.com
• Llama Scope: Extracting Features from Llama 3.1-8B with SAEsarxiv.org