Monet: Mixture of Monosemantic Experts for Transformers Explained — LessWrong

Note: This is an exposition of the recent preprint "Monet: Mixture of Monosemantic Experts for Transformers". I wrote this exposition as my project for the ARBOx program, I was not involved in writing the paper. Any errors are my own. Thank you to @David Quarel for his excellent comments and suggestions. TL;DR: MONET is a novel neural network architecture that achieves interpretability by design rather than through post-hoc analysis. Using a specialized Mixture of Experts (MoE) architecture with ~250k experts per layer, MONET encourages individual components to learn specific, interpretable tasks (e.g., Python coding, biology knowledge, or toxicity generation). This enables selective removal of capabilities without harming performance in other domains. I show that MONET can be framed as a standard transformer MLP with two key modifications: (1) forced sparse activations in the hidden layer and (2) dynamic reshuffling of the hidden layer. This interpretation helps explain how MONET achi

x Monet: Mixture of Monosemantic Experts for Transformers Explained — LessWrong Academic Papers Interpretability (ML & AI) Transformers AI Frontpage 31 Monet: Mixture of Monosemantic Experts for Transformers Explained by CalebMaresca 25th Jan 2025 13 min read 2 31 Note: This is an exposition of the recent preprint " Monet: Mixture of Monosemantic Experts for Transformers ". I wrote this exposition as my project for the ARBOx program , I was not involved in writing the paper. Any errors are my own. Thank you to @David Quarel for his excellent comments and suggestions. TL;DR: MONET is a novel ne

saved by

• Idhant Gulati

related reading

• Towards Monosemanticity: Decomposing Language Models With Dictionary Learningtransformer-circuits.pub
• A Mathematical Framework for Transformer Circuitstransformer-circuits.pub
• Very Simple MoE Intro1a3orn.com
• Papers I’ve read this week, Mixture of Experts editionfinbarrtimbers.substack.com
• Transformer Circuits Threadtransformer-circuits.pub
• Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnettransformer-circuits.pub
• An Extremely Opinionated Annotated List of My Favourite Mechanistic Interpretability Papers v2 — AI Alignment Forumalignmentforum.org
• Mixture of experts - Wikipediaen.wikipedia.org
• [2202.08906] ST-MoE: Designing Stable and Transferable Sparse Expert Modelsar5iv.labs.arxiv.org
• Softmax Linear Unitstransformer-circuits.pub
• Transformer Explainer: LLM Transformer Model Visually Explainedpoloclub.github.io
• Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnettransformer-circuits.pub