AI ALIGNMENT FORUM
Tags
AF

Sparse Autoencoders (SAEs)

•

Applied to Are SAE features from the Base Model still meaningful to LLaVA? by Shan Chen 5d ago

•

Applied to Are SAE features from the Base Model still meaningful to LLaVA? by Shan Chen 5d ago

•

Applied to Mechanistic Interpretability of Llama 3.2 with Sparse Autoencoders by PaulPauls 18d ago

•

Applied to Analyzing how SAE features evolve across a forward pass by bensenberner 1mo ago

•

Applied to SAEs are highly dataset dependent: a case study on the refusal direction by Connor Kissane 1mo ago

•

Applied to Evolutionary prompt optimization for SAE feature visualization by neverix 1mo ago

•

Applied to SAE Probing: What is it good for? Absolutely something! by Subhash Kantamneni 1mo ago

•

Applied to A suite of Vision Sparse Autoencoders by Louka Ewington-Pitsos 1mo ago

•

Applied to SAEs you can See: Applying Sparse Autoencoders to Clustering by Robert_AIZI 2mo ago

•

Applied to On the Practical Applications of Interpretability by Ruben Bloom 2mo ago

•

Applied to It's important to know when to stop: Mechanistic Exploration of Gemma 2 List Generation by Gerard Boxo 2mo ago

•

Applied to Standard SAEs Might Be Incoherent: A Choosing Problem & A “Concise” Solution by Kola Ayonrinde 2mo ago

•

Applied to SAE features for refusal and sycophancy steering vectors by neverix 2mo ago

•

Applied to HDBSCAN is Surprisingly Effective at Finding Interpretable Clusters of the SAE Decoder Matrix by Jaehyuk Lim 2mo ago

•

Applied to Domain-specific SAEs by jacob_drori 2mo ago

•

Applied to An X-Ray is Worth 15 Features: Sparse Autoencoders for Interpretable Radiology Report Generation by hugofry 2mo ago

•

Applied to Exploring SAE features in LLMs with definition trees and token lists by mwatkins 2mo ago

•

Applied to Interpretability of SAE Features Representing Check in ChessGPT by Jonathan Kutasov 2mo ago

•

Applied to Toy Models of Feature Absorption in SAEs by chanind 2mo ago

•

Applied to Do Sparse Autoencoders (SAEs) transfer across base and finetuned language models? by Raymond Arnold 2mo ago