engels-2025-iclr-publisher
IN premise — summaries/2026/08/24/engels-2024-not-all-features-linear-s4-s-parse-autoencoders-find-multi-d-imensional-features.md
Created 2026-08-25T02:58:01+00:00
The paper 'Not All Language Model Features Are Linear' (Engels et al.) was published at ICLR 2025.
Summary
Engels and colleagues had their work on the geometry of language model features accepted for publication at ICLR 2025, a top-tier machine learning conference. This matters because it establishes that the paper passed peer review at a major venue, giving any downstream reasoning about its findings a credibility baseline rather than resting on an unreviewed preprint.