https://rbellec.github.io/blog/en/categories/interpretability/Recent content in Interpretability on Raphaël BellecHugo -- gohugo.ioen© 2026 Raphaël BellecWed, 02 Sep 2026 00:00:00 +0000https://rbellec.github.io/blog/en/posts/llm-toki-pona-and-interpretability/Wed, 02 Sep 2026 00:00:00 +0000https://rbellec.github.io/blog/en/posts/llm-toki-pona-and-interpretability/<p><em>Translated from the <a href="https://rbellec.github.io/blog/posts/llm-toki-pona-and-interpretability/" target="_blank" rel="noreferrer">French original</a> with AI assistance, then reviewed by the author.</em></p> <p>Could a language of 140 words help us better understand what happens inside an LLM? In every discipline that deals with complex systems, a <em>simple</em> model has driven enormous advances: <em>E. coli</em> and the fruit fly in biology, the harmonic oscillator in physics.</p>