Interpretability researcher maxsloef leaves Goodfire to study how LLM personas form
maxsloef · x · 2026-11-01
Interpretability researcher maxsloef announced he has left Goodfire. He outlined the layerwise picture of LLMs — early layers translate tokens into concepts, middle layers compute with them, late layers translate concepts back into tokens — and said his real interest is understanding how a model's character/persona is decided, aiming at interventions that could edit model personality.
More from Companies & People
- Chamath's word count of Anthropic's risk factors: compute tops the list, customers barely mentioned — rohanpaul_ai · 2026-11-01
- Apple CEO: 'We don't want to ship AI for AI's sake' — TechCrunch AI · 2026-11-01
- ControlAI Sponsors The Infographics Show's "AI Bubble" Videos, Sparking Bias Allegations — jessi_cata · 2026-11-01
- Ackman: Brain-computer interfaces will restore sight in years; warns neoclouds are leveraged plays — elonmusk · 2026-11-01
- DeepMind's Gemini Carbon launch lands with a whimper as community shrugs — brandon_galang · 2026-11-01
- Google Maps exists because Bret Taylor rewrote all its code in one weekend — DynamicWebPaige · 2026-11-01