Steering vector researchers compare notes on model identity arithmetic and thrashing
Sauers_ · x · 2026-10-24
@voooooogel notes observing similar 'identity arithmetic' behavior with steering vectors and asks whether Sauers has seen models thrash in these setups; Sauers replies that 'they seem to thrash less' and says he's looking into it. A substantive exchange between researchers on model behavior under steering.
More from Research
- RoboSimStudio Turns One Phone Video Into an Editable Robot Physics Twin in Under an Hour — lukas_m_ziegler · 2026-10-24
- Methodologist argues placebo observables are a bad sign for IQ as a measurement — RexDouglass · 2026-10-24
- 'LLMs as Human Proxies' paper lands Nature Computational Science cover — iyadrahwan · 2026-10-24
- DOE's Genesis Mission Lists 297 Projects, to Release Open-Weight Foundation Models — ingliguori · 2026-10-24
- Kornia 0.9.0 ships a conventions page targeting LLM coding pitfalls — kornia_foss · 2026-10-24
- Fleuret: teacher forcing turns time iteration into depth iteration, and GPUs love it — francoisfleuret · 2026-10-24