J-space, open models — v2
- Anthropic's Verbalizable-Workspace paper showed, on one closed model family, that a model's middle layers carry a dictionary of directions that causally steer its output.
- It left the natural next questions open: how far forward in time the steering reaches, when the structure forms during training, whether it transfers between models, and how it scales.
- We measured all four on open models, then two follow-ups the results forced on us.
Unverified
- Anthropic's Verbalizable-Workspace paper showed, on one closed model family, that a model's middle layers carry a dictionary of directions that causally steer its output.
- It left the natural next questions open: how far forward in time the steering reaches, when the structure forms during training, whether it transfers between models, and how it scales.
- We measured all four on open models, then two follow-ups the results forced on us.
Sources: Eliebak