claim
active
claim:loss-level-interventions-let-the-model-decide-how-to-explain-the-loss-avoiding-incentive-to-fight-the-interventionLoss-level interventions let the model decide how to explain the loss, avoiding incentive to fight the intervention
Contrast with layer-specific steering.
Source paper
extracted_from(2026) · Leon Bergen · Usha Bhalla · Sidharth Baskaran · Max Loeffler +13
Neighborhood — ranked by edge-count
Papers (1)
paper
Findings (1)
finding
- Table 10.
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- Extrapolation of scaling predictive models to AGI.
- Splitting the loss into predictive-only and novel-residual terms breaks the competition dynamic.hypothesis0.750Proposed training fix inspired by Kalman filters.
- Interpretation of effectiveness tail result.
- Additional synthetic example of pernicious divergence from balanced subspaces
- Mechanistic interpretation of how activation steering induces deception through the model's reasoning process
- §3 Discussion.
- Central claim motivating the product.
- Acknowledges the confound of not explicitly instructing models to track wealth, yet points to reasoning gaps given code agents avoid errors without prompts.