claim
active
claim:post-training-interventions-need-a-notion-of-concept-structure-hierarchy-rather-than-flat-independent-axesPost-training interventions need a notion of concept structure (hierarchy) rather than flat independent axes
Discussion conclusion.
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
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.
- Central interpretive claim and motivation for future work
- Is the Assistant Axis formed during post-training or inherited from representations learned during pre-training?question0.775Motivates the base model steering experiments in §3.2.2
- Validation finding from Lu et al. 2026 supporting PSM's claim about pre-training persona structure
- Context-independent transformation.
- Observation of catastrophic performance drop when steering certain concepts.
- Base pretrained models show high false positive rates and achieve no net task performance on concept injection detection; post-training essential for introspection.
- Post-training causes the Assistant's perspective to take over increasing J-space capacity.claim0.760Base vs post-trained comparison.
- Interpretive claim from Case Study II about the distinction between correlational probes and causal interventions