ESR Rate (metric)

Primary metric: percentage of responses containing multiple attempts that successfully improve on the first attempt

Neighborhood — ranked by edge-count

concept

Endogenous Steering Resistance
implements
The central phenomenon introduced by this paper: inference-time recovery from irrelevant activation steering in LLMs

2-hop · via this concept's ideas

Where ideas in this concept connect to the rest of the corpus — the same concept, an analogy, or a restatement elsewhere.

cosine ≥ 0.65 · no typed edge

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

ESR Testing Pipelinemethod0.768
Three-step protocol: (1) object-level prompting, (2) SAE-latent steering, (3) judge model scoring of attempts
Scale-Dependent ESRconcept0.751
The observed pattern that ESR appears predominantly in the largest model tested, suggesting scale-dependence
Multi-Attempt Rate (metric)concept0.748
Secondary metric: percentage of responses containing multiple attempts, separating surface from actual self-correction
Reflection rateconcept0.742
Ratio of reflection steps to total reasoning steps, used to quantify reflection behavior
Answer Switching Rate (ASR)method0.739
Key evaluation metric: proportion of inputs for which an intervention successfully flips model output
Implicit ESRconcept0.729
Form of ESR occurring without explicit verbal self-interruption markers, not captured by current metrics
Explicit ESRconcept0.725
The form of ESR focused on in this paper, measured by verbal self-interruption phrases as segment boundaries
TC-accept rate metricmethod0.723
Fraction of trade challenges resolved by accepting the face-down offer rather than countering.