Features have life history. And we should care

arXiv:2605.18789v1 Announce Type: new
Abstract: Features in language models have life history: they emerge, persist, and die during training, yet the importance of that history remains largely unexplored. We find evidence of a persistent representational backbone, which we identify in Pythia-160M and -410M as the carrier scaffold: $sim50$ sparse features with stable life histories, around which the model’s representational structure organises. It has four properties. emph(i)~emphIt assembles early: features emerge, die, and reorganise $sim40!times$ faster in the first $1%$ of training than afterwards, and the scaffold is already largely fixed by then. emph(ii)~emphIt is load-bearing: joint cross-layer ablation identifies the carriers as far more load-bearing than any count-matched non-scaffold population, a gap invisible to per-firing single-feature methods. emph(iii)~emphFunction precedes direction: which features will become carriers is already predictable from training-onset firing patterns alone, correctly distinguishing future carriers from non-carriers in $4$ of $5$ cases, before the geometry has settled. emph(iv)~emphIt seeds subsequent development: by the end of training, scaffold carriers have recruited $64%$ of all active features into the scaffold hierarchy. Life history is consistent with a two-phase account of training: selection appears to largely determine the scaffold in the first $1%$; the remaining $99%$ appears to calibrate geometry around a substrate already set.