claim
active
claim:planned-words-are-represented-by-ordinary-word-features-rather-than-planning-specific-featuresPlanned words are represented by ordinary word features rather than planning-specific features.
Model thinks about planned words like read words.
Neighborhood — ranked by edge-count
Papers (1)
paper
Claims (1)
claim
- Planning in poems.
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.
- May reflect CLT gaps or non-planning.
- Steering planned-word features only affects the final token when applied at the newline planning token.finding0.754Planning is localized.
- Tested via interventions at the newline planning site.
- Quantitative planning intervention result.
- All models exhibit above-baseline representation of the think word when instructed to think about itfinding0.742In the intentional control experiment, all tested models show above-zero cosine similarity to the think word's concept vector.
- Decoder cosine similarity maps onto concept similarity.
- Illustrative finding from Lindsey et al. 2025 / Hanna and Ameisen 2026 showing intention-like representations in attention streams
- Personifies texts and asks about their desirous relational drives.