concept
active
concept:thinkedit-interpretable-weight-editing-to-mitigate-overly-short-thinking-in-reasoning-models-sun-et-al-2025bThinkEdit: Interpretable weight editing to mitigate overly short thinking in reasoning models (Sun et al., 2025b)
Related work identifying neurons controlling short-thinking in reasoning models via weight editing
Neighborhood — ranked by edge-count
Papers (1)
paper
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.
- All models exhibit above-baseline representation of the think word when instructed to think about itfinding0.750In the intentional control experiment, all tested models show above-zero cosine similarity to the think word's concept vector.
- Inference-Time Intervention: Eliciting Truthful Answers from a Language Model (Li et al., 2023)concept0.738Safety intervention that relies on activation modification, which ESR might undermine
- Reframes the traditional life/mind hierarchy, supporting panpsychist implications.
- Wait, we don't need to Wait! Removing thinking tokens improves reasoning efficiency (Wang et al., 2025a)concept0.733Baseline comparison method paper for reflection token suppression
- Proof of concept.
- A better basis (e.g., factoring tokens into languages or parts of speech) may yield a sparse interpretable modelhypothesis0.722Same model can be dense in one basis and sparse in another.
- Practical implication.
- Mechanism for how the model modulates representation strength.