# SEAL: Steerable Reasoning Calibration of Large Language Models for Free

Authors: Runjin Chen, Zhenyu Zhang, Junyuan Hong, Souvik Kundu, Zhangyang Wang.

Separates execution, reflection, and transition steps in written CoT using hidden representations. Adds a steering vector at thought boundaries to adjust how often the model reflects or changes direction, then measures accuracy and token use.

First public: 2025-04-07.
ArXiv initial submission: 2025-04-07.

[Original paper](https://arxiv.org/abs/2504.07986) · [PDF](https://arxiv.org/pdf/2504.07986) · [HTML](https://arxiv.org/html/2504.07986v3) · [Reading-list entry](https://latent-implicit-thinking.github.io/papers/2504.07986/)
[Preprint BibTeX](https://latent-implicit-thinking.github.io/papers/2504.07986/citation.bib) · [Preprint CSL-JSON](https://latent-implicit-thinking.github.io/papers/2504.07986/citation.csl.json) · [JSON record](https://latent-implicit-thinking.github.io/papers/2504.07986/record.json)

## Approaches

- [Analysis and evaluation](https://latent-implicit-thinking.github.io/topics/implicit-reasoning-evaluation/): Measure how well these methods work, investigate what their internal computation does, and identify where they fail. This includes experiments, studies of model internals, and mathematical results.

Reading-list version 0.15.2; literature checked through 2026-11-03.