# Do Halting Heads Really Save Compute in Looped Language Models?

Authors: Tsung-Han Wu, Minsuk Heo, Kenny Tan.

Compares the stopping probabilities predicted by halting heads with the iterations actually executed. Useful early exits require training every iteration to predict tokens and a strong penalty favoring early stopping; batches that wait for their slowest token erase most of the advantage.

First public: 2026-10-23.
ArXiv initial submission: 2026-10-23.

[Original paper](https://arxiv.org/abs/2610.32246) · [PDF](https://arxiv.org/pdf/2610.32246) · [Reading-list entry](https://latent-implicit-thinking.github.io/papers/2610.32246/)
[Preprint BibTeX](https://latent-implicit-thinking.github.io/papers/2610.32246/citation.bib) · [Preprint CSL-JSON](https://latent-implicit-thinking.github.io/papers/2610.32246/citation.csl.json) · [JSON record](https://latent-implicit-thinking.github.io/papers/2610.32246/record.json)

## Approaches

- [Recurrent reasoning](https://latent-implicit-thinking.github.io/topics/recurrent-reasoning/): Repeat parts of the model or keep intermediate information in memory while working through a problem. Some methods also learn how many passes to run.
- [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 0.11.0; literature checked through 2026-10-26.