# AlterNet: Adaptive-Depth Mixture-of-Experts with Learned Halting for Elastic Inference

Authors: Alexander Kastius.

Routes each sequence through shallow experts repeatedly, using the current hidden state and depth to choose experts. Trains expert-specific stopping signals alongside the language model so computation can end at different depths.

First public: 2026-11-05.
ArXiv initial submission: 2026-11-05.

[Original paper](https://arxiv.org/abs/2611.08741) · [PDF](https://arxiv.org/pdf/2611.08741) · [HTML](https://arxiv.org/html/2611.08741v1) · [Reading-list entry](https://latent-implicit-thinking.github.io/papers/2611.08741/)
[Preprint BibTeX](https://latent-implicit-thinking.github.io/papers/2611.08741/citation.bib) · [Preprint CSL-JSON](https://latent-implicit-thinking.github.io/papers/2611.08741/citation.csl.json) · [JSON record](https://latent-implicit-thinking.github.io/papers/2611.08741/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.

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