# Polymorphic Universal Transformer

Authors: Yilong Chen, Zitian Gao, Yihao Xiao, Jason Klein Liu, Xinyu Yang, Yifan Luo, Tingwen Liu, Haoming Luo, Zhengmao Ye, Ran Tao, Bryan Dai.

Varies which feed-forward units are active across loops of a shared transformer and strengthens nonlinear attention. Studies how this affects hidden-state rank, depth scaling, and the computation needed for reasoning.

First public: 2026-06-23.

[Original paper](https://aclanthology.org/2026.acl-long.1809/) · [Reading-list entry](https://latent-implicit-thinking.github.io/papers/chen2026polymorphic/)
[BibTeX](https://latent-implicit-thinking.github.io/papers/chen2026polymorphic/citation.bib) · [CSL-JSON](https://latent-implicit-thinking.github.io/papers/chen2026polymorphic/citation.csl.json) · [JSON record](https://latent-implicit-thinking.github.io/papers/chen2026polymorphic/record.json)

Date note: ACL Anthology release date. [Source](https://github.com/acl-org/acl-anthology/commit/ba4f39e29e78167a188821ad9e5ce796a7f8b09a).

## Published version

Polymorphic Universal Transformer. Yilong Chen, Zitian Gao, Yihao Xiao, Jason Klein Liu, Xinyu Yang, Yifan Luo, Tingwen Liu, Haoming Luo, Zhengmao Ye, Ran Tao, Bryan Dai. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), 2026.
[Publication](https://aclanthology.org/2026.acl-long.1809/) · [BibTeX](https://latent-implicit-thinking.github.io/papers/chen2026polymorphic/publication.bib) · [CSL-JSON](https://latent-implicit-thinking.github.io/papers/chen2026polymorphic/publication.csl.json)
[Publication record](https://aclanthology.org/2026.acl-long.1809.pdf).

## 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 0.15.2; literature checked through 2026-11-03.