# TAC-CoT: Tool-Empowered Implicit Chain-of-Thought via Interleaving Explicit Tokens

Authors: Juseon Do, Sangmin Bae, Jeesu Jung, Hwanjun Song.

Interleaves continuous hidden-state reasoning with written tool calls and observations. Learns from explicit tool-use traces through reconstruction and self-distillation, and chooses when to call a tool or answer after each reasoning block.

First public: 2026-11-01.
ArXiv initial submission: 2026-11-01.

[Original paper](https://arxiv.org/abs/2611.01926) · [PDF](https://arxiv.org/pdf/2611.01926) · [HTML](https://arxiv.org/html/2611.01926v1) · [Reading-list entry](https://latent-implicit-thinking.github.io/papers/2611.01926/)
[Preprint BibTeX](https://latent-implicit-thinking.github.io/papers/2611.01926/citation.bib) · [Preprint CSL-JSON](https://latent-implicit-thinking.github.io/papers/2611.01926/citation.csl.json) · [JSON record](https://latent-implicit-thinking.github.io/papers/2611.01926/record.json)

## Published version

TAC-CoT: Tool-Empowered Implicit Chain-of-Thought via Interleaving Explicit Tokens. Juseon Do, Sangmin Bae, Jeesu Jung, Hwanjun Song. Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, 2026.
[Publication](https://aclanthology.org/2026.emnlp-main.1436/) · [BibTeX](https://latent-implicit-thinking.github.io/papers/2611.01926/publication.bib) · [CSL-JSON](https://latent-implicit-thinking.github.io/papers/2611.01926/publication.csl.json)
[Publication record](https://aclanthology.org/2026.emnlp-main.1436/).

## Approaches

- [Internalized reasoning](https://latent-implicit-thinking.github.io/topics/internalized-reasoning/): Learn from worked solutions so that some or all of the reasoning can later happen without writing those steps. Training may use a separate teacher model or progressively remove written steps from the model being trained.
- [Continuous-state reasoning](https://latent-implicit-thinking.github.io/topics/continuous-states/): Use vectors—lists of numbers—to carry intermediate reasoning from one step to the next, instead of selecting a text token at every step. Some methods use the model's hidden states; others combine several candidate token representations.

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