# How Much Gradient Does Truncated Backprop Lose in Looped Transformers?

Authors: Aur Shalev Merin.

Measures how much of the weight gradient is dropped when looped language models backpropagate through only their final iterations. Tracks how this changes during fine-tuning and tests whether the measurement predicts the resulting training loss.

First public: 2026-10-20.
ArXiv initial submission: 2026-10-20.

[Original paper](https://arxiv.org/abs/2610.28871) · [PDF](https://arxiv.org/pdf/2610.28871) · [HTML](https://arxiv.org/html/2610.28871v1) · [Reading-list entry](https://latent-implicit-thinking.github.io/papers/2610.28871/)
[Preprint BibTeX](https://latent-implicit-thinking.github.io/papers/2610.28871/citation.bib) · [Preprint CSL-JSON](https://latent-implicit-thinking.github.io/papers/2610.28871/citation.csl.json) · [JSON record](https://latent-implicit-thinking.github.io/papers/2610.28871/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.