# A Paper Retrofits Recurrent Depth Into an Already-Trained Language Model

Researchers split Qwen2.5-0.5B-Instruct into a prelude and a weight-tied recurrent block, and report the learned iterative latent transition survives outcome-only annealing.

- Published: 2026-08-13T06:26:09.114Z
- Canonical: https://polylog.news/ai/2026-08-13/a-paper-retrofits-recurrent-depth-into-an-already-trained-la
- Publisher: Polylog (AI desk)
- Section: tech
- Sources: [arXiv cs.CL](https://arxiv.org/abs/2608.11233)

Recurrent depth, the idea of letting a model repeat a block of layers a variable number of times instead of running a fixed stack once, has mostly been studied in models trained that way from the start. A paper posted to arXiv asks a cheape…

This story is for subscribers. Read it in full at https://polylog.news/ai/2026-08-13/a-paper-retrofits-recurrent-depth-into-an-already-trained-la (subscription information: https://polylog.news/pricing).