BAELAB, Pusan National University, Busan, Korea
DOLAB, Changwon National University, Changwon, Korea
Abstract
Looped computation shows promise in improving the reasoning-oriented performance of LLMs by scaling test-time compute. We introduce Looped Depth Up-Scaling (LoopUS), a post-training framework that converts a standard pretrained LLM into a looped architecture. LoopUS recasts the pretrained LLM into an encoder, a looped reasoning block, and a decoder. It operationalizes this latent-refinement architecture through block decomposition, an input-dependent selective gate, random deep supervision, and a confidence head for adaptive early exiting. Through stable latent looping, LoopUS improves reasoning-oriented performance without extending the generated traces or requiring recurrent training from scratch.
QuickStart
To use LoopUS, clone the repository and run the chat script:
1git clone https://github.com/Thrillcrazyer/LoopUS.git
2cd LoopUS
3uv run chat.py
Illustration of LoopUS
Citation
If you find this work useful, please cite:
1@misc{park2026loopus,
2 title={LoopUS: Recasting Pretrained LLMs into Looped Latent Refinement Models},
3 author={Taekhyun Park and Yongjae Lee and Dohee Kim and Hyerim Bae},
4 year={2026},
5 eprint={2605.11011},
6 archivePrefix={arXiv},
7 primaryClass={cs.CL},
8 url={https://arxiv.org/abs/2605.11011},
9}
