Towards Looped Models Done Right, Part II: Rethinking at Fixed Points

Paper recorded by Signals 4 on 2026-10-05 in cs.LG. Abstract reproduced from arXiv; link to the original below.

Published 2026-10-05 on arXiv · recorded by Signals 4 on 2026-10-06

Category: cs.LG · 机器学习 · first seen 2026-10-06

Abstract

Every recurrence of a looped language model adds cost in training, decoding, prefill, and reinforcement learning (RL). The closer recurrent states get to fixed points, the less the path to them matters. This enables truncated backpropagation in training; terminal key-value (KV) sharing for decoding with almost no loss in accuracy; a distilled student that prefills up to 1.79x faster; and RL update

Read on arXiv →