SOURCE-LINKED INTELLIGENCE

LoopCD: Loop-wise Contrastive Decoding for Improving Reasoning in Looped Language Models

arXiv · AI, language, vision and robotics · article · Sep 21, 2026 · UTC

Looped Language Models (LoopLMs) perform "latent reasoning" by recursively refining internal latent representations with shared weights, offering a more effective alternative to explicit verbal reasoning. Despite their effectiveness, we find that LoopLMs remain prone to loop instability: unstable refinement across iterations can produce localized uncertain "hard" tokens associated with reasoning errors. To address this, we propose LoopCD, loop-wise contrastive decoding that enhances the reasoning performance of LoopLMs by intervening on these tokens at inference time. Specifically, we exploit

Read original source ↗ Open in workspace

recordType

paper

region

Global