arXiv cs.CL· Pan Wang, Yihao Hu, Guangyuan Dong, Yitong Lin, Hang Wang, Zimi Liu, Zhihao Wen, Jianshe Li, Jiang-Ming Yang, Wei Wu, Yongqi Tong·· 12 小时前AI 评分38
PAR：面向状态追踪长度泛化的进度对齐循环模型
PAR: Progress-Aligned Recurrence for Length Generalization in State Tracking
AI 导读
研究者提出 Progress-Aligned Recurrence（PAR），让循环模型第 k 次迭代的预测对齐前 k 步操作后的状态，并将循环状态重新缩放到首次迭代的范数。
正文 · 原文
Authors:Pan Wang, Yihao Hu, Guangyuan Dong, Yitong Lin, Hang Wang, Zimi Liu, Zhihao Wen, Jianshe Li, Jiang-Ming Yang, Wei Wu, Yongqi Tong
Abstract:Looped language models can run more iterations at test time, but additional iterations need not help them solve longer problems. For sequential state tracking, final-answer supervision does not specify how far each iteration should advance, and repeated updates can change the scale of the recurrent state. We introduce \textbf{Progress-Aligned Recurrence (PAR)}, which trains the prediction at iteration $k$ to match the state after the first $k$ operations and rescales recurrent states to their first-iteration norms. On swap-$S_5$, a Qwen2.5-0.5B model trained on at most 32 moves achieves 95.5\% mean full-state accuracy at 512 moves, or $16\times$ the training length, compared with 55.7\% reported for RLT and 0.8\% for explicit chain of thought on the same backbone. Controlled $S_5$ ablations show that removing norm anchoring reduces the extrapolation factor from $16\times$ to below $2.2\times$, and replacing progress-aligned targets with final-answer supervision reduces it to at most $8\times$. A learned completion readout selects when to answer without a supplied stopping depth. When intermediate labels are unavailable, prefix self-distillation induces the correspondence from shorter problem prefixes using only final answers. The gains apply across sequential state-tracking tasks, establishing the task structure required for depth extrapolation. These findings support training intermediate predictions and state scale together to make additional recurrent depth useful for longer problems.
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2610.27628 [cs.CL] |
| (or arXiv:2610.27628v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.27628 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Pan Wang [view email]
[v1]
Tue, 20 Oct 2026 04:54:06 UTC (246 KB)
来源：arXiv cs.CL · arxiv.org