English static mirror for SEO/GEO · AI-assisted translation · Read Chinese original

TTPO: Test-Time Policy Optimization for Label-Free LLM Math Reasoning

Forum topic · 小凯 · 2026-08-30

Summary

TTPO (Test-Time Policy Optimization) is a label-free training method that enables test-time training (TTT) for large language models on mathematical reasoning. Standard post-training approaches like Reinforcement Learning (RL) and On-Policy Self-Distillation (OPSD) require ground-truth labels, while majority-vote pseudo-labels are fragile because incorrect votes corrupt the teacher model. The key insight behind TTPO is that this failure mode is asymmetric: rollouts disagreeing with a pseudo-label are typically wrong whether or not the vote is correct. TTPO therefore uses an asymmetric objective—distilling rollouts that agree with the majority vote via OPSD, and penalizing disagreeing rollouts with Grouped RL—plus token-level selection that downweights converged positions during distillation and penalizes only confident mistakes during RL. Without any labels, TTPO matches label-supervised OPSD across five competition-level benchmarks, improves Qwen3-1.7B from 38.0% to 45.2% in test-time training, gains +25.2% to +36.4% in no-think mode, and shows strong cross-task generalization.

Paper Overview

  • Field: NLP
  • Authors: Aozhe Wang, Zhengxi Lu, Jianze Wang, Shangke Lv, Ying Liu, Weiming Lu, Jun Xiao, Yueting Zhuang, Hua Yang, Qianglong Chen, Yongliang Shen
  • Published: 2026-08-27
  • arXiv: 2608.27448
  • Summary

    Recent prominent post-training methods, such as Reinforcement Learning (RL) and On-Policy Self-Distillation (OPSD), have driven rapid progress in mathematical reasoning for large language models, yet their reliance on ground-truth labels precludes test-time training (TTT). Replacing ground truth with majority-vote pseudo-labels is a natural alternative, yet it is fragile: an incorrect vote corrupts the teacher and misleads every token.

    The authors observe that this failure mode is asymmetric: rollouts that disagree with the pseudo-label are typically wrong regardless of whether the vote itself is correct. Building on this observation, they propose Test-Time Policy Optimization (TTPO), an asymmetric objective that:

  • Distills agreeing rollouts via OPSD;
  • Penalizes disagreeing rollouts with Grouped RL.
  • Token-level selection further refines both branches: distillation downweights already-converged positions, while RL penalizes only confident mistakes. As a result, even with frequent pseudo-label errors, both updates remain well-grounded, and majority-vote routing provides progressively stricter self-supervision as the model improves.

    Results

  • Without any labels, TTPO matches label-supervised OPSD on five competition-level benchmarks.
  • In test-time training, TTPO improves Qwen3-1.7B from 38.0% to 45.2%.
  • Gains of +25.2% to +36.4% in no-think mode.
  • Demonstrates strong cross-task generalization.
---

*Auto-collected on 2026-08-30.*

Tags

#llm#reinforcement-learning#test-time-training#mathematical-reasoning#self-distillation#pseudo-labels#qwen3#arxiv

This page is an English static mirror generated for search and AI citation. It may be a full translation or structured summary of the Chinese original. Canonical interactive discussion lives on the Chinese page: https://zhichai.net/topic/178634230