arXiv cs.CL· Chengyi Yang, Zhishang Xiang, Yunbo Tang, Zongpei Teng, Chengsong Huang, Fei Long, Yuhan Liu, Jinsong Su·· 12 小时前AI 评分31
TTCS：面向自进化的测试时课程合成框架
TTCS: Test-Time Curriculum Synthesis for Self-Evolving
AI 导读
TTCS 是一个协同进化的测试时训练框架，从同一预训练模型初始化出题器与推理求解器两个策略，通过迭代优化让出题器按求解器当前能力生成递进难度的题目变体。求解器利用原始测试题与合成题多次采样得到的自一致性奖励更新自身，其反馈又反过来引导出题器对齐当前能力。实验显示 TTCS 在困难数学基准上持续提升推理能力，并可迁移到通用领域任务，代码已公开。
正文 · 原文
Abstract:Test-Time Training offers a promising way to improve the reasoning ability of large language models (LLMs) by adapting the model using only the test questions. However, existing methods struggle with difficult reasoning problems for two reasons: raw test questions are often too difficult to yield high-quality pseudo-labels, and the limited size of test sets makes continuous online updates prone to instability. To address these limitations, we propose TTCS, a co-evolving test-time training framework. Specifically, TTCS initializes two policies from the same pretrained model: a question synthesizer and a reasoning solver. These policies evolve through iterative optimization: the synthesizer generates progressively challenging question variants conditioned on the test questions, creating a structured curriculum tailored to the solver's current capability, while the solver updates itself using self-consistency rewards computed from multiple sampled responses on both original test and synthetic questions. Crucially, the solver's feedback guides the synthesizer to generate questions aligned with the model's current capability, and the generated question variants in turn stabilize the solver's test-time training. Experiments show that TTCS consistently strengthens the reasoning ability on challenging mathematical benchmarks and transfers to general-domain tasks across different LLM backbones, highlighting a scalable path towards dynamically constructing test-time curricula for self-evolving. Our code and implementation details are available at this https URL.
| Comments: | NeurIPS 2026 Poster |
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) |
| Cite as: | arXiv:2601.22628 [cs.LG] |
| (or arXiv:2601.22628v3 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2601.22628 arXiv-issued DOI via DataCite |
Submission history
From: Chengyi Yang [view email]
[v1]
Fri, 30 Jan 2026 06:38:02 UTC (1,038 KB)
[v2]
Sat, 17 Oct 2026 02:55:42 UTC (1,031 KB)
[v3]
Tue, 20 Oct 2026 08:25:47 UTC (1,031 KB)
来源：arXiv cs.CL · arxiv.org