TTVS: Boosting Self-Exploring Reinforcement Learning via Test-time Variational Synthesis

Despite significant advances in Large Reasoning Models (LRMs) driven by reinforcement learning with verifiable rewards (RLVR), this paradigm is fundamentally limited in specialized or novel domains where such supervision is prohibitively expensive or unavailable, posing a key challenge for test-time adaptation. While existing test-time methods offer a potential solution, they are constrained by learning from static query sets, risking overfitting to textual patterns. To address this gap, we introduce Test-Time Variational Synthesis (TTVS), a novel framework that enables LRMs to self-evolve by dynamically augmenting the training stream from unlabeled test queries. TTVS comprises two synergistic modules: (1) Online Variational Synthesis, which transforms static test queries into a dynamic stream of diverse, semantically-equivalent variations, enforcing the model to learn underlying problem logic rather than superficial patterns; (2) Test-time Hybrid Exploration, which balances accuracy-driven exploitation with consistency-driven exploration across synthetic variants. Extensive experiments show TTVS yields superior performance across eight model architectures. Notably, using only unlabeled test-time data, TTVS not only surpasses other test-time adaptation methods but also outperforms state-of-the-art supervised RL-based techniques trained on vast, high-quality labeled data.

Think How to Think: Mitigating Overthinking with Autonomous Difficulty Cognition in Large Reasoning Models

Recent Long Reasoning Models(LRMs) have demonstrated remarkable capabilities in handling complex reasoning tasks, but are hindered by excessive overthinking. To explore its essence, our empirical analysis reveals that LRMs are primarily limited to recognizing task properties (i.e., difficulty levels) like humans before solving the problem, leading to a one-size-fits-all reasoning process. Inspired by this, a pressing and natural question emerges: Can we bootstrap such ability to further alleviate the overthinking phenomenon in LRMs? In this paper, we propose Think-How-to-Think (TH2T), a novel two-stage fine-tuning strategy that progressively inspires LRMs' difficulty cognition and redundancy cognition. First, we introduce difficulty-hypnosis in the prefixes of model outputs to intervene in the internal reasoning trajectory. Combined with a heterogeneous short and long reasoning dataset, the trained model enhances its sensitivity to task difficulty, enabling native, differentiated reasoning strategies across various tasks. Second, we further extend redundancy-hypnosis to the internal reasoning process, guiding the model to identify redundant structures within the reasoning steps and generate more concise reasoning outputs. Experiments on 7B/14B/32B models demonstrate that TH2T significantly reduces inference costs (more than 70% on easy tasks and 40% on hard tasks) while maintaining performance stability. The resulting outputs exhibit clear difficulty-aware capabilities and reduced redundancy (e.g., reflection).