AdapThink: Adaptive Thinking Preferences for Reasoning Language Model
This addresses efficiency issues in reasoning language models for applications like mathematical problem-solving, though it appears incremental as it builds on existing RL-based post-training methods.
The paper tackles the problem of inefficient reasoning in language models by proposing AdapThink, an adaptive post-training framework that dynamically adjusts thinking preferences based on question complexity and model confidence, achieving improved efficiency while maintaining performance on mathematical reasoning datasets.
Reinforcement Learning (RL)-based post-training has significantly advanced the complex reasoning capabilities of language models, fostering sophisticated self-reflection processes. However, this ``slow thinking'' paradigm presents a critical challenge to reasoning efficiency: models may expend excessive computation on simple questions and shift reasoning prematurely for complex ones. Previous mechanisms typically rely on static length budgets or predefined rules, lacking the adaptability for varying question complexities and models' evolving capabilities. To this end, we propose AdapThink, an adaptive post-training framework designed to induce more efficient thinking while maintaining the performance of reasoning language models. Specifically, AdapThink incorporates two key mechanisms: 1) A group-relative reward function that leverages model confidence and response's characteristic to dynamically adjust the preference of reflection-related transition words without resorting to a fixed length preference. 2) A diversity-aware sampling mechanism that balances the training group's solution accuracy with reasoning diversity via an entropy-guided score. Experiments on several mathematical reasoning datasets with DeepSeek-distilled models demonstrate AdapThink's advantages in enabling adaptive reasoning patterns and mitigating the inefficiencies.