与 DistilQwen 一起思考：四个蒸馏推理与奖励模型系列的故事
Thinking with DistilQwen: A Tale of Four Distilled Reasoning and Reward Model Series
- Shanghai Jiao Tong University（上海交通大学）
- Alibaba Cloud Computing（阿里云计算）
机构由 AI 辅助整理，请以论文原文为准。
AI总结：
本文扩展了 DistilQwen 模型家族，提出四个面向工业需求的蒸馏推理与奖励模型系列，在多个基准上兼具高效推理与强性能，并在阿里云 PAI 平台提供可扩展支持。
AI中文摘要：
近期，对支持实际应用的小型高效推理模型的需求，推动了在推理性能和推理速度之间取得平衡的知识蒸馏技术的发展。在本文中，我们通过引入专门为满足工业需求而设计的四个模型系列，进一步扩展了从 Qwen 模型初始化的 DistilQwen 模型家族。该蒸馏模型集合包括：(1) 针对需要高准确率的推理任务进行优化的慢思考模型；(2) 两个自适应思考模型系列，能够根据输入任务动态调整推理策略，以在多样化场景中最大化效率；以及 (3) 蒸馏奖励模型，利用蒸馏知识支持推理模型的进一步强化学习。在多个基准上的综合评估表明，这些模型兼具高推理效率和强大的推理性能，以及蒸馏奖励模型的实用价值。我们进一步表明，这些模型通过在阿里云 PAI（Platform for Artificial Intelligence）平台上提供可扩展的训练和推理功能，为行业从业者提供支持。
英文摘要：
Recently, the demand for small and efficient reasoning models to support real-world applications has driven the development of knowledge distillation techniques that balance reasoning performance and inference speed. In this paper, we further extend the DistilQwen model family, initialized from the Qwen models, by introducing four model series specifically designed to meet industrial requirements. The distilled model collection comprises: (1) slow-thinking models, optimized for reasoning tasks that require high accuracy; (2) two series of adaptive-thinking models, which dynamically adjust reasoning strategies based on input tasks to maximize efficiency across diverse scenarios; and (3) distilled reward models, which enable further reinforcement learning of reasoning models using distilled knowledge. Comprehensive evaluations across multiple benchmarks demonstrate both high inference efficiency and strong reasoning performance for these models, as well as the practical utility of distilled reward models. We further show that these models support industry practitioners by providing scalable training and inference functionalities on the Alibaba Cloud PAI (Platform for Artificial Intelligence) platform.
