Lvfang Tao 陶履方

Lvfang Tao has been a researcher at Tencent since 2022, after receiving his Master of Science degree from the School of Electronic and Computer Engineering (SECE) at Peking University. His most recent work focuses on mid-training and post-training for large language models, including synthetic data, on-policy distillation, RLVR, and agentic reinforcement learning. Since 2019, he has led projects on from-scratch and continual pretraining of both small and large language models across text, speech, and image modalities. He is also a core contributor to the reinforcement-learning platform AI Arena, which serves the education and research community as well as game developers.

陶履方，2022 年毕业于北京大学信息工程学院并获得理学硕士学位，同年加入腾讯担任算法研究员。作为 腾讯开悟平台 算法方向的核心贡献者，研究方向包括强化学习环境、智能体与平台系统。近期工作主要围绕大模型的中训练与后训练阶段，主导项目涵盖合成数据、on-policy 蒸馏、RLVR、Tool-Use / Agentic RL 等技术方向，相关成果已在公开竞赛与内部产品中得到验证。自 2019 年起，在文本、语音和图像等模态上积累了大规模预训练与增量预训练的经验。