Algorithm Engineer, China Telecom
Working on Large Language Models / Multimodal LLMs / LLM Agents
I am an AI Research & Development Engineer at Tianyi Cloud, China Telecom, specializing in multimodal large language models. My research focuses on trustworthy multimodal learning, particularly hallucination mitigation and alignment techniques such as RLHF for large vision-language models, and currently working on agent memory system and AI agents products.
During my academic studies, I contributed to the development of the MiniCPM-V/o series and streaming interactive multimodal models, advancing efficient and interactive multimodal systems. My research work has been published in prestigious conferences including CVPR and ICLR, covering multimodal alignment, efficient model design and open-source multimodal learning.
My long-term research goal is to build reliable, efficient and practical multimodal systems, addressing real-world challenges in industrial cloud scenarios and promoting the application of responsible multimodal artificial intelligence.
Contribute in the development of MiniCPM-V 2.5, MiniCPM-o 2.6
Large Language Models / Multimodal LLMs
Collected a large-scale RLHF dataset for mitigating hallucination in Multimodal LLMs
Multimodal LLMs / RLHF / Hallucination